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 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() = default; 76 77 static cl::opt<unsigned> AsmMacroMaxNestingDepth( 78 "asm-macro-max-nesting-depth", cl::init(20), cl::Hidden, 79 cl::desc("The maximum nesting depth allowed for assembly macros.")); 80 81 namespace { 82 83 /// Helper types for tracking macro definitions. 84 typedef std::vector<AsmToken> MCAsmMacroArgument; 85 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 86 87 /// Helper class for storing information about an active macro 88 /// instantiation. 89 struct MacroInstantiation { 90 /// The location of the instantiation. 91 SMLoc InstantiationLoc; 92 93 /// The buffer where parsing should resume upon instantiation completion. 94 int ExitBuffer; 95 96 /// The location where parsing should resume upon instantiation completion. 97 SMLoc ExitLoc; 98 99 /// The depth of TheCondStack at the start of the instantiation. 100 size_t CondStackDepth; 101 102 public: 103 MacroInstantiation(SMLoc IL, int EB, SMLoc EL, size_t CondStackDepth); 104 }; 105 106 struct ParseStatementInfo { 107 /// The parsed operands from the last parsed statement. 108 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 109 110 /// The opcode from the last parsed instruction. 111 unsigned Opcode = ~0U; 112 113 /// Was there an error parsing the inline assembly? 114 bool ParseError = false; 115 116 SmallVectorImpl<AsmRewrite> *AsmRewrites = nullptr; 117 118 ParseStatementInfo() = delete; 119 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 120 : AsmRewrites(rewrites) {} 121 }; 122 123 /// The concrete assembly parser instance. 124 class AsmParser : public MCAsmParser { 125 private: 126 AsmLexer Lexer; 127 MCContext &Ctx; 128 MCStreamer &Out; 129 const MCAsmInfo &MAI; 130 SourceMgr &SrcMgr; 131 SourceMgr::DiagHandlerTy SavedDiagHandler; 132 void *SavedDiagContext; 133 std::unique_ptr<MCAsmParserExtension> PlatformParser; 134 135 /// This is the current buffer index we're lexing from as managed by the 136 /// SourceMgr object. 137 unsigned CurBuffer; 138 139 AsmCond TheCondState; 140 std::vector<AsmCond> TheCondStack; 141 142 /// maps directive names to handler methods in parser 143 /// extensions. Extensions register themselves in this map by calling 144 /// addDirectiveHandler. 145 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 146 147 /// Stack of active macro instantiations. 148 std::vector<MacroInstantiation*> ActiveMacros; 149 150 /// List of bodies of anonymous macros. 151 std::deque<MCAsmMacro> MacroLikeBodies; 152 153 /// Boolean tracking whether macro substitution is enabled. 154 unsigned MacrosEnabledFlag : 1; 155 156 /// Keeps track of how many .macro's have been instantiated. 157 unsigned NumOfMacroInstantiations; 158 159 /// The values from the last parsed cpp hash file line comment if any. 160 struct CppHashInfoTy { 161 StringRef Filename; 162 int64_t LineNumber; 163 SMLoc Loc; 164 unsigned Buf; 165 CppHashInfoTy() : Filename(), LineNumber(0), Loc(), Buf(0) {} 166 }; 167 CppHashInfoTy CppHashInfo; 168 169 /// The filename from the first cpp hash file line comment, if any. 170 StringRef FirstCppHashFilename; 171 172 /// List of forward directional labels for diagnosis at the end. 173 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 174 175 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 176 unsigned AssemblerDialect = ~0U; 177 178 /// is Darwin compatibility enabled? 179 bool IsDarwin = false; 180 181 /// Are we parsing ms-style inline assembly? 182 bool ParsingInlineAsm = false; 183 184 /// Did we already inform the user about inconsistent MD5 usage? 185 bool ReportedInconsistentMD5 = false; 186 187 // Is alt macro mode enabled. 188 bool AltMacroMode = false; 189 190 public: 191 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 192 const MCAsmInfo &MAI, unsigned CB); 193 AsmParser(const AsmParser &) = delete; 194 AsmParser &operator=(const AsmParser &) = delete; 195 ~AsmParser() override; 196 197 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 198 199 void addDirectiveHandler(StringRef Directive, 200 ExtensionDirectiveHandler Handler) override { 201 ExtensionDirectiveMap[Directive] = Handler; 202 } 203 204 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 205 DirectiveKindMap[Directive] = DirectiveKindMap[Alias]; 206 } 207 208 /// @name MCAsmParser Interface 209 /// { 210 211 SourceMgr &getSourceManager() override { return SrcMgr; } 212 MCAsmLexer &getLexer() override { return Lexer; } 213 MCContext &getContext() override { return Ctx; } 214 MCStreamer &getStreamer() override { return Out; } 215 216 CodeViewContext &getCVContext() { return Ctx.getCVContext(); } 217 218 unsigned getAssemblerDialect() override { 219 if (AssemblerDialect == ~0U) 220 return MAI.getAssemblerDialect(); 221 else 222 return AssemblerDialect; 223 } 224 void setAssemblerDialect(unsigned i) override { 225 AssemblerDialect = i; 226 } 227 228 void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override; 229 bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override; 230 bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override; 231 232 const AsmToken &Lex() override; 233 234 void setParsingInlineAsm(bool V) override { 235 ParsingInlineAsm = V; 236 // When parsing MS inline asm, we must lex 0b1101 and 0ABCH as binary and 237 // hex integer literals. 238 Lexer.setLexMasmIntegers(V); 239 } 240 bool isParsingInlineAsm() override { return ParsingInlineAsm; } 241 242 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 243 unsigned &NumOutputs, unsigned &NumInputs, 244 SmallVectorImpl<std::pair<void *,bool>> &OpDecls, 245 SmallVectorImpl<std::string> &Constraints, 246 SmallVectorImpl<std::string> &Clobbers, 247 const MCInstrInfo *MII, const MCInstPrinter *IP, 248 MCAsmParserSemaCallback &SI) override; 249 250 bool parseExpression(const MCExpr *&Res); 251 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 252 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override; 253 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 254 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 255 SMLoc &EndLoc) override; 256 bool parseAbsoluteExpression(int64_t &Res) override; 257 258 /// Parse a floating point expression using the float \p Semantics 259 /// and set \p Res to the value. 260 bool parseRealValue(const fltSemantics &Semantics, APInt &Res); 261 262 /// Parse an identifier or string (as a quoted identifier) 263 /// and set \p Res to the identifier contents. 264 bool parseIdentifier(StringRef &Res) override; 265 void eatToEndOfStatement() override; 266 267 bool checkForValidSection() override; 268 269 /// } 270 271 private: 272 bool parseStatement(ParseStatementInfo &Info, 273 MCAsmParserSemaCallback *SI); 274 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 275 bool parseCppHashLineFilenameComment(SMLoc L); 276 277 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 278 ArrayRef<MCAsmMacroParameter> Parameters); 279 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 280 ArrayRef<MCAsmMacroParameter> Parameters, 281 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 282 SMLoc L); 283 284 /// Are macros enabled in the parser? 285 bool areMacrosEnabled() {return MacrosEnabledFlag;} 286 287 /// Control a flag in the parser that enables or disables macros. 288 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 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 /// Process the specified file for the .incbin directive. 325 /// This returns true on failure. 326 bool processIncbinFile(const std::string &Filename, int64_t Skip = 0, 327 const MCExpr *Count = nullptr, SMLoc Loc = SMLoc()); 328 329 /// Reset the current lexer position to that given by \p Loc. The 330 /// current token is not set; clients should ensure Lex() is called 331 /// subsequently. 332 /// 333 /// \param InBuffer If not 0, should be the known buffer id that contains the 334 /// location. 335 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 336 337 /// Parse up to the end of statement and a return the contents from the 338 /// current token until the end of the statement; the current token on exit 339 /// will be either the EndOfStatement or EOF. 340 StringRef parseStringToEndOfStatement() override; 341 342 /// Parse until the end of a statement or a comma is encountered, 343 /// return the contents from the current token up to the end or comma. 344 StringRef parseStringToComma(); 345 346 bool parseAssignment(StringRef Name, bool allow_redef, 347 bool NoDeadStrip = false); 348 349 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 350 MCBinaryExpr::Opcode &Kind); 351 352 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 353 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 354 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 355 356 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 357 358 bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName); 359 bool parseCVFileId(int64_t &FileId, StringRef DirectiveName); 360 361 // Generic (target and platform independent) directive parsing. 362 enum DirectiveKind { 363 DK_NO_DIRECTIVE, // Placeholder 364 DK_SET, 365 DK_EQU, 366 DK_EQUIV, 367 DK_ASCII, 368 DK_ASCIZ, 369 DK_STRING, 370 DK_BYTE, 371 DK_SHORT, 372 DK_RELOC, 373 DK_VALUE, 374 DK_2BYTE, 375 DK_LONG, 376 DK_INT, 377 DK_4BYTE, 378 DK_QUAD, 379 DK_8BYTE, 380 DK_OCTA, 381 DK_DC, 382 DK_DC_A, 383 DK_DC_B, 384 DK_DC_D, 385 DK_DC_L, 386 DK_DC_S, 387 DK_DC_W, 388 DK_DC_X, 389 DK_DCB, 390 DK_DCB_B, 391 DK_DCB_D, 392 DK_DCB_L, 393 DK_DCB_S, 394 DK_DCB_W, 395 DK_DCB_X, 396 DK_DS, 397 DK_DS_B, 398 DK_DS_D, 399 DK_DS_L, 400 DK_DS_P, 401 DK_DS_S, 402 DK_DS_W, 403 DK_DS_X, 404 DK_SINGLE, 405 DK_FLOAT, 406 DK_DOUBLE, 407 DK_ALIGN, 408 DK_ALIGN32, 409 DK_BALIGN, 410 DK_BALIGNW, 411 DK_BALIGNL, 412 DK_P2ALIGN, 413 DK_P2ALIGNW, 414 DK_P2ALIGNL, 415 DK_ORG, 416 DK_FILL, 417 DK_ENDR, 418 DK_BUNDLE_ALIGN_MODE, 419 DK_BUNDLE_LOCK, 420 DK_BUNDLE_UNLOCK, 421 DK_ZERO, 422 DK_EXTERN, 423 DK_GLOBL, 424 DK_GLOBAL, 425 DK_LAZY_REFERENCE, 426 DK_NO_DEAD_STRIP, 427 DK_SYMBOL_RESOLVER, 428 DK_PRIVATE_EXTERN, 429 DK_REFERENCE, 430 DK_WEAK_DEFINITION, 431 DK_WEAK_REFERENCE, 432 DK_WEAK_DEF_CAN_BE_HIDDEN, 433 DK_COLD, 434 DK_COMM, 435 DK_COMMON, 436 DK_LCOMM, 437 DK_ABORT, 438 DK_INCLUDE, 439 DK_INCBIN, 440 DK_CODE16, 441 DK_CODE16GCC, 442 DK_REPT, 443 DK_IRP, 444 DK_IRPC, 445 DK_IF, 446 DK_IFEQ, 447 DK_IFGE, 448 DK_IFGT, 449 DK_IFLE, 450 DK_IFLT, 451 DK_IFNE, 452 DK_IFB, 453 DK_IFNB, 454 DK_IFC, 455 DK_IFEQS, 456 DK_IFNC, 457 DK_IFNES, 458 DK_IFDEF, 459 DK_IFNDEF, 460 DK_IFNOTDEF, 461 DK_ELSEIF, 462 DK_ELSE, 463 DK_ENDIF, 464 DK_SPACE, 465 DK_SKIP, 466 DK_FILE, 467 DK_LINE, 468 DK_LOC, 469 DK_STABS, 470 DK_CV_FILE, 471 DK_CV_FUNC_ID, 472 DK_CV_INLINE_SITE_ID, 473 DK_CV_LOC, 474 DK_CV_LINETABLE, 475 DK_CV_INLINE_LINETABLE, 476 DK_CV_DEF_RANGE, 477 DK_CV_STRINGTABLE, 478 DK_CV_STRING, 479 DK_CV_FILECHECKSUMS, 480 DK_CV_FILECHECKSUM_OFFSET, 481 DK_CV_FPO_DATA, 482 DK_CFI_SECTIONS, 483 DK_CFI_STARTPROC, 484 DK_CFI_ENDPROC, 485 DK_CFI_DEF_CFA, 486 DK_CFI_DEF_CFA_OFFSET, 487 DK_CFI_ADJUST_CFA_OFFSET, 488 DK_CFI_DEF_CFA_REGISTER, 489 DK_CFI_OFFSET, 490 DK_CFI_REL_OFFSET, 491 DK_CFI_PERSONALITY, 492 DK_CFI_LSDA, 493 DK_CFI_REMEMBER_STATE, 494 DK_CFI_RESTORE_STATE, 495 DK_CFI_SAME_VALUE, 496 DK_CFI_RESTORE, 497 DK_CFI_ESCAPE, 498 DK_CFI_RETURN_COLUMN, 499 DK_CFI_SIGNAL_FRAME, 500 DK_CFI_UNDEFINED, 501 DK_CFI_REGISTER, 502 DK_CFI_WINDOW_SAVE, 503 DK_CFI_B_KEY_FRAME, 504 DK_MACROS_ON, 505 DK_MACROS_OFF, 506 DK_ALTMACRO, 507 DK_NOALTMACRO, 508 DK_MACRO, 509 DK_EXITM, 510 DK_ENDM, 511 DK_ENDMACRO, 512 DK_PURGEM, 513 DK_SLEB128, 514 DK_ULEB128, 515 DK_ERR, 516 DK_ERROR, 517 DK_WARNING, 518 DK_PRINT, 519 DK_ADDRSIG, 520 DK_ADDRSIG_SYM, 521 DK_END 522 }; 523 524 /// Maps directive name --> DirectiveKind enum, for 525 /// directives parsed by this class. 526 StringMap<DirectiveKind> DirectiveKindMap; 527 528 // Codeview def_range type parsing. 529 enum CVDefRangeType { 530 CVDR_DEFRANGE = 0, // Placeholder 531 CVDR_DEFRANGE_REGISTER, 532 CVDR_DEFRANGE_FRAMEPOINTER_REL, 533 CVDR_DEFRANGE_SUBFIELD_REGISTER, 534 CVDR_DEFRANGE_REGISTER_REL 535 }; 536 537 /// Maps Codeview def_range types --> CVDefRangeType enum, for 538 /// Codeview def_range types parsed by this class. 539 StringMap<CVDefRangeType> CVDefRangeTypeMap; 540 541 // ".ascii", ".asciz", ".string" 542 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 543 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 544 bool parseDirectiveValue(StringRef IDVal, 545 unsigned Size); // ".byte", ".long", ... 546 bool parseDirectiveOctaValue(StringRef IDVal); // ".octa", ... 547 bool parseDirectiveRealValue(StringRef IDVal, 548 const fltSemantics &); // ".single", ... 549 bool parseDirectiveFill(); // ".fill" 550 bool parseDirectiveZero(); // ".zero" 551 // ".set", ".equ", ".equiv" 552 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 553 bool parseDirectiveOrg(); // ".org" 554 // ".align{,32}", ".p2align{,w,l}" 555 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 556 557 // ".file", ".line", ".loc", ".stabs" 558 bool parseDirectiveFile(SMLoc DirectiveLoc); 559 bool parseDirectiveLine(); 560 bool parseDirectiveLoc(); 561 bool parseDirectiveStabs(); 562 563 // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable", 564 // ".cv_inline_linetable", ".cv_def_range", ".cv_string" 565 bool parseDirectiveCVFile(); 566 bool parseDirectiveCVFuncId(); 567 bool parseDirectiveCVInlineSiteId(); 568 bool parseDirectiveCVLoc(); 569 bool parseDirectiveCVLinetable(); 570 bool parseDirectiveCVInlineLinetable(); 571 bool parseDirectiveCVDefRange(); 572 bool parseDirectiveCVString(); 573 bool parseDirectiveCVStringTable(); 574 bool parseDirectiveCVFileChecksums(); 575 bool parseDirectiveCVFileChecksumOffset(); 576 bool parseDirectiveCVFPOData(); 577 578 // .cfi directives 579 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 580 bool parseDirectiveCFIWindowSave(); 581 bool parseDirectiveCFISections(); 582 bool parseDirectiveCFIStartProc(); 583 bool parseDirectiveCFIEndProc(); 584 bool parseDirectiveCFIDefCfaOffset(); 585 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 586 bool parseDirectiveCFIAdjustCfaOffset(); 587 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 588 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 589 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 590 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 591 bool parseDirectiveCFIRememberState(); 592 bool parseDirectiveCFIRestoreState(); 593 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 594 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 595 bool parseDirectiveCFIEscape(); 596 bool parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc); 597 bool parseDirectiveCFISignalFrame(); 598 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 599 600 // macro directives 601 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 602 bool parseDirectiveExitMacro(StringRef Directive); 603 bool parseDirectiveEndMacro(StringRef Directive); 604 bool parseDirectiveMacro(SMLoc DirectiveLoc); 605 bool parseDirectiveMacrosOnOff(StringRef Directive); 606 // alternate macro mode directives 607 bool parseDirectiveAltmacro(StringRef Directive); 608 // ".bundle_align_mode" 609 bool parseDirectiveBundleAlignMode(); 610 // ".bundle_lock" 611 bool parseDirectiveBundleLock(); 612 // ".bundle_unlock" 613 bool parseDirectiveBundleUnlock(); 614 615 // ".space", ".skip" 616 bool parseDirectiveSpace(StringRef IDVal); 617 618 // ".dcb" 619 bool parseDirectiveDCB(StringRef IDVal, unsigned Size); 620 bool parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &); 621 // ".ds" 622 bool parseDirectiveDS(StringRef IDVal, unsigned Size); 623 624 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 625 bool parseDirectiveLEB128(bool Signed); 626 627 /// Parse a directive like ".globl" which 628 /// accepts a single symbol (which should be a label or an external). 629 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 630 631 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 632 633 bool parseDirectiveAbort(); // ".abort" 634 bool parseDirectiveInclude(); // ".include" 635 bool parseDirectiveIncbin(); // ".incbin" 636 637 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 638 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 639 // ".ifb" or ".ifnb", depending on ExpectBlank. 640 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 641 // ".ifc" or ".ifnc", depending on ExpectEqual. 642 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 643 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 644 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 645 // ".ifdef" or ".ifndef", depending on expect_defined 646 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 647 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 648 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 649 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 650 bool parseEscapedString(std::string &Data) override; 651 652 const MCExpr *applyModifierToExpr(const MCExpr *E, 653 MCSymbolRefExpr::VariantKind Variant); 654 655 // Macro-like directives 656 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 657 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 658 raw_svector_ostream &OS); 659 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 660 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 661 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 662 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 663 664 // "_emit" or "__emit" 665 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 666 size_t Len); 667 668 // "align" 669 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 670 671 // "end" 672 bool parseDirectiveEnd(SMLoc DirectiveLoc); 673 674 // ".err" or ".error" 675 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 676 677 // ".warning" 678 bool parseDirectiveWarning(SMLoc DirectiveLoc); 679 680 // .print <double-quotes-string> 681 bool parseDirectivePrint(SMLoc DirectiveLoc); 682 683 // Directives to support address-significance tables. 684 bool parseDirectiveAddrsig(); 685 bool parseDirectiveAddrsigSym(); 686 687 void initializeDirectiveKindMap(); 688 void initializeCVDefRangeTypeMap(); 689 }; 690 691 } // end anonymous namespace 692 693 namespace llvm { 694 695 extern MCAsmParserExtension *createDarwinAsmParser(); 696 extern MCAsmParserExtension *createELFAsmParser(); 697 extern MCAsmParserExtension *createCOFFAsmParser(); 698 extern MCAsmParserExtension *createWasmAsmParser(); 699 700 } // end namespace llvm 701 702 enum { DEFAULT_ADDRSPACE = 0 }; 703 704 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 705 const MCAsmInfo &MAI, unsigned CB = 0) 706 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 707 CurBuffer(CB ? CB : SM.getMainFileID()), MacrosEnabledFlag(true) { 708 HadError = false; 709 // Save the old handler. 710 SavedDiagHandler = SrcMgr.getDiagHandler(); 711 SavedDiagContext = SrcMgr.getDiagContext(); 712 // Set our own handler which calls the saved handler. 713 SrcMgr.setDiagHandler(DiagHandler, this); 714 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 715 716 // Initialize the platform / file format parser. 717 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 718 case MCObjectFileInfo::IsCOFF: 719 PlatformParser.reset(createCOFFAsmParser()); 720 break; 721 case MCObjectFileInfo::IsMachO: 722 PlatformParser.reset(createDarwinAsmParser()); 723 IsDarwin = true; 724 break; 725 case MCObjectFileInfo::IsELF: 726 PlatformParser.reset(createELFAsmParser()); 727 break; 728 case MCObjectFileInfo::IsWasm: 729 PlatformParser.reset(createWasmAsmParser()); 730 break; 731 case MCObjectFileInfo::IsXCOFF: 732 report_fatal_error( 733 "Need to implement createXCOFFAsmParser for XCOFF format."); 734 break; 735 } 736 737 PlatformParser->Initialize(*this); 738 initializeDirectiveKindMap(); 739 initializeCVDefRangeTypeMap(); 740 741 NumOfMacroInstantiations = 0; 742 } 743 744 AsmParser::~AsmParser() { 745 assert((HadError || ActiveMacros.empty()) && 746 "Unexpected active macro instantiation!"); 747 748 // Restore the saved diagnostics handler and context for use during 749 // finalization. 750 SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext); 751 } 752 753 void AsmParser::printMacroInstantiations() { 754 // Print the active macro instantiation stack. 755 for (std::vector<MacroInstantiation *>::const_reverse_iterator 756 it = ActiveMacros.rbegin(), 757 ie = ActiveMacros.rend(); 758 it != ie; ++it) 759 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 760 "while in macro instantiation"); 761 } 762 763 void AsmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) { 764 printPendingErrors(); 765 printMessage(L, SourceMgr::DK_Note, Msg, Range); 766 printMacroInstantiations(); 767 } 768 769 bool AsmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) { 770 if(getTargetParser().getTargetOptions().MCNoWarn) 771 return false; 772 if (getTargetParser().getTargetOptions().MCFatalWarnings) 773 return Error(L, Msg, Range); 774 printMessage(L, SourceMgr::DK_Warning, Msg, Range); 775 printMacroInstantiations(); 776 return false; 777 } 778 779 bool AsmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) { 780 HadError = true; 781 printMessage(L, SourceMgr::DK_Error, Msg, Range); 782 printMacroInstantiations(); 783 return true; 784 } 785 786 bool AsmParser::enterIncludeFile(const std::string &Filename) { 787 std::string IncludedFile; 788 unsigned NewBuf = 789 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 790 if (!NewBuf) 791 return true; 792 793 CurBuffer = NewBuf; 794 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 795 return false; 796 } 797 798 /// Process the specified .incbin file by searching for it in the include paths 799 /// then just emitting the byte contents of the file to the streamer. This 800 /// returns true on failure. 801 bool AsmParser::processIncbinFile(const std::string &Filename, int64_t Skip, 802 const MCExpr *Count, SMLoc Loc) { 803 std::string IncludedFile; 804 unsigned NewBuf = 805 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 806 if (!NewBuf) 807 return true; 808 809 // Pick up the bytes from the file and emit them. 810 StringRef Bytes = SrcMgr.getMemoryBuffer(NewBuf)->getBuffer(); 811 Bytes = Bytes.drop_front(Skip); 812 if (Count) { 813 int64_t Res; 814 if (!Count->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 815 return Error(Loc, "expected absolute expression"); 816 if (Res < 0) 817 return Warning(Loc, "negative count has no effect"); 818 Bytes = Bytes.take_front(Res); 819 } 820 getStreamer().EmitBytes(Bytes); 821 return false; 822 } 823 824 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 825 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 826 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 827 Loc.getPointer()); 828 } 829 830 const AsmToken &AsmParser::Lex() { 831 if (Lexer.getTok().is(AsmToken::Error)) 832 Error(Lexer.getErrLoc(), Lexer.getErr()); 833 834 // if it's a end of statement with a comment in it 835 if (getTok().is(AsmToken::EndOfStatement)) { 836 // if this is a line comment output it. 837 if (!getTok().getString().empty() && getTok().getString().front() != '\n' && 838 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 839 Out.addExplicitComment(Twine(getTok().getString())); 840 } 841 842 const AsmToken *tok = &Lexer.Lex(); 843 844 // Parse comments here to be deferred until end of next statement. 845 while (tok->is(AsmToken::Comment)) { 846 if (MAI.preserveAsmComments()) 847 Out.addExplicitComment(Twine(tok->getString())); 848 tok = &Lexer.Lex(); 849 } 850 851 if (tok->is(AsmToken::Eof)) { 852 // If this is the end of an included file, pop the parent file off the 853 // include stack. 854 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 855 if (ParentIncludeLoc != SMLoc()) { 856 jumpToLoc(ParentIncludeLoc); 857 return Lex(); 858 } 859 } 860 861 return *tok; 862 } 863 864 bool AsmParser::enabledGenDwarfForAssembly() { 865 // Check whether the user specified -g. 866 if (!getContext().getGenDwarfForAssembly()) 867 return false; 868 // If we haven't encountered any .file directives (which would imply that 869 // the assembler source was produced with debug info already) then emit one 870 // describing the assembler source file itself. 871 if (getContext().getGenDwarfFileNumber() == 0) { 872 // Use the first #line directive for this, if any. It's preprocessed, so 873 // there is no checksum, and of course no source directive. 