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 a parser for assembly files similar to gas syntax. 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/SmallSet.h" 19 #include "llvm/ADT/SmallString.h" 20 #include "llvm/ADT/SmallVector.h" 21 #include "llvm/ADT/StringExtras.h" 22 #include "llvm/ADT/StringMap.h" 23 #include "llvm/ADT/StringRef.h" 24 #include "llvm/ADT/Twine.h" 25 #include "llvm/BinaryFormat/Dwarf.h" 26 #include "llvm/DebugInfo/CodeView/SymbolRecord.h" 27 #include "llvm/MC/MCAsmInfo.h" 28 #include "llvm/MC/MCCodeView.h" 29 #include "llvm/MC/MCContext.h" 30 #include "llvm/MC/MCDirectives.h" 31 #include "llvm/MC/MCDwarf.h" 32 #include "llvm/MC/MCExpr.h" 33 #include "llvm/MC/MCInstPrinter.h" 34 #include "llvm/MC/MCInstrDesc.h" 35 #include "llvm/MC/MCInstrInfo.h" 36 #include "llvm/MC/MCObjectFileInfo.h" 37 #include "llvm/MC/MCParser/AsmCond.h" 38 #include "llvm/MC/MCParser/AsmLexer.h" 39 #include "llvm/MC/MCParser/MCAsmLexer.h" 40 #include "llvm/MC/MCParser/MCAsmParser.h" 41 #include "llvm/MC/MCParser/MCAsmParserExtension.h" 42 #include "llvm/MC/MCParser/MCAsmParserUtils.h" 43 #include "llvm/MC/MCParser/MCParsedAsmOperand.h" 44 #include "llvm/MC/MCParser/MCTargetAsmParser.h" 45 #include "llvm/MC/MCRegisterInfo.h" 46 #include "llvm/MC/MCSection.h" 47 #include "llvm/MC/MCStreamer.h" 48 #include "llvm/MC/MCSymbol.h" 49 #include "llvm/MC/MCTargetOptions.h" 50 #include "llvm/MC/MCValue.h" 51 #include "llvm/Support/Casting.h" 52 #include "llvm/Support/CommandLine.h" 53 #include "llvm/Support/ErrorHandling.h" 54 #include "llvm/Support/MD5.h" 55 #include "llvm/Support/MathExtras.h" 56 #include "llvm/Support/MemoryBuffer.h" 57 #include "llvm/Support/SMLoc.h" 58 #include "llvm/Support/SourceMgr.h" 59 #include "llvm/Support/raw_ostream.h" 60 #include <algorithm> 61 #include <cassert> 62 #include <cctype> 63 #include <climits> 64 #include <cstddef> 65 #include <cstdint> 66 #include <deque> 67 #include <memory> 68 #include <sstream> 69 #include <string> 70 #include <tuple> 71 #include <utility> 72 #include <vector> 73 74 using namespace llvm; 75 76 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() = default; 77 78 extern cl::opt<unsigned> AsmMacroMaxNestingDepth; 79 80 namespace { 81 82 /// Helper types for tracking macro definitions. 83 typedef std::vector<AsmToken> MCAsmMacroArgument; 84 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 85 86 /// Helper class for storing information about an active macro 87 /// instantiation. 88 struct MacroInstantiation { 89 /// The location of the instantiation. 90 SMLoc InstantiationLoc; 91 92 /// The buffer where parsing should resume upon instantiation completion. 93 unsigned ExitBuffer; 94 95 /// The location where parsing should resume upon instantiation completion. 96 SMLoc ExitLoc; 97 98 /// The depth of TheCondStack at the start of the instantiation. 99 size_t CondStackDepth; 100 }; 101 102 struct ParseStatementInfo { 103 /// The parsed operands from the last parsed statement. 104 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 105 106 /// The opcode from the last parsed instruction. 107 unsigned Opcode = ~0U; 108 109 /// Was there an error parsing the inline assembly? 110 bool ParseError = false; 111 112 SmallVectorImpl<AsmRewrite> *AsmRewrites = nullptr; 113 114 ParseStatementInfo() = delete; 115 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 116 : AsmRewrites(rewrites) {} 117 }; 118 119 /// The concrete assembly parser instance. 120 class AsmParser : public MCAsmParser { 121 private: 122 AsmLexer Lexer; 123 MCContext &Ctx; 124 MCStreamer &Out; 125 const MCAsmInfo &MAI; 126 SourceMgr &SrcMgr; 127 SourceMgr::DiagHandlerTy SavedDiagHandler; 128 void *SavedDiagContext; 129 std::unique_ptr<MCAsmParserExtension> PlatformParser; 130 SMLoc StartTokLoc; 131 132 /// This is the current buffer index we're lexing from as managed by the 133 /// SourceMgr object. 134 unsigned CurBuffer; 135 136 AsmCond TheCondState; 137 std::vector<AsmCond> TheCondStack; 138 139 /// maps directive names to handler methods in parser 140 /// extensions. Extensions register themselves in this map by calling 141 /// addDirectiveHandler. 142 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 143 144 /// Stack of active macro instantiations. 145 std::vector<MacroInstantiation*> ActiveMacros; 146 147 /// List of bodies of anonymous macros. 148 std::deque<MCAsmMacro> MacroLikeBodies; 149 150 /// Boolean tracking whether macro substitution is enabled. 151 unsigned MacrosEnabledFlag : 1; 152 153 /// Keeps track of how many .macro's have been instantiated. 154 unsigned NumOfMacroInstantiations; 155 156 /// The values from the last parsed cpp hash file line comment if any. 157 struct CppHashInfoTy { 158 StringRef Filename; 159 int64_t LineNumber; 160 SMLoc Loc; 161 unsigned Buf; 162 CppHashInfoTy() : Filename(), LineNumber(0), Loc(), Buf(0) {} 163 }; 164 CppHashInfoTy CppHashInfo; 165 166 /// The filename from the first cpp hash file line comment, if any. 167 StringRef FirstCppHashFilename; 168 169 /// List of forward directional labels for diagnosis at the end. 170 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 171 172 SmallSet<StringRef, 2> LTODiscardSymbols; 173 174 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 175 unsigned AssemblerDialect = ~0U; 176 177 /// is Darwin compatibility enabled? 178 bool IsDarwin = false; 179 180 /// Are we parsing ms-style inline assembly? 181 bool ParsingMSInlineAsm = false; 182 183 /// Did we already inform the user about inconsistent MD5 usage? 184 bool ReportedInconsistentMD5 = false; 185 186 // Is alt macro mode enabled. 187 bool AltMacroMode = false; 188 189 public: 190 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 191 const MCAsmInfo &MAI, unsigned CB); 192 AsmParser(const AsmParser &) = delete; 193 AsmParser &operator=(const AsmParser &) = delete; 194 ~AsmParser() override; 195 196 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 197 198 void addDirectiveHandler(StringRef Directive, 199 ExtensionDirectiveHandler Handler) override { 200 ExtensionDirectiveMap[Directive] = Handler; 201 } 202 203 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 204 DirectiveKindMap[Directive.lower()] = DirectiveKindMap[Alias.lower()]; 205 } 206 207 /// @name MCAsmParser Interface 208 /// { 209 210 SourceMgr &getSourceManager() override { return SrcMgr; } 211 MCAsmLexer &getLexer() override { return Lexer; } 212 MCContext &getContext() override { return Ctx; } 213 MCStreamer &getStreamer() override { return Out; } 214 215 CodeViewContext &getCVContext() { return Ctx.getCVContext(); } 216 217 unsigned getAssemblerDialect() override { 218 if (AssemblerDialect == ~0U) 219 return MAI.getAssemblerDialect(); 220 else 221 return AssemblerDialect; 222 } 223 void setAssemblerDialect(unsigned i) override { 224 AssemblerDialect = i; 225 } 226 227 void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override; 228 bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override; 229 bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override; 230 231 const AsmToken &Lex() override; 232 233 void setParsingMSInlineAsm(bool V) override { 234 ParsingMSInlineAsm = V; 235 // When parsing MS inline asm, we must lex 0b1101 and 0ABCH as binary and 236 // hex integer literals. 