874 if (!FirstCppHashFilename.empty()) 875 getContext().setMCLineTableRootFile(/*CUID=*/0, 876 getContext().getCompilationDir(), 877 FirstCppHashFilename, 878 /*Cksum=*/None, /*Source=*/None); 879 const MCDwarfFile &RootFile = 880 getContext().getMCDwarfLineTable(/*CUID=*/0).getRootFile(); 881 getContext().setGenDwarfFileNumber(getStreamer().EmitDwarfFileDirective( 882 /*CUID=*/0, getContext().getCompilationDir(), RootFile.Name, 883 RootFile.Checksum, RootFile.Source)); 884 } 885 return true; 886 } 887 888 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 889 // Create the initial section, if requested. 890 if (!NoInitialTextSection) 891 Out.InitSections(false); 892 893 // Prime the lexer. 894 Lex(); 895 896 HadError = false; 897 AsmCond StartingCondState = TheCondState; 898 SmallVector<AsmRewrite, 4> AsmStrRewrites; 899 900 // If we are generating dwarf for assembly source files save the initial text 901 // section. (Don't use enabledGenDwarfForAssembly() here, as we aren't 902 // emitting any actual debug info yet and haven't had a chance to parse any 903 // embedded .file directives.) 904 if (getContext().getGenDwarfForAssembly()) { 905 MCSection *Sec = getStreamer().getCurrentSectionOnly(); 906 if (!Sec->getBeginSymbol()) { 907 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 908 getStreamer().EmitLabel(SectionStartSym); 909 Sec->setBeginSymbol(SectionStartSym); 910 } 911 bool InsertResult = getContext().addGenDwarfSection(Sec); 912 assert(InsertResult && ".text section should not have debug info yet"); 913 (void)InsertResult; 914 } 915 916 // While we have input, parse each statement. 917 while (Lexer.isNot(AsmToken::Eof)) { 918 ParseStatementInfo Info(&AsmStrRewrites); 919 bool Parsed = parseStatement(Info, nullptr); 920 921 // If we have a Lexer Error we are on an Error Token. Load in Lexer Error 922 // for printing ErrMsg via Lex() only if no (presumably better) parser error 923 // exists. 924 if (Parsed && !hasPendingError() && Lexer.getTok().is(AsmToken::Error)) { 925 Lex(); 926 } 927 928 // parseStatement returned true so may need to emit an error. 929 printPendingErrors(); 930 931 // Skipping to the next line if needed. 932 if (Parsed && !getLexer().isAtStartOfStatement()) 933 eatToEndOfStatement(); 934 } 935 936 getTargetParser().onEndOfFile(); 937 printPendingErrors(); 938 939 // All errors should have been emitted. 940 assert(!hasPendingError() && "unexpected error from parseStatement"); 941 942 getTargetParser().flushPendingInstructions(getStreamer()); 943 944 if (TheCondState.TheCond != StartingCondState.TheCond || 945 TheCondState.Ignore != StartingCondState.Ignore) 946 printError(getTok().getLoc(), "unmatched .ifs or .elses"); 947 // Check to see there are no empty DwarfFile slots. 948 const auto &LineTables = getContext().getMCDwarfLineTables(); 949 if (!LineTables.empty()) { 950 unsigned Index = 0; 951 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 952 if (File.Name.empty() && Index != 0) 953 printError(getTok().getLoc(), "unassigned file number: " + 954 Twine(Index) + 955 " for .file directives"); 956 ++Index; 957 } 958 } 959 960 // Check to see that all assembler local symbols were actually defined. 961 // Targets that don't do subsections via symbols may not want this, though, 962 // so conservatively exclude them. Only do this if we're finalizing, though, 963 // as otherwise we won't necessarilly have seen everything yet. 964 if (!NoFinalize) { 965 if (MAI.hasSubsectionsViaSymbols()) { 966 for (const auto &TableEntry : getContext().getSymbols()) { 967 MCSymbol *Sym = TableEntry.getValue(); 968 // Variable symbols may not be marked as defined, so check those 969 // explicitly. If we know it's a variable, we have a definition for 970 // the purposes of this check. 971 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 972 // FIXME: We would really like to refer back to where the symbol was 973 // first referenced for a source location. We need to add something 974 // to track that. Currently, we just point to the end of the file. 975 printError(getTok().getLoc(), "assembler local symbol '" + 976 Sym->getName() + "' not defined"); 977 } 978 } 979 980 // Temporary symbols like the ones for directional jumps don't go in the 981 // symbol table. They also need to be diagnosed in all (final) cases. 982 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 983 if (std::get<2>(LocSym)->isUndefined()) { 984 // Reset the state of any "# line file" directives we've seen to the 985 // context as it was at the diagnostic site. 986 CppHashInfo = std::get<1>(LocSym); 987 printError(std::get<0>(LocSym), "directional label undefined"); 988 } 989 } 990 } 991 992 // Finalize the output stream if there are no errors and if the client wants 993 // us to. 994 if (!HadError && !NoFinalize) 995 Out.Finish(); 996 997 return HadError || getContext().hadError(); 998 } 999 1000 bool AsmParser::checkForValidSection() { 1001 if (!ParsingInlineAsm && !getStreamer().getCurrentSectionOnly()) { 1002 Out.InitSections(false); 1003 return Error(getTok().getLoc(), 1004 "expected section directive before assembly directive"); 1005 } 1006 return false; 1007 } 1008 1009 /// Throw away the rest of the line for testing purposes. 1010 void AsmParser::eatToEndOfStatement() { 1011 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1012 Lexer.Lex(); 1013 1014 // Eat EOL. 1015 if (Lexer.is(AsmToken::EndOfStatement)) 1016 Lexer.Lex(); 1017 } 1018 1019 StringRef AsmParser::parseStringToEndOfStatement() { 1020 const char *Start = getTok().getLoc().getPointer(); 1021 1022 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1023 Lexer.Lex(); 1024 1025 const char *End = getTok().getLoc().getPointer(); 1026 return StringRef(Start, End - Start); 1027 } 1028 1029 StringRef AsmParser::parseStringToComma() { 1030 const char *Start = getTok().getLoc().getPointer(); 1031 1032 while (Lexer.isNot(AsmToken::EndOfStatement) && 1033 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 1034 Lexer.Lex(); 1035 1036 const char *End = getTok().getLoc().getPointer(); 1037 return StringRef(Start, End - Start); 1038 } 1039 1040 /// Parse a paren expression and return it. 1041 /// NOTE: This assumes the leading '(' has already been consumed. 1042 /// 1043 /// parenexpr ::= expr) 1044 /// 1045 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1046 if (parseExpression(Res)) 1047 return true; 1048 if (Lexer.isNot(AsmToken::RParen)) 1049 return TokError("expected ')' in parentheses expression"); 1050 EndLoc = Lexer.getTok().getEndLoc(); 1051 Lex(); 1052 return false; 1053 } 1054 1055 /// Parse a bracket expression and return it. 1056 /// NOTE: This assumes the leading '[' has already been consumed. 1057 /// 1058 /// bracketexpr ::= expr] 1059 /// 1060 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1061 if (parseExpression(Res)) 1062 return true; 1063 EndLoc = getTok().getEndLoc(); 1064 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 1065 return true; 1066 return false; 1067 } 1068 1069 /// Parse a primary expression and return it. 1070 /// primaryexpr ::= (parenexpr 1071 /// primaryexpr ::= symbol 1072 /// primaryexpr ::= number 1073 /// primaryexpr ::= '.' 1074 /// primaryexpr ::= ~,+,- primaryexpr 1075 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1076 SMLoc FirstTokenLoc = getLexer().getLoc(); 1077 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 1078 switch (FirstTokenKind) { 1079 default: 1080 return TokError("unknown token in expression"); 1081 // If we have an error assume that we've already handled it. 1082 case AsmToken::Error: 1083 return true; 1084 case AsmToken::Exclaim: 1085 Lex(); // Eat the operator. 1086 if (parsePrimaryExpr(Res, EndLoc)) 1087 return true; 1088 Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc); 1089 return false; 1090 case AsmToken::Dollar: 1091 case AsmToken::At: 1092 case AsmToken::String: 1093 case AsmToken::Identifier: { 1094 StringRef Identifier; 1095 if (parseIdentifier(Identifier)) { 1096 // We may have failed but $ may be a valid token. 1097 if (getTok().is(AsmToken::Dollar)) { 1098 if (Lexer.getMAI().getDollarIsPC()) { 1099 Lex(); 1100 // This is a '$' reference, which references the current PC. Emit a 1101 // temporary label to the streamer and refer to it. 1102 MCSymbol *Sym = Ctx.createTempSymbol(); 1103 Out.EmitLabel(Sym); 1104 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 1105 getContext()); 1106 EndLoc = FirstTokenLoc; 1107 return false; 1108 } 1109 return Error(FirstTokenLoc, "invalid token in expression"); 1110 } 1111 } 1112 // Parse symbol variant 1113 std::pair<StringRef, StringRef> Split; 1114 if (!MAI.useParensForSymbolVariant()) { 1115 if (FirstTokenKind == AsmToken::String) { 1116 if (Lexer.is(AsmToken::At)) { 1117 Lex(); // eat @ 1118 SMLoc AtLoc = getLexer().getLoc(); 1119 StringRef VName; 1120 if (parseIdentifier(VName)) 1121 return Error(AtLoc, "expected symbol variant after '@'"); 1122 1123 Split = std::make_pair(Identifier, VName); 1124 } 1125 } else { 1126 Split = Identifier.split('@'); 1127 } 1128 } else if (Lexer.is(AsmToken::LParen)) { 1129 Lex(); // eat '('. 1130 StringRef VName; 1131 parseIdentifier(VName); 1132 // eat ')'. 1133 if (parseToken(AsmToken::RParen, 1134 "unexpected token in variant, expected ')'")) 1135 return true; 1136 Split = std::make_pair(Identifier, VName); 1137 } 1138 1139 EndLoc = SMLoc::getFromPointer(Identifier.end()); 1140 1141 // This is a symbol reference. 1142 StringRef SymbolName = Identifier; 1143 if (SymbolName.empty()) 1144 return Error(getLexer().getLoc(), "expected a symbol reference"); 1145 1146 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1147 1148 // Lookup the symbol variant if used. 1149 if (!Split.second.empty()) { 1150 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1151 if (Variant != MCSymbolRefExpr::VK_Invalid) { 1152 SymbolName = Split.first; 1153 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 1154 Variant = MCSymbolRefExpr::VK_None; 1155 } else { 1156 return Error(SMLoc::getFromPointer(Split.second.begin()), 1157 "invalid variant '" + Split.second + "'"); 1158 } 1159 } 1160 1161 MCSymbol *Sym = getContext().getInlineAsmLabel(SymbolName); 1162 if (!Sym) 1163 Sym = getContext().getOrCreateSymbol(SymbolName); 1164 1165 // If this is an absolute variable reference, substitute it now to preserve 1166 // semantics in the face of reassignment. 1167 if (Sym->isVariable()) { 1168 auto V = Sym->getVariableValue(/*SetUsed*/ false); 1169 bool DoInline = isa<MCConstantExpr>(V) && !Variant; 1170 if (auto TV = dyn_cast<MCTargetExpr>(V)) 1171 DoInline = TV->inlineAssignedExpr(); 1172 if (DoInline) { 1173 if (Variant) 1174 return Error(EndLoc, "unexpected modifier on variable reference"); 1175 Res = Sym->getVariableValue(/*SetUsed*/ false); 1176 return false; 1177 } 1178 } 1179 1180 // Otherwise create a symbol ref. 1181 Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc); 1182 return false; 1183 } 1184 case AsmToken::BigNum: 1185 return TokError("literal value out of range for directive"); 1186 case AsmToken::Integer: { 1187 SMLoc Loc = getTok().getLoc(); 1188 int64_t IntVal = getTok().getIntVal(); 1189 Res = MCConstantExpr::create(IntVal, getContext()); 1190 EndLoc = Lexer.getTok().getEndLoc(); 1191 Lex(); // Eat token. 1192 // Look for 'b' or 'f' following an Integer as a directional label 1193 if (Lexer.getKind() == AsmToken::Identifier) { 1194 StringRef IDVal = getTok().getString(); 1195 // Lookup the symbol variant if used. 1196 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 1197 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1198 if (Split.first.size() != IDVal.size()) { 1199 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1200 if (Variant == MCSymbolRefExpr::VK_Invalid) 1201 return TokError("invalid variant '" + Split.second + "'"); 1202 IDVal = Split.first; 1203 } 1204 if (IDVal == "f" || IDVal == "b") { 1205 MCSymbol *Sym = 1206 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 1207 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 1208 if (IDVal == "b" && Sym->isUndefined()) 1209 return Error(Loc, "directional label undefined"); 1210 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 1211 EndLoc = Lexer.getTok().getEndLoc(); 1212 Lex(); // Eat identifier. 1213 } 1214 } 1215 return false; 1216 } 1217 case AsmToken::Real: { 1218 APFloat RealVal(APFloat::IEEEdouble(), getTok().getString()); 1219 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1220 Res = MCConstantExpr::create(IntVal, getContext()); 1221 EndLoc = Lexer.getTok().getEndLoc(); 1222 Lex(); // Eat token. 1223 return false; 1224 } 1225 case AsmToken::Dot: { 1226 // This is a '.' reference, which references the current PC. Emit a 1227 // temporary label to the streamer and refer to it. 1228 MCSymbol *Sym = Ctx.createTempSymbol(); 1229 Out.EmitLabel(Sym); 1230 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1231 EndLoc = Lexer.getTok().getEndLoc(); 1232 Lex(); // Eat identifier. 1233 return false; 1234 } 1235 case AsmToken::LParen: 1236 Lex(); // Eat the '('. 1237 return parseParenExpr(Res, EndLoc); 1238 case AsmToken::LBrac: 1239 if (!PlatformParser->HasBracketExpressions()) 1240 return TokError("brackets expression not supported on this target"); 1241 Lex(); // Eat the '['. 1242 return parseBracketExpr(Res, EndLoc); 1243 case AsmToken::Minus: 1244 Lex(); // Eat the operator. 1245 if (parsePrimaryExpr(Res, EndLoc)) 1246 return true; 1247 Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc); 1248 return false; 1249 case AsmToken::Plus: 1250 Lex(); // Eat the operator. 1251 if (parsePrimaryExpr(Res, EndLoc)) 1252 return true; 1253 Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc); 1254 return false; 1255 case AsmToken::Tilde: 1256 Lex(); // Eat the operator. 1257 if (parsePrimaryExpr(Res, EndLoc)) 1258 return true; 1259 Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc); 1260 return false; 1261 // MIPS unary expression operators. The lexer won't generate these tokens if 1262 // MCAsmInfo::HasMipsExpressions is false for the target. 1263 case AsmToken::PercentCall16: 1264 case AsmToken::PercentCall_Hi: 1265 case AsmToken::PercentCall_Lo: 1266 case AsmToken::PercentDtprel_Hi: 1267 case AsmToken::PercentDtprel_Lo: 1268 case AsmToken::PercentGot: 1269 case AsmToken::PercentGot_Disp: 1270 case AsmToken::PercentGot_Hi: 1271 case AsmToken::PercentGot_Lo: 1272 case AsmToken::PercentGot_Ofst: 1273 case AsmToken::PercentGot_Page: 1274 case AsmToken::PercentGottprel: 1275 case AsmToken::PercentGp_Rel: 1276 case AsmToken::PercentHi: 1277 case AsmToken::PercentHigher: 1278 case AsmToken::PercentHighest: 1279 case AsmToken::PercentLo: 1280 case AsmToken::PercentNeg: 1281 case AsmToken::PercentPcrel_Hi: 1282 case AsmToken::PercentPcrel_Lo: 1283 case AsmToken::PercentTlsgd: 1284 case AsmToken::PercentTlsldm: 1285 case AsmToken::PercentTprel_Hi: 1286 case AsmToken::PercentTprel_Lo: 1287 Lex(); // Eat the operator. 1288 if (Lexer.isNot(AsmToken::LParen)) 1289 return TokError("expected '(' after operator"); 1290 Lex(); // Eat the operator. 1291 if (parseExpression(Res, EndLoc)) 1292 return true; 1293 if (Lexer.isNot(AsmToken::RParen)) 1294 return TokError("expected ')'"); 1295 Lex(); // Eat the operator. 1296 Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx); 1297 return !Res; 1298 } 1299 } 1300 1301 bool AsmParser::parseExpression(const MCExpr *&Res) { 1302 SMLoc EndLoc; 1303 return parseExpression(Res, EndLoc); 1304 } 1305 1306 const MCExpr * 1307 AsmParser::applyModifierToExpr(const MCExpr *E, 1308 MCSymbolRefExpr::VariantKind Variant) { 1309 // Ask the target implementation about this expression first. 1310 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 1311 if (NewE) 1312 return NewE; 1313 // Recurse over the given expression, rebuilding it to apply the given variant 1314 // if there is exactly one symbol. 1315 switch (E->getKind()) { 1316 case MCExpr::Target: 1317 case MCExpr::Constant: 1318 return nullptr; 1319 1320 case MCExpr::SymbolRef: { 1321 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 1322 1323 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 1324 TokError("invalid variant on expression '" + getTok().getIdentifier() + 1325 "' (already modified)"); 1326 return E; 1327 } 1328 1329 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1330 } 1331 1332 case MCExpr::Unary: { 1333 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1334 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1335 if (!Sub) 1336 return nullptr; 1337 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1338 } 1339 1340 case MCExpr::Binary: { 1341 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1342 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1343 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1344 1345 if (!LHS && !RHS) 1346 return nullptr; 1347 1348 if (!LHS) 1349 LHS = BE->getLHS(); 1350 if (!RHS) 1351 RHS = BE->getRHS(); 1352 1353 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1354 } 1355 } 1356 1357 llvm_unreachable("Invalid expression kind!"); 1358 } 1359 1360 /// This function checks if the next token is <string> type or arithmetic. 1361 /// string that begin with character '<' must end with character '>'. 1362 /// otherwise it is arithmetics. 1363 /// If the function returns a 'true' value, 1364 /// the End argument will be filled with the last location pointed to the '>' 1365 /// character. 1366 1367 /// There is a gap between the AltMacro's documentation and the single quote 1368 /// implementation. GCC does not fully support this feature and so we will not 1369 /// support it. 1370 /// TODO: Adding single quote as a string. 1371 static bool isAltmacroString(SMLoc &StrLoc, SMLoc &EndLoc) { 1372 assert((StrLoc.getPointer() != nullptr) && 1373 "Argument to the function cannot be a NULL value"); 1374 const char *CharPtr = StrLoc.getPointer(); 1375 while ((*CharPtr != '>') && (*CharPtr != '\n') && (*CharPtr != '\r') && 1376 (*CharPtr != '\0')) { 1377 if (*CharPtr == '!') 1378 CharPtr++; 1379 CharPtr++; 1380 } 1381 if (*CharPtr == '>') { 1382 EndLoc = StrLoc.getFromPointer(CharPtr + 1); 1383 return true; 1384 } 1385 return false; 1386 } 1387 1388 /// creating a string without the escape characters '!'. 1389 static std::string altMacroString(StringRef AltMacroStr) { 1390 std::string Res; 1391 for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) { 1392 if (AltMacroStr[Pos] == '!') 1393 Pos++; 1394 Res += AltMacroStr[Pos]; 1395 } 1396 return Res; 1397 } 1398 1399 /// Parse an expression and return it. 1400 /// 1401 /// expr ::= expr &&,|| expr -> lowest. 1402 /// expr ::= expr |,^,&,! expr 1403 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1404 /// expr ::= expr <<,>> expr 1405 /// expr ::= expr +,- expr 1406 /// expr ::= expr *,/,% expr -> highest. 1407 /// expr ::= primaryexpr 1408 /// 1409 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1410 // Parse the expression. 1411 Res = nullptr; 1412 if (getTargetParser().parsePrimaryExpr(Res, EndLoc) || 1413 parseBinOpRHS(1, Res, EndLoc)) 1414 return true; 1415 1416 // As a special case, we support 'a op b @ modifier' by rewriting the 1417 // expression to include the modifier. This is inefficient, but in general we 1418 // expect users to use 'a@modifier op b'. 1419 if (Lexer.getKind() == AsmToken::At) { 1420 Lex(); 1421 1422 if (Lexer.isNot(AsmToken::Identifier)) 1423 return TokError("unexpected symbol modifier following '@'"); 1424 1425 MCSymbolRefExpr::VariantKind Variant = 1426 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1427 if (Variant == MCSymbolRefExpr::VK_Invalid) 1428 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1429 1430 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1431 if (!ModifiedRes) { 1432 return TokError("invalid modifier '" + getTok().getIdentifier() + 1433 "' (no symbols present)"); 1434 } 1435 1436 Res = ModifiedRes; 1437 Lex(); 1438 } 1439 1440 // Try to constant fold it up front, if possible. Do not exploit 1441 // assembler here. 1442 int64_t Value; 1443 if (Res->evaluateAsAbsolute(Value)) 1444 Res = MCConstantExpr::create(Value, getContext()); 1445 1446 return false; 1447 } 1448 1449 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1450 Res = nullptr; 1451 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1452 } 1453 1454 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1455 SMLoc &EndLoc) { 1456 if (parseParenExpr(Res, EndLoc)) 1457 return true; 1458 1459 for (; ParenDepth > 0; --ParenDepth) { 1460 if (parseBinOpRHS(1, Res, EndLoc)) 1461 return true; 1462 1463 // We don't Lex() the last RParen. 1464 // This is the same behavior as parseParenExpression(). 1465 if (ParenDepth - 1 > 0) { 1466 EndLoc = getTok().getEndLoc(); 1467 if (parseToken(AsmToken::RParen, 1468 "expected ')' in parentheses expression")) 1469 return true; 1470 } 1471 } 1472 return false; 1473 } 1474 1475 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1476 const MCExpr *Expr; 1477 1478 SMLoc StartLoc = Lexer.getLoc(); 1479 if (parseExpression(Expr)) 1480 return true; 1481 1482 if (!Expr->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 1483 return Error(StartLoc, "expected absolute expression"); 1484 1485 return false; 1486 } 1487 1488 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1489 MCBinaryExpr::Opcode &Kind, 1490 bool ShouldUseLogicalShr) { 1491 switch (K) { 1492 default: 1493 return 0; // not a binop. 1494 1495 // Lowest Precedence: &&, || 1496 case AsmToken::AmpAmp: 1497 Kind = MCBinaryExpr::LAnd; 1498 return 1; 1499 case AsmToken::PipePipe: 1500 Kind = MCBinaryExpr::LOr; 1501 return 1; 1502 1503 // Low Precedence: |, &, ^ 1504 // 1505 // FIXME: gas seems to support '!' as an infix operator? 1506 case AsmToken::Pipe: 1507 Kind = MCBinaryExpr::Or; 1508 return 2; 1509 case AsmToken::Caret: 1510 Kind = MCBinaryExpr::Xor; 1511 return 2; 1512 case AsmToken::Amp: 1513 Kind = MCBinaryExpr::And; 1514 return 2; 1515 1516 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1517 case AsmToken::EqualEqual: 1518 Kind = MCBinaryExpr::EQ; 1519 return 3; 1520 case AsmToken::ExclaimEqual: 1521 case AsmToken::LessGreater: 1522 Kind = MCBinaryExpr::NE; 1523 return 3; 1524 case AsmToken::Less: 1525 Kind = MCBinaryExpr::LT; 1526 return 3; 1527 case AsmToken::LessEqual: 1528 Kind = MCBinaryExpr::LTE; 1529 return 3; 1530 case AsmToken::Greater: 1531 Kind = MCBinaryExpr::GT; 1532 return 3; 1533 case AsmToken::GreaterEqual: 1534 Kind = MCBinaryExpr::GTE; 1535 return 3; 1536 1537 // Intermediate Precedence: <<, >> 1538 case AsmToken::LessLess: 1539 Kind = MCBinaryExpr::Shl; 1540 return 4; 1541 case AsmToken::GreaterGreater: 1542 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1543 return 4; 1544 1545 // High Intermediate Precedence: +, - 1546 case AsmToken::Plus: 1547 Kind = MCBinaryExpr::Add; 1548 return 5; 1549 case AsmToken::Minus: 1550 Kind = MCBinaryExpr::Sub; 1551 return 5; 1552 1553 // Highest Precedence: *, /, % 1554 case AsmToken::Star: 1555 Kind = MCBinaryExpr::Mul; 1556 return 6; 1557 case AsmToken::Slash: 1558 Kind = MCBinaryExpr::Div; 1559 return 6; 1560 case AsmToken::Percent: 1561 Kind = MCBinaryExpr::Mod; 1562 return 6; 1563 } 1564 } 1565 1566 static unsigned getGNUBinOpPrecedence(AsmToken::TokenKind K, 1567 MCBinaryExpr::Opcode &Kind, 1568 bool ShouldUseLogicalShr) { 1569 switch (K) { 1570 default: 1571 return 0; // not a binop. 1572 1573 // Lowest Precedence: &&, || 1574 case AsmToken::AmpAmp: 1575 Kind = MCBinaryExpr::LAnd; 1576 return 2; 1577 case AsmToken::PipePipe: 1578 Kind = MCBinaryExpr::LOr; 1579 return 1; 1580 1581 // Low Precedence: ==, !=, <>, <, <=, >, >= 1582 case AsmToken::EqualEqual: 1583 Kind = MCBinaryExpr::EQ; 1584 return 3; 1585 case AsmToken::ExclaimEqual: 1586 case AsmToken::LessGreater: 1587 Kind = MCBinaryExpr::NE; 1588 return 3; 1589 case AsmToken::Less: 1590 Kind = MCBinaryExpr::LT; 1591 return 3; 1592 case AsmToken::LessEqual: 1593 Kind = MCBinaryExpr::LTE; 1594 return 3; 1595 case AsmToken::Greater: 1596 Kind = MCBinaryExpr::GT; 1597 return 3; 1598 case AsmToken::GreaterEqual: 1599 Kind = MCBinaryExpr::GTE; 1600 return 3; 1601 1602 // Low Intermediate Precedence: +, - 1603 case AsmToken::Plus: 1604 Kind = MCBinaryExpr::Add; 1605 return 4; 1606 case AsmToken::Minus: 1607 Kind = MCBinaryExpr::Sub; 1608 return 4; 1609 1610 // High Intermediate Precedence: |, &, ^ 1611 // 1612 // FIXME: gas seems to support '!' as an infix operator? 1613 case AsmToken::Pipe: 1614 Kind = MCBinaryExpr::Or; 1615 return 5; 1616 case AsmToken::Caret: 1617 Kind = MCBinaryExpr::Xor; 1618 return 5; 1619 case AsmToken::Amp: 1620 Kind = MCBinaryExpr::And; 1621 return 5; 1622 1623 // Highest Precedence: *, /, %, <<, >> 1624 case AsmToken::Star: 1625 Kind = MCBinaryExpr::Mul; 1626 return 6; 1627 case AsmToken::Slash: 1628 Kind = MCBinaryExpr::Div; 1629 return 6; 1630 case AsmToken::Percent: 1631 Kind = MCBinaryExpr::Mod; 1632 return 6; 1633 case AsmToken::LessLess: 1634 Kind = MCBinaryExpr::Shl; 1635 return 6; 1636 case AsmToken::GreaterGreater: 1637 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1638 return 6; 1639 } 1640 } 1641 1642 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1643 MCBinaryExpr::Opcode &Kind) { 1644 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1645 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1646 : getGNUBinOpPrecedence(K, Kind, ShouldUseLogicalShr); 1647 } 1648 1649 /// Parse all binary operators with precedence >= 'Precedence'. 1650 /// Res contains the LHS of the expression on input. 1651 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1652 SMLoc &EndLoc) { 1653 SMLoc StartLoc = Lexer.getLoc(); 1654 while (true) { 1655 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1656 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1657 1658 // If the next token is lower precedence than we are allowed to eat, return 1659 // successfully with what we ate already. 1660 if (TokPrec < Precedence) 1661 return false; 1662 1663 Lex(); 1664 1665 // Eat the next primary expression. 1666 const MCExpr *RHS; 1667 if (getTargetParser().parsePrimaryExpr(RHS, EndLoc)) 1668 return true; 1669 1670 // If BinOp binds less tightly with RHS than the operator after RHS, let 1671 // the pending operator take RHS as its LHS. 1672 MCBinaryExpr::Opcode Dummy; 1673 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1674 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1675 return true; 1676 1677 // Merge LHS and RHS according to operator. 1678 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc); 1679 } 1680 } 1681 1682 /// ParseStatement: 1683 /// ::= EndOfStatement 1684 /// ::= Label* Directive ...Operands... EndOfStatement 1685 /// ::= Label* Identifier OperandList* EndOfStatement 1686 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1687 MCAsmParserSemaCallback *SI) { 1688 assert(!hasPendingError() && "parseStatement started with pending error"); 1689 // Eat initial spaces and comments 1690 while (Lexer.is(AsmToken::Space)) 1691 Lex(); 1692 if (Lexer.is(AsmToken::EndOfStatement)) { 1693 // if this is a line comment we can drop it safely 1694 if (getTok().getString().empty() || getTok().getString().front() == '\r' || 1695 getTok().getString().front() == '\n') 1696 Out.AddBlankLine(); 1697 Lex(); 1698 return false; 1699 } 1700 // Statements always start with an identifier. 1701 AsmToken ID = getTok(); 1702 SMLoc IDLoc = ID.getLoc(); 1703 StringRef IDVal; 1704 int64_t LocalLabelVal = -1; 1705 if (Lexer.is(AsmToken::HashDirective)) 1706 return parseCppHashLineFilenameComment(IDLoc); 1707 // Allow an integer followed by a ':' as a directional local label. 1708 if (Lexer.is(AsmToken::Integer)) { 1709 LocalLabelVal = getTok().getIntVal(); 1710 if (LocalLabelVal < 0) { 1711 if (!TheCondState.Ignore) { 1712 Lex(); // always eat a token 1713 return Error(IDLoc, "unexpected token at start of statement"); 1714 } 1715 IDVal = ""; 1716 } else { 1717 IDVal = getTok().getString(); 1718 Lex(); // Consume the integer token to be used as an identifier token. 1719 if (Lexer.getKind() != AsmToken::Colon) { 1720 if (!TheCondState.Ignore) { 1721 Lex(); // always eat a token 1722 return Error(IDLoc, "unexpected token at start of statement"); 1723 } 1724 } 1725 } 1726 } else if (Lexer.is(AsmToken::Dot)) { 1727 // Treat '.' as a valid identifier in this context. 1728 Lex(); 1729 IDVal = "."; 1730 } else if (Lexer.is(AsmToken::LCurly)) { 1731 // Treat '{' as a valid identifier in this context. 1732 Lex(); 1733 IDVal = "{"; 1734 1735 } else if (Lexer.is(AsmToken::RCurly)) { 1736 // Treat '}' as a valid identifier in this context. 1737 Lex(); 1738 IDVal = "}"; 1739 } else if (Lexer.is(AsmToken::Star) && 1740 getTargetParser().starIsStartOfStatement()) { 1741 // Accept '*' as a valid start of statement. 1742 Lex(); 1743 IDVal = "*"; 1744 } else if (parseIdentifier(IDVal)) { 1745 if (!TheCondState.Ignore) { 1746 Lex(); // always eat a token 1747 return Error(IDLoc, "unexpected token at start of statement"); 1748 } 1749 IDVal = ""; 1750 } 1751 1752 // Handle conditional assembly here before checking for skipping. We 1753 // have to do this so that .endif isn't skipped in a ".if 0" block for 1754 // example. 1755 StringMap<DirectiveKind>::const_iterator DirKindIt = 1756 DirectiveKindMap.find(IDVal); 1757 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1758 1759 ? DK_NO_DIRECTIVE 1760 : DirKindIt->getValue(); 1761 switch (DirKind) { 1762 default: 1763 break; 1764 case DK_IF: 1765 case DK_IFEQ: 1766 case DK_IFGE: 1767 case DK_IFGT: 1768 case DK_IFLE: 1769 case DK_IFLT: 1770 case DK_IFNE: 1771 return parseDirectiveIf(IDLoc, DirKind); 1772 case DK_IFB: 1773 return parseDirectiveIfb(IDLoc, true); 1774 case DK_IFNB: 1775 return parseDirectiveIfb(IDLoc, false); 1776 case DK_IFC: 1777 return parseDirectiveIfc(IDLoc, true); 1778 case DK_IFEQS: 1779 return parseDirectiveIfeqs(IDLoc, true); 1780 case DK_IFNC: 1781 return parseDirectiveIfc(IDLoc, false); 1782 case DK_IFNES: 1783 return parseDirectiveIfeqs(IDLoc, false); 1784 case DK_IFDEF: 1785 return parseDirectiveIfdef(IDLoc, true); 1786 case DK_IFNDEF: 1787 case DK_IFNOTDEF: 1788 return parseDirectiveIfdef(IDLoc, false); 1789 case DK_ELSEIF: 1790 return parseDirectiveElseIf(IDLoc); 1791 case DK_ELSE: 1792 return parseDirectiveElse(IDLoc); 1793 case DK_ENDIF: 1794 return parseDirectiveEndIf(IDLoc); 1795 } 1796 1797 // Ignore the statement if in the middle of inactive conditional 1798 // (e.g. ".if 0"). 1799 if (TheCondState.Ignore) { 1800 eatToEndOfStatement(); 1801 return false; 1802 } 1803 1804 // FIXME: Recurse on local labels? 1805 1806 // See what kind of statement we have. 1807 switch (Lexer.getKind()) { 1808 case AsmToken::Colon: { 1809 if (!getTargetParser().isLabel(ID)) 1810 break; 1811 if (checkForValidSection()) 1812 return true; 1813 1814 // identifier ':' -> Label. 1815 Lex(); 1816 1817 // Diagnose attempt to use '.' as a label. 1818 if (IDVal == ".") 1819 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1820 1821 // Diagnose attempt to use a variable as a label. 1822 // 1823 // FIXME: Diagnostics. Note the location of the definition as a label. 1824 // FIXME: This doesn't diagnose assignment to a symbol which has been 1825 // implicitly marked as external. 1826 MCSymbol *Sym; 1827 if (LocalLabelVal == -1) { 1828 if (ParsingInlineAsm && SI) { 1829 StringRef RewrittenLabel = 1830 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1831 assert(!RewrittenLabel.empty() && 1832 "We should have an internal name here."); 1833 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1834 RewrittenLabel); 1835 IDVal = RewrittenLabel; 1836 } 1837 Sym = getContext().getOrCreateSymbol(IDVal); 1838 } else 1839 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1840 // End of Labels should be treated as end of line for lexing 1841 // purposes but that information is not available to the Lexer who 1842 // does not understand Labels. This may cause us to see a Hash 1843 // here instead of a preprocessor line comment. 1844 if (getTok().is(AsmToken::Hash)) { 1845 StringRef CommentStr = parseStringToEndOfStatement(); 1846 Lexer.Lex(); 1847 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1848 } 1849 1850 // Consume any end of statement token, if present, to avoid spurious 1851 // AddBlankLine calls(). 1852 if (getTok().is(AsmToken::EndOfStatement)) { 1853 Lex(); 1854 } 1855 1856 getTargetParser().doBeforeLabelEmit(Sym); 1857 1858 // Emit the label. 1859 if (!getTargetParser().isParsingInlineAsm()) 1860 Out.EmitLabel(Sym, IDLoc); 1861 1862 // If we are generating dwarf for assembly source files then gather the 1863 // info to make a dwarf label entry for this label if needed. 1864 if (enabledGenDwarfForAssembly()) 1865 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1866 IDLoc); 1867 1868 getTargetParser().onLabelParsed(Sym); 1869 1870 return false; 1871 } 1872 1873 case AsmToken::Equal: 1874 if (!getTargetParser().equalIsAsmAssignment()) 1875 break; 1876 // identifier '=' ... -> assignment statement 1877 Lex(); 1878 1879 return parseAssignment(IDVal, true); 1880 1881 default: // Normal instruction or directive. 1882 break; 1883 } 1884 1885 // If macros are enabled, check to see if this is a macro instantiation. 1886 if (areMacrosEnabled()) 1887 if (const MCAsmMacro *M = getContext().lookupMacro(IDVal)) { 1888 return handleMacroEntry(M, IDLoc); 1889 } 1890 1891 // Otherwise, we have a normal instruction or directive. 1892 1893 // Directives start with "." 1894 if (IDVal.startswith(".") && IDVal != ".") { 1895 // There are several entities interested in parsing directives: 1896 // 1897 // 1. The target-specific assembly parser. Some directives are target 1898 // specific or may potentially behave differently on certain targets. 1899 // 2. Asm parser extensions. For example, platform-specific parsers 1900 // (like the ELF parser) register themselves as extensions. 1901 // 3. The generic directive parser implemented by this class. These are 1902 // all the directives that behave in a target and platform independent 1903 // manner, or at least have a default behavior that's shared between 1904 // all targets and platforms. 1905 1906 getTargetParser().flushPendingInstructions(getStreamer()); 1907 1908 SMLoc StartTokLoc = getTok().getLoc(); 1909 bool TPDirectiveReturn = getTargetParser().ParseDirective(ID); 1910 1911 if (hasPendingError()) 1912 return true; 1913 // Currently the return value should be true if we are 1914 // uninterested but as this is at odds with the standard parsing 1915 // convention (return true = error) we have instances of a parsed 1916 // directive that fails returning true as an error. Catch these 1917 // cases as best as possible errors here. 1918 if (TPDirectiveReturn && StartTokLoc != getTok().getLoc()) 1919 return true; 1920 // Return if we did some parsing or believe we succeeded. 1921 if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc()) 1922 return false; 1923 1924 // Next, check the extension directive map to see if any extension has 1925 // registered itself to parse this directive. 1926 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1927 ExtensionDirectiveMap.lookup(IDVal); 1928 if (Handler.first) 1929 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1930 1931 // Finally, if no one else is interested in this directive, it must be 1932 // generic and familiar to this class. 1933 switch (DirKind) { 1934 default: 1935 break; 1936 case DK_SET: 1937 case DK_EQU: 1938 return parseDirectiveSet(IDVal, true); 1939 case DK_EQUIV: 1940 return parseDirectiveSet(IDVal, false); 1941 case DK_ASCII: 1942 return parseDirectiveAscii(IDVal, false); 1943 case DK_ASCIZ: 1944 case DK_STRING: 1945 return parseDirectiveAscii(IDVal, true); 1946 case DK_BYTE: 1947 case DK_DC_B: 1948 return parseDirectiveValue(IDVal, 1); 1949 case DK_DC: 1950 case DK_DC_W: 1951 case DK_SHORT: 1952 case DK_VALUE: 1953 case DK_2BYTE: 1954 return parseDirectiveValue(IDVal, 2); 1955 case DK_LONG: 1956 case DK_INT: 1957 case DK_4BYTE: 1958 case DK_DC_L: 1959 return parseDirectiveValue(IDVal, 4); 1960 case DK_QUAD: 1961 case DK_8BYTE: 1962 return parseDirectiveValue(IDVal, 8); 1963 case DK_DC_A: 1964 return parseDirectiveValue( 1965 IDVal, getContext().getAsmInfo()->getCodePointerSize()); 1966 case DK_OCTA: 1967 return parseDirectiveOctaValue(IDVal); 1968 case DK_SINGLE: 1969 case DK_FLOAT: 1970 case DK_DC_S: 1971 return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle()); 1972 case DK_DOUBLE: 1973 case DK_DC_D: 1974 return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble()); 1975 case DK_ALIGN: { 1976 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1977 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1978 } 1979 case DK_ALIGN32: { 1980 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1981 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1982 } 1983 case DK_BALIGN: 1984 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1985 case DK_BALIGNW: 1986 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1987 case DK_BALIGNL: 1988 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 1989 case DK_P2ALIGN: 1990 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 1991 case DK_P2ALIGNW: 1992 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 1993 case DK_P2ALIGNL: 1994 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 1995 case DK_ORG: 1996 return parseDirectiveOrg(); 1997 case DK_FILL: 1998 return parseDirectiveFill(); 1999 case DK_ZERO: 2000 return parseDirectiveZero(); 2001 case DK_EXTERN: 2002 eatToEndOfStatement(); // .extern is the default, ignore it. 2003 return false; 2004 case DK_GLOBL: 2005 case DK_GLOBAL: 2006 return parseDirectiveSymbolAttribute(MCSA_Global); 2007 case DK_LAZY_REFERENCE: 2008 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 2009 case DK_NO_DEAD_STRIP: 2010 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 2011 case DK_SYMBOL_RESOLVER: 2012 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 2013 case DK_PRIVATE_EXTERN: 2014 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 2015 case DK_REFERENCE: 2016 return parseDirectiveSymbolAttribute(MCSA_Reference); 2017 case DK_WEAK_DEFINITION: 2018 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 2019 case DK_WEAK_REFERENCE: 2020 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 2021 case DK_WEAK_DEF_CAN_BE_HIDDEN: 2022 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 2023 case DK_COLD: 2024 return parseDirectiveSymbolAttribute(MCSA_Cold); 2025 case DK_COMM: 2026 case DK_COMMON: 2027 return parseDirectiveComm(/*IsLocal=*/false); 2028 case DK_LCOMM: 2029 return parseDirectiveComm(/*IsLocal=*/true); 2030 case DK_ABORT: 2031 return parseDirectiveAbort(); 2032 case DK_INCLUDE: 2033 return parseDirectiveInclude(); 2034 case DK_INCBIN: 2035 return parseDirectiveIncbin(); 2036 case DK_CODE16: 2037 case DK_CODE16GCC: 2038 return TokError(Twine(IDVal) + 2039 " not currently supported for this target"); 2040 case DK_REPT: 2041 return parseDirectiveRept(IDLoc, IDVal); 2042 case DK_IRP: 2043 return parseDirectiveIrp(IDLoc); 2044 case DK_IRPC: 2045 return parseDirectiveIrpc(IDLoc); 2046 case DK_ENDR: 2047 return parseDirectiveEndr(IDLoc); 2048 case DK_BUNDLE_ALIGN_MODE: 2049 return parseDirectiveBundleAlignMode(); 2050 case DK_BUNDLE_LOCK: 2051 return parseDirectiveBundleLock(); 2052 case DK_BUNDLE_UNLOCK: 2053 return parseDirectiveBundleUnlock(); 2054 case DK_SLEB128: 2055 return parseDirectiveLEB128(true); 2056 case DK_ULEB128: 2057 return parseDirectiveLEB128(false); 2058 case DK_SPACE: 2059 case DK_SKIP: 2060 return parseDirectiveSpace(IDVal); 2061 case DK_FILE: 2062 return parseDirectiveFile(IDLoc); 2063 case DK_LINE: 2064 return parseDirectiveLine(); 2065 case DK_LOC: 2066 return parseDirectiveLoc(); 2067 case DK_STABS: 2068 return parseDirectiveStabs(); 2069 case DK_CV_FILE: 2070 return parseDirectiveCVFile(); 2071 case DK_CV_FUNC_ID: 2072 return parseDirectiveCVFuncId(); 2073 case DK_CV_INLINE_SITE_ID: 2074 return parseDirectiveCVInlineSiteId(); 2075 case DK_CV_LOC: 2076 return parseDirectiveCVLoc(); 2077 case DK_CV_LINETABLE: 2078 return parseDirectiveCVLinetable(); 2079 case DK_CV_INLINE_LINETABLE: 2080 return parseDirectiveCVInlineLinetable(); 2081 case DK_CV_DEF_RANGE: 2082 return parseDirectiveCVDefRange(); 2083 case DK_CV_STRING: 2084 return parseDirectiveCVString(); 2085 case DK_CV_STRINGTABLE: 2086 return parseDirectiveCVStringTable(); 2087 case DK_CV_FILECHECKSUMS: 2088 return parseDirectiveCVFileChecksums(); 2089 case DK_CV_FILECHECKSUM_OFFSET: 2090 return parseDirectiveCVFileChecksumOffset(); 2091 case DK_CV_FPO_DATA: 2092 return parseDirectiveCVFPOData(); 2093 case DK_CFI_SECTIONS: 2094 return parseDirectiveCFISections(); 2095 case DK_CFI_STARTPROC: 2096 return parseDirectiveCFIStartProc(); 2097 case DK_CFI_ENDPROC: 2098 return parseDirectiveCFIEndProc(); 2099 case DK_CFI_DEF_CFA: 2100 return parseDirectiveCFIDefCfa(IDLoc); 2101 case DK_CFI_DEF_CFA_OFFSET: 2102 return parseDirectiveCFIDefCfaOffset(); 2103 case DK_CFI_ADJUST_CFA_OFFSET: 2104 return parseDirectiveCFIAdjustCfaOffset(); 2105 case DK_CFI_DEF_CFA_REGISTER: 2106 return parseDirectiveCFIDefCfaRegister(IDLoc); 2107 case DK_CFI_OFFSET: 2108 return parseDirectiveCFIOffset(IDLoc); 2109 case DK_CFI_REL_OFFSET: 2110 return parseDirectiveCFIRelOffset(IDLoc); 2111 case DK_CFI_PERSONALITY: 2112 return parseDirectiveCFIPersonalityOrLsda(true); 2113 case DK_CFI_LSDA: 2114 return parseDirectiveCFIPersonalityOrLsda(false); 2115 case DK_CFI_REMEMBER_STATE: 2116 return parseDirectiveCFIRememberState(); 2117 case DK_CFI_RESTORE_STATE: 2118 return parseDirectiveCFIRestoreState(); 2119 case DK_CFI_SAME_VALUE: 2120 return parseDirectiveCFISameValue(IDLoc); 2121 case DK_CFI_RESTORE: 2122 return parseDirectiveCFIRestore(IDLoc); 2123 case DK_CFI_ESCAPE: 2124 return parseDirectiveCFIEscape(); 2125 case DK_CFI_RETURN_COLUMN: 2126 return parseDirectiveCFIReturnColumn(IDLoc); 2127 case DK_CFI_SIGNAL_FRAME: 2128 return parseDirectiveCFISignalFrame(); 2129 case DK_CFI_UNDEFINED: 2130 return parseDirectiveCFIUndefined(IDLoc); 2131 case DK_CFI_REGISTER: 2132 return parseDirectiveCFIRegister(IDLoc); 2133 case DK_CFI_WINDOW_SAVE: 2134 return parseDirectiveCFIWindowSave(); 2135 case DK_MACROS_ON: 2136 case DK_MACROS_OFF: 2137 return parseDirectiveMacrosOnOff(IDVal); 2138 case DK_MACRO: 2139 return parseDirectiveMacro(IDLoc); 2140 case DK_ALTMACRO: 2141 case DK_NOALTMACRO: 2142 return parseDirectiveAltmacro(IDVal); 2143 case DK_EXITM: 2144 return parseDirectiveExitMacro(IDVal); 2145 case DK_ENDM: 2146 case DK_ENDMACRO: 2147 return parseDirectiveEndMacro(IDVal); 2148 case DK_PURGEM: 2149 return parseDirectivePurgeMacro(IDLoc); 2150 case DK_END: 2151 return parseDirectiveEnd(IDLoc); 2152 case DK_ERR: 2153 return parseDirectiveError(IDLoc, false); 2154 case DK_ERROR: 2155 return parseDirectiveError(IDLoc, true); 2156 case DK_WARNING: 2157 return parseDirectiveWarning(IDLoc); 2158 case DK_RELOC: 2159 return parseDirectiveReloc(IDLoc); 2160 case DK_DCB: 2161 case DK_DCB_W: 2162 return parseDirectiveDCB(IDVal, 2); 2163 case DK_DCB_B: 2164 return parseDirectiveDCB(IDVal, 1); 2165 case DK_DCB_D: 2166 return parseDirectiveRealDCB(IDVal, APFloat::IEEEdouble()); 2167 case DK_DCB_L: 2168 return parseDirectiveDCB(IDVal, 4); 2169 case DK_DCB_S: 2170 return parseDirectiveRealDCB(IDVal, APFloat::IEEEsingle()); 2171 case DK_DC_X: 2172 case DK_DCB_X: 2173 return TokError(Twine(IDVal) + 2174 " not currently supported for this target"); 2175 case DK_DS: 2176 case DK_DS_W: 2177 return parseDirectiveDS(IDVal, 2); 2178 case DK_DS_B: 2179 return parseDirectiveDS(IDVal, 1); 2180 case DK_DS_D: 2181 return parseDirectiveDS(IDVal, 8); 2182 case DK_DS_L: 2183 case DK_DS_S: 2184 return parseDirectiveDS(IDVal, 4); 2185 case DK_DS_P: 2186 case DK_DS_X: 2187 return parseDirectiveDS(IDVal, 12); 2188 case DK_PRINT: 2189 return parseDirectivePrint(IDLoc); 2190 case DK_ADDRSIG: 2191 return parseDirectiveAddrsig(); 2192 case DK_ADDRSIG_SYM: 2193 return parseDirectiveAddrsigSym(); 2194 } 2195 2196 return Error(IDLoc, "unknown directive"); 2197 } 2198 2199 // __asm _emit or __asm __emit 2200 if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 2201 IDVal == "_EMIT" || IDVal == "__EMIT")) 2202 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 2203 2204 // __asm align 2205 if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 2206 return parseDirectiveMSAlign(IDLoc, Info); 2207 2208 if (ParsingInlineAsm && (IDVal == "even" || IDVal == "EVEN")) 2209 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 2210 if (checkForValidSection()) 2211 return true; 2212 2213 // Canonicalize the opcode to lower case. 2214 std::string OpcodeStr = IDVal.lower(); 2215 ParseInstructionInfo IInfo(Info.AsmRewrites); 2216 bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 2217 Info.ParsedOperands); 2218 Info.ParseError = ParseHadError; 2219 2220 // Dump the parsed representation, if requested. 2221 if (getShowParsedOperands()) { 2222 SmallString<256> Str; 2223 raw_svector_ostream OS(Str); 2224 OS << "parsed instruction: ["; 2225 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 2226 if (i != 0) 2227 OS << ", "; 2228 Info.ParsedOperands[i]->print(OS); 2229 } 2230 OS << "]"; 2231 2232 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 2233 } 2234 2235 // Fail even if ParseInstruction erroneously returns false. 2236 if (hasPendingError() || ParseHadError) 2237 return true; 2238 2239 // If we are generating dwarf for the current section then generate a .loc 2240 // directive for the instruction. 2241 if (!ParseHadError && enabledGenDwarfForAssembly() && 2242 getContext().getGenDwarfSectionSyms().count( 2243 getStreamer().getCurrentSectionOnly())) { 2244 unsigned Line; 2245 if (ActiveMacros.empty()) 2246 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 2247 else 2248 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 2249 ActiveMacros.front()->ExitBuffer); 2250 2251 // If we previously parsed a cpp hash file line comment then make sure the 2252 // current Dwarf File is for the CppHashFilename if not then emit the 2253 // Dwarf File table for it and adjust the line number for the .loc. 2254 if (!CppHashInfo.Filename.empty()) { 2255 unsigned FileNumber = getStreamer().EmitDwarfFileDirective( 2256 0, StringRef(), CppHashInfo.Filename); 2257 getContext().setGenDwarfFileNumber(FileNumber); 2258 2259 unsigned CppHashLocLineNo = 2260 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 2261 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 2262 } 2263 2264 getStreamer().EmitDwarfLocDirective( 2265 getContext().getGenDwarfFileNumber(), Line, 0, 2266 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 2267 StringRef()); 2268 } 2269 2270 // If parsing succeeded, match the instruction. 2271 if (!ParseHadError) { 2272 uint64_t ErrorInfo; 2273 if (getTargetParser().MatchAndEmitInstruction( 2274 IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo, 2275 getTargetParser().isParsingInlineAsm())) 2276 return true; 2277 } 2278 return false; 2279 } 2280 2281 // Parse and erase curly braces marking block start/end 2282 bool 2283 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 2284 // Identify curly brace marking block start/end 2285 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 2286 return false; 2287 2288 SMLoc StartLoc = Lexer.getLoc(); 2289 Lex(); // Eat the brace 2290 if (Lexer.is(AsmToken::EndOfStatement)) 2291 Lex(); // Eat EndOfStatement following the brace 2292 2293 // Erase the block start/end brace from the output asm string 2294 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 2295 StartLoc.getPointer()); 2296 return true; 2297 } 2298 2299 /// parseCppHashLineFilenameComment as this: 2300 /// ::= # number "filename" 2301 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L) { 2302 Lex(); // Eat the hash token. 2303 // Lexer only ever emits HashDirective if it fully formed if it's 2304 // done the checking already so this is an internal error. 2305 assert(getTok().is(AsmToken::Integer) && 2306 "Lexing Cpp line comment: Expected Integer"); 2307 int64_t LineNumber = getTok().getIntVal(); 2308 Lex(); 2309 assert(getTok().is(AsmToken::String) && 2310 "Lexing Cpp line comment: Expected String"); 2311 StringRef Filename = getTok().getString(); 2312 Lex(); 2313 2314 // Get rid of the enclosing quotes. 2315 Filename = Filename.substr(1, Filename.size() - 2); 2316 2317 // Save the SMLoc, Filename and LineNumber for later use by diagnostics 2318 // and possibly DWARF file info. 2319 CppHashInfo.Loc = L; 2320 CppHashInfo.Filename = Filename; 2321 CppHashInfo.LineNumber = LineNumber; 2322 CppHashInfo.Buf = CurBuffer; 2323 if (FirstCppHashFilename.empty()) 2324 FirstCppHashFilename = Filename; 2325 return false; 2326 } 2327 2328 /// will use the last parsed cpp hash line filename comment 2329 /// for the Filename and LineNo if any in the diagnostic. 2330 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 2331 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 2332 raw_ostream &OS = errs(); 2333 2334 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 2335 SMLoc DiagLoc = Diag.getLoc(); 2336 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2337 unsigned CppHashBuf = 2338 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 2339 2340 // Like SourceMgr::printMessage() we need to print the include stack if any 2341 // before printing the message. 2342 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2343 if (!Parser->SavedDiagHandler && DiagCurBuffer && 2344 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 2345 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 2346 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 2347 } 2348 2349 // If we have not parsed a cpp hash line filename comment or the source 2350 // manager changed or buffer changed (like in a nested include) then just 2351 // print the normal diagnostic using its Filename and LineNo. 