237 Lexer.setLexMasmIntegers(V); 238 } 239 bool isParsingMSInlineAsm() override { return ParsingMSInlineAsm; } 240 241 bool discardLTOSymbol(StringRef Name) const override { 242 return LTODiscardSymbols.contains(Name); 243 } 244 245 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 246 unsigned &NumOutputs, unsigned &NumInputs, 247 SmallVectorImpl<std::pair<void *,bool>> &OpDecls, 248 SmallVectorImpl<std::string> &Constraints, 249 SmallVectorImpl<std::string> &Clobbers, 250 const MCInstrInfo *MII, const MCInstPrinter *IP, 251 MCAsmParserSemaCallback &SI) override; 252 253 bool parseExpression(const MCExpr *&Res); 254 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 255 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc, 256 AsmTypeInfo *TypeInfo) override; 257 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 258 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 259 SMLoc &EndLoc) override; 260 bool parseAbsoluteExpression(int64_t &Res) override; 261 262 /// Parse a floating point expression using the float \p Semantics 263 /// and set \p Res to the value. 264 bool parseRealValue(const fltSemantics &Semantics, APInt &Res); 265 266 /// Parse an identifier or string (as a quoted identifier) 267 /// and set \p Res to the identifier contents. 268 bool parseIdentifier(StringRef &Res) override; 269 void eatToEndOfStatement() override; 270 271 bool checkForValidSection() override; 272 273 /// } 274 275 private: 276 bool parseStatement(ParseStatementInfo &Info, 277 MCAsmParserSemaCallback *SI); 278 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 279 bool parseCppHashLineFilenameComment(SMLoc L, bool SaveLocInfo = true); 280 281 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 282 ArrayRef<MCAsmMacroParameter> Parameters); 283 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 284 ArrayRef<MCAsmMacroParameter> Parameters, 285 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 286 SMLoc L); 287 288 /// Are macros enabled in the parser? 289 bool areMacrosEnabled() {return MacrosEnabledFlag;} 290 291 /// Control a flag in the parser that enables or disables macros. 292 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 293 294 /// Are we inside a macro instantiation? 295 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 296 297 /// Handle entry to macro instantiation. 298 /// 299 /// \param M The macro. 300 /// \param NameLoc Instantiation location. 301 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 302 303 /// Handle exit from macro instantiation. 304 void handleMacroExit(); 305 306 /// Extract AsmTokens for a macro argument. 307 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 308 309 /// Parse all macro arguments for a given macro. 310 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 311 312 void printMacroInstantiations(); 313 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 314 SMRange Range = None) const { 315 ArrayRef<SMRange> Ranges(Range); 316 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 317 } 318 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 319 320 /// Should we emit DWARF describing this assembler source? (Returns false if 321 /// the source has .file directives, which means we don't want to generate 322 /// info describing the assembler source itself.) 323 bool enabledGenDwarfForAssembly(); 324 325 /// Enter the specified file. This returns true on failure. 326 bool enterIncludeFile(const std::string &Filename); 327 328 /// Process the specified file for the .incbin directive. 329 /// This returns true on failure. 330 bool processIncbinFile(const std::string &Filename, int64_t Skip = 0, 331 const MCExpr *Count = nullptr, SMLoc Loc = SMLoc()); 332 333 /// Reset the current lexer position to that given by \p Loc. The 334 /// current token is not set; clients should ensure Lex() is called 335 /// subsequently. 336 /// 337 /// \param InBuffer If not 0, should be the known buffer id that contains the 338 /// location. 339 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 340 341 /// Parse up to the end of statement and a return the contents from the 342 /// current token until the end of the statement; the current token on exit 343 /// will be either the EndOfStatement or EOF. 344 StringRef parseStringToEndOfStatement() override; 345 346 /// Parse until the end of a statement or a comma is encountered, 347 /// return the contents from the current token up to the end or comma. 348 StringRef parseStringToComma(); 349 350 bool parseAssignment(StringRef Name, bool allow_redef, 351 bool NoDeadStrip = false); 352 353 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 354 MCBinaryExpr::Opcode &Kind); 355 356 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 357 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 358 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 359 360 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 361 362 bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName); 363 bool parseCVFileId(int64_t &FileId, StringRef DirectiveName); 364 365 // Generic (target and platform independent) directive parsing. 366 enum DirectiveKind { 367 DK_NO_DIRECTIVE, // Placeholder 368 DK_SET, 369 DK_EQU, 370 DK_EQUIV, 371 DK_ASCII, 372 DK_ASCIZ, 373 DK_STRING, 374 DK_BYTE, 375 DK_SHORT, 376 DK_RELOC, 377 DK_VALUE, 378 DK_2BYTE, 379 DK_LONG, 380 DK_INT, 381 DK_4BYTE, 382 DK_QUAD, 383 DK_8BYTE, 384 DK_OCTA, 385 DK_DC, 386 DK_DC_A, 387 DK_DC_B, 388 DK_DC_D, 389 DK_DC_L, 390 DK_DC_S, 391 DK_DC_W, 392 DK_DC_X, 393 DK_DCB, 394 DK_DCB_B, 395 DK_DCB_D, 396 DK_DCB_L, 397 DK_DCB_S, 398 DK_DCB_W, 399 DK_DCB_X, 400 DK_DS, 401 DK_DS_B, 402 DK_DS_D, 403 DK_DS_L, 404 DK_DS_P, 405 DK_DS_S, 406 DK_DS_W, 407 DK_DS_X, 408 DK_SINGLE, 409 DK_FLOAT, 410 DK_DOUBLE, 411 DK_ALIGN, 412 DK_ALIGN32, 413 DK_BALIGN, 414 DK_BALIGNW, 415 DK_BALIGNL, 416 DK_P2ALIGN, 417 DK_P2ALIGNW, 418 DK_P2ALIGNL, 419 DK_ORG, 420 DK_FILL, 421 DK_ENDR, 422 DK_BUNDLE_ALIGN_MODE, 423 DK_BUNDLE_LOCK, 424 DK_BUNDLE_UNLOCK, 425 DK_ZERO, 426 DK_EXTERN, 427 DK_GLOBL, 428 DK_GLOBAL, 429 DK_LAZY_REFERENCE, 430 DK_NO_DEAD_STRIP, 431 DK_SYMBOL_RESOLVER, 432 DK_PRIVATE_EXTERN, 433 DK_REFERENCE, 434 DK_WEAK_DEFINITION, 435 DK_WEAK_REFERENCE, 436 DK_WEAK_DEF_CAN_BE_HIDDEN, 437 DK_COLD, 438 DK_COMM, 439 DK_COMMON, 440 DK_LCOMM, 441 DK_ABORT, 442 DK_INCLUDE, 443 DK_INCBIN, 444 DK_CODE16, 445 DK_CODE16GCC, 446 DK_REPT, 447 DK_IRP, 448 DK_IRPC, 449 DK_IF, 450 DK_IFEQ, 451 DK_IFGE, 452 DK_IFGT, 453 DK_IFLE, 454 DK_IFLT, 455 DK_IFNE, 456 DK_IFB, 457 DK_IFNB, 458 DK_IFC, 459 DK_IFEQS, 460 DK_IFNC, 461 DK_IFNES, 462 DK_IFDEF, 463 DK_IFNDEF, 464 DK_IFNOTDEF, 465 DK_ELSEIF, 466 DK_ELSE, 467 DK_ENDIF, 468 DK_SPACE, 469 DK_SKIP, 470 DK_FILE, 471 DK_LINE, 472 DK_LOC, 473 DK_STABS, 474 DK_CV_FILE, 475 DK_CV_FUNC_ID, 476 DK_CV_INLINE_SITE_ID, 477 DK_CV_LOC, 478 DK_CV_LINETABLE, 479 DK_CV_INLINE_LINETABLE, 480 DK_CV_DEF_RANGE, 481 DK_CV_STRINGTABLE, 482 DK_CV_STRING, 483 DK_CV_FILECHECKSUMS, 484 DK_CV_FILECHECKSUM_OFFSET, 485 DK_CV_FPO_DATA, 486 DK_CFI_SECTIONS, 487 DK_CFI_STARTPROC, 488 DK_CFI_ENDPROC, 489 DK_CFI_DEF_CFA, 490 DK_CFI_DEF_CFA_OFFSET, 491 DK_CFI_ADJUST_CFA_OFFSET, 492 DK_CFI_DEF_CFA_REGISTER, 493 DK_CFI_OFFSET, 494 DK_CFI_REL_OFFSET, 495 DK_CFI_PERSONALITY, 496 DK_CFI_LSDA, 497 DK_CFI_REMEMBER_STATE, 498 DK_CFI_RESTORE_STATE, 499 DK_CFI_SAME_VALUE, 500 DK_CFI_RESTORE, 501 DK_CFI_ESCAPE, 502 DK_CFI_RETURN_COLUMN, 503 DK_CFI_SIGNAL_FRAME, 504 DK_CFI_UNDEFINED, 505 DK_CFI_REGISTER, 506 DK_CFI_WINDOW_SAVE, 507 DK_CFI_B_KEY_FRAME, 508 DK_MACROS_ON, 509 DK_MACROS_OFF, 510 DK_ALTMACRO, 511 DK_NOALTMACRO, 512 DK_MACRO, 513 DK_EXITM, 514 DK_ENDM, 515 DK_ENDMACRO, 516 DK_PURGEM, 517 DK_SLEB128, 518 DK_ULEB128, 519 DK_ERR, 520 DK_ERROR, 521 DK_WARNING, 522 DK_PRINT, 523 DK_ADDRSIG, 524 DK_ADDRSIG_SYM, 525 DK_PSEUDO_PROBE, 526 DK_LTO_DISCARD, 527 DK_END 528 }; 529 530 /// Maps directive name --> DirectiveKind enum, for 531 /// directives parsed by this class. 