2352 if (!Parser->CppHashInfo.LineNumber || &DiagSrcMgr != &Parser->SrcMgr || 2353 DiagBuf != CppHashBuf) { 2354 if (Parser->SavedDiagHandler) 2355 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2356 else 2357 Diag.print(nullptr, OS); 2358 return; 2359 } 2360 2361 // Use the CppHashFilename and calculate a line number based on the 2362 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 2363 // for the diagnostic. 2364 const std::string &Filename = Parser->CppHashInfo.Filename; 2365 2366 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 2367 int CppHashLocLineNo = 2368 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 2369 int LineNo = 2370 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 2371 2372 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 2373 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 2374 Diag.getLineContents(), Diag.getRanges()); 2375 2376 if (Parser->SavedDiagHandler) 2377 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 2378 else 2379 NewDiag.print(nullptr, OS); 2380 } 2381 2382 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2383 // difference being that that function accepts '@' as part of identifiers and 2384 // we can't do that. AsmLexer.cpp should probably be changed to handle 2385 // '@' as a special case when needed. 2386 static bool isIdentifierChar(char c) { 2387 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2388 c == '.'; 2389 } 2390 2391 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2392 ArrayRef<MCAsmMacroParameter> Parameters, 2393 ArrayRef<MCAsmMacroArgument> A, 2394 bool EnableAtPseudoVariable, SMLoc L) { 2395 unsigned NParameters = Parameters.size(); 2396 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2397 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2398 return Error(L, "Wrong number of arguments"); 2399 2400 // A macro without parameters is handled differently on Darwin: 2401 // gas accepts no arguments and does no substitutions 2402 while (!Body.empty()) { 2403 // Scan for the next substitution. 2404 std::size_t End = Body.size(), Pos = 0; 2405 for (; Pos != End; ++Pos) { 2406 // Check for a substitution or escape. 2407 if (IsDarwin && !NParameters) { 2408 // This macro has no parameters, look for $0, $1, etc. 2409 if (Body[Pos] != '$' || Pos + 1 == End) 2410 continue; 2411 2412 char Next = Body[Pos + 1]; 2413 if (Next == '$' || Next == 'n' || 2414 isdigit(static_cast<unsigned char>(Next))) 2415 break; 2416 } else { 2417 // This macro has parameters, look for \foo, \bar, etc. 2418 if (Body[Pos] == '\\' && Pos + 1 != End) 2419 break; 2420 } 2421 } 2422 2423 // Add the prefix. 2424 OS << Body.slice(0, Pos); 2425 2426 // Check if we reached the end. 2427 if (Pos == End) 2428 break; 2429 2430 if (IsDarwin && !NParameters) { 2431 switch (Body[Pos + 1]) { 2432 // $$ => $ 2433 case '$': 2434 OS << '$'; 2435 break; 2436 2437 // $n => number of arguments 2438 case 'n': 2439 OS << A.size(); 2440 break; 2441 2442 // $[0-9] => argument 2443 default: { 2444 // Missing arguments are ignored. 2445 unsigned Index = Body[Pos + 1] - '0'; 2446 if (Index >= A.size()) 2447 break; 2448 2449 // Otherwise substitute with the token values, with spaces eliminated. 2450 for (const AsmToken &Token : A[Index]) 2451 OS << Token.getString(); 2452 break; 2453 } 2454 } 2455 Pos += 2; 2456 } else { 2457 unsigned I = Pos + 1; 2458 2459 // Check for the \@ pseudo-variable. 2460 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2461 ++I; 2462 else 2463 while (isIdentifierChar(Body[I]) && I + 1 != End) 2464 ++I; 2465 2466 const char *Begin = Body.data() + Pos + 1; 2467 StringRef Argument(Begin, I - (Pos + 1)); 2468 unsigned Index = 0; 2469 2470 if (Argument == "@") { 2471 OS << NumOfMacroInstantiations; 2472 Pos += 2; 2473 } else { 2474 for (; Index < NParameters; ++Index) 2475 if (Parameters[Index].Name == Argument) 2476 break; 2477 2478 if (Index == NParameters) { 2479 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2480 Pos += 3; 2481 else { 2482 OS << '\\' << Argument; 2483 Pos = I; 2484 } 2485 } else { 2486 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2487 for (const AsmToken &Token : A[Index]) 2488 // For altmacro mode, you can write '%expr'. 2489 // The prefix '%' evaluates the expression 'expr' 2490 // and uses the result as a string (e.g. replace %(1+2) with the 2491 // string "3"). 2492 // Here, we identify the integer token which is the result of the 2493 // absolute expression evaluation and replace it with its string 2494 // representation. 2495 if (AltMacroMode && Token.getString().front() == '%' && 2496 Token.is(AsmToken::Integer)) 2497 // Emit an integer value to the buffer. 2498 OS << Token.getIntVal(); 2499 // Only Token that was validated as a string and begins with '<' 2500 // is considered altMacroString!!! 2501 else if (AltMacroMode && Token.getString().front() == '<' && 2502 Token.is(AsmToken::String)) { 2503 OS << altMacroString(Token.getStringContents()); 2504 } 2505 // We expect no quotes around the string's contents when 2506 // parsing for varargs. 2507 else if (Token.isNot(AsmToken::String) || VarargParameter) 2508 OS << Token.getString(); 2509 else 2510 OS << Token.getStringContents(); 2511 2512 Pos += 1 + Argument.size(); 2513 } 2514 } 2515 } 2516 // Update the scan point. 2517 Body = Body.substr(Pos); 2518 } 2519 2520 return false; 2521 } 2522 2523 MacroInstantiation::MacroInstantiation(SMLoc IL, int EB, SMLoc EL, 2524 size_t CondStackDepth) 2525 : InstantiationLoc(IL), ExitBuffer(EB), ExitLoc(EL), 2526 CondStackDepth(CondStackDepth) {} 2527 2528 static bool isOperator(AsmToken::TokenKind kind) { 2529 switch (kind) { 2530 default: 2531 return false; 2532 case AsmToken::Plus: 2533 case AsmToken::Minus: 2534 case AsmToken::Tilde: 2535 case AsmToken::Slash: 2536 case AsmToken::Star: 2537 case AsmToken::Dot: 2538 case AsmToken::Equal: 2539 case AsmToken::EqualEqual: 2540 case AsmToken::Pipe: 2541 case AsmToken::PipePipe: 2542 case AsmToken::Caret: 2543 case AsmToken::Amp: 2544 case AsmToken::AmpAmp: 2545 case AsmToken::Exclaim: 2546 case AsmToken::ExclaimEqual: 2547 case AsmToken::Less: 2548 case AsmToken::LessEqual: 2549 case AsmToken::LessLess: 2550 case AsmToken::LessGreater: 2551 case AsmToken::Greater: 2552 case AsmToken::GreaterEqual: 2553 case AsmToken::GreaterGreater: 2554 return true; 2555 } 2556 } 2557 2558 namespace { 2559 2560 class AsmLexerSkipSpaceRAII { 2561 public: 2562 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2563 Lexer.setSkipSpace(SkipSpace); 2564 } 2565 2566 ~AsmLexerSkipSpaceRAII() { 2567 Lexer.setSkipSpace(true); 2568 } 2569 2570 private: 2571 AsmLexer &Lexer; 2572 }; 2573 2574 } // end anonymous namespace 2575 2576 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2577 2578 if (Vararg) { 2579 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2580 StringRef Str = parseStringToEndOfStatement(); 2581 MA.emplace_back(AsmToken::String, Str); 2582 } 2583 return false; 2584 } 2585 2586 unsigned ParenLevel = 0; 2587 2588 // Darwin doesn't use spaces to delmit arguments. 2589 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2590 2591 bool SpaceEaten; 2592 2593 while (true) { 2594 SpaceEaten = false; 2595 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2596 return TokError("unexpected token in macro instantiation"); 2597 2598 if (ParenLevel == 0) { 2599 2600 if (Lexer.is(AsmToken::Comma)) 2601 break; 2602 2603 if (Lexer.is(AsmToken::Space)) { 2604 SpaceEaten = true; 2605 Lexer.Lex(); // Eat spaces 2606 } 2607 2608 // Spaces can delimit parameters, but could also be part an expression. 2609 // If the token after a space is an operator, add the token and the next 2610 // one into this argument 2611 if (!IsDarwin) { 2612 if (isOperator(Lexer.getKind())) { 2613 MA.push_back(getTok()); 2614 Lexer.Lex(); 2615 2616 // Whitespace after an operator can be ignored. 2617 if (Lexer.is(AsmToken::Space)) 2618 Lexer.Lex(); 2619 2620 continue; 2621 } 2622 } 2623 if (SpaceEaten) 2624 break; 2625 } 2626 2627 // handleMacroEntry relies on not advancing the lexer here 2628 // to be able to fill in the remaining default parameter values 2629 if (Lexer.is(AsmToken::EndOfStatement)) 2630 break; 2631 2632 // Adjust the current parentheses level. 2633 if (Lexer.is(AsmToken::LParen)) 2634 ++ParenLevel; 2635 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2636 --ParenLevel; 2637 2638 // Append the token to the current argument list. 2639 MA.push_back(getTok()); 2640 Lexer.Lex(); 2641 } 2642 2643 if (ParenLevel != 0) 2644 return TokError("unbalanced parentheses in macro argument"); 2645 return false; 2646 } 2647 2648 // Parse the macro instantiation arguments. 2649 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2650 MCAsmMacroArguments &A) { 2651 const unsigned NParameters = M ? M->Parameters.size() : 0; 2652 bool NamedParametersFound = false; 2653 SmallVector<SMLoc, 4> FALocs; 2654 2655 A.resize(NParameters); 2656 FALocs.resize(NParameters); 2657 2658 // Parse two kinds of macro invocations: 2659 // - macros defined without any parameters accept an arbitrary number of them 2660 // - macros defined with parameters accept at most that many of them 2661 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2662 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2663 ++Parameter) { 2664 SMLoc IDLoc = Lexer.getLoc(); 2665 MCAsmMacroParameter FA; 2666 2667 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2668 if (parseIdentifier(FA.Name)) 2669 return Error(IDLoc, "invalid argument identifier for formal argument"); 2670 2671 if (Lexer.isNot(AsmToken::Equal)) 2672 return TokError("expected '=' after formal parameter identifier"); 2673 2674 Lex(); 2675 2676 NamedParametersFound = true; 2677 } 2678 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2679 2680 if (NamedParametersFound && FA.Name.empty()) 2681 return Error(IDLoc, "cannot mix positional and keyword arguments"); 2682 2683 SMLoc StrLoc = Lexer.getLoc(); 2684 SMLoc EndLoc; 2685 if (AltMacroMode && Lexer.is(AsmToken::Percent)) { 2686 const MCExpr *AbsoluteExp; 2687 int64_t Value; 2688 /// Eat '%' 2689 Lex(); 2690 if (parseExpression(AbsoluteExp, EndLoc)) 2691 return false; 2692 if (!AbsoluteExp->evaluateAsAbsolute(Value, 2693 getStreamer().getAssemblerPtr())) 2694 return Error(StrLoc, "expected absolute expression"); 2695 const char *StrChar = StrLoc.getPointer(); 2696 const char *EndChar = EndLoc.getPointer(); 2697 AsmToken newToken(AsmToken::Integer, 2698 StringRef(StrChar, EndChar - StrChar), Value); 2699 FA.Value.push_back(newToken); 2700 } else if (AltMacroMode && Lexer.is(AsmToken::Less) && 2701 isAltmacroString(StrLoc, EndLoc)) { 2702 const char *StrChar = StrLoc.getPointer(); 2703 const char *EndChar = EndLoc.getPointer(); 2704 jumpToLoc(EndLoc, CurBuffer); 2705 /// Eat from '<' to '>' 2706 Lex(); 2707 AsmToken newToken(AsmToken::String, 2708 StringRef(StrChar, EndChar - StrChar)); 2709 FA.Value.push_back(newToken); 2710 } else if(parseMacroArgument(FA.Value, Vararg)) 2711 return true; 2712 2713 unsigned PI = Parameter; 2714 if (!FA.Name.empty()) { 2715 unsigned FAI = 0; 2716 for (FAI = 0; FAI < NParameters; ++FAI) 2717 if (M->Parameters[FAI].Name == FA.Name) 2718 break; 2719 2720 if (FAI >= NParameters) { 2721 assert(M && "expected macro to be defined"); 2722 return Error(IDLoc, "parameter named '" + FA.Name + 2723 "' does not exist for macro '" + M->Name + "'"); 2724 } 2725 PI = FAI; 2726 } 2727 2728 if (!FA.Value.empty()) { 2729 if (A.size() <= PI) 2730 A.resize(PI + 1); 2731 A[PI] = FA.Value; 2732 2733 if (FALocs.size() <= PI) 2734 FALocs.resize(PI + 1); 2735 2736 FALocs[PI] = Lexer.getLoc(); 2737 } 2738 2739 // At the end of the statement, fill in remaining arguments that have 2740 // default values. If there aren't any, then the next argument is 2741 // required but missing 2742 if (Lexer.is(AsmToken::EndOfStatement)) { 2743 bool Failure = false; 2744 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2745 if (A[FAI].empty()) { 2746 if (M->Parameters[FAI].Required) { 2747 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2748 "missing value for required parameter " 2749 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2750 Failure = true; 2751 } 2752 2753 if (!M->Parameters[FAI].Value.empty()) 2754 A[FAI] = M->Parameters[FAI].Value; 2755 } 2756 } 2757 return Failure; 2758 } 2759 2760 if (Lexer.is(AsmToken::Comma)) 2761 Lex(); 2762 } 2763 2764 return TokError("too many positional arguments"); 2765 } 2766 2767 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2768 // Arbitrarily limit macro nesting depth (default matches 'as'). We can 2769 // eliminate this, although we should protect against infinite loops. 2770 unsigned MaxNestingDepth = AsmMacroMaxNestingDepth; 2771 if (ActiveMacros.size() == MaxNestingDepth) { 2772 std::ostringstream MaxNestingDepthError; 2773 MaxNestingDepthError << "macros cannot be nested more than " 2774 << MaxNestingDepth << " levels deep." 2775 << " Use -asm-macro-max-nesting-depth to increase " 2776 "this limit."; 2777 return TokError(MaxNestingDepthError.str()); 2778 } 2779 2780 MCAsmMacroArguments A; 2781 if (parseMacroArguments(M, A)) 2782 return true; 2783 2784 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2785 // to hold the macro body with substitutions. 2786 SmallString<256> Buf; 2787 StringRef Body = M->Body; 2788 raw_svector_ostream OS(Buf); 2789 2790 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2791 return true; 2792 2793 // We include the .endmacro in the buffer as our cue to exit the macro 2794 // instantiation. 2795 OS << ".endmacro\n"; 2796 2797 std::unique_ptr<MemoryBuffer> Instantiation = 2798 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2799 2800 // Create the macro instantiation object and add to the current macro 2801 // instantiation stack. 2802 MacroInstantiation *MI = new MacroInstantiation( 2803 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 2804 ActiveMacros.push_back(MI); 2805 2806 ++NumOfMacroInstantiations; 2807 2808 // Jump to the macro instantiation and prime the lexer. 2809 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2810 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2811 Lex(); 2812 2813 return false; 2814 } 2815 2816 void AsmParser::handleMacroExit() { 2817 // Jump to the EndOfStatement we should return to, and consume it. 2818 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2819 Lex(); 2820 2821 // Pop the instantiation entry. 2822 delete ActiveMacros.back(); 2823 ActiveMacros.pop_back(); 2824 } 2825 2826 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2827 bool NoDeadStrip) { 2828 MCSymbol *Sym; 2829 const MCExpr *Value; 2830 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2831 Value)) 2832 return true; 2833 2834 if (!Sym) { 2835 // In the case where we parse an expression starting with a '.', we will 2836 // not generate an error, nor will we create a symbol. In this case we 2837 // should just return out. 2838 return false; 2839 } 2840 2841 // Do the assignment. 2842 Out.EmitAssignment(Sym, Value); 2843 if (NoDeadStrip) 2844 Out.EmitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2845 2846 return false; 2847 } 2848 2849 /// parseIdentifier: 2850 /// ::= identifier 2851 /// ::= string 2852 bool AsmParser::parseIdentifier(StringRef &Res) { 2853 // The assembler has relaxed rules for accepting identifiers, in particular we 2854 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2855 // separate tokens. At this level, we have already lexed so we cannot (currently) 2856 // handle this as a context dependent token, instead we detect adjacent tokens 2857 // and return the combined identifier. 2858 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2859 SMLoc PrefixLoc = getLexer().getLoc(); 2860 2861 // Consume the prefix character, and check for a following identifier. 2862 2863 AsmToken Buf[1]; 2864 Lexer.peekTokens(Buf, false); 2865 2866 if (Buf[0].isNot(AsmToken::Identifier)) 2867 return true; 2868 2869 // We have a '$' or '@' followed by an identifier, make sure they are adjacent. 2870 if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer()) 2871 return true; 2872 2873 // eat $ or @ 2874 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2875 // Construct the joined identifier and consume the token. 2876 Res = 2877 StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1); 2878 Lex(); // Parser Lex to maintain invariants. 2879 return false; 2880 } 2881 2882 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2883 return true; 2884 2885 Res = getTok().getIdentifier(); 2886 2887 Lex(); // Consume the identifier token. 2888 2889 return false; 2890 } 2891 2892 /// parseDirectiveSet: 2893 /// ::= .equ identifier ',' expression 2894 /// ::= .equiv identifier ',' expression 2895 /// ::= .set identifier ',' expression 2896 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2897 StringRef Name; 2898 if (check(parseIdentifier(Name), "expected identifier") || 2899 parseToken(AsmToken::Comma) || parseAssignment(Name, allow_redef, true)) 2900 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2901 return false; 2902 } 2903 2904 bool AsmParser::parseEscapedString(std::string &Data) { 2905 if (check(getTok().isNot(AsmToken::String), "expected string")) 2906 return true; 2907 2908 Data = ""; 2909 StringRef Str = getTok().getStringContents(); 2910 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2911 if (Str[i] != '\\') { 2912 Data += Str[i]; 2913 continue; 2914 } 2915 2916 // Recognize escaped characters. Note that this escape semantics currently 2917 // loosely follows Darwin 'as'. 2918 ++i; 2919 if (i == e) 2920 return TokError("unexpected backslash at end of string"); 2921 2922 // Recognize hex sequences similarly to GNU 'as'. 2923 if (Str[i] == 'x' || Str[i] == 'X') { 2924 size_t length = Str.size(); 2925 if (i + 1 >= length || !isHexDigit(Str[i + 1])) 2926 return TokError("invalid hexadecimal escape sequence"); 2927 2928 // Consume hex characters. GNU 'as' reads all hexadecimal characters and 2929 // then truncates to the lower 16 bits. Seems reasonable. 2930 unsigned Value = 0; 2931 while (i + 1 < length && isHexDigit(Str[i + 1])) 2932 Value = Value * 16 + hexDigitValue(Str[++i]); 2933 2934 Data += (unsigned char)(Value & 0xFF); 2935 continue; 2936 } 2937 2938 // Recognize octal sequences. 2939 if ((unsigned)(Str[i] - '0') <= 7) { 2940 // Consume up to three octal characters. 2941 unsigned Value = Str[i] - '0'; 2942 2943 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2944 ++i; 2945 Value = Value * 8 + (Str[i] - '0'); 2946 2947 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2948 ++i; 2949 Value = Value * 8 + (Str[i] - '0'); 2950 } 2951 } 2952 2953 if (Value > 255) 2954 return TokError("invalid octal escape sequence (out of range)"); 2955 2956 Data += (unsigned char)Value; 2957 continue; 2958 } 2959 2960 // Otherwise recognize individual escapes. 2961 switch (Str[i]) { 2962 default: 2963 // Just reject invalid escape sequences for now. 2964 return TokError("invalid escape sequence (unrecognized character)"); 2965 2966 case 'b': Data += '\b'; break; 2967 case 'f': Data += '\f'; break; 2968 case 'n': Data += '\n'; break; 2969 case 'r': Data += '\r'; break; 2970 case 't': Data += '\t'; break; 2971 case '"': Data += '"'; break; 2972 case '\\': Data += '\\'; break; 2973 } 2974 } 2975 2976 Lex(); 2977 return false; 2978 } 2979 2980 /// parseDirectiveAscii: 2981 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2982 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2983 auto parseOp = [&]() -> bool { 2984 std::string Data; 2985 if (checkForValidSection() || parseEscapedString(Data)) 2986 return true; 2987 getStreamer().EmitBytes(Data); 2988 if (ZeroTerminated) 2989 getStreamer().EmitBytes(StringRef("\0", 1)); 2990 return false; 2991 }; 2992 2993 if (parseMany(parseOp)) 2994 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2995 return false; 2996 } 2997 2998 /// parseDirectiveReloc 2999 /// ::= .reloc expression , identifier [ , expression ] 3000 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 3001 const MCExpr *Offset; 3002 const MCExpr *Expr = nullptr; 3003 int64_t OffsetValue; 3004 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 3005 3006 if (parseExpression(Offset)) 3007 return true; 3008 3009 if ((Offset->evaluateAsAbsolute(OffsetValue, 3010 getStreamer().getAssemblerPtr()) && 3011 check(OffsetValue < 0, OffsetLoc, "expression is negative")) || 3012 (check(Offset->getKind() != llvm::MCExpr::Constant && 3013 Offset->getKind() != llvm::MCExpr::SymbolRef, 3014 OffsetLoc, "expected non-negative number or a label")) || 3015 (parseToken(AsmToken::Comma, "expected comma") || 3016 check(getTok().isNot(AsmToken::Identifier), "expected relocation name"))) 3017 return true; 3018 3019 SMLoc NameLoc = Lexer.getTok().getLoc(); 3020 StringRef Name = Lexer.getTok().getIdentifier(); 3021 Lex(); 3022 3023 if (Lexer.is(AsmToken::Comma)) { 3024 Lex(); 3025 SMLoc ExprLoc = Lexer.getLoc(); 3026 if (parseExpression(Expr)) 3027 return true; 3028 3029 MCValue Value; 3030 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 3031 return Error(ExprLoc, "expression must be relocatable"); 3032 } 3033 3034 if (parseToken(AsmToken::EndOfStatement, 3035 "unexpected token in .reloc directive")) 3036 return true; 3037 3038 const MCTargetAsmParser &MCT = getTargetParser(); 3039 const MCSubtargetInfo &STI = MCT.getSTI(); 3040 if (getStreamer().EmitRelocDirective(*Offset, Name, Expr, DirectiveLoc, STI)) 3041 return Error(NameLoc, "unknown relocation name"); 3042 3043 return false; 3044 } 3045 3046 /// parseDirectiveValue 3047 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 3048 bool AsmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) { 3049 auto parseOp = [&]() -> bool { 3050 const MCExpr *Value; 3051 SMLoc ExprLoc = getLexer().getLoc(); 3052 if (checkForValidSection() || parseExpression(Value)) 3053 return true; 3054 // Special case constant expressions to match code generator. 3055 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3056 assert(Size <= 8 && "Invalid size"); 3057 uint64_t IntValue = MCE->getValue(); 3058 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 3059 return Error(ExprLoc, "out of range literal value"); 3060 getStreamer().EmitIntValue(IntValue, Size); 3061 } else 3062 getStreamer().EmitValue(Value, Size, ExprLoc); 3063 return false; 3064 }; 3065 3066 if (parseMany(parseOp)) 3067 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3068 return false; 3069 } 3070 3071 static bool parseHexOcta(AsmParser &Asm, uint64_t &hi, uint64_t &lo) { 3072 if (Asm.getTok().isNot(AsmToken::Integer) && 3073 Asm.getTok().isNot(AsmToken::BigNum)) 3074 return Asm.TokError("unknown token in expression"); 3075 SMLoc ExprLoc = Asm.getTok().getLoc(); 3076 APInt IntValue = Asm.getTok().getAPIntVal(); 3077 Asm.Lex(); 3078 if (!IntValue.isIntN(128)) 3079 return Asm.Error(ExprLoc, "out of range literal value"); 3080 if (!IntValue.isIntN(64)) { 3081 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 3082 lo = IntValue.getLoBits(64).getZExtValue(); 3083 } else { 3084 hi = 0; 3085 lo = IntValue.getZExtValue(); 3086 } 3087 return false; 3088 } 3089 3090 /// ParseDirectiveOctaValue 3091 /// ::= .octa [ hexconstant (, hexconstant)* ] 3092 3093 bool AsmParser::parseDirectiveOctaValue(StringRef IDVal) { 3094 auto parseOp = [&]() -> bool { 3095 if (checkForValidSection()) 3096 return true; 3097 uint64_t hi, lo; 3098 if (parseHexOcta(*this, hi, lo)) 3099 return true; 3100 if (MAI.isLittleEndian()) { 3101 getStreamer().EmitIntValue(lo, 8); 3102 getStreamer().EmitIntValue(hi, 8); 3103 } else { 3104 getStreamer().EmitIntValue(hi, 8); 3105 getStreamer().EmitIntValue(lo, 8); 3106 } 3107 return false; 3108 }; 3109 3110 if (parseMany(parseOp)) 3111 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3112 return false; 3113 } 3114 3115 bool AsmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) { 3116 // We don't truly support arithmetic on floating point expressions, so we 3117 // have to manually parse unary prefixes. 3118 bool IsNeg = false; 3119 if (getLexer().is(AsmToken::Minus)) { 3120 Lexer.Lex(); 3121 IsNeg = true; 3122 } else if (getLexer().is(AsmToken::Plus)) 3123 Lexer.Lex(); 3124 3125 if (Lexer.is(AsmToken::Error)) 3126 return TokError(Lexer.getErr()); 3127 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 3128 Lexer.isNot(AsmToken::Identifier)) 3129 return TokError("unexpected token in directive"); 3130 3131 // Convert to an APFloat. 3132 APFloat Value(Semantics); 3133 StringRef IDVal = getTok().getString(); 3134 if (getLexer().is(AsmToken::Identifier)) { 3135 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 3136 Value = APFloat::getInf(Semantics); 3137 else if (!IDVal.compare_lower("nan")) 3138 Value = APFloat::getNaN(Semantics, false, ~0); 3139 else 3140 return TokError("invalid floating point literal"); 3141 } else if (Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) == 3142 APFloat::opInvalidOp) 3143 return TokError("invalid floating point literal"); 3144 if (IsNeg) 3145 Value.changeSign(); 3146 3147 // Consume the numeric token. 3148 Lex(); 3149 3150 Res = Value.bitcastToAPInt(); 3151 3152 return false; 3153 } 3154 3155 /// parseDirectiveRealValue 3156 /// ::= (.single | .double) [ expression (, expression)* ] 3157 bool AsmParser::parseDirectiveRealValue(StringRef IDVal, 3158 const fltSemantics &Semantics) { 3159 auto parseOp = [&]() -> bool { 3160 APInt AsInt; 3161 if (checkForValidSection() || parseRealValue(Semantics, AsInt)) 3162 return true; 3163 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 3164 AsInt.getBitWidth() / 8); 3165 return false; 3166 }; 3167 3168 if (parseMany(parseOp)) 3169 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3170 return false; 3171 } 3172 3173 /// parseDirectiveZero 3174 /// ::= .zero expression 3175 bool AsmParser::parseDirectiveZero() { 3176 SMLoc NumBytesLoc = Lexer.getLoc(); 3177 const MCExpr *NumBytes; 3178 if (checkForValidSection() || parseExpression(NumBytes)) 3179 return true; 3180 3181 int64_t Val = 0; 3182 if (getLexer().is(AsmToken::Comma)) { 3183 Lex(); 3184 if (parseAbsoluteExpression(Val)) 3185 return true; 3186 } 3187 3188 if (parseToken(AsmToken::EndOfStatement, 3189 "unexpected token in '.zero' directive")) 3190 return true; 3191 getStreamer().emitFill(*NumBytes, Val, NumBytesLoc); 3192 3193 return false; 3194 } 3195 3196 /// parseDirectiveFill 3197 /// ::= .fill expression [ , expression [ , expression ] ] 3198 bool AsmParser::parseDirectiveFill() { 3199 SMLoc NumValuesLoc = Lexer.getLoc(); 3200 const MCExpr *NumValues; 3201 if (checkForValidSection() || parseExpression(NumValues)) 3202 return true; 3203 3204 int64_t FillSize = 1; 3205 int64_t FillExpr = 0; 3206 3207 SMLoc SizeLoc, ExprLoc; 3208 3209 if (parseOptionalToken(AsmToken::Comma)) { 3210 SizeLoc = getTok().getLoc(); 3211 if (parseAbsoluteExpression(FillSize)) 3212 return true; 3213 if (parseOptionalToken(AsmToken::Comma)) { 3214 ExprLoc = getTok().getLoc(); 3215 if (parseAbsoluteExpression(FillExpr)) 3216 return true; 3217 } 3218 } 3219 if (parseToken(AsmToken::EndOfStatement, 3220 "unexpected token in '.fill' directive")) 3221 return true; 3222 3223 if (FillSize < 0) { 3224 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 3225 return false; 3226 } 3227 if (FillSize > 8) { 3228 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 3229 FillSize = 8; 3230 } 3231 3232 if (!isUInt<32>(FillExpr) && FillSize > 4) 3233 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 3234 3235 getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc); 3236 3237 return false; 3238 } 3239 3240 /// parseDirectiveOrg 3241 /// ::= .org expression [ , expression ] 3242 bool AsmParser::parseDirectiveOrg() { 3243 const MCExpr *Offset; 3244 SMLoc OffsetLoc = Lexer.getLoc(); 3245 if (checkForValidSection() || parseExpression(Offset)) 3246 return true; 3247 3248 // Parse optional fill expression. 3249 int64_t FillExpr = 0; 3250 if (parseOptionalToken(AsmToken::Comma)) 3251 if (parseAbsoluteExpression(FillExpr)) 3252 return addErrorSuffix(" in '.org' directive"); 3253 if (parseToken(AsmToken::EndOfStatement)) 3254 return addErrorSuffix(" in '.org' directive"); 3255 3256 getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc); 3257 return false; 3258 } 3259 3260 /// parseDirectiveAlign 3261 /// ::= {.align, ...} expression [ , expression [ , expression ]] 3262 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 3263 SMLoc AlignmentLoc = getLexer().getLoc(); 3264 int64_t Alignment; 3265 SMLoc MaxBytesLoc; 3266 bool HasFillExpr = false; 3267 int64_t FillExpr = 0; 3268 int64_t MaxBytesToFill = 0; 3269 3270 auto parseAlign = [&]() -> bool { 3271 if (parseAbsoluteExpression(Alignment)) 3272 return true; 3273 if (parseOptionalToken(AsmToken::Comma)) { 3274 // The fill expression can be omitted while specifying a maximum number of 3275 // alignment bytes, e.g: 3276 // .align 3,,4 3277 if (getTok().isNot(AsmToken::Comma)) { 3278 HasFillExpr = true; 3279 if (parseAbsoluteExpression(FillExpr)) 3280 return true; 3281 } 3282 if (parseOptionalToken(AsmToken::Comma)) 3283 if (parseTokenLoc(MaxBytesLoc) || 3284 parseAbsoluteExpression(MaxBytesToFill)) 3285 return true; 3286 } 3287 return parseToken(AsmToken::EndOfStatement); 3288 }; 3289 3290 if (checkForValidSection()) 3291 return addErrorSuffix(" in directive"); 3292 // Ignore empty '.p2align' directives for GNU-as compatibility 3293 if (IsPow2 && (ValueSize == 1) && getTok().is(AsmToken::EndOfStatement)) { 3294 Warning(AlignmentLoc, "p2align directive with no operand(s) is ignored"); 3295 return parseToken(AsmToken::EndOfStatement); 3296 } 3297 if (parseAlign()) 3298 return addErrorSuffix(" in directive"); 3299 3300 // Always emit an alignment here even if we thrown an error. 3301 bool ReturnVal = false; 3302 3303 // Compute alignment in bytes. 3304 if (IsPow2) { 3305 // FIXME: Diagnose overflow. 3306 if (Alignment >= 32) { 3307 ReturnVal |= Error(AlignmentLoc, "invalid alignment value"); 3308 Alignment = 31; 3309 } 3310 3311 Alignment = 1ULL << Alignment; 3312 } else { 3313 // Reject alignments that aren't either a power of two or zero, 3314 // for gas compatibility. Alignment of zero is silently rounded 3315 // up to one. 3316 if (Alignment == 0) 3317 Alignment = 1; 3318 if (!isPowerOf2_64(Alignment)) 3319 ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2"); 3320 } 3321 3322 // Diagnose non-sensical max bytes to align. 3323 if (MaxBytesLoc.isValid()) { 3324 if (MaxBytesToFill < 1) { 3325 ReturnVal |= Error(MaxBytesLoc, 3326 "alignment directive can never be satisfied in this " 3327 "many bytes, ignoring maximum bytes expression"); 3328 MaxBytesToFill = 0; 3329 } 3330 3331 if (MaxBytesToFill >= Alignment) { 3332 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 3333 "has no effect"); 3334 MaxBytesToFill = 0; 3335 } 3336 } 3337 3338 // Check whether we should use optimal code alignment for this .align 3339 // directive. 3340 const MCSection *Section = getStreamer().getCurrentSectionOnly(); 3341 assert(Section && "must have section to emit alignment"); 3342 bool UseCodeAlign = Section->UseCodeAlign(); 3343 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 3344 ValueSize == 1 && UseCodeAlign) { 3345 getStreamer().EmitCodeAlignment(Alignment, MaxBytesToFill); 3346 } else { 3347 // FIXME: Target specific behavior about how the "extra" bytes are filled. 3348 getStreamer().EmitValueToAlignment(Alignment, FillExpr, ValueSize, 3349 MaxBytesToFill); 3350 } 3351 3352 return ReturnVal; 3353 } 3354 3355 /// parseDirectiveFile 3356 /// ::= .file filename 3357 /// ::= .file number [directory] filename [md5 checksum] [source source-text] 3358 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 3359 // FIXME: I'm not sure what this is. 3360 int64_t FileNumber = -1; 3361 if (getLexer().is(AsmToken::Integer)) { 3362 FileNumber = getTok().getIntVal(); 3363 Lex(); 3364 3365 if (FileNumber < 0) 3366 return TokError("negative file number"); 3367 } 3368 3369 std::string Path; 3370 3371 // Usually the directory and filename together, otherwise just the directory. 3372 // Allow the strings to have escaped octal character sequence. 3373 if (check(getTok().isNot(AsmToken::String), 3374 "unexpected token in '.file' directive") || 3375 parseEscapedString(Path)) 3376 return true; 3377 3378 StringRef Directory; 3379 StringRef Filename; 3380 std::string FilenameData; 3381 if (getLexer().is(AsmToken::String)) { 3382 if (check(FileNumber == -1, 3383 "explicit path specified, but no file number") || 3384 parseEscapedString(FilenameData)) 3385 return true; 3386 Filename = FilenameData; 3387 Directory = Path; 3388 } else { 3389 Filename = Path; 3390 } 3391 3392 uint64_t MD5Hi, MD5Lo; 3393 bool HasMD5 = false; 3394 3395 Optional<StringRef> Source; 3396 bool HasSource = false; 3397 std::string SourceString; 3398 3399 while (!parseOptionalToken(AsmToken::EndOfStatement)) { 3400 StringRef Keyword; 3401 if (check(getTok().isNot(AsmToken::Identifier), 3402 "unexpected token in '.file' directive") || 3403 parseIdentifier(Keyword)) 3404 return true; 3405 if (Keyword == "md5") { 3406 HasMD5 = true; 3407 if (check(FileNumber == -1, 3408 "MD5 checksum specified, but no file number") || 3409 parseHexOcta(*this, MD5Hi, MD5Lo)) 3410 return true; 3411 } else if (Keyword == "source") { 3412 HasSource = true; 3413 if (check(FileNumber == -1, 3414 "source specified, but no file number") || 3415 check(getTok().isNot(AsmToken::String), 3416 "unexpected token in '.file' directive") || 3417 parseEscapedString(SourceString)) 3418 return true; 3419 } else { 3420 return TokError("unexpected token in '.file' directive"); 3421 } 3422 } 3423 3424 if (FileNumber == -1) { 3425 // Ignore the directive if there is no number and the target doesn't support 3426 // numberless .file directives. This allows some portability of assembler 3427 // between different object file formats. 3428 if (getContext().getAsmInfo()->hasSingleParameterDotFile()) 3429 getStreamer().EmitFileDirective(Filename); 3430 } else { 3431 // In case there is a -g option as well as debug info from directive .file, 3432 // we turn off the -g option, directly use the existing debug info instead. 3433 // Throw away any implicit file table for the assembler source. 3434 if (Ctx.getGenDwarfForAssembly()) { 3435 Ctx.getMCDwarfLineTable(0).resetFileTable(); 3436 Ctx.setGenDwarfForAssembly(false); 3437 } 3438 3439 Optional<MD5::MD5Result> CKMem; 3440 if (HasMD5) { 3441 MD5::MD5Result Sum; 3442 for (unsigned i = 0; i != 8; ++i) { 3443 Sum.Bytes[i] = uint8_t(MD5Hi >> ((7 - i) * 8)); 3444 Sum.Bytes[i + 8] = uint8_t(MD5Lo >> ((7 - i) * 8)); 3445 } 3446 CKMem = Sum; 3447 } 3448 if (HasSource) { 3449 char *SourceBuf = static_cast<char *>(Ctx.allocate(SourceString.size())); 3450 memcpy(SourceBuf, SourceString.data(), SourceString.size()); 3451 Source = StringRef(SourceBuf, SourceString.size()); 3452 } 3453 if (FileNumber == 0) { 3454 if (Ctx.getDwarfVersion() < 5) 3455 return Warning(DirectiveLoc, "file 0 not supported prior to DWARF-5"); 3456 getStreamer().emitDwarfFile0Directive(Directory, Filename, CKMem, Source); 3457 } else { 3458 Expected<unsigned> FileNumOrErr = getStreamer().tryEmitDwarfFileDirective( 3459 FileNumber, Directory, Filename, CKMem, Source); 3460 if (!FileNumOrErr) 3461 return Error(DirectiveLoc, toString(FileNumOrErr.takeError())); 3462 } 3463 // Alert the user if there are some .file directives with MD5 and some not. 3464 // But only do that once. 3465 if (!ReportedInconsistentMD5 && !Ctx.isDwarfMD5UsageConsistent(0)) { 3466 ReportedInconsistentMD5 = true; 3467 return Warning(DirectiveLoc, "inconsistent use of MD5 checksums"); 3468 } 3469 } 3470 3471 return false; 3472 } 3473 3474 /// parseDirectiveLine 3475 /// ::= .line [number] 3476 bool AsmParser::parseDirectiveLine() { 3477 int64_t LineNumber; 3478 if (getLexer().is(AsmToken::Integer)) { 3479 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3480 return true; 3481 (void)LineNumber; 3482 // FIXME: Do something with the .line. 3483 } 3484 if (parseToken(AsmToken::EndOfStatement, 3485 "unexpected token in '.line' directive")) 3486 return true; 3487 3488 return false; 3489 } 3490 3491 /// parseDirectiveLoc 3492 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3493 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3494 /// The first number is a file number, must have been previously assigned with 3495 /// a .file directive, the second number is the line number and optionally the 3496 /// third number is a column position (zero if not specified). The remaining 3497 /// optional items are .loc sub-directives. 3498 bool AsmParser::parseDirectiveLoc() { 3499 int64_t FileNumber = 0, LineNumber = 0; 3500 SMLoc Loc = getTok().getLoc(); 3501 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3502 check(FileNumber < 1 && Ctx.getDwarfVersion() < 5, Loc, 3503 "file number less than one in '.loc' directive") || 3504 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3505 "unassigned file number in '.loc' directive")) 3506 return true; 3507 3508 // optional 3509 if (getLexer().is(AsmToken::Integer)) { 3510 LineNumber = getTok().getIntVal(); 3511 if (LineNumber < 0) 3512 return TokError("line number less than zero in '.loc' directive"); 3513 Lex(); 3514 } 3515 3516 int64_t ColumnPos = 0; 3517 if (getLexer().is(AsmToken::Integer)) { 3518 ColumnPos = getTok().getIntVal(); 3519 if (ColumnPos < 0) 3520 return TokError("column position less than zero in '.loc' directive"); 3521 Lex(); 3522 } 3523 3524 unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0; 3525 unsigned Isa = 0; 3526 int64_t Discriminator = 0; 3527 3528 auto parseLocOp = [&]() -> bool { 3529 StringRef Name; 3530 SMLoc Loc = getTok().getLoc(); 3531 if (parseIdentifier(Name)) 3532 return TokError("unexpected token in '.loc' directive"); 3533 3534 if (Name == "basic_block") 3535 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3536 else if (Name == "prologue_end") 3537 Flags |= DWARF2_FLAG_PROLOGUE_END; 3538 else if (Name == "epilogue_begin") 3539 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3540 else if (Name == "is_stmt") { 3541 Loc = getTok().getLoc(); 3542 const MCExpr *Value; 3543 if (parseExpression(Value)) 3544 return true; 3545 // The expression must be the constant 0 or 1. 3546 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3547 int Value = MCE->getValue(); 3548 if (Value == 0) 3549 Flags &= ~DWARF2_FLAG_IS_STMT; 3550 else if (Value == 1) 3551 Flags |= DWARF2_FLAG_IS_STMT; 3552 else 3553 return Error(Loc, "is_stmt value not 0 or 1"); 3554 } else { 3555 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3556 } 3557 } else if (Name == "isa") { 3558 Loc = getTok().getLoc(); 3559 const MCExpr *Value; 3560 if (parseExpression(Value)) 3561 return true; 3562 // The expression must be a constant greater or equal to 0. 3563 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3564 int Value = MCE->getValue(); 3565 if (Value < 0) 3566 return Error(Loc, "isa number less than zero"); 3567 Isa = Value; 3568 } else { 3569 return Error(Loc, "isa number not a constant value"); 3570 } 3571 } else if (Name == "discriminator") { 3572 if (parseAbsoluteExpression(Discriminator)) 3573 return true; 3574 } else { 3575 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3576 } 3577 return false; 3578 }; 3579 3580 if (parseMany(parseLocOp, false /*hasComma*/)) 3581 return true; 3582 3583 getStreamer().EmitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3584 Isa, Discriminator, StringRef()); 3585 3586 return false; 3587 } 3588 3589 /// parseDirectiveStabs 3590 /// ::= .stabs string, number, number, number 3591 bool AsmParser::parseDirectiveStabs() { 3592 return TokError("unsupported directive '.stabs'"); 3593 } 3594 3595 /// parseDirectiveCVFile 3596 /// ::= .cv_file number filename [checksum] [checksumkind] 3597 bool AsmParser::parseDirectiveCVFile() { 3598 SMLoc FileNumberLoc = getTok().getLoc(); 3599 int64_t FileNumber; 3600 std::string Filename; 3601 std::string Checksum; 3602 int64_t ChecksumKind = 0; 3603 3604 if (parseIntToken(FileNumber, 3605 "expected file number in '.cv_file' directive") || 3606 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3607 check(getTok().isNot(AsmToken::String), 3608 "unexpected token in '.cv_file' directive") || 3609 parseEscapedString(Filename)) 3610 return true; 3611 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3612 if (check(getTok().isNot(AsmToken::String), 3613 "unexpected token in '.cv_file' directive") || 3614 parseEscapedString(Checksum) || 3615 parseIntToken(ChecksumKind, 3616 "expected checksum kind in '.cv_file' directive") || 3617 parseToken(AsmToken::EndOfStatement, 3618 "unexpected token in '.cv_file' directive")) 3619 return true; 3620 } 3621 3622 Checksum = fromHex(Checksum); 3623 void *CKMem = Ctx.allocate(Checksum.size(), 1); 3624 memcpy(CKMem, Checksum.data(), Checksum.size()); 3625 ArrayRef<uint8_t> ChecksumAsBytes(reinterpret_cast<const uint8_t *>(CKMem), 3626 Checksum.size()); 3627 3628 if (!getStreamer().EmitCVFileDirective(FileNumber, Filename, ChecksumAsBytes, 3629 static_cast<uint8_t>(ChecksumKind))) 3630 return Error(FileNumberLoc, "file number already allocated"); 3631 3632 return false; 3633 } 3634 3635 bool AsmParser::parseCVFunctionId(int64_t &FunctionId, 3636 StringRef DirectiveName) { 3637 SMLoc Loc; 3638 return parseTokenLoc(Loc) || 3639 parseIntToken(FunctionId, "expected function id in '" + DirectiveName + 3640 "' directive") || 3641 check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc, 3642 "expected function id within range [0, UINT_MAX)"); 3643 } 3644 3645 bool AsmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) { 3646 SMLoc Loc; 3647 return parseTokenLoc(Loc) || 3648 parseIntToken(FileNumber, "expected integer in '" + DirectiveName + 3649 "' directive") || 3650 check(FileNumber < 1, Loc, "file number less than one in '" + 3651 DirectiveName + "' directive") || 3652 check(!getCVContext().isValidFileNumber(FileNumber), Loc, 3653 "unassigned file number in '" + DirectiveName + "' directive"); 3654 } 3655 3656 /// parseDirectiveCVFuncId 3657 /// ::= .cv_func_id FunctionId 3658 /// 3659 /// Introduces a function ID that can be used with .cv_loc. 3660 bool AsmParser::parseDirectiveCVFuncId() { 3661 SMLoc FunctionIdLoc = getTok().getLoc(); 3662 int64_t FunctionId; 3663 3664 if (parseCVFunctionId(FunctionId, ".cv_func_id") || 3665 parseToken(AsmToken::EndOfStatement, 3666 "unexpected token in '.cv_func_id' directive")) 3667 return true; 3668 3669 if (!getStreamer().EmitCVFuncIdDirective(FunctionId)) 3670 return Error(FunctionIdLoc, "function id already allocated"); 3671 3672 return false; 3673 } 3674 3675 /// parseDirectiveCVInlineSiteId 3676 /// ::= .cv_inline_site_id FunctionId 3677 /// "within" IAFunc 3678 /// "inlined_at" IAFile IALine [IACol] 3679 /// 3680 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined 3681 /// at" source location information for use in the line table of the caller, 3682 /// whether the caller is a real function or another inlined call site. 3683 bool AsmParser::parseDirectiveCVInlineSiteId() { 3684 SMLoc FunctionIdLoc = getTok().getLoc(); 3685 int64_t FunctionId; 3686 int64_t IAFunc; 3687 int64_t IAFile; 3688 int64_t IALine; 3689 int64_t IACol = 0; 3690 3691 // FunctionId 3692 if (parseCVFunctionId(FunctionId, ".cv_inline_site_id")) 3693 return true; 3694 3695 // "within" 3696 if (check((getLexer().isNot(AsmToken::Identifier) || 3697 getTok().getIdentifier() != "within"), 3698 "expected 'within' identifier in '.cv_inline_site_id' directive")) 3699 return true; 3700 Lex(); 3701 3702 // IAFunc 3703 if (parseCVFunctionId(IAFunc, ".cv_inline_site_id")) 3704 return true; 3705 3706 // "inlined_at" 3707 if (check((getLexer().isNot(AsmToken::Identifier) || 3708 getTok().getIdentifier() != "inlined_at"), 3709 "expected 'inlined_at' identifier in '.cv_inline_site_id' " 3710 "directive") ) 3711 return true; 3712 Lex(); 3713 3714 // IAFile IALine 3715 if (parseCVFileId(IAFile, ".cv_inline_site_id") || 3716 parseIntToken(IALine, "expected line number after 'inlined_at'")) 3717 return true; 3718 3719 // [IACol] 3720 if (getLexer().is(AsmToken::Integer)) { 3721 IACol = getTok().getIntVal(); 3722 Lex(); 3723 } 3724 3725 if (parseToken(AsmToken::EndOfStatement, 3726 "unexpected token in '.cv_inline_site_id' directive")) 3727 return true; 3728 3729 if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile, 3730 IALine, IACol, FunctionIdLoc)) 3731 return Error(FunctionIdLoc, "function id already allocated"); 3732 3733 return false; 3734 } 3735 3736 /// parseDirectiveCVLoc 3737 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3738 /// [is_stmt VALUE] 3739 /// The first number is a file number, must have been previously assigned with 3740 /// a .file directive, the second number is the line number and optionally the 3741 /// third number is a column position (zero if not specified). The remaining 3742 /// optional items are .loc sub-directives. 3743 bool AsmParser::parseDirectiveCVLoc() { 3744 SMLoc DirectiveLoc = getTok().getLoc(); 3745 int64_t FunctionId, FileNumber; 3746 if (parseCVFunctionId(FunctionId, ".cv_loc") || 3747 parseCVFileId(FileNumber, ".cv_loc")) 3748 return true; 3749 3750 int64_t LineNumber = 0; 3751 if (getLexer().is(AsmToken::Integer)) { 3752 LineNumber = getTok().getIntVal(); 3753 if (LineNumber < 0) 3754 return TokError("line number less than zero in '.cv_loc' directive"); 3755 Lex(); 3756 } 3757 3758 int64_t ColumnPos = 0; 3759 if (getLexer().is(AsmToken::Integer)) { 3760 ColumnPos = getTok().getIntVal(); 3761 if (ColumnPos < 0) 3762 return TokError("column position less than zero in '.cv_loc' directive"); 3763 Lex(); 3764 } 3765 3766 bool PrologueEnd = false; 3767 uint64_t IsStmt = 0; 3768 3769 auto parseOp = [&]() -> bool { 3770 StringRef Name; 3771 SMLoc Loc = getTok().getLoc(); 3772 if (parseIdentifier(Name)) 3773 return TokError("unexpected token in '.cv_loc' directive"); 3774 if (Name == "prologue_end") 3775 PrologueEnd = true; 3776 else if (Name == "is_stmt") { 3777 Loc = getTok().getLoc(); 3778 const MCExpr *Value; 3779 if (parseExpression(Value)) 3780 return true; 3781 // The expression must be the constant 0 or 1. 