532 StringMap<DirectiveKind> DirectiveKindMap; 533 534 // Codeview def_range type parsing. 535 enum CVDefRangeType { 536 CVDR_DEFRANGE = 0, // Placeholder 537 CVDR_DEFRANGE_REGISTER, 538 CVDR_DEFRANGE_FRAMEPOINTER_REL, 539 CVDR_DEFRANGE_SUBFIELD_REGISTER, 540 CVDR_DEFRANGE_REGISTER_REL 541 }; 542 543 /// Maps Codeview def_range types --> CVDefRangeType enum, for 544 /// Codeview def_range types parsed by this class. 545 StringMap<CVDefRangeType> CVDefRangeTypeMap; 546 547 // ".ascii", ".asciz", ".string" 548 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 549 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 550 bool parseDirectiveValue(StringRef IDVal, 551 unsigned Size); // ".byte", ".long", ... 552 bool parseDirectiveOctaValue(StringRef IDVal); // ".octa", ... 553 bool parseDirectiveRealValue(StringRef IDVal, 554 const fltSemantics &); // ".single", ... 555 bool parseDirectiveFill(); // ".fill" 556 bool parseDirectiveZero(); // ".zero" 557 // ".set", ".equ", ".equiv" 558 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 559 bool parseDirectiveOrg(); // ".org" 560 // ".align{,32}", ".p2align{,w,l}" 561 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 562 563 // ".file", ".line", ".loc", ".stabs" 564 bool parseDirectiveFile(SMLoc DirectiveLoc); 565 bool parseDirectiveLine(); 566 bool parseDirectiveLoc(); 567 bool parseDirectiveStabs(); 568 569 // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable", 570 // ".cv_inline_linetable", ".cv_def_range", ".cv_string" 571 bool parseDirectiveCVFile(); 572 bool parseDirectiveCVFuncId(); 573 bool parseDirectiveCVInlineSiteId(); 574 bool parseDirectiveCVLoc(); 575 bool parseDirectiveCVLinetable(); 576 bool parseDirectiveCVInlineLinetable(); 577 bool parseDirectiveCVDefRange(); 578 bool parseDirectiveCVString(); 579 bool parseDirectiveCVStringTable(); 580 bool parseDirectiveCVFileChecksums(); 581 bool parseDirectiveCVFileChecksumOffset(); 582 bool parseDirectiveCVFPOData(); 583 584 // .cfi directives 585 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 586 bool parseDirectiveCFIWindowSave(); 587 bool parseDirectiveCFISections(); 588 bool parseDirectiveCFIStartProc(); 589 bool parseDirectiveCFIEndProc(); 590 bool parseDirectiveCFIDefCfaOffset(); 591 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 592 bool parseDirectiveCFIAdjustCfaOffset(); 593 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 594 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 595 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 596 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 597 bool parseDirectiveCFIRememberState(); 598 bool parseDirectiveCFIRestoreState(); 599 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 600 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 601 bool parseDirectiveCFIEscape(); 602 bool parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc); 603 bool parseDirectiveCFISignalFrame(); 604 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 605 606 // macro directives 607 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 608 bool parseDirectiveExitMacro(StringRef Directive); 609 bool parseDirectiveEndMacro(StringRef Directive); 610 bool parseDirectiveMacro(SMLoc DirectiveLoc); 611 bool parseDirectiveMacrosOnOff(StringRef Directive); 612 // alternate macro mode directives 613 bool parseDirectiveAltmacro(StringRef Directive); 614 // ".bundle_align_mode" 615 bool parseDirectiveBundleAlignMode(); 616 // ".bundle_lock" 617 bool parseDirectiveBundleLock(); 618 // ".bundle_unlock" 619 bool parseDirectiveBundleUnlock(); 620 621 // ".space", ".skip" 622 bool parseDirectiveSpace(StringRef IDVal); 623 624 // ".dcb" 625 bool parseDirectiveDCB(StringRef IDVal, unsigned Size); 626 bool parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &); 627 // ".ds" 628 bool parseDirectiveDS(StringRef IDVal, unsigned Size); 629 630 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 631 bool parseDirectiveLEB128(bool Signed); 632 633 /// Parse a directive like ".globl" which 634 /// accepts a single symbol (which should be a label or an external). 635 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 636 637 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 638 639 bool parseDirectiveAbort(); // ".abort" 640 bool parseDirectiveInclude(); // ".include" 641 bool parseDirectiveIncbin(); // ".incbin" 642 643 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 644 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 645 // ".ifb" or ".ifnb", depending on ExpectBlank. 646 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 647 // ".ifc" or ".ifnc", depending on ExpectEqual. 648 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 649 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 650 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 651 // ".ifdef" or ".ifndef", depending on expect_defined 652 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 653 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 654 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 655 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 656 bool parseEscapedString(std::string &Data) override; 657 bool parseAngleBracketString(std::string &Data) override; 658 659 const MCExpr *applyModifierToExpr(const MCExpr *E, 660 MCSymbolRefExpr::VariantKind Variant); 661 662 // Macro-like directives 663 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 664 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 665 raw_svector_ostream &OS); 666 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 667 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 668 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 669 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 670 671 // "_emit" or "__emit" 672 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 673 size_t Len); 674 675 // "align" 676 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 677 678 // "end" 679 bool parseDirectiveEnd(SMLoc DirectiveLoc); 680 681 // ".err" or ".error" 682 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 683 684 // ".warning" 685 bool parseDirectiveWarning(SMLoc DirectiveLoc); 686 687 // .print <double-quotes-string> 688 bool parseDirectivePrint(SMLoc DirectiveLoc); 689 690 // .pseudoprobe 691 bool parseDirectivePseudoProbe(); 692 693 // ".lto_discard" 694 bool parseDirectiveLTODiscard(); 695 696 // Directives to support address-significance tables. 697 bool parseDirectiveAddrsig(); 698 bool parseDirectiveAddrsigSym(); 699 700 void initializeDirectiveKindMap(); 701 void initializeCVDefRangeTypeMap(); 702 }; 703 704 } // end anonymous namespace 705 706 namespace llvm { 707 708 extern MCAsmParserExtension *createDarwinAsmParser(); 709 extern MCAsmParserExtension *createELFAsmParser(); 710 extern MCAsmParserExtension *createCOFFAsmParser(); 711 extern MCAsmParserExtension *createWasmAsmParser(); 712 713 } // end namespace llvm 714 715 enum { DEFAULT_ADDRSPACE = 0 }; 716 717 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 718 const MCAsmInfo &MAI, unsigned CB = 0) 719 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 720 CurBuffer(CB ? CB : SM.getMainFileID()), MacrosEnabledFlag(true) { 721 HadError = false; 722 // Save the old handler. 723 SavedDiagHandler = SrcMgr.getDiagHandler(); 724 SavedDiagContext = SrcMgr.getDiagContext(); 725 // Set our own handler which calls the saved handler. 726 SrcMgr.setDiagHandler(DiagHandler, this); 727 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 728 // Make MCStreamer aware of the StartTokLoc for locations in diagnostics. 729 Out.setStartTokLocPtr(&StartTokLoc); 730 731 // Initialize the platform / file format parser. 732 switch (Ctx.getObjectFileType()) { 733 case MCContext::IsCOFF: 734 PlatformParser.reset(createCOFFAsmParser()); 735 break; 736 case MCContext::IsMachO: 737 PlatformParser.reset(createDarwinAsmParser()); 738 IsDarwin = true; 739 break; 740 case MCContext::IsELF: 741 PlatformParser.reset(createELFAsmParser()); 742 break; 743 case MCContext::IsWasm: 744 PlatformParser.reset(createWasmAsmParser()); 745 break; 746 case MCContext::IsXCOFF: 747 report_fatal_error( 748 "Need to implement createXCOFFAsmParser for XCOFF format."); 749 break; 750 } 751 752 PlatformParser->Initialize(*this); 753 initializeDirectiveKindMap(); 754 initializeCVDefRangeTypeMap(); 755 756 NumOfMacroInstantiations = 0; 757 } 758 759 AsmParser::~AsmParser() { 760 assert((HadError || ActiveMacros.empty()) && 761 "Unexpected active macro instantiation!"); 762 763 // Remove MCStreamer's reference to the parser SMLoc. 764 Out.setStartTokLocPtr(nullptr); 765 // Restore the saved diagnostics handler and context for use during 766 // finalization. 767 SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext); 768 } 769 770 void AsmParser::printMacroInstantiations() { 771 // Print the active macro instantiation stack. 772 for (std::vector<MacroInstantiation *>::const_reverse_iterator 773 it = ActiveMacros.rbegin(), 774 ie = ActiveMacros.rend(); 775 it != ie; ++it) 776 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 777 "while in macro instantiation"); 778 } 779 780 void AsmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) { 781 printPendingErrors(); 782 printMessage(L, SourceMgr::DK_Note, Msg, Range); 783 printMacroInstantiations(); 784 } 785 786 bool AsmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) { 787 if(getTargetParser().getTargetOptions().MCNoWarn) 788 return false; 789 if (getTargetParser().getTargetOptions().MCFatalWarnings) 790 return Error(L, Msg, Range); 791 printMessage(L, SourceMgr::DK_Warning, Msg, Range); 792 printMacroInstantiations(); 793 return false; 794 } 795 796 bool AsmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) { 797 HadError = true; 798 printMessage(L, SourceMgr::DK_Error, Msg, Range); 799 printMacroInstantiations(); 800 return true; 801 } 802 803 bool AsmParser::enterIncludeFile(const std::string &Filename) { 804 std::string IncludedFile; 805 unsigned NewBuf = 806 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 807 if (!NewBuf) 808 return true; 809 810 CurBuffer = NewBuf; 811 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 812 return false; 813 } 814 815 /// Process the specified .incbin file by searching for it in the include paths 816 /// then just emitting the byte contents of the file to the streamer. This 817 /// returns true on failure. 818 bool AsmParser::processIncbinFile(const std::string &Filename, int64_t Skip, 819 const MCExpr *Count, SMLoc Loc) { 820 std::string IncludedFile; 821 unsigned NewBuf = 822 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 823 if (!NewBuf) 824 return true; 825 826 // Pick up the bytes from the file and emit them. 827 StringRef Bytes = SrcMgr.getMemoryBuffer(NewBuf)->getBuffer(); 828 Bytes = Bytes.drop_front(Skip); 829 if (Count) { 830 int64_t Res; 831 if (!Count->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 832 return Error(Loc, "expected absolute expression"); 833 if (Res < 0) 834 return Warning(Loc, "negative count has no effect"); 835 Bytes = Bytes.take_front(Res); 836 } 837 getStreamer().emitBytes(Bytes); 838 return false; 839 } 840 841 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 842 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 843 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 844 Loc.getPointer()); 845 } 846 847 const AsmToken &AsmParser::Lex() { 848 if (Lexer.getTok().is(AsmToken::Error)) 849 Error(Lexer.getErrLoc(), Lexer.getErr()); 850 851 // if it's a end of statement with a comment in it 852 if (getTok().is(AsmToken::EndOfStatement)) { 853 // if this is a line comment output it. 854 if (!getTok().getString().empty() && getTok().getString().front() != '\n' && 855 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 856 Out.addExplicitComment(Twine(getTok().getString())); 857 } 858 859 const AsmToken *tok = &Lexer.Lex(); 860 861 // Parse comments here to be deferred until end of next statement. 862 while (tok->is(AsmToken::Comment)) { 863 if (MAI.preserveAsmComments()) 864 Out.addExplicitComment(Twine(tok->getString())); 865 tok = &Lexer.Lex(); 866 } 867 868 if (tok->is(AsmToken::Eof)) { 869 // If this is the end of an included file, pop the parent file off the 870 // include stack. 871 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 872 if (ParentIncludeLoc != SMLoc()) { 873 jumpToLoc(ParentIncludeLoc); 874 return Lex(); 875 } 876 } 877 878 return *tok; 879 } 880 881 bool AsmParser::enabledGenDwarfForAssembly() { 882 // Check whether the user specified -g. 883 if (!getContext().getGenDwarfForAssembly()) 884 return false; 885 // If we haven't encountered any .file directives (which would imply that 886 // the assembler source was produced with debug info already) then emit one 887 // describing the assembler source file itself. 888 if (getContext().getGenDwarfFileNumber() == 0) { 889 // Use the first #line directive for this, if any. It's preprocessed, so 890 // there is no checksum, and of course no source directive. 891 if (!FirstCppHashFilename.empty()) 892 getContext().setMCLineTableRootFile(/*CUID=*/0, 893 getContext().getCompilationDir(), 894 FirstCppHashFilename, 895 /*Cksum=*/None, /*Source=*/None); 896 const MCDwarfFile &RootFile = 897 getContext().getMCDwarfLineTable(/*CUID=*/0).getRootFile(); 898 getContext().setGenDwarfFileNumber(getStreamer().emitDwarfFileDirective( 899 /*CUID=*/0, getContext().getCompilationDir(), RootFile.Name, 900 RootFile.Checksum, RootFile.Source)); 901 } 902 return true; 903 } 904 905 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 906 LTODiscardSymbols.clear(); 907 908 // Create the initial section, if requested. 909 if (!NoInitialTextSection) 910 Out.InitSections(false); 911 912 // Prime the lexer. 913 Lex(); 914 915 HadError = false; 916 AsmCond StartingCondState = TheCondState; 917 SmallVector<AsmRewrite, 4> AsmStrRewrites; 918 919 // If we are generating dwarf for assembly source files save the initial text 920 // section. (Don't use enabledGenDwarfForAssembly() here, as we aren't 921 // emitting any actual debug info yet and haven't had a chance to parse any 922 // embedded .file directives.) 923 if (getContext().getGenDwarfForAssembly()) { 924 MCSection *Sec = getStreamer().getCurrentSectionOnly(); 925 if (!Sec->getBeginSymbol()) { 926 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 927 getStreamer().emitLabel(SectionStartSym); 928 Sec->setBeginSymbol(SectionStartSym); 929 } 930 bool InsertResult = getContext().addGenDwarfSection(Sec); 931 assert(InsertResult && ".text section should not have debug info yet"); 932 (void)InsertResult; 933 } 934 935 getTargetParser().onBeginOfFile(); 936 937 // While we have input, parse each statement. 938 while (Lexer.isNot(AsmToken::Eof)) { 939 ParseStatementInfo Info(&AsmStrRewrites); 940 bool Parsed = parseStatement(Info, nullptr); 941 942 // If we have a Lexer Error we are on an Error Token. Load in Lexer Error 943 // for printing ErrMsg via Lex() only if no (presumably better) parser error 944 // exists. 945 if (Parsed && !hasPendingError() && Lexer.getTok().is(AsmToken::Error)) { 946 Lex(); 947 } 948 949 // parseStatement returned true so may need to emit an error. 950 printPendingErrors(); 951 952 // Skipping to the next line if needed. 953 if (Parsed && !getLexer().isAtStartOfStatement()) 954 eatToEndOfStatement(); 955 } 956 957 getTargetParser().onEndOfFile(); 958 printPendingErrors(); 959 960 // All errors should have been emitted. 961 assert(!hasPendingError() && "unexpected error from parseStatement"); 962 963 getTargetParser().flushPendingInstructions(getStreamer()); 964 965 if (TheCondState.TheCond != StartingCondState.TheCond || 966 TheCondState.Ignore != StartingCondState.Ignore) 967 printError(getTok().getLoc(), "unmatched .ifs or .elses"); 968 // Check to see there are no empty DwarfFile slots. 969 const auto &LineTables = getContext().getMCDwarfLineTables(); 970 if (!LineTables.empty()) { 971 unsigned Index = 0; 972 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 973 if (File.Name.empty() && Index != 0) 974 printError(getTok().getLoc(), "unassigned file number: " + 975 Twine(Index) + 976 " for .file directives"); 977 ++Index; 978 } 979 } 980 981 // Check to see that all assembler local symbols were actually defined. 982 // Targets that don't do subsections via symbols may not want this, though, 983 // so conservatively exclude them. Only do this if we're finalizing, though, 984 // as otherwise we won't necessarilly have seen everything yet. 985 if (!NoFinalize) { 986 if (MAI.hasSubsectionsViaSymbols()) { 987 for (const auto &TableEntry : getContext().getSymbols()) { 988 MCSymbol *Sym = TableEntry.getValue(); 989 // Variable symbols may not be marked as defined, so check those 990 // explicitly. If we know it's a variable, we have a definition for 991 // the purposes of this check. 992 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 993 // FIXME: We would really like to refer back to where the symbol was 994 // first referenced for a source location. We need to add something 995 // to track that. Currently, we just point to the end of the file. 996 printError(getTok().getLoc(), "assembler local symbol '" + 997 Sym->getName() + "' not defined"); 998 } 999 } 1000 1001 // Temporary symbols like the ones for directional jumps don't go in the 1002 // symbol table. They also need to be diagnosed in all (final) cases. 1003 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 1004 if (std::get<2>(LocSym)->isUndefined()) { 1005 // Reset the state of any "# line file" directives we've seen to the 1006 // context as it was at the diagnostic site. 1007 CppHashInfo = std::get<1>(LocSym); 1008 printError(std::get<0>(LocSym), "directional label undefined"); 1009 } 1010 } 1011 } 1012 1013 // Finalize the output stream if there are no errors and if the client wants 1014 // us to. 1015 if (!HadError && !NoFinalize) 1016 Out.Finish(Lexer.getLoc()); 1017 1018 return HadError || getContext().hadError(); 1019 } 1020 1021 bool AsmParser::checkForValidSection() { 1022 if (!ParsingMSInlineAsm && !getStreamer().getCurrentSectionOnly()) { 1023 Out.InitSections(false); 1024 return Error(getTok().getLoc(), 1025 "expected section directive before assembly directive"); 1026 } 1027 return false; 1028 } 1029 1030 /// Throw away the rest of the line for testing purposes. 1031 void AsmParser::eatToEndOfStatement() { 1032 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1033 Lexer.Lex(); 1034 1035 // Eat EOL. 1036 if (Lexer.is(AsmToken::EndOfStatement)) 1037 Lexer.Lex(); 1038 } 1039 1040 StringRef AsmParser::parseStringToEndOfStatement() { 1041 const char *Start = getTok().getLoc().getPointer(); 1042 1043 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1044 Lexer.Lex(); 1045 1046 const char *End = getTok().getLoc().getPointer(); 1047 return StringRef(Start, End - Start); 1048 } 1049 1050 StringRef AsmParser::parseStringToComma() { 1051 const char *Start = getTok().getLoc().getPointer(); 1052 1053 while (Lexer.isNot(AsmToken::EndOfStatement) && 1054 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 1055 Lexer.Lex(); 1056 1057 const char *End = getTok().getLoc().getPointer(); 1058 return StringRef(Start, End - Start); 1059 } 1060 1061 /// Parse a paren expression and return it. 1062 /// NOTE: This assumes the leading '(' has already been consumed. 1063 /// 1064 /// parenexpr ::= expr) 1065 /// 1066 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1067 if (parseExpression(Res)) 1068 return true; 1069 if (Lexer.isNot(AsmToken::RParen)) 1070 return TokError("expected ')' in parentheses expression"); 1071 EndLoc = Lexer.getTok().getEndLoc(); 1072 Lex(); 1073 return false; 1074 } 1075 1076 /// Parse a bracket expression and return it. 1077 /// NOTE: This assumes the leading '[' has already been consumed. 1078 /// 1079 /// bracketexpr ::= expr] 1080 /// 1081 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1082 if (parseExpression(Res)) 1083 return true; 1084 EndLoc = getTok().getEndLoc(); 1085 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 1086 return true; 1087 return false; 1088 } 1089 1090 /// Parse a primary expression and return it. 1091 /// primaryexpr ::= (parenexpr 1092 /// primaryexpr ::= symbol 1093 /// primaryexpr ::= number 1094 /// primaryexpr ::= '.' 1095 /// primaryexpr ::= ~,+,- primaryexpr 1096 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc, 1097 AsmTypeInfo *TypeInfo) { 1098 SMLoc FirstTokenLoc = getLexer().getLoc(); 1099 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 1100 switch (FirstTokenKind) { 1101 default: 1102 return TokError("unknown token in expression"); 1103 // If we have an error assume that we've already handled it. 1104 case AsmToken::Error: 1105 return true; 1106 case AsmToken::Exclaim: 1107 Lex(); // Eat the operator. 1108 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1109 return true; 1110 Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc); 1111 return false; 1112 case AsmToken::Dollar: 1113 case AsmToken::Star: 1114 case AsmToken::At: 1115 case AsmToken::String: 1116 case AsmToken::Identifier: { 1117 StringRef Identifier; 1118 if (parseIdentifier(Identifier)) { 1119 // We may have failed but '$'|'*' may be a valid token in context of 1120 // the current PC. 1121 if (getTok().is(AsmToken::Dollar) || getTok().is(AsmToken::Star)) { 1122 bool ShouldGenerateTempSymbol = false; 1123 if ((getTok().is(AsmToken::Dollar) && MAI.getDollarIsPC()) || 1124 (getTok().is(AsmToken::Star) && MAI.getStarIsPC())) 1125 ShouldGenerateTempSymbol = true; 1126 1127 if (!ShouldGenerateTempSymbol) 1128 return Error(FirstTokenLoc, "invalid token in expression"); 1129 1130 // Eat the '$'|'*' token. 1131 Lex(); 1132 // This is either a '$'|'*' reference, which references the current PC. 1133 // Emit a temporary label to the streamer and refer to it. 1134 MCSymbol *Sym = Ctx.createTempSymbol(); 1135 Out.emitLabel(Sym); 1136 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 1137 getContext()); 1138 EndLoc = FirstTokenLoc; 1139 return false; 1140 } 1141 } 1142 // Parse symbol variant 1143 std::pair<StringRef, StringRef> Split; 1144 if (!MAI.useParensForSymbolVariant()) { 1145 if (FirstTokenKind == AsmToken::String) { 1146 if (Lexer.is(AsmToken::At)) { 1147 Lex(); // eat @ 1148 SMLoc AtLoc = getLexer().getLoc(); 1149 StringRef VName; 1150 if (parseIdentifier(VName)) 1151 return Error(AtLoc, "expected symbol variant after '@'"); 1152 1153 Split = std::make_pair(Identifier, VName); 1154 } 1155 } else { 1156 Split = Identifier.split('@'); 1157 } 1158 } else if (Lexer.is(AsmToken::LParen)) { 1159 Lex(); // eat '('. 1160 StringRef VName; 1161 parseIdentifier(VName); 1162 // eat ')'. 1163 if (parseToken(AsmToken::RParen, 1164 "unexpected token in variant, expected ')'")) 1165 return true; 1166 Split = std::make_pair(Identifier, VName); 1167 } 1168 1169 EndLoc = SMLoc::getFromPointer(Identifier.end()); 1170 1171 // This is a symbol reference. 1172 StringRef SymbolName = Identifier; 1173 if (SymbolName.empty()) 1174 return Error(getLexer().getLoc(), "expected a symbol reference"); 1175 1176 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1177 1178 // Lookup the symbol variant if used. 1179 if (!Split.second.empty()) { 1180 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1181 if (Variant != MCSymbolRefExpr::VK_Invalid) { 1182 SymbolName = Split.first; 1183 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 1184 Variant = MCSymbolRefExpr::VK_None; 1185 } else { 1186 return Error(SMLoc::getFromPointer(Split.second.begin()), 1187 "invalid variant '" + Split.second + "'"); 1188 } 1189 } 1190 1191 MCSymbol *Sym = getContext().getInlineAsmLabel(SymbolName); 1192 if (!Sym) 1193 Sym = getContext().getOrCreateSymbol(SymbolName); 1194 1195 // If this is an absolute variable reference, substitute it now to preserve 1196 // semantics in the face of reassignment. 1197 if (Sym->isVariable()) { 1198 auto V = Sym->getVariableValue(/*SetUsed*/ false); 1199 bool DoInline = isa<MCConstantExpr>(V) && !Variant; 1200 if (auto TV = dyn_cast<MCTargetExpr>(V)) 1201 DoInline = TV->inlineAssignedExpr(); 1202 if (DoInline) { 1203 if (Variant) 1204 return Error(EndLoc, "unexpected modifier on variable reference"); 1205 Res = Sym->getVariableValue(/*SetUsed*/ false); 1206 return false; 1207 } 1208 } 1209 1210 // Otherwise create a symbol ref. 1211 Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc); 1212 return false; 1213 } 1214 case AsmToken::BigNum: 1215 return TokError("literal value out of range for directive"); 1216 case AsmToken::Integer: { 1217 SMLoc Loc = getTok().getLoc(); 1218 int64_t IntVal = getTok().getIntVal(); 1219 Res = MCConstantExpr::create(IntVal, getContext()); 1220 EndLoc = Lexer.getTok().getEndLoc(); 1221 Lex(); // Eat token. 1222 // Look for 'b' or 'f' following an Integer as a directional label 1223 if (Lexer.getKind() == AsmToken::Identifier) { 1224 StringRef IDVal = getTok().getString(); 1225 // Lookup the symbol variant if used. 1226 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 1227 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1228 if (Split.first.size() != IDVal.size()) { 1229 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1230 if (Variant == MCSymbolRefExpr::VK_Invalid) 1231 return TokError("invalid variant '" + Split.second + "'"); 1232 IDVal = Split.first; 1233 } 1234 if (IDVal == "f" || IDVal == "b") { 1235 MCSymbol *Sym = 1236 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 1237 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 1238 if (IDVal == "b" && Sym->isUndefined()) 1239 return Error(Loc, "directional label undefined"); 1240 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 1241 EndLoc = Lexer.getTok().getEndLoc(); 1242 Lex(); // Eat identifier. 1243 } 1244 } 1245 return false; 1246 } 1247 case AsmToken::Real: { 1248 APFloat RealVal(APFloat::IEEEdouble(), getTok().getString()); 1249 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1250 Res = MCConstantExpr::create(IntVal, getContext()); 1251 EndLoc = Lexer.getTok().getEndLoc(); 1252 Lex(); // Eat token. 1253 return false; 1254 } 1255 case AsmToken::Dot: { 1256 if (!MAI.getDotIsPC()) 1257 return TokError("cannot use . as current PC"); 1258 1259 // This is a '.' reference, which references the current PC. Emit a 1260 // temporary label to the streamer and refer to it. 1261 MCSymbol *Sym = Ctx.createTempSymbol(); 1262 Out.emitLabel(Sym); 1263 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1264 EndLoc = Lexer.getTok().getEndLoc(); 1265 Lex(); // Eat identifier. 1266 return false; 1267 } 1268 case AsmToken::LParen: 1269 Lex(); // Eat the '('. 1270 return parseParenExpr(Res, EndLoc); 1271 case AsmToken::LBrac: 1272 if (!PlatformParser->HasBracketExpressions()) 1273 return TokError("brackets expression not supported on this target"); 1274 Lex(); // Eat the '['. 1275 return parseBracketExpr(Res, EndLoc); 1276 case AsmToken::Minus: 1277 Lex(); // Eat the operator. 1278 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1279 return true; 1280 Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc); 1281 return false; 1282 case AsmToken::Plus: 1283 Lex(); // Eat the operator. 1284 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1285 return true; 1286 Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc); 1287 return false; 1288 case AsmToken::Tilde: 1289 Lex(); // Eat the operator. 1290 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1291 return true; 1292 Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc); 1293 return false; 1294 // MIPS unary expression operators. The lexer won't generate these tokens if 1295 // MCAsmInfo::HasMipsExpressions is false for the target. 1296 case AsmToken::PercentCall16: 1297 case AsmToken::PercentCall_Hi: 1298 case AsmToken::PercentCall_Lo: 1299 case AsmToken::PercentDtprel_Hi: 1300 case AsmToken::PercentDtprel_Lo: 1301 case AsmToken::PercentGot: 1302 case AsmToken::PercentGot_Disp: 1303 case AsmToken::PercentGot_Hi: 1304 case AsmToken::PercentGot_Lo: 1305 case AsmToken::PercentGot_Ofst: 1306 case AsmToken::PercentGot_Page: 1307 case AsmToken::PercentGottprel: 1308 case AsmToken::PercentGp_Rel: 1309 case AsmToken::PercentHi: 1310 case AsmToken::PercentHigher: 1311 case AsmToken::PercentHighest: 1312 case AsmToken::PercentLo: 1313 case AsmToken::PercentNeg: 1314 case AsmToken::PercentPcrel_Hi: 1315 case AsmToken::PercentPcrel_Lo: 1316 case AsmToken::PercentTlsgd: 1317 case AsmToken::PercentTlsldm: 1318 case AsmToken::PercentTprel_Hi: 1319 case AsmToken::PercentTprel_Lo: 1320 Lex(); // Eat the operator. 1321 if (Lexer.isNot(AsmToken::LParen)) 1322 return TokError("expected '(' after operator"); 1323 Lex(); // Eat the operator. 1324 if (parseExpression(Res, EndLoc)) 1325 return true; 1326 if (Lexer.isNot(AsmToken::RParen)) 1327 return TokError("expected ')'"); 1328 Lex(); // Eat the operator. 1329 Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx); 1330 return !Res; 1331 } 1332 } 1333 1334 bool AsmParser::parseExpression(const MCExpr *&Res) { 1335 SMLoc EndLoc; 1336 return parseExpression(Res, EndLoc); 1337 } 1338 1339 const MCExpr * 1340 AsmParser::applyModifierToExpr(const MCExpr *E, 1341 MCSymbolRefExpr::VariantKind Variant) { 1342 // Ask the target implementation about this expression first. 1343 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 1344 if (NewE) 1345 return NewE; 1346 // Recurse over the given expression, rebuilding it to apply the given variant 1347 // if there is exactly one symbol. 1348 switch (E->getKind()) { 1349 case MCExpr::Target: 1350 case MCExpr::Constant: 1351 return nullptr; 1352 1353 case MCExpr::SymbolRef: { 1354 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 1355 1356 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 1357 TokError("invalid variant on expression '" + getTok().getIdentifier() + 1358 "' (already modified)"); 1359 return E; 1360 } 1361 1362 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1363 } 1364 1365 case MCExpr::Unary: { 1366 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1367 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1368 if (!Sub) 1369 return nullptr; 1370 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1371 } 1372 1373 case MCExpr::Binary: { 1374 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1375 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1376 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1377 1378 if (!LHS && !RHS) 1379 return nullptr; 1380 1381 if (!LHS) 1382 LHS = BE->getLHS(); 1383 if (!RHS) 1384 RHS = BE->getRHS(); 1385 1386 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1387 } 1388 } 1389 1390 llvm_unreachable("Invalid expression kind!"); 1391 } 1392 1393 /// This function checks if the next token is <string> type or arithmetic. 1394 /// string that begin with character '<' must end with character '>'. 1395 /// otherwise it is arithmetics. 1396 /// If the function returns a 'true' value, 1397 /// the End argument will be filled with the last location pointed to the '>' 1398 /// character. 1399 1400 /// There is a gap between the AltMacro's documentation and the single quote 1401 /// implementation. GCC does not fully support this feature and so we will not 1402 /// support it. 1403 /// TODO: Adding single quote as a string. 1404 static bool isAngleBracketString(SMLoc &StrLoc, SMLoc &EndLoc) { 1405 assert((StrLoc.getPointer() != nullptr) && 1406 "Argument to the function cannot be a NULL value"); 1407 const char *CharPtr = StrLoc.getPointer(); 1408 while ((*CharPtr != '>') && (*CharPtr != '\n') && (*CharPtr != '\r') && 1409 (*CharPtr != '\0')) { 1410 if (*CharPtr == '!') 1411 CharPtr++; 1412 CharPtr++; 1413 } 1414 if (*CharPtr == '>') { 1415 EndLoc = StrLoc.getFromPointer(CharPtr + 1); 1416 return true; 1417 } 1418 return false; 1419 } 1420 1421 /// creating a string without the escape characters '!'. 1422 static std::string angleBracketString(StringRef AltMacroStr) { 1423 std::string Res; 1424 for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) { 1425 if (AltMacroStr[Pos] == '!') 1426 Pos++; 1427 Res += AltMacroStr[Pos]; 1428 } 1429 return Res; 1430 } 1431 1432 /// Parse an expression and return it. 1433 /// 1434 /// expr ::= expr &&,|| expr -> lowest. 1435 /// expr ::= expr |,^,&,! expr 1436 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1437 /// expr ::= expr <<,>> expr 1438 /// expr ::= expr +,- expr 1439 /// expr ::= expr *,/,% expr -> highest. 1440 /// expr ::= primaryexpr 1441 /// 1442 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1443 // Parse the expression. 1444 Res = nullptr; 1445 if (getTargetParser().parsePrimaryExpr(Res, EndLoc) || 1446 parseBinOpRHS(1, Res, EndLoc)) 1447 return true; 1448 1449 // As a special case, we support 'a op b @ modifier' by rewriting the 1450 // expression to include the modifier. This is inefficient, but in general we 1451 // expect users to use 'a@modifier op b'. 1452 if (Lexer.getKind() == AsmToken::At) { 1453 Lex(); 1454 1455 if (Lexer.isNot(AsmToken::Identifier)) 1456 return TokError("unexpected symbol modifier following '@'"); 1457 1458 MCSymbolRefExpr::VariantKind Variant = 1459 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1460 if (Variant == MCSymbolRefExpr::VK_Invalid) 1461 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1462 1463 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1464 if (!ModifiedRes) { 1465 return TokError("invalid modifier '" + getTok().getIdentifier() + 1466 "' (no symbols present)"); 1467 } 1468 1469 Res = ModifiedRes; 1470 Lex(); 1471 } 1472 1473 // Try to constant fold it up front, if possible. Do not exploit 1474 // assembler here. 1475 int64_t Value; 1476 if (Res->evaluateAsAbsolute(Value)) 1477 Res = MCConstantExpr::create(Value, getContext()); 1478 1479 return false; 1480 } 1481 1482 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1483 Res = nullptr; 1484 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1485 } 1486 1487 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1488 SMLoc &EndLoc) { 1489 if (parseParenExpr(Res, EndLoc)) 1490 return true; 1491 1492 for (; ParenDepth > 0; --ParenDepth) { 1493 if (parseBinOpRHS(1, Res, EndLoc)) 1494 return true; 1495 1496 // We don't Lex() the last RParen. 1497 // This is the same behavior as parseParenExpression(). 1498 if (ParenDepth - 1 > 0) { 1499 EndLoc = getTok().getEndLoc(); 1500 if (parseToken(AsmToken::RParen, 1501 "expected ')' in parentheses expression")) 1502 return true; 1503 } 1504 } 1505 return false; 1506 } 1507 1508 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1509 const MCExpr *Expr; 1510 1511 SMLoc StartLoc = Lexer.getLoc(); 1512 if (parseExpression(Expr)) 1513 return true; 1514 1515 if (!Expr->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 1516 return Error(StartLoc, "expected absolute expression"); 1517 1518 return false; 1519 } 1520 1521 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1522 MCBinaryExpr::Opcode &Kind, 1523 bool ShouldUseLogicalShr) { 1524 switch (K) { 1525 default: 1526 return 0; // not a binop. 1527 1528 // Lowest Precedence: &&, || 1529 case AsmToken::AmpAmp: 1530 Kind = MCBinaryExpr::LAnd; 1531 return 1; 1532 case AsmToken::PipePipe: 1533 Kind = MCBinaryExpr::LOr; 1534 return 1; 1535 1536 // Low Precedence: |, &, ^ 1537 case AsmToken::Pipe: 1538 Kind = MCBinaryExpr::Or; 1539 return 2; 1540 case AsmToken::Caret: 1541 Kind = MCBinaryExpr::Xor; 1542 return 2; 1543 case AsmToken::Amp: 1544 Kind = MCBinaryExpr::And; 1545 return 2; 1546 1547 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1548 case AsmToken::EqualEqual: 1549 Kind = MCBinaryExpr::EQ; 1550 return 3; 1551 case AsmToken::ExclaimEqual: 1552 case AsmToken::LessGreater: 1553 Kind = MCBinaryExpr::NE; 1554 return 3; 1555 case AsmToken::Less: 1556 Kind = MCBinaryExpr::LT; 1557 return 3; 1558 case AsmToken::LessEqual: 1559 Kind = MCBinaryExpr::LTE; 1560 return 3; 1561 case AsmToken::Greater: 1562 Kind = MCBinaryExpr::GT; 1563 return 3; 1564 case AsmToken::GreaterEqual: 1565 Kind = MCBinaryExpr::GTE; 1566 return 3; 1567 1568 // Intermediate Precedence: <<, >> 1569 case AsmToken::LessLess: 1570 Kind = MCBinaryExpr::Shl; 1571 return 4; 1572 case AsmToken::GreaterGreater: 1573 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1574 return 4; 1575 1576 // High Intermediate Precedence: +, - 1577 case AsmToken::Plus: 1578 Kind = MCBinaryExpr::Add; 1579 return 5; 1580 case AsmToken::Minus: 1581 Kind = MCBinaryExpr::Sub; 1582 return 5; 1583 1584 // Highest Precedence: *, /, % 1585 case AsmToken::Star: 1586 Kind = MCBinaryExpr::Mul; 1587 return 6; 1588 case AsmToken::Slash: 1589 Kind = MCBinaryExpr::Div; 1590 return 6; 1591 case AsmToken::Percent: 1592 Kind = MCBinaryExpr::Mod; 1593 return 6; 1594 } 1595 } 1596 1597 static unsigned getGNUBinOpPrecedence(const MCAsmInfo &MAI, 1598 AsmToken::TokenKind K, 1599 MCBinaryExpr::Opcode &Kind, 1600 bool ShouldUseLogicalShr) { 1601 switch (K) { 1602 default: 1603 return 0; // not a binop. 1604 1605 // Lowest Precedence: &&, || 1606 case AsmToken::AmpAmp: 1607 Kind = MCBinaryExpr::LAnd; 1608 return 2; 1609 case AsmToken::PipePipe: 1610 Kind = MCBinaryExpr::LOr; 1611 return 1; 1612 1613 // Low Precedence: ==, !=, <>, <, <=, >, >= 1614 case AsmToken::EqualEqual: 1615 Kind = MCBinaryExpr::EQ; 1616 return 3; 1617 case AsmToken::ExclaimEqual: 1618 case AsmToken::LessGreater: 1619 Kind = MCBinaryExpr::NE; 1620 return 3; 1621 case AsmToken::Less: 1622 Kind = MCBinaryExpr::LT; 1623 return 3; 1624 case AsmToken::LessEqual: 1625 Kind = MCBinaryExpr::LTE; 1626 return 3; 1627 case AsmToken::Greater: 1628 Kind = MCBinaryExpr::GT; 1629 return 3; 1630 case AsmToken::GreaterEqual: 1631 Kind = MCBinaryExpr::GTE; 1632 return 3; 1633 1634 // Low Intermediate Precedence: +, - 1635 case AsmToken::Plus: 1636 Kind = MCBinaryExpr::Add; 1637 return 4; 1638 case AsmToken::Minus: 1639 Kind = MCBinaryExpr::Sub; 1640 return 4; 1641 1642 // High Intermediate Precedence: |, !, &, ^ 1643 // 1644 case AsmToken::Pipe: 1645 Kind = MCBinaryExpr::Or; 1646 return 5; 1647 case AsmToken::Exclaim: 1648 // Hack to support ARM compatible aliases (implied 'sp' operand in 'srs*' 1649 // instructions like 'srsda #31!') and not parse ! as an infix operator. 1650 if (MAI.getCommentString() == "@") 1651 return 0; 1652 Kind = MCBinaryExpr::OrNot; 1653 return 5; 1654 case AsmToken::Caret: 1655 Kind = MCBinaryExpr::Xor; 1656 return 5; 1657 case AsmToken::Amp: 1658 Kind = MCBinaryExpr::And; 1659 return 5; 1660 1661 // Highest Precedence: *, /, %, <<, >> 1662 case AsmToken::Star: 1663 Kind = MCBinaryExpr::Mul; 1664 return 6; 1665 case AsmToken::Slash: 1666 Kind = MCBinaryExpr::Div; 1667 return 6; 1668 case AsmToken::Percent: 1669 Kind = MCBinaryExpr::Mod; 1670 return 6; 1671 case AsmToken::LessLess: 1672 Kind = MCBinaryExpr::Shl; 1673 return 6; 1674 case AsmToken::GreaterGreater: 1675 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1676 return 6; 1677 } 1678 } 1679 1680 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1681 MCBinaryExpr::Opcode &Kind) { 1682 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1683 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1684 : getGNUBinOpPrecedence(MAI, K, Kind, ShouldUseLogicalShr); 1685 } 1686 1687 /// Parse all binary operators with precedence >= 'Precedence'. 1688 /// Res contains the LHS of the expression on input. 1689 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1690 SMLoc &EndLoc) { 1691 SMLoc StartLoc = Lexer.getLoc(); 1692 while (true) { 1693 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1694 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1695 1696 // If the next token is lower precedence than we are allowed to eat, return 1697 // successfully with what we ate already. 1698 if (TokPrec < Precedence) 1699 return false; 1700 1701 Lex(); 1702 1703 // Eat the next primary expression. 1704 const MCExpr *RHS; 1705 if (getTargetParser().parsePrimaryExpr(RHS, EndLoc)) 1706 return true; 1707 1708 // If BinOp binds less tightly with RHS than the operator after RHS, let 1709 // the pending operator take RHS as its LHS. 1710 MCBinaryExpr::Opcode Dummy; 1711 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1712 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1713 return true; 1714 1715 // Merge LHS and RHS according to operator. 1716 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc); 1717 } 1718 } 1719 1720 /// ParseStatement: 1721 /// ::= EndOfStatement 1722 /// ::= Label* Directive ...Operands... EndOfStatement 1723 /// ::= Label* Identifier OperandList* EndOfStatement 1724 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1725 MCAsmParserSemaCallback *SI) { 1726 assert(!hasPendingError() && "parseStatement started with pending error"); 1727 // Eat initial spaces and comments 1728 while (Lexer.is(AsmToken::Space)) 1729 Lex(); 1730 if (Lexer.is(AsmToken::EndOfStatement)) { 1731 // if this is a line comment we can drop it safely 1732 if (getTok().getString().empty() || getTok().getString().front() == '\r' || 1733 getTok().getString().front() == '\n') 1734 Out.AddBlankLine(); 1735 Lex(); 1736 return false; 1737 } 1738 // Statements always start with an identifier. 1739 AsmToken ID = getTok(); 1740 SMLoc IDLoc = ID.getLoc(); 1741 StringRef IDVal; 1742 int64_t LocalLabelVal = -1; 1743 StartTokLoc = ID.getLoc(); 1744 if (Lexer.is(AsmToken::HashDirective)) 1745 return parseCppHashLineFilenameComment(IDLoc, 1746 !isInsideMacroInstantiation()); 1747 1748 // Allow an integer followed by a ':' as a directional local label. 1749 if (Lexer.is(AsmToken::Integer)) { 1750 LocalLabelVal = getTok().getIntVal(); 1751 if (LocalLabelVal < 0) { 1752 if (!TheCondState.Ignore) { 1753 Lex(); // always eat a token 1754 return Error(IDLoc, "unexpected token at start of statement"); 1755 } 1756 IDVal = ""; 1757 } else { 1758 IDVal = getTok().getString(); 1759 Lex(); // Consume the integer token to be used as an identifier token. 1760 if (Lexer.getKind() != AsmToken::Colon) { 1761 if (!TheCondState.Ignore) { 1762 Lex(); // always eat a token 1763 return Error(IDLoc, "unexpected token at start of statement"); 1764 } 1765 } 1766 } 1767 } else if (Lexer.is(AsmToken::Dot)) { 1768 // Treat '.' as a valid identifier in this context. 1769 Lex(); 1770 IDVal = "."; 1771 } else if (Lexer.is(AsmToken::LCurly)) { 1772 // Treat '{' as a valid identifier in this context. 1773 Lex(); 1774 IDVal = "{"; 1775 1776 } else if (Lexer.is(AsmToken::RCurly)) { 1777 // Treat '}' as a valid identifier in this context. 1778 Lex(); 1779 IDVal = "}"; 1780 } else if (Lexer.is(AsmToken::Star) && 1781 getTargetParser().starIsStartOfStatement()) { 1782 // Accept '*' as a valid start of statement. 1783 Lex(); 1784 IDVal = "*"; 1785 } else if (parseIdentifier(IDVal)) { 1786 if (!TheCondState.Ignore) { 1787 Lex(); // always eat a token 1788 return Error(IDLoc, "unexpected token at start of statement"); 1789 } 1790 IDVal = ""; 1791 } 1792 1793 // Handle conditional assembly here before checking for skipping. We 1794 // have to do this so that .endif isn't skipped in a ".if 0" block for 1795 // example. 1796 StringMap<DirectiveKind>::const_iterator DirKindIt = 1797 DirectiveKindMap.find(IDVal.lower()); 1798 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1799 ? DK_NO_DIRECTIVE 1800 : DirKindIt->getValue(); 1801 switch (DirKind) { 1802 default: 1803 break; 1804 case DK_IF: 1805 case DK_IFEQ: 1806 case DK_IFGE: 1807 case DK_IFGT: 1808 case DK_IFLE: 1809 case DK_IFLT: 1810 case DK_IFNE: 1811 return parseDirectiveIf(IDLoc, DirKind); 1812 case DK_IFB: 1813 return parseDirectiveIfb(IDLoc, true); 1814 case DK_IFNB: 1815 return parseDirectiveIfb(IDLoc, false); 1816 case DK_IFC: 1817 return parseDirectiveIfc(IDLoc, true); 1818 case DK_IFEQS: 1819 return parseDirectiveIfeqs(IDLoc, true); 1820 case DK_IFNC: 1821 return parseDirectiveIfc(IDLoc, false); 1822 case DK_IFNES: 1823 return parseDirectiveIfeqs(