3782 IsStmt = ~0ULL; 3783 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3784 IsStmt = MCE->getValue(); 3785 3786 if (IsStmt > 1) 3787 return Error(Loc, "is_stmt value not 0 or 1"); 3788 } else { 3789 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3790 } 3791 return false; 3792 }; 3793 3794 if (parseMany(parseOp, false /*hasComma*/)) 3795 return true; 3796 3797 getStreamer().EmitCVLocDirective(FunctionId, FileNumber, LineNumber, 3798 ColumnPos, PrologueEnd, IsStmt, StringRef(), 3799 DirectiveLoc); 3800 return false; 3801 } 3802 3803 /// parseDirectiveCVLinetable 3804 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3805 bool AsmParser::parseDirectiveCVLinetable() { 3806 int64_t FunctionId; 3807 StringRef FnStartName, FnEndName; 3808 SMLoc Loc = getTok().getLoc(); 3809 if (parseCVFunctionId(FunctionId, ".cv_linetable") || 3810 parseToken(AsmToken::Comma, 3811 "unexpected token in '.cv_linetable' directive") || 3812 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3813 "expected identifier in directive") || 3814 parseToken(AsmToken::Comma, 3815 "unexpected token in '.cv_linetable' directive") || 3816 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3817 "expected identifier in directive")) 3818 return true; 3819 3820 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3821 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3822 3823 getStreamer().EmitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3824 return false; 3825 } 3826 3827 /// parseDirectiveCVInlineLinetable 3828 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3829 bool AsmParser::parseDirectiveCVInlineLinetable() { 3830 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3831 StringRef FnStartName, FnEndName; 3832 SMLoc Loc = getTok().getLoc(); 3833 if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") || 3834 parseTokenLoc(Loc) || 3835 parseIntToken( 3836 SourceFileId, 3837 "expected SourceField in '.cv_inline_linetable' directive") || 3838 check(SourceFileId <= 0, Loc, 3839 "File id less than zero in '.cv_inline_linetable' directive") || 3840 parseTokenLoc(Loc) || 3841 parseIntToken( 3842 SourceLineNum, 3843 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3844 check(SourceLineNum < 0, Loc, 3845 "Line number less than zero in '.cv_inline_linetable' directive") || 3846 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3847 "expected identifier in directive") || 3848 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3849 "expected identifier in directive")) 3850 return true; 3851 3852 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3853 return true; 3854 3855 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3856 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3857 getStreamer().EmitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3858 SourceLineNum, FnStartSym, 3859 FnEndSym); 3860 return false; 3861 } 3862 3863 void AsmParser::initializeCVDefRangeTypeMap() { 3864 CVDefRangeTypeMap["reg"] = CVDR_DEFRANGE_REGISTER; 3865 CVDefRangeTypeMap["frame_ptr_rel"] = CVDR_DEFRANGE_FRAMEPOINTER_REL; 3866 CVDefRangeTypeMap["subfield_reg"] = CVDR_DEFRANGE_SUBFIELD_REGISTER; 3867 CVDefRangeTypeMap["reg_rel"] = CVDR_DEFRANGE_REGISTER_REL; 3868 } 3869 3870 /// parseDirectiveCVDefRange 3871 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3872 bool AsmParser::parseDirectiveCVDefRange() { 3873 SMLoc Loc; 3874 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3875 while (getLexer().is(AsmToken::Identifier)) { 3876 Loc = getLexer().getLoc(); 3877 StringRef GapStartName; 3878 if (parseIdentifier(GapStartName)) 3879 return Error(Loc, "expected identifier in directive"); 3880 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3881 3882 Loc = getLexer().getLoc(); 3883 StringRef GapEndName; 3884 if (parseIdentifier(GapEndName)) 3885 return Error(Loc, "expected identifier in directive"); 3886 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3887 3888 Ranges.push_back({GapStartSym, GapEndSym}); 3889 } 3890 3891 StringRef CVDefRangeTypeStr; 3892 if (parseToken( 3893 AsmToken::Comma, 3894 "expected comma before def_range type in .cv_def_range directive") || 3895 parseIdentifier(CVDefRangeTypeStr)) 3896 return Error(Loc, "expected def_range type in directive"); 3897 3898 StringMap<CVDefRangeType>::const_iterator CVTypeIt = 3899 CVDefRangeTypeMap.find(CVDefRangeTypeStr); 3900 CVDefRangeType CVDRType = (CVTypeIt == CVDefRangeTypeMap.end()) 3901 ? CVDR_DEFRANGE 3902 : CVTypeIt->getValue(); 3903 switch (CVDRType) { 3904 case CVDR_DEFRANGE_REGISTER: { 3905 int64_t DRRegister; 3906 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3907 ".cv_def_range directive") || 3908 parseAbsoluteExpression(DRRegister)) 3909 return Error(Loc, "expected register number"); 3910 3911 codeview::DefRangeRegisterHeader DRHdr; 3912 DRHdr.Register = DRRegister; 3913 DRHdr.MayHaveNoName = 0; 3914 getStreamer().EmitCVDefRangeDirective(Ranges, DRHdr); 3915 break; 3916 } 3917 case CVDR_DEFRANGE_FRAMEPOINTER_REL: { 3918 int64_t DROffset; 3919 if (parseToken(AsmToken::Comma, 3920 "expected comma before offset in .cv_def_range directive") || 3921 parseAbsoluteExpression(DROffset)) 3922 return Error(Loc, "expected offset value"); 3923 3924 codeview::DefRangeFramePointerRelHeader DRHdr; 3925 DRHdr.Offset = DROffset; 3926 getStreamer().EmitCVDefRangeDirective(Ranges, DRHdr); 3927 break; 3928 } 3929 case CVDR_DEFRANGE_SUBFIELD_REGISTER: { 3930 int64_t DRRegister; 3931 int64_t DROffsetInParent; 3932 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3933 ".cv_def_range directive") || 3934 parseAbsoluteExpression(DRRegister)) 3935 return Error(Loc, "expected register number"); 3936 if (parseToken(AsmToken::Comma, 3937 "expected comma before offset in .cv_def_range directive") || 3938 parseAbsoluteExpression(DROffsetInParent)) 3939 return Error(Loc, "expected offset value"); 3940 3941 codeview::DefRangeSubfieldRegisterHeader DRHdr; 3942 DRHdr.Register = DRRegister; 3943 DRHdr.MayHaveNoName = 0; 3944 DRHdr.OffsetInParent = DROffsetInParent; 3945 getStreamer().EmitCVDefRangeDirective(Ranges, DRHdr); 3946 break; 3947 } 3948 case CVDR_DEFRANGE_REGISTER_REL: { 3949 int64_t DRRegister; 3950 int64_t DRFlags; 3951 int64_t DRBasePointerOffset; 3952 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3953 ".cv_def_range directive") || 3954 parseAbsoluteExpression(DRRegister)) 3955 return Error(Loc, "expected register value"); 3956 if (parseToken( 3957 AsmToken::Comma, 3958 "expected comma before flag value in .cv_def_range directive") || 3959 parseAbsoluteExpression(DRFlags)) 3960 return Error(Loc, "expected flag value"); 3961 if (parseToken(AsmToken::Comma, "expected comma before base pointer offset " 3962 "in .cv_def_range directive") || 3963 parseAbsoluteExpression(DRBasePointerOffset)) 3964 return Error(Loc, "expected base pointer offset value"); 3965 3966 codeview::DefRangeRegisterRelHeader DRHdr; 3967 DRHdr.Register = DRRegister; 3968 DRHdr.Flags = DRFlags; 3969 DRHdr.BasePointerOffset = DRBasePointerOffset; 3970 getStreamer().EmitCVDefRangeDirective(Ranges, DRHdr); 3971 break; 3972 } 3973 default: 3974 return Error(Loc, "unexpected def_range type in .cv_def_range directive"); 3975 } 3976 return true; 3977 } 3978 3979 /// parseDirectiveCVString 3980 /// ::= .cv_stringtable "string" 3981 bool AsmParser::parseDirectiveCVString() { 3982 std::string Data; 3983 if (checkForValidSection() || parseEscapedString(Data)) 3984 return addErrorSuffix(" in '.cv_string' directive"); 3985 3986 // Put the string in the table and emit the offset. 3987 std::pair<StringRef, unsigned> Insertion = 3988 getCVContext().addToStringTable(Data); 3989 getStreamer().EmitIntValue(Insertion.second, 4); 3990 return false; 3991 } 3992 3993 /// parseDirectiveCVStringTable 3994 /// ::= .cv_stringtable 3995 bool AsmParser::parseDirectiveCVStringTable() { 3996 getStreamer().EmitCVStringTableDirective(); 3997 return false; 3998 } 3999 4000 /// parseDirectiveCVFileChecksums 4001 /// ::= .cv_filechecksums 4002 bool AsmParser::parseDirectiveCVFileChecksums() { 4003 getStreamer().EmitCVFileChecksumsDirective(); 4004 return false; 4005 } 4006 4007 /// parseDirectiveCVFileChecksumOffset 4008 /// ::= .cv_filechecksumoffset fileno 4009 bool AsmParser::parseDirectiveCVFileChecksumOffset() { 4010 int64_t FileNo; 4011 if (parseIntToken(FileNo, "expected identifier in directive")) 4012 return true; 4013 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 4014 return true; 4015 getStreamer().EmitCVFileChecksumOffsetDirective(FileNo); 4016 return false; 4017 } 4018 4019 /// parseDirectiveCVFPOData 4020 /// ::= .cv_fpo_data procsym 4021 bool AsmParser::parseDirectiveCVFPOData() { 4022 SMLoc DirLoc = getLexer().getLoc(); 4023 StringRef ProcName; 4024 if (parseIdentifier(ProcName)) 4025 return TokError("expected symbol name"); 4026 if (parseEOL("unexpected tokens")) 4027 return addErrorSuffix(" in '.cv_fpo_data' directive"); 4028 MCSymbol *ProcSym = getContext().getOrCreateSymbol(ProcName); 4029 getStreamer().EmitCVFPOData(ProcSym, DirLoc); 4030 return false; 4031 } 4032 4033 /// parseDirectiveCFISections 4034 /// ::= .cfi_sections section [, section] 4035 bool AsmParser::parseDirectiveCFISections() { 4036 StringRef Name; 4037 bool EH = false; 4038 bool Debug = false; 4039 4040 if (parseIdentifier(Name)) 4041 return TokError("Expected an identifier"); 4042 4043 if (Name == ".eh_frame") 4044 EH = true; 4045 else if (Name == ".debug_frame") 4046 Debug = true; 4047 4048 if (getLexer().is(AsmToken::Comma)) { 4049 Lex(); 4050 4051 if (parseIdentifier(Name)) 4052 return TokError("Expected an identifier"); 4053 4054 if (Name == ".eh_frame") 4055 EH = true; 4056 else if (Name == ".debug_frame") 4057 Debug = true; 4058 } 4059 4060 getStreamer().EmitCFISections(EH, Debug); 4061 return false; 4062 } 4063 4064 /// parseDirectiveCFIStartProc 4065 /// ::= .cfi_startproc [simple] 4066 bool AsmParser::parseDirectiveCFIStartProc() { 4067 StringRef Simple; 4068 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4069 if (check(parseIdentifier(Simple) || Simple != "simple", 4070 "unexpected token") || 4071 parseToken(AsmToken::EndOfStatement)) 4072 return addErrorSuffix(" in '.cfi_startproc' directive"); 4073 } 4074 4075 // TODO(kristina): Deal with a corner case of incorrect diagnostic context 4076 // being produced if this directive is emitted as part of preprocessor macro 4077 // expansion which can *ONLY* happen if Clang's cc1as is the API consumer. 4078 // Tools like llvm-mc on the other hand are not affected by it, and report 4079 // correct context information. 4080 getStreamer().EmitCFIStartProc(!Simple.empty(), Lexer.getLoc()); 4081 return false; 4082 } 4083 4084 /// parseDirectiveCFIEndProc 4085 /// ::= .cfi_endproc 4086 bool AsmParser::parseDirectiveCFIEndProc() { 4087 getStreamer().EmitCFIEndProc(); 4088 return false; 4089 } 4090 4091 /// parse register name or number. 4092 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 4093 SMLoc DirectiveLoc) { 4094 unsigned RegNo; 4095 4096 if (getLexer().isNot(AsmToken::Integer)) { 4097 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 4098 return true; 4099 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 4100 } else 4101 return parseAbsoluteExpression(Register); 4102 4103 return false; 4104 } 4105 4106 /// parseDirectiveCFIDefCfa 4107 /// ::= .cfi_def_cfa register, offset 4108 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 4109 int64_t Register = 0, Offset = 0; 4110 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4111 parseToken(AsmToken::Comma, "unexpected token in directive") || 4112 parseAbsoluteExpression(Offset)) 4113 return true; 4114 4115 getStreamer().EmitCFIDefCfa(Register, Offset); 4116 return false; 4117 } 4118 4119 /// parseDirectiveCFIDefCfaOffset 4120 /// ::= .cfi_def_cfa_offset offset 4121 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 4122 int64_t Offset = 0; 4123 if (parseAbsoluteExpression(Offset)) 4124 return true; 4125 4126 getStreamer().EmitCFIDefCfaOffset(Offset); 4127 return false; 4128 } 4129 4130 /// parseDirectiveCFIRegister 4131 /// ::= .cfi_register register, register 4132 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 4133 int64_t Register1 = 0, Register2 = 0; 4134 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || 4135 parseToken(AsmToken::Comma, "unexpected token in directive") || 4136 parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 4137 return true; 4138 4139 getStreamer().EmitCFIRegister(Register1, Register2); 4140 return false; 4141 } 4142 4143 /// parseDirectiveCFIWindowSave 4144 /// ::= .cfi_window_save 4145 bool AsmParser::parseDirectiveCFIWindowSave() { 4146 getStreamer().EmitCFIWindowSave(); 4147 return false; 4148 } 4149 4150 /// parseDirectiveCFIAdjustCfaOffset 4151 /// ::= .cfi_adjust_cfa_offset adjustment 4152 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 4153 int64_t Adjustment = 0; 4154 if (parseAbsoluteExpression(Adjustment)) 4155 return true; 4156 4157 getStreamer().EmitCFIAdjustCfaOffset(Adjustment); 4158 return false; 4159 } 4160 4161 /// parseDirectiveCFIDefCfaRegister 4162 /// ::= .cfi_def_cfa_register register 4163 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 4164 int64_t Register = 0; 4165 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4166 return true; 4167 4168 getStreamer().EmitCFIDefCfaRegister(Register); 4169 return false; 4170 } 4171 4172 /// parseDirectiveCFIOffset 4173 /// ::= .cfi_offset register, offset 4174 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 4175 int64_t Register = 0; 4176 int64_t Offset = 0; 4177 4178 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4179 parseToken(AsmToken::Comma, "unexpected token in directive") || 4180 parseAbsoluteExpression(Offset)) 4181 return true; 4182 4183 getStreamer().EmitCFIOffset(Register, Offset); 4184 return false; 4185 } 4186 4187 /// parseDirectiveCFIRelOffset 4188 /// ::= .cfi_rel_offset register, offset 4189 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 4190 int64_t Register = 0, Offset = 0; 4191 4192 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4193 parseToken(AsmToken::Comma, "unexpected token in directive") || 4194 parseAbsoluteExpression(Offset)) 4195 return true; 4196 4197 getStreamer().EmitCFIRelOffset(Register, Offset); 4198 return false; 4199 } 4200 4201 static bool isValidEncoding(int64_t Encoding) { 4202 if (Encoding & ~0xff) 4203 return false; 4204 4205 if (Encoding == dwarf::DW_EH_PE_omit) 4206 return true; 4207 4208 const unsigned Format = Encoding & 0xf; 4209 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 4210 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 4211 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 4212 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 4213 return false; 4214 4215 const unsigned Application = Encoding & 0x70; 4216 if (Application != dwarf::DW_EH_PE_absptr && 4217 Application != dwarf::DW_EH_PE_pcrel) 4218 return false; 4219 4220 return true; 4221 } 4222 4223 /// parseDirectiveCFIPersonalityOrLsda 4224 /// IsPersonality true for cfi_personality, false for cfi_lsda 4225 /// ::= .cfi_personality encoding, [symbol_name] 4226 /// ::= .cfi_lsda encoding, [symbol_name] 4227 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 4228 int64_t Encoding = 0; 4229 if (parseAbsoluteExpression(Encoding)) 4230 return true; 4231 if (Encoding == dwarf::DW_EH_PE_omit) 4232 return false; 4233 4234 StringRef Name; 4235 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 4236 parseToken(AsmToken::Comma, "unexpected token in directive") || 4237 check(parseIdentifier(Name), "expected identifier in directive")) 4238 return true; 4239 4240 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4241 4242 if (IsPersonality) 4243 getStreamer().EmitCFIPersonality(Sym, Encoding); 4244 else 4245 getStreamer().EmitCFILsda(Sym, Encoding); 4246 return false; 4247 } 4248 4249 /// parseDirectiveCFIRememberState 4250 /// ::= .cfi_remember_state 4251 bool AsmParser::parseDirectiveCFIRememberState() { 4252 getStreamer().EmitCFIRememberState(); 4253 return false; 4254 } 4255 4256 /// parseDirectiveCFIRestoreState 4257 /// ::= .cfi_remember_state 4258 bool AsmParser::parseDirectiveCFIRestoreState() { 4259 getStreamer().EmitCFIRestoreState(); 4260 return false; 4261 } 4262 4263 /// parseDirectiveCFISameValue 4264 /// ::= .cfi_same_value register 4265 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 4266 int64_t Register = 0; 4267 4268 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4269 return true; 4270 4271 getStreamer().EmitCFISameValue(Register); 4272 return false; 4273 } 4274 4275 /// parseDirectiveCFIRestore 4276 /// ::= .cfi_restore register 4277 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 4278 int64_t Register = 0; 4279 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4280 return true; 4281 4282 getStreamer().EmitCFIRestore(Register); 4283 return false; 4284 } 4285 4286 /// parseDirectiveCFIEscape 4287 /// ::= .cfi_escape expression[,...] 4288 bool AsmParser::parseDirectiveCFIEscape() { 4289 std::string Values; 4290 int64_t CurrValue; 4291 if (parseAbsoluteExpression(CurrValue)) 4292 return true; 4293 4294 Values.push_back((uint8_t)CurrValue); 4295 4296 while (getLexer().is(AsmToken::Comma)) { 4297 Lex(); 4298 4299 if (parseAbsoluteExpression(CurrValue)) 4300 return true; 4301 4302 Values.push_back((uint8_t)CurrValue); 4303 } 4304 4305 getStreamer().EmitCFIEscape(Values); 4306 return false; 4307 } 4308 4309 /// parseDirectiveCFIReturnColumn 4310 /// ::= .cfi_return_column register 4311 bool AsmParser::parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc) { 4312 int64_t Register = 0; 4313 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4314 return true; 4315 getStreamer().EmitCFIReturnColumn(Register); 4316 return false; 4317 } 4318 4319 /// parseDirectiveCFISignalFrame 4320 /// ::= .cfi_signal_frame 4321 bool AsmParser::parseDirectiveCFISignalFrame() { 4322 if (parseToken(AsmToken::EndOfStatement, 4323 "unexpected token in '.cfi_signal_frame'")) 4324 return true; 4325 4326 getStreamer().EmitCFISignalFrame(); 4327 return false; 4328 } 4329 4330 /// parseDirectiveCFIUndefined 4331 /// ::= .cfi_undefined register 4332 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 4333 int64_t Register = 0; 4334 4335 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4336 return true; 4337 4338 getStreamer().EmitCFIUndefined(Register); 4339 return false; 4340 } 4341 4342 /// parseDirectiveAltmacro 4343 /// ::= .altmacro 4344 /// ::= .noaltmacro 4345 bool AsmParser::parseDirectiveAltmacro(StringRef Directive) { 4346 if (getLexer().isNot(AsmToken::EndOfStatement)) 4347 return TokError("unexpected token in '" + Directive + "' directive"); 4348 AltMacroMode = (Directive == ".altmacro"); 4349 return false; 4350 } 4351 4352 /// parseDirectiveMacrosOnOff 4353 /// ::= .macros_on 4354 /// ::= .macros_off 4355 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 4356 if (parseToken(AsmToken::EndOfStatement, 4357 "unexpected token in '" + Directive + "' directive")) 4358 return true; 4359 4360 setMacrosEnabled(Directive == ".macros_on"); 4361 return false; 4362 } 4363 4364 /// parseDirectiveMacro 4365 /// ::= .macro name[,] [parameters] 4366 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 4367 StringRef Name; 4368 if (parseIdentifier(Name)) 4369 return TokError("expected identifier in '.macro' directive"); 4370 4371 if (getLexer().is(AsmToken::Comma)) 4372 Lex(); 4373 4374 MCAsmMacroParameters Parameters; 4375 while (getLexer().isNot(AsmToken::EndOfStatement)) { 4376 4377 if (!Parameters.empty() && Parameters.back().Vararg) 4378 return Error(Lexer.getLoc(), 4379 "Vararg parameter '" + Parameters.back().Name + 4380 "' should be last one in the list of parameters."); 4381 4382 MCAsmMacroParameter Parameter; 4383 if (parseIdentifier(Parameter.Name)) 4384 return TokError("expected identifier in '.macro' directive"); 4385 4386 // Emit an error if two (or more) named parameters share the same name 4387 for (const MCAsmMacroParameter& CurrParam : Parameters) 4388 if (CurrParam.Name.equals(Parameter.Name)) 4389 return TokError("macro '" + Name + "' has multiple parameters" 4390 " named '" + Parameter.Name + "'"); 4391 4392 if (Lexer.is(AsmToken::Colon)) { 4393 Lex(); // consume ':' 4394 4395 SMLoc QualLoc; 4396 StringRef Qualifier; 4397 4398 QualLoc = Lexer.getLoc(); 4399 if (parseIdentifier(Qualifier)) 4400 return Error(QualLoc, "missing parameter qualifier for " 4401 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4402 4403 if (Qualifier == "req") 4404 Parameter.Required = true; 4405 else if (Qualifier == "vararg") 4406 Parameter.Vararg = true; 4407 else 4408 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 4409 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 4410 } 4411 4412 if (getLexer().is(AsmToken::Equal)) { 4413 Lex(); 4414 4415 SMLoc ParamLoc; 4416 4417 ParamLoc = Lexer.getLoc(); 4418 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 4419 return true; 4420 4421 if (Parameter.Required) 4422 Warning(ParamLoc, "pointless default value for required parameter " 4423 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4424 } 4425 4426 Parameters.push_back(std::move(Parameter)); 4427 4428 if (getLexer().is(AsmToken::Comma)) 4429 Lex(); 4430 } 4431 4432 // Eat just the end of statement. 4433 Lexer.Lex(); 4434 4435 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 4436 AsmToken EndToken, StartToken = getTok(); 4437 unsigned MacroDepth = 0; 4438 // Lex the macro definition. 4439 while (true) { 4440 // Ignore Lexing errors in macros. 4441 while (Lexer.is(AsmToken::Error)) { 4442 Lexer.Lex(); 4443 } 4444 4445 // Check whether we have reached the end of the file. 4446 if (getLexer().is(AsmToken::Eof)) 4447 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 4448 4449 // Otherwise, check whether we have reach the .endmacro. 4450 if (getLexer().is(AsmToken::Identifier)) { 4451 if (getTok().getIdentifier() == ".endm" || 4452 getTok().getIdentifier() == ".endmacro") { 4453 if (MacroDepth == 0) { // Outermost macro. 4454 EndToken = getTok(); 4455 Lexer.Lex(); 4456 if (getLexer().isNot(AsmToken::EndOfStatement)) 4457 return TokError("unexpected token in '" + EndToken.getIdentifier() + 4458 "' directive"); 4459 break; 4460 } else { 4461 // Otherwise we just found the end of an inner macro. 4462 --MacroDepth; 4463 } 4464 } else if (getTok().getIdentifier() == ".macro") { 4465 // We allow nested macros. Those aren't instantiated until the outermost 4466 // macro is expanded so just ignore them for now. 4467 ++MacroDepth; 4468 } 4469 } 4470 4471 // Otherwise, scan til the end of the statement. 4472 eatToEndOfStatement(); 4473 } 4474 4475 if (getContext().lookupMacro(Name)) { 4476 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 4477 } 4478 4479 const char *BodyStart = StartToken.getLoc().getPointer(); 4480 const char *BodyEnd = EndToken.getLoc().getPointer(); 4481 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4482 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 4483 MCAsmMacro Macro(Name, Body, std::move(Parameters)); 4484 DEBUG_WITH_TYPE("asm-macros", dbgs() << "Defining new macro:\n"; 4485 Macro.dump()); 4486 getContext().defineMacro(Name, std::move(Macro)); 4487 return false; 4488 } 4489 4490 /// checkForBadMacro 4491 /// 4492 /// With the support added for named parameters there may be code out there that 4493 /// is transitioning from positional parameters. In versions of gas that did 4494 /// not support named parameters they would be ignored on the macro definition. 4495 /// But to support both styles of parameters this is not possible so if a macro 4496 /// definition has named parameters but does not use them and has what appears 4497 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 4498 /// warning that the positional parameter found in body which have no effect. 4499 /// Hoping the developer will either remove the named parameters from the macro 4500 /// definition so the positional parameters get used if that was what was 4501 /// intended or change the macro to use the named parameters. It is possible 4502 /// this warning will trigger when the none of the named parameters are used 4503 /// and the strings like $1 are infact to simply to be passed trough unchanged. 4504 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 4505 StringRef Body, 4506 ArrayRef<MCAsmMacroParameter> Parameters) { 4507 // If this macro is not defined with named parameters the warning we are 4508 // checking for here doesn't apply. 4509 unsigned NParameters = Parameters.size(); 4510 if (NParameters == 0) 4511 return; 4512 4513 bool NamedParametersFound = false; 4514 bool PositionalParametersFound = false; 4515 4516 // Look at the body of the macro for use of both the named parameters and what 4517 // are likely to be positional parameters. This is what expandMacro() is 4518 // doing when it finds the parameters in the body. 4519 while (!Body.empty()) { 4520 // Scan for the next possible parameter. 4521 std::size_t End = Body.size(), Pos = 0; 4522 for (; Pos != End; ++Pos) { 4523 // Check for a substitution or escape. 4524 // This macro is defined with parameters, look for \foo, \bar, etc. 4525 if (Body[Pos] == '\\' && Pos + 1 != End) 4526 break; 4527 4528 // This macro should have parameters, but look for $0, $1, ..., $n too. 4529 if (Body[Pos] != '$' || Pos + 1 == End) 4530 continue; 4531 char Next = Body[Pos + 1]; 4532 if (Next == '$' || Next == 'n' || 4533 isdigit(static_cast<unsigned char>(Next))) 4534 break; 4535 } 4536 4537 // Check if we reached the end. 4538 if (Pos == End) 4539 break; 4540 4541 if (Body[Pos] == '$') { 4542 switch (Body[Pos + 1]) { 4543 // $$ => $ 4544 case '$': 4545 break; 4546 4547 // $n => number of arguments 4548 case 'n': 4549 PositionalParametersFound = true; 4550 break; 4551 4552 // $[0-9] => argument 4553 default: { 4554 PositionalParametersFound = true; 4555 break; 4556 } 4557 } 4558 Pos += 2; 4559 } else { 4560 unsigned I = Pos + 1; 4561 while (isIdentifierChar(Body[I]) && I + 1 != End) 4562 ++I; 4563 4564 const char *Begin = Body.data() + Pos + 1; 4565 StringRef Argument(Begin, I - (Pos + 1)); 4566 unsigned Index = 0; 4567 for (; Index < NParameters; ++Index) 4568 if (Parameters[Index].Name == Argument) 4569 break; 4570 4571 if (Index == NParameters) { 4572 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 4573 Pos += 3; 4574 else { 4575 Pos = I; 4576 } 4577 } else { 4578 NamedParametersFound = true; 4579 Pos += 1 + Argument.size(); 4580 } 4581 } 4582 // Update the scan point. 4583 Body = Body.substr(Pos); 4584 } 4585 4586 if (!NamedParametersFound && PositionalParametersFound) 4587 Warning(DirectiveLoc, "macro defined with named parameters which are not " 4588 "used in macro body, possible positional parameter " 4589 "found in body which will have no effect"); 4590 } 4591 4592 /// parseDirectiveExitMacro 4593 /// ::= .exitm 4594 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 4595 if (parseToken(AsmToken::EndOfStatement, 4596 "unexpected token in '" + Directive + "' directive")) 4597 return true; 4598 4599 if (!isInsideMacroInstantiation()) 4600 return TokError("unexpected '" + Directive + "' in file, " 4601 "no current macro definition"); 4602 4603 // Exit all conditionals that are active in the current macro. 4604 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 4605 TheCondState = TheCondStack.back(); 4606 TheCondStack.pop_back(); 4607 } 4608 4609 handleMacroExit(); 4610 return false; 4611 } 4612 4613 /// parseDirectiveEndMacro 4614 /// ::= .endm 4615 /// ::= .endmacro 4616 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 4617 if (getLexer().isNot(AsmToken::EndOfStatement)) 4618 return TokError("unexpected token in '" + Directive + "' directive"); 4619 4620 // If we are inside a macro instantiation, terminate the current 4621 // instantiation. 4622 if (isInsideMacroInstantiation()) { 4623 handleMacroExit(); 4624 return false; 4625 } 4626 4627 // Otherwise, this .endmacro is a stray entry in the file; well formed 4628 // .endmacro directives are handled during the macro definition parsing. 4629 return TokError("unexpected '" + Directive + "' in file, " 4630 "no current macro definition"); 4631 } 4632 4633 /// parseDirectivePurgeMacro 4634 /// ::= .purgem 4635 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 4636 StringRef Name; 4637 SMLoc Loc; 4638 if (parseTokenLoc(Loc) || 4639 check(parseIdentifier(Name), Loc, 4640 "expected identifier in '.purgem' directive") || 4641 parseToken(AsmToken::EndOfStatement, 4642 "unexpected token in '.purgem' directive")) 4643 return true; 4644 4645 if (!getContext().lookupMacro(Name)) 4646 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 4647 4648 getContext().undefineMacro(Name); 4649 DEBUG_WITH_TYPE("asm-macros", dbgs() 4650 << "Un-defining macro: " << Name << "\n"); 4651 return false; 4652 } 4653 4654 /// parseDirectiveBundleAlignMode 4655 /// ::= {.bundle_align_mode} expression 4656 bool AsmParser::parseDirectiveBundleAlignMode() { 4657 // Expect a single argument: an expression that evaluates to a constant 4658 // in the inclusive range 0-30. 4659 SMLoc ExprLoc = getLexer().getLoc(); 4660 int64_t AlignSizePow2; 4661 if (checkForValidSection() || parseAbsoluteExpression(AlignSizePow2) || 4662 parseToken(AsmToken::EndOfStatement, "unexpected token after expression " 4663 "in '.bundle_align_mode' " 4664 "directive") || 4665 check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc, 4666 "invalid bundle alignment size (expected between 0 and 30)")) 4667 return true; 4668 4669 // Because of AlignSizePow2's verified range we can safely truncate it to 4670 // unsigned. 4671 getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 4672 return false; 4673 } 4674 4675 /// parseDirectiveBundleLock 4676 /// ::= {.bundle_lock} [align_to_end] 4677 bool AsmParser::parseDirectiveBundleLock() { 4678 if (checkForValidSection()) 4679 return true; 4680 bool AlignToEnd = false; 4681 4682 StringRef Option; 4683 SMLoc Loc = getTok().getLoc(); 4684 const char *kInvalidOptionError = 4685 "invalid option for '.bundle_lock' directive"; 4686 4687 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4688 if (check(parseIdentifier(Option), Loc, kInvalidOptionError) || 4689 check(Option != "align_to_end", Loc, kInvalidOptionError) || 4690 parseToken(AsmToken::EndOfStatement, 4691 "unexpected token after '.bundle_lock' directive option")) 4692 return true; 4693 AlignToEnd = true; 4694 } 4695 4696 getStreamer().EmitBundleLock(AlignToEnd); 4697 return false; 4698 } 4699 4700 /// parseDirectiveBundleLock 4701 /// ::= {.bundle_lock} 4702 bool AsmParser::parseDirectiveBundleUnlock() { 4703 if (checkForValidSection() || 4704 parseToken(AsmToken::EndOfStatement, 4705 "unexpected token in '.bundle_unlock' directive")) 4706 return true; 4707 4708 getStreamer().EmitBundleUnlock(); 4709 return false; 4710 } 4711 4712 /// parseDirectiveSpace 4713 /// ::= (.skip | .space) expression [ , expression ] 4714 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4715 SMLoc NumBytesLoc = Lexer.getLoc(); 4716 const MCExpr *NumBytes; 4717 if (checkForValidSection() || parseExpression(NumBytes)) 4718 return true; 4719 4720 int64_t FillExpr = 0; 4721 if (parseOptionalToken(AsmToken::Comma)) 4722 if (parseAbsoluteExpression(FillExpr)) 4723 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4724 if (parseToken(AsmToken::EndOfStatement)) 4725 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4726 4727 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4728 getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc); 4729 4730 return false; 4731 } 4732 4733 /// parseDirectiveDCB 4734 /// ::= .dcb.{b, l, w} expression, expression 4735 bool AsmParser::parseDirectiveDCB(StringRef IDVal, unsigned Size) { 4736 SMLoc NumValuesLoc = Lexer.getLoc(); 4737 int64_t NumValues; 4738 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4739 return true; 4740 4741 if (NumValues < 0) { 4742 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4743 return false; 4744 } 4745 4746 if (parseToken(AsmToken::Comma, 4747 "unexpected token in '" + Twine(IDVal) + "' directive")) 4748 return true; 4749 4750 const MCExpr *Value; 4751 SMLoc ExprLoc = getLexer().getLoc(); 4752 if (parseExpression(Value)) 4753 return true; 4754 4755 // Special case constant expressions to match code generator. 4756 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 4757 assert(Size <= 8 && "Invalid size"); 4758 uint64_t IntValue = MCE->getValue(); 4759 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 4760 return Error(ExprLoc, "literal value out of range for directive"); 4761 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4762 getStreamer().EmitIntValue(IntValue, Size); 4763 } else { 4764 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4765 getStreamer().EmitValue(Value, Size, ExprLoc); 4766 } 4767 4768 if (parseToken(AsmToken::EndOfStatement, 4769 "unexpected token in '" + Twine(IDVal) + "' directive")) 4770 return true; 4771 4772 return false; 4773 } 4774 4775 /// parseDirectiveRealDCB 4776 /// ::= .dcb.{d, s} expression, expression 4777 bool AsmParser::parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &Semantics) { 4778 SMLoc NumValuesLoc = Lexer.getLoc(); 4779 int64_t NumValues; 4780 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4781 return true; 4782 4783 if (NumValues < 0) { 4784 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4785 return false; 4786 } 4787 4788 if (parseToken(AsmToken::Comma, 4789 "unexpected token in '" + Twine(IDVal) + "' directive")) 4790 return true; 4791 4792 APInt AsInt; 4793 if (parseRealValue(Semantics, AsInt)) 4794 return true; 4795 4796 if (parseToken(AsmToken::EndOfStatement, 4797 "unexpected token in '" + Twine(IDVal) + "' directive")) 4798 return true; 4799 4800 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4801 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 4802 AsInt.getBitWidth() / 8); 4803 4804 return false; 4805 } 4806 4807 /// parseDirectiveDS 4808 /// ::= .ds.{b, d, l, p, s, w, x} expression 4809 bool AsmParser::parseDirectiveDS(StringRef IDVal, unsigned Size) { 4810 SMLoc NumValuesLoc = Lexer.getLoc(); 4811 int64_t NumValues; 4812 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4813 return true; 4814 4815 if (NumValues < 0) { 4816 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4817 return false; 4818 } 4819 4820 if (parseToken(AsmToken::EndOfStatement, 4821 "unexpected token in '" + Twine(IDVal) + "' directive")) 4822 return true; 4823 4824 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4825 getStreamer().emitFill(Size, 0); 4826 4827 return false; 4828 } 4829 4830 /// parseDirectiveLEB128 4831 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4832 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4833 if (checkForValidSection()) 4834 return true; 4835 4836 auto parseOp = [&]() -> bool { 4837 const MCExpr *Value; 4838 if (parseExpression(Value)) 4839 return true; 4840 if (Signed) 4841 getStreamer().EmitSLEB128Value(Value); 4842 else 4843 getStreamer().EmitULEB128Value(Value); 4844 return false; 4845 }; 4846 4847 if (parseMany(parseOp)) 4848 return addErrorSuffix(" in directive"); 4849 4850 return false; 4851 } 4852 4853 /// parseDirectiveSymbolAttribute 4854 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4855 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4856 auto parseOp = [&]() -> bool { 4857 StringRef Name; 4858 SMLoc Loc = getTok().getLoc(); 4859 if (parseIdentifier(Name)) 4860 return Error(Loc, "expected identifier"); 4861 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4862 4863 // Assembler local symbols don't make any sense here. Complain loudly. 4864 if (Sym->isTemporary()) 4865 return Error(Loc, "non-local symbol required"); 4866 4867 if (!getStreamer().EmitSymbolAttribute(Sym, Attr)) 4868 return Error(Loc, "unable to emit symbol attribute"); 4869 return false; 4870 }; 4871 4872 if (parseMany(parseOp)) 4873 return addErrorSuffix(" in directive"); 4874 return false; 4875 } 4876 4877 /// parseDirectiveComm 4878 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4879 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4880 if (checkForValidSection()) 4881 return true; 4882 4883 SMLoc IDLoc = getLexer().getLoc(); 4884 StringRef Name; 4885 if (parseIdentifier(Name)) 4886 return TokError("expected identifier in directive"); 4887 4888 // Handle the identifier as the key symbol. 4889 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4890 4891 if (getLexer().isNot(AsmToken::Comma)) 4892 return TokError("unexpected token in directive"); 4893 Lex(); 4894 4895 int64_t Size; 4896 SMLoc SizeLoc = getLexer().getLoc(); 4897 if (parseAbsoluteExpression(Size)) 4898 return true; 4899 4900 int64_t Pow2Alignment = 0; 4901 SMLoc Pow2AlignmentLoc; 4902 if (getLexer().is(AsmToken::Comma)) { 4903 Lex(); 4904 Pow2AlignmentLoc = getLexer().getLoc(); 4905 if (parseAbsoluteExpression(Pow2Alignment)) 4906 return true; 4907 4908 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4909 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4910 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4911 4912 // If this target takes alignments in bytes (not log) validate and convert. 4913 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4914 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4915 if (!isPowerOf2_64(Pow2Alignment)) 4916 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4917 Pow2Alignment = Log2_64(Pow2Alignment); 4918 } 4919 } 4920 4921 if (parseToken(AsmToken::EndOfStatement, 4922 "unexpected token in '.comm' or '.lcomm' directive")) 4923 return true; 4924 4925 // NOTE: a size of zero for a .comm should create a undefined symbol 4926 // but a size of .lcomm creates a bss symbol of size zero. 4927 if (Size < 0) 4928 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4929 "be less than zero"); 4930 4931 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4932 // may internally end up wanting an alignment in bytes. 4933 // FIXME: Diagnose overflow. 4934 if (Pow2Alignment < 0) 4935 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4936 "alignment, can't be less than zero"); 4937 4938 Sym->redefineIfPossible(); 4939 if (!Sym->isUndefined()) 4940 return Error(IDLoc, "invalid symbol redefinition"); 4941 4942 // Create the Symbol as a common or local common with Size and Pow2Alignment 4943 if (IsLocal) { 4944 getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4945 return false; 4946 } 4947 4948 getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4949 return false; 4950 } 4951 4952 /// parseDirectiveAbort 4953 /// ::= .abort [... message ...] 4954 bool AsmParser::parseDirectiveAbort() { 4955 // FIXME: Use loc from directive. 4956 SMLoc Loc = getLexer().getLoc(); 4957 4958 StringRef Str = parseStringToEndOfStatement(); 4959 if (parseToken(AsmToken::EndOfStatement, 4960 "unexpected token in '.abort' directive")) 4961 return true; 4962 4963 if (Str.empty()) 4964 return Error(Loc, ".abort detected. Assembly stopping."); 4965 else 4966 return Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 4967 // FIXME: Actually abort assembly here. 4968 4969 return false; 4970 } 4971 4972 /// parseDirectiveInclude 4973 /// ::= .include "filename" 4974 bool AsmParser::parseDirectiveInclude() { 4975 // Allow the strings to have escaped octal character sequence. 4976 std::string Filename; 4977 SMLoc IncludeLoc = getTok().getLoc(); 4978 4979 if (check(getTok().isNot(AsmToken::String), 4980 "expected string in '.include' directive") || 4981 parseEscapedString(Filename) || 4982 check(getTok().isNot(AsmToken::EndOfStatement), 4983 "unexpected token in '.include' directive") || 4984 // Attempt to switch the lexer to the included file before consuming the 4985 // end of statement to avoid losing it when we switch. 4986 check(enterIncludeFile(Filename), IncludeLoc, 4987 "Could not find include file '" + Filename + "'")) 4988 return true; 4989 4990 return false; 4991 } 4992 4993 /// parseDirectiveIncbin 4994 /// ::= .incbin "filename" [ , skip [ , count ] ] 4995 bool AsmParser::parseDirectiveIncbin() { 4996 // Allow the strings to have escaped octal character sequence. 4997 std::string Filename; 4998 SMLoc IncbinLoc = getTok().getLoc(); 4999 if (check(getTok().isNot(AsmToken::String), 5000 "expected string in '.incbin' directive") || 5001 parseEscapedString(Filename)) 5002 return true; 5003 5004 int64_t Skip = 0; 5005 const MCExpr *Count = nullptr; 5006 SMLoc SkipLoc, CountLoc; 5007 if (parseOptionalToken(AsmToken::Comma)) { 5008 // The skip expression can be omitted while specifying the count, e.g: 5009 // .incbin "filename",,4 5010 if (getTok().isNot(AsmToken::Comma)) { 5011 if (parseTokenLoc(SkipLoc) || parseAbsoluteExpression(Skip)) 5012 return true; 5013 } 5014 if (parseOptionalToken(AsmToken::Comma)) { 5015 CountLoc = getTok().getLoc(); 5016 if (parseExpression(Count)) 5017 return true; 5018 } 5019 } 5020 5021 if (parseToken(AsmToken::EndOfStatement, 5022 "unexpected token in '.incbin' directive")) 5023 return true; 5024 5025 if (check(Skip < 0, SkipLoc, "skip is negative")) 5026 return true; 5027 5028 // Attempt to process the included file. 5029 if (processIncbinFile(Filename, Skip, Count, CountLoc)) 5030 return Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 5031 return false; 5032 } 5033 5034 /// parseDirectiveIf 5035 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 5036 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 5037 TheCondStack.push_back(TheCondState); 5038 TheCondState.TheCond = AsmCond::IfCond; 5039 if (TheCondState.Ignore) { 5040 eatToEndOfStatement(); 5041 } else { 5042 int64_t ExprValue; 5043 if (parseAbsoluteExpression(ExprValue) || 5044 parseToken(AsmToken::EndOfStatement, 5045 "unexpected token in '.if' directive")) 5046 return true; 5047 5048 switch (DirKind) { 5049 default: 5050 llvm_unreachable("unsupported directive"); 5051 case DK_IF: 5052 case DK_IFNE: 5053 break; 5054 case DK_IFEQ: 5055 ExprValue = ExprValue == 0; 5056 break; 5057 case DK_IFGE: 5058 ExprValue = ExprValue >= 0; 5059 break; 5060 case DK_IFGT: 5061 ExprValue = ExprValue > 0; 5062 break; 5063 case DK_IFLE: 5064 ExprValue = ExprValue <= 0; 5065 break; 5066 case DK_IFLT: 5067 ExprValue = ExprValue < 0; 5068 break; 5069 } 5070 5071 TheCondState.CondMet = ExprValue; 5072 TheCondState.Ignore = !TheCondState.CondMet; 5073 } 5074 5075 return false; 5076 } 5077 5078 /// parseDirectiveIfb 5079 /// ::= .ifb string 5080 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 5081 TheCondStack.push_back(TheCondState); 5082 TheCondState.TheCond = AsmCond::IfCond; 5083 5084 if (TheCondState.Ignore) { 5085 eatToEndOfStatement(); 5086 } else { 5087 StringRef Str = parseStringToEndOfStatement(); 5088 5089 if (parseToken(AsmToken::EndOfStatement, 5090 "unexpected token in '.ifb' directive")) 5091 return true; 5092 5093 TheCondState.CondMet = ExpectBlank == Str.empty(); 5094 TheCondState.Ignore = !TheCondState.CondMet; 5095 } 5096 5097 return false; 5098 } 5099 5100 /// parseDirectiveIfc 5101 /// ::= .ifc string1, string2 5102 /// ::= .ifnc string1, string2 5103 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 5104 TheCondStack.push_back(TheCondState); 5105 TheCondState.TheCond = AsmCond::IfCond; 5106 5107 if (TheCondState.Ignore) { 5108 eatToEndOfStatement(); 5109 } else { 5110 StringRef Str1 = parseStringToComma(); 5111 5112 if (parseToken(AsmToken::Comma, "unexpected token in '.ifc' directive")) 5113 return true; 5114 5115 StringRef Str2 = parseStringToEndOfStatement(); 5116 5117 if (parseToken(AsmToken::EndOfStatement, 5118 "unexpected token in '.ifc' directive")) 5119 return true; 5120 5121 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 5122 TheCondState.Ignore = !TheCondState.CondMet; 5123 } 5124 5125 return false; 5126 } 5127 5128 /// parseDirectiveIfeqs 5129 /// ::= .ifeqs string1, string2 5130 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 5131 if (Lexer.isNot(AsmToken::String)) { 5132 if (ExpectEqual) 5133 return TokError("expected string parameter for '.ifeqs' directive"); 5134 return TokError("expected string parameter for '.ifnes' directive"); 5135 } 5136 5137 StringRef String1 = getTok().getStringContents(); 5138 Lex(); 5139 5140 if (Lexer.isNot(AsmToken::Comma)) { 5141 if (ExpectEqual) 5142 return TokError( 5143 "expected comma after first string for '.ifeqs' directive"); 5144 return TokError("expected comma after first string for '.ifnes' directive"); 5145 } 5146 5147 Lex(); 5148 5149 if (Lexer.isNot(AsmToken::String)) { 5150 if (ExpectEqual) 5151 return TokError("expected string parameter for '.ifeqs' directive"); 5152 return TokError("expected string parameter for '.ifnes' directive"); 5153 } 5154 5155 StringRef String2 = getTok().getStringContents(); 5156 Lex(); 5157 5158 TheCondStack.push_back(TheCondState); 5159 TheCondState.TheCond = AsmCond::IfCond; 5160 TheCondState.CondMet = ExpectEqual == (String1 == String2); 5161 TheCondState.Ignore = !TheCondState.CondMet; 5162 5163 return false; 5164 } 5165 5166 /// parseDirectiveIfdef 5167 /// ::= .ifdef symbol 5168 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 5169 StringRef Name; 5170 TheCondStack.push_back(TheCondState); 5171 TheCondState.TheCond = AsmCond::IfCond; 5172 5173 if (TheCondState.Ignore) { 5174 eatToEndOfStatement(); 5175 } else { 5176 if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") || 5177 parseToken(AsmToken::EndOfStatement, "unexpected token in '.ifdef'")) 5178 return true; 5179 5180 MCSymbol *Sym = getContext().lookupSymbol(Name); 5181 5182 if (expect_defined) 5183 TheCondState.CondMet = (Sym && !Sym->isUndefined(false)); 5184 else 5185 TheCondState.CondMet = (!Sym || Sym->isUndefined(false)); 5186 TheCondState.Ignore = !TheCondState.CondMet; 5187 } 5188 5189 return false; 5190 } 5191 5192 /// parseDirectiveElseIf 5193 /// ::= .elseif expression 5194 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 5195 if (TheCondState.TheCond != AsmCond::IfCond && 5196 TheCondState.TheCond != AsmCond::ElseIfCond) 5197 return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an" 5198 " .if or an .elseif"); 5199 TheCondState.TheCond = AsmCond::ElseIfCond; 5200 5201 bool LastIgnoreState = false; 5202 if (!TheCondStack.empty()) 5203 LastIgnoreState = TheCondStack.back().Ignore; 5204 if (LastIgnoreState || TheCondState.CondMet) { 5205 TheCondState.Ignore = true; 5206 eatToEndOfStatement(); 5207 } else { 5208 int64_t ExprValue; 5209 if (parseAbsoluteExpression(ExprValue)) 5210 return true; 5211 5212 if (parseToken(AsmToken::EndOfStatement, 5213 "unexpected token in '.elseif' directive")) 5214 return true; 5215 5216 TheCondState.CondMet = ExprValue; 5217 TheCondState.Ignore = !TheCondState.CondMet; 5218 } 5219 5220 return false; 5221 } 5222 5223 /// parseDirectiveElse 5224 /// ::= .else 5225 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 5226 if (parseToken(AsmToken::EndOfStatement, 5227 "unexpected token in '.else' directive")) 5228 return true; 5229 5230 if (TheCondState.TheCond != AsmCond::IfCond && 5231 TheCondState.TheCond != AsmCond::ElseIfCond) 5232 return Error(DirectiveLoc, "Encountered a .else that doesn't follow " 5233 " an .if or an .elseif"); 5234 TheCondState.TheCond = AsmCond::ElseCond; 5235 bool LastIgnoreState = false; 5236 if (!TheCondStack.empty()) 5237 LastIgnoreState = TheCondStack.back().Ignore; 5238 if (LastIgnoreState || TheCondState.CondMet) 5239 TheCondState.Ignore = true; 5240 else 5241 TheCondState.Ignore = false; 5242 5243 return false; 5244 } 5245 5246 /// parseDirectiveEnd 5247 /// ::= .end 5248 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 5249 if (parseToken(AsmToken::EndOfStatement, 5250 "unexpected token in '.end' directive")) 5251 return true; 5252 5253 while (Lexer.isNot(AsmToken::Eof)) 5254 Lexer.Lex(); 5255 5256 return false; 5257 } 5258 5259 /// parseDirectiveError 5260 /// ::= .err 5261 /// ::= .error [string] 5262 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 5263 if (!TheCondStack.empty()) { 5264 if (TheCondStack.back().Ignore) { 5265 eatToEndOfStatement(); 5266 return false; 5267 } 5268 } 5269 5270 if (!WithMessage) 5271 return Error(L, ".err encountered"); 5272 5273 StringRef Message = ".error directive invoked in source file"; 5274 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5275 if (Lexer.isNot(AsmToken::String)) 5276 return TokError(".error argument must be a string"); 5277 5278 Message = getTok().getStringContents(); 5279 Lex(); 5280 } 5281 5282 return Error(L, Message); 5283 } 5284 5285 /// parseDirectiveWarning 5286 /// ::= .warning [string] 5287 bool AsmParser::parseDirectiveWarning(SMLoc L) { 5288 if (!TheCondStack.empty()) { 5289 if (TheCondStack.back().Ignore) { 5290 eatToEndOfStatement(); 5291 return false; 5292 } 5293 } 5294 5295 StringRef Message = ".warning directive invoked in source file"; 5296 5297 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 5298 if (Lexer.isNot(AsmToken::String)) 5299 return TokError(".warning argument must be a string"); 5300 5301 Message = getTok().getStringContents(); 5302 Lex(); 5303 if (parseToken(AsmToken::EndOfStatement, 5304 "expected end of statement in '.warning' directive")) 5305 return true; 5306 } 5307 5308 return Warning(L, Message); 5309 } 5310 5311 /// parseDirectiveEndIf 5312 /// ::= .endif 5313 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 5314 if (parseToken(AsmToken::EndOfStatement, 5315 "unexpected token in '.endif' directive")) 5316 return true; 5317 5318 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 5319 return Error(DirectiveLoc, "Encountered a .endif that doesn't follow " 5320 "an .if or .else"); 5321 if (!TheCondStack.empty()) { 5322 TheCondState = TheCondStack.back(); 5323 TheCondStack.pop_back(); 5324 } 5325 5326 return false; 5327 } 5328 5329 void AsmParser::initializeDirectiveKindMap() { 5330 DirectiveKindMap[".set"] = DK_SET; 5331 DirectiveKindMap[".equ"] = DK_EQU; 5332 DirectiveKindMap[".equiv"] = DK_EQUIV; 5333 DirectiveKindMap[".ascii"] = DK_ASCII; 5334 DirectiveKindMap[".asciz"] = DK_ASCIZ; 5335 DirectiveKindMap[".string"] = DK_STRING; 5336 DirectiveKindMap[".byte"] = DK_BYTE; 5337 DirectiveKindMap[".short"] = DK_SHORT; 5338 DirectiveKindMap[".value"] = DK_VALUE; 5339 DirectiveKindMap[".2byte"] = DK_2BYTE; 5340 DirectiveKindMap[".long"] = DK_LONG; 5341 DirectiveKindMap[".int"] = DK_INT; 5342 DirectiveKindMap[".4byte"] = DK_4BYTE; 5343 DirectiveKindMap[".quad"] = DK_QUAD; 5344 DirectiveKindMap[".8byte"] = DK_8BYTE; 5345 DirectiveKindMap[".octa"] = DK_OCTA; 5346 DirectiveKindMap[".single"] = DK_SINGLE; 5347 DirectiveKindMap[".float"] = DK_FLOAT; 5348 DirectiveKindMap[".double"] = DK_DOUBLE; 5349 DirectiveKindMap[".align"] = DK_ALIGN; 5350 DirectiveKindMap[".align32"] = DK_ALIGN32; 5351 DirectiveKindMap[".balign"] = DK_BALIGN; 5352 DirectiveKindMap[".balignw"] = DK_BALIGNW; 5353 DirectiveKindMap[".balignl"] = DK_BALIGNL; 5354 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 5355 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 5356 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 5357 DirectiveKindMap[".org"] = DK_ORG; 5358 DirectiveKindMap[".fill"] = DK_FILL; 5359 DirectiveKindMap[".zero"] = DK_ZERO; 5360 DirectiveKindMap[".extern"] = DK_EXTERN; 5361 DirectiveKindMap[".globl"] = DK_GLOBL; 5362 DirectiveKindMap[".global"] = DK_GLOBAL; 5363 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 5364 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 5365 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 5366 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 5367 DirectiveKindMap[".reference"] = DK_REFERENCE; 5368 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 5369 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 5370 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 5371 DirectiveKindMap[".cold"] = DK_COLD; 5372 DirectiveKindMap[".comm"] = DK_COMM; 5373 DirectiveKindMap[".common"] = DK_COMMON; 5374 DirectiveKindMap[".lcomm"] = DK_LCOMM; 5375 DirectiveKindMap[".abort"] = DK_ABORT; 5376 DirectiveKindMap[".include"] = DK_INCLUDE; 5377 DirectiveKindMap[".incbin"] = DK_INCBIN; 5378 DirectiveKindMap[".code16"] = DK_CODE16; 5379 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 5380 DirectiveKindMap[".rept"] = DK_REPT; 5381 DirectiveKindMap[".rep"] = DK_REPT; 5382 DirectiveKindMap[".irp"] = DK_IRP; 5383 DirectiveKindMap[".irpc"] = DK_IRPC; 5384 DirectiveKindMap[".endr"] = DK_ENDR; 5385 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 5386 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 5387 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 5388 DirectiveKindMap[".if"] = DK_IF; 5389 DirectiveKindMap[".ifeq"] = DK_IFEQ; 5390 DirectiveKindMap[".ifge"] = DK_IFGE; 5391 DirectiveKindMap[".ifgt"] = DK_IFGT; 5392 DirectiveKindMap[".ifle"] = DK_IFLE; 5393 DirectiveKindMap[".iflt"] = DK_IFLT; 5394 DirectiveKindMap[".ifne"] = DK_IFNE; 5395 DirectiveKindMap[".ifb"] = DK_IFB; 5396 DirectiveKindMap[".ifnb"] = DK_IFNB; 5397 DirectiveKindMap[".ifc"] = DK_IFC; 5398 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 5399 DirectiveKindMap[".ifnc"] = DK_IFNC; 5400 DirectiveKindMap[".ifnes"] = DK_IFNES; 5401 DirectiveKindMap[".ifdef"] = DK_IFDEF; 5402 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 5403 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 5404 DirectiveKindMap[".elseif"] = DK_ELSEIF; 5405 DirectiveKindMap[".else"] = DK_ELSE; 5406 DirectiveKindMap[".end"] = DK_END; 5407 DirectiveKindMap[".endif"] = DK_ENDIF; 5408 DirectiveKindMap[".skip"] = DK_SKIP; 5409 DirectiveKindMap[".space"] = DK_SPACE; 5410 DirectiveKindMap[".file"] = DK_FILE; 5411 DirectiveKindMap[".line"] = DK_LINE; 5412 DirectiveKindMap[".loc"] = DK_LOC; 5413 DirectiveKindMap[".stabs"] = DK_STABS; 5414 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 5415 DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID; 5416 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 5417 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 5418 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 5419 DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID; 5420 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 5421 DirectiveKindMap[".cv_string"] = DK_CV_STRING; 5422 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 5423 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 5424 DirectiveKindMap[".cv_filechecksumoffset"] = DK_CV_FILECHECKSUM_OFFSET; 5425 DirectiveKindMap[".cv_fpo_data"] = DK_CV_FPO_DATA; 5426 DirectiveKindMap[".sleb128"] = DK_SLEB128; 5427 DirectiveKindMap[".uleb128"] = DK_ULEB128; 5428 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 5429 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 5430 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 5431 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 5432 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 5433 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 5434 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 5435 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 5436 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 5437 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 5438 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 5439 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 5440 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 5441 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 5442 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 5443 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 5444 DirectiveKindMap[".cfi_return_column"] = DK_CFI_RETURN_COLUMN; 5445 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 5446 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 5447 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 5448 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 5449 DirectiveKindMap[".cfi_b_key_frame"] = DK_CFI_B_KEY_FRAME; 5450 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 5451 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 5452 DirectiveKindMap[".macro"] = DK_MACRO; 5453 DirectiveKindMap[".exitm"] = DK_EXITM; 5454 DirectiveKindMap[".endm"] = DK_ENDM; 5455 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 5456 DirectiveKindMap[".purgem"] = DK_PURGEM; 5457 DirectiveKindMap[".err"] = DK_ERR; 5458 DirectiveKindMap[".error"] = DK_ERROR; 5459 DirectiveKindMap[".warning"] = DK_WARNING; 5460 DirectiveKindMap[".altmacro"] = DK_ALTMACRO; 5461 DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO; 5462 DirectiveKindMap[".reloc"] = DK_RELOC; 5463 DirectiveKindMap[".dc"] = DK_DC; 5464 DirectiveKindMap[".dc.a"] = DK_DC_A; 5465 DirectiveKindMap[".dc.b"] = DK_DC_B; 5466 DirectiveKindMap[".dc.d"] = DK_DC_D; 5467 DirectiveKindMap[".dc.l"] = DK_DC_L; 5468 DirectiveKindMap[".dc.s"] = DK_DC_S; 5469 DirectiveKindMap[".dc.w"] = DK_DC_W; 5470 DirectiveKindMap[".dc.x"] = DK_DC_X; 5471 DirectiveKindMap[".dcb"] = DK_DCB; 5472 DirectiveKindMap[".dcb.b"] = DK_DCB_B; 5473 DirectiveKindMap[".dcb.d"] = DK_DCB_D; 5474 DirectiveKindMap[".dcb.l"] = DK_DCB_L; 5475 DirectiveKindMap[".dcb.s"] = DK_DCB_S; 5476 DirectiveKindMap[".dcb.w"] = DK_DCB_W; 5477 DirectiveKindMap[".dcb.x"] = DK_DCB_X; 5478 DirectiveKindMap[".ds"] = DK_DS; 5479 DirectiveKindMap[".ds.b"] = DK_DS_B; 5480 DirectiveKindMap[".ds.d"] = DK_DS_D; 5481 DirectiveKindMap[".ds.l"] = DK_DS_L; 5482 DirectiveKindMap[".ds.p"] = DK_DS_P; 5483 DirectiveKindMap[".ds.s"] = DK_DS_S; 5484 DirectiveKindMap[".ds.w"] = DK_DS_W; 5485 DirectiveKindMap[".ds.x"] = DK_DS_X; 5486 DirectiveKindMap[".print"] = DK_PRINT; 5487 DirectiveKindMap[".addrsig"] = DK_ADDRSIG; 5488 DirectiveKindMap[".addrsig_sym"] = DK_ADDRSIG_SYM; 5489 } 5490 5491 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 5492 AsmToken EndToken, StartToken = getTok(); 5493 5494 unsigned NestLevel = 0; 5495 while (true) { 5496 // Check whether we have reached the end of the file. 5497 if (getLexer().is(AsmToken::Eof)) { 5498 printError(DirectiveLoc, "no matching '.endr' in definition"); 5499 return nullptr; 5500 } 5501 5502 if (Lexer.is(AsmToken::Identifier) && 5503 (getTok().getIdentifier() == ".rep" || 5504 getTok().getIdentifier() == ".rept" || 5505 getTok().getIdentifier() == ".irp" || 5506 getTok().getIdentifier() == ".irpc")) { 5507 ++NestLevel; 5508 } 5509 5510 // Otherwise, check whether we have reached the .endr. 5511 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 5512 if (NestLevel == 0) { 5513 EndToken = getTok(); 5514 Lex(); 5515 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5516 printError(getTok().getLoc(), 5517 "unexpected token in '.endr' directive"); 5518 return nullptr; 5519 } 5520 break; 5521 } 5522 --NestLevel; 5523 } 5524 5525 // Otherwise, scan till the end of the statement. 5526 eatToEndOfStatement(); 5527 } 5528 5529 const char *BodyStart = StartToken.getLoc().getPointer(); 5530 const char *BodyEnd = EndToken.getLoc().getPointer(); 5531 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 5532 5533 // We Are Anonymous. 5534 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 5535 return &MacroLikeBodies.back(); 5536 } 5537 5538 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 5539 raw_svector_ostream &OS) { 5540 OS << ".endr\n"; 5541 5542 std::unique_ptr<MemoryBuffer> Instantiation = 5543 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 5544 5545 // Create the macro instantiation object and add to the current macro 5546 // instantiation stack. 5547 MacroInstantiation *MI = new MacroInstantiation( 5548 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 5549 ActiveMacros.push_back(MI); 5550 5551 // Jump to the macro instantiation and prime the lexer. 5552 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 5553 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 5554 Lex(); 5555 } 5556 5557 /// parseDirectiveRept 5558 /// ::= .rep | .rept count 5559 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 5560 const MCExpr *CountExpr; 5561 SMLoc CountLoc = getTok().getLoc(); 5562 if (parseExpression(CountExpr)) 5563 return true; 5564 5565 int64_t Count; 5566 if (!CountExpr->evaluateAsAbsolute(Count, getStreamer().getAssemblerPtr())) { 5567 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 5568 } 5569 5570 if (check(Count < 0, CountLoc, "Count is negative") || 5571 parseToken(AsmToken::EndOfStatement, 5572 "unexpected token in '" + Dir + "' directive")) 5573 return true; 5574 5575 // Lex the rept definition. 5576 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5577 if (!M) 5578 return true; 5579 5580 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5581 // to hold the macro body with substitutions. 5582 SmallString<256> Buf; 5583 raw_svector_ostream OS(Buf); 5584 while (Count--) { 5585 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 5586 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 5587 return true; 5588 } 5589 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5590 5591 return false; 5592 } 5593 5594 /// parseDirectiveIrp 5595 /// ::= .irp symbol,values 5596 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 5597 MCAsmMacroParameter Parameter; 5598 MCAsmMacroArguments A; 5599 if (check(parseIdentifier(Parameter.Name), 5600 "expected identifier in '.irp' directive") || 5601 parseToken(AsmToken::Comma, "expected comma in '.irp' directive") || 5602 parseMacroArguments(nullptr, A) || 5603 parseToken(AsmToken::EndOfStatement, "expected End of Statement")) 5604 return true; 5605 5606 // Lex the irp definition. 5607 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5608 if (!M) 5609 return true; 5610 5611 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5612 // to hold the macro body with substitutions. 5613 SmallString<256> Buf; 5614 raw_svector_ostream OS(Buf); 5615 5616 for (const MCAsmMacroArgument &Arg : A) { 5617 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 5618 // This is undocumented, but GAS seems to support it. 5619 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5620 return true; 5621 } 5622 5623 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5624 5625 return false; 5626 } 5627 5628 /// parseDirectiveIrpc 5629 /// ::= .irpc symbol,values 5630 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 5631 MCAsmMacroParameter Parameter; 5632 MCAsmMacroArguments A; 5633 5634 if (check(parseIdentifier(Parameter.Name), 5635 "expected identifier in '.irpc' directive") || 5636 parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") || 5637 parseMacroArguments(nullptr, A)) 5638 return true; 5639 5640 if (A.size() != 1 || A.front().size() != 1) 5641 return TokError("unexpected token in '.irpc' directive"); 5642 5643 // Eat the end of statement. 5644 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5645 return true; 5646 5647 // Lex the irpc definition. 5648 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5649 if (!M) 5650 return true; 5651 5652 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5653 // to hold the macro body with substitutions. 5654 SmallString<256> Buf; 5655 raw_svector_ostream OS(Buf); 5656 5657 StringRef Values = A.front().front().getString(); 5658 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 5659 MCAsmMacroArgument Arg; 5660 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 5661 5662 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 5663 // This is undocumented, but GAS seems to support it. 5664 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5665 return true; 5666 } 5667 5668 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5669 5670 return false; 5671 } 5672 5673 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 5674 if (ActiveMacros.empty()) 5675 return TokError("unmatched '.endr' directive"); 5676 5677 // The only .repl that should get here are the ones created by 5678 // instantiateMacroLikeBody. 5679 assert(getLexer().is(AsmToken::EndOfStatement)); 5680 5681 handleMacroExit(); 5682 return false; 5683 } 5684 5685 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 5686 size_t Len) { 5687 const MCExpr *Value; 5688 SMLoc ExprLoc = getLexer().getLoc(); 5689 if (parseExpression(Value)) 5690 return true; 5691 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5692 if (!MCE) 5693 return Error(ExprLoc, "unexpected expression in _emit"); 5694 uint64_t IntValue = MCE->getValue(); 5695 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 5696 return Error(ExprLoc, "literal value out of range for directive"); 5697 5698 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 5699 return false; 5700 } 5701 5702 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 5703 const MCExpr *Value; 5704 SMLoc ExprLoc = getLexer().getLoc(); 5705 if (parseExpression(Value)) 5706 return true; 5707 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5708 if (!MCE) 5709 return Error(ExprLoc, "unexpected expression in align"); 5710 uint64_t IntValue = MCE->getValue(); 5711 if (!isPowerOf2_64(IntValue)) 5712 return Error(ExprLoc, "literal value not a power of two greater then zero"); 5713 5714 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 5715 return false; 5716 } 5717 5718 bool AsmParser::parseDirectivePrint(SMLoc DirectiveLoc) { 5719 const AsmToken StrTok = getTok(); 5720 Lex(); 5721 if (StrTok.isNot(AsmToken::String) || StrTok.getString().front() != '"') 5722 return Error(DirectiveLoc, "expected double quoted string after .print"); 5723 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5724 return true; 5725 llvm::outs() << StrTok.getStringContents() << '\n'; 5726 return false; 5727 } 5728 5729 bool AsmParser::parseDirectiveAddrsig() { 5730 getStreamer().EmitAddrsig(); 5731 return false; 5732 } 5733 5734 bool AsmParser::parseDirectiveAddrsigSym() { 5735 StringRef Name; 5736 if (check(parseIdentifier(Name), 5737 "expected identifier in '.addrsig_sym' directive")) 5738 return true; 5739 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 5740 getStreamer().EmitAddrsigSym(Sym); 5741 return false; 5742 } 5743 5744 // We are comparing pointers, but the pointers are relative to a single string. 5745 // Thus, this should always be deterministic. 5746 static int rewritesSort(const AsmRewrite *AsmRewriteA, 5747 const AsmRewrite *AsmRewriteB) { 5748 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 5749 return -1; 5750 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 5751 return 1; 5752 5753 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 5754 // rewrite to the same location. Make sure the SizeDirective rewrite is 5755 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 5756 // ensures the sort algorithm is stable. 5757 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 5758 AsmRewritePrecedence[AsmRewriteB->Kind]) 5759 return -1; 5760 5761 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 5762 AsmRewritePrecedence[AsmRewriteB->Kind]) 5763 return 1; 5764 llvm_unreachable("Unstable rewrite sort."); 5765 } 5766 5767 bool AsmParser::parseMSInlineAsm( 5768 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 5769 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls, 5770 SmallVectorImpl<std::string> &Constraints, 5771 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 5772 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 5773 SmallVector<void *, 4> InputDecls; 5774 SmallVector<void *, 4> OutputDecls; 5775 SmallVector<bool, 4> InputDeclsAddressOf; 5776 SmallVector<bool, 4> OutputDeclsAddressOf; 5777 SmallVector<std::string, 4> InputConstraints; 5778 SmallVector<std::string, 4> OutputConstraints; 5779 SmallVector<unsigned, 4> ClobberRegs; 5780 5781 SmallVector<AsmRewrite, 4> AsmStrRewrites; 5782 5783 // Prime the lexer. 5784 Lex(); 5785 5786 // While we have input, parse each statement. 5787 unsigned InputIdx = 0; 5788 unsigned OutputIdx = 0; 5789 while (getLexer().isNot(AsmToken::Eof)) { 5790 // Parse curly braces marking block start/end 5791 if (parseCurlyBlockScope(AsmStrRewrites)) 5792 continue; 5793 5794 ParseStatementInfo Info(&AsmStrRewrites); 5795 bool StatementErr = parseStatement(Info, &SI); 5796 5797 if (StatementErr || Info.ParseError) { 5798 // Emit pending errors if any exist. 5799 printPendingErrors(); 5800 return true; 5801 } 5802 5803 // No pending error should exist here. 5804 assert(!hasPendingError() && "unexpected error from parseStatement"); 5805 5806 if (Info.Opcode == ~0U) 5807 continue; 5808 5809 const MCInstrDesc &Desc = MII->get(Info.Opcode); 5810 5811 // Build the list of clobbers, outputs and inputs. 5812 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 5813 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 5814 5815 // Immediate. 5816 if (Operand.isImm()) 5817 continue; 5818 5819 // Register operand. 5820 if (Operand.isReg() && !Operand.needAddressOf() && 5821 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5822 unsigned NumDefs = Desc.getNumDefs(); 5823 // Clobber. 5824 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5825 ClobberRegs.push_back(Operand.getReg()); 5826 continue; 5827 } 5828 5829 // Expr/Input or Output. 5830 StringRef SymName = Operand.getSymName(); 5831 if (SymName.empty()) 5832 continue; 5833 5834 void *OpDecl = Operand.getOpDecl(); 5835 if (!OpDecl) 5836 continue; 5837 5838 bool isOutput = (i == 1) && Desc.mayStore(); 5839 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5840 if (isOutput) { 5841 ++InputIdx; 5842 OutputDecls.push_back(OpDecl); 5843 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5844 OutputConstraints.push_back(("=" + Operand.getConstraint()).str()); 5845 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5846 } else { 5847 InputDecls.push_back(OpDecl); 5848 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5849 InputConstraints.push_back(Operand.getConstraint().str()); 5850 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5851 } 5852 } 5853 5854 // Consider implicit defs to be clobbers. Think of cpuid and push. 5855 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5856 Desc.getNumImplicitDefs()); 5857 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 5858 } 5859 5860 // Set the number of Outputs and Inputs. 5861 NumOutputs = OutputDecls.size(); 5862 NumInputs = InputDecls.size(); 5863 5864 // Set the unique clobbers. 5865 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5866 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5867 ClobberRegs.end()); 5868 Clobbers.assign(ClobberRegs.size(), std::string()); 5869 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5870 raw_string_ostream OS(Clobbers[I]); 5871 IP->printRegName(OS, ClobberRegs[I]); 5872 } 5873 5874 // Merge the various outputs and inputs. Output are expected first. 5875 if (NumOutputs || NumInputs) { 5876 unsigned NumExprs = NumOutputs + NumInputs; 5877 OpDecls.resize(NumExprs); 5878 Constraints.resize(NumExprs); 5879 for (unsigned i = 0; i < NumOutputs; ++i) { 5880 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5881 Constraints[i] = OutputConstraints[i]; 5882 } 5883 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5884 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5885 Constraints[j] = InputConstraints[i]; 5886 } 5887 } 5888 5889 // Build the IR assembly string. 5890 std::string AsmStringIR; 5891 raw_string_ostream OS(AsmStringIR); 5892 StringRef ASMString = 5893 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5894 const char *AsmStart = ASMString.begin(); 5895 const char *AsmEnd = ASMString.end(); 5896 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 5897 for (const AsmRewrite &AR : AsmStrRewrites) { 5898 AsmRewriteKind Kind = AR.Kind; 5899 5900 const char *Loc = AR.Loc.getPointer(); 5901 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 5902 5903 // Emit everything up to the immediate/expression. 5904 if (unsigned Len = Loc - AsmStart) 5905 OS << StringRef(AsmStart, Len); 5906 5907 // Skip the original expression. 5908 if (Kind == AOK_Skip) { 5909 AsmStart = Loc + AR.Len; 5910 continue; 5911 } 5912 5913 unsigned AdditionalSkip = 0; 5914 // Rewrite expressions in $N notation. 5915 switch (Kind) { 5916 default: 5917 break; 5918 case AOK_IntelExpr: 5919 assert(AR.IntelExp.isValid() && "cannot write invalid intel expression"); 5920 if (AR.IntelExp.NeedBracs) 5921 OS << "["; 5922 if (AR.IntelExp.hasBaseReg()) 5923 OS << AR.IntelExp.BaseReg; 5924 if (AR.IntelExp.hasIndexReg()) 5925 OS << (AR.IntelExp.hasBaseReg() ? " + " : "") 5926 << AR.IntelExp.IndexReg; 5927 if (AR.IntelExp.Scale > 1) 5928 OS << " * $$" << AR.IntelExp.Scale; 5929 if (AR.IntelExp.Imm || !AR.IntelExp.hasRegs()) 5930 OS << (AR.IntelExp.hasRegs() ? " + $$" : "$$") << AR.IntelExp.Imm; 5931 if (AR.IntelExp.NeedBracs) 5932 OS << "]"; 5933 break; 5934 case AOK_Label: 5935 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 5936 break; 5937 case AOK_Input: 5938 OS << '$' << InputIdx++; 5939 break; 5940 case AOK_Output: 5941 OS << '$' << OutputIdx++; 5942 break; 5943 case AOK_SizeDirective: 5944 switch (AR.Val) { 5945 default: break; 5946 case 8: OS << "byte ptr "; break; 5947 case 16: OS << "word ptr "; break; 5948 case 32: OS << "dword ptr "; break; 5949 case 64: OS << "qword ptr "; break; 5950 case 80: OS << "xword ptr "; break; 5951 case 128: OS << "xmmword ptr "; break; 5952 case 256: OS << "ymmword ptr "; break; 5953 } 5954 break; 5955 case AOK_Emit: 5956 OS << ".byte"; 5957 break; 5958 case AOK_Align: { 5959 // MS alignment directives are measured in bytes. If the native assembler 5960 // measures alignment in bytes, we can pass it straight through. 5961 OS << ".align"; 5962 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 5963 break; 5964 5965 // Alignment is in log2 form, so print that instead and skip the original 5966 // immediate. 5967 unsigned Val = AR.Val; 5968 OS << ' ' << Val; 5969 assert(Val < 10 && "Expected alignment less then 2^10."); 5970 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 5971 break; 5972 } 5973 case AOK_EVEN: 5974 OS << ".even"; 5975 break; 5976 case AOK_EndOfStatement: 5977 OS << "\n\t"; 5978 break; 5979 } 5980 5981 // Skip the original expression. 5982 AsmStart = Loc + AR.Len + AdditionalSkip; 5983 } 5984 5985 // Emit the remainder of the asm string. 5986 if (AsmStart != AsmEnd) 5987 OS << StringRef(AsmStart, AsmEnd - AsmStart); 5988 5989 AsmString = OS.str(); 5990 return false; 5991 } 5992 5993 namespace llvm { 5994 namespace MCParserUtils { 5995 5996 /// Returns whether the given symbol is used anywhere in the given expression, 5997 /// or subexpressions. 5998 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 5999 switch (Value->getKind()) { 6000 case MCExpr::Binary: { 6001 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 6002 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 6003 isSymbolUsedInExpression(Sym, BE->getRHS()); 6004 } 6005 case MCExpr::Target: 6006 case MCExpr::Constant: 6007 return false; 6008 case MCExpr::SymbolRef: { 6009 const MCSymbol &S = 6010 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 6011 if (S.isVariable()) 6012 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 6013 return &S == Sym; 6014 } 6015 case MCExpr::Unary: 6016 return isSymbolUsedInExpression( 6017 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 6018 } 6019 6020 llvm_unreachable("Unknown expr kind!"); 6021 } 6022 6023 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 6024 MCAsmParser &Parser, MCSymbol *&Sym, 6025 const MCExpr *&Value) { 6026 6027 // FIXME: Use better location, we should use proper tokens. 6028 SMLoc EqualLoc = Parser.getTok().getLoc(); 6029 if (Parser.parseExpression(Value)) 6030 return Parser.TokError("missing expression"); 6031 6032 // Note: we don't count b as used in "a = b". This is to allow 6033 // a = b 6034 // b = c 6035 6036 if (Parser.parseToken(AsmToken::EndOfStatement)) 6037 return true; 6038 6039 // Validate that the LHS is allowed to be a variable (either it has not been 6040 // used as a symbol, or it is an absolute symbol). 6041 Sym = Parser.getContext().lookupSymbol(Name); 6042 if (Sym) { 6043 // Diagnose assignment to a label. 6044 // 6045 // FIXME: Diagnostics. Note the location of the definition as a label. 6046 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 6047 if (isSymbolUsedInExpression(Sym, Value)) 6048 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 6049 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 6050 !Sym->isVariable()) 6051 ; // Allow redefinitions of undefined symbols only used in directives. 6052 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 6053 ; // Allow redefinitions of variables that haven't yet been used. 6054 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 6055 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 6056 else if (!Sym->isVariable()) 6057 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 6058 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 6059 return Parser.Error(EqualLoc, 6060 "invalid reassignment of non-absolute variable '" + 6061 Name + "'"); 6062 } else if (Name == ".") { 6063 Parser.getStreamer().emitValueToOffset(Value, 0, EqualLoc); 6064 return false; 6065 } else 6066 Sym = Parser.getContext().getOrCreateSymbol(Name); 6067 6068 Sym->setRedefinable(allow_redef); 6069 6070 return false; 6071 } 6072 6073 } // end namespace MCParserUtils 6074 } // end namespace llvm 6075 6076 /// Create an MCAsmParser instance. 6077 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 6078 MCStreamer &Out, const MCAsmInfo &MAI, 6079 unsigned CB) { 6080 return new AsmParser(SM, C, Out, MAI, CB); 6081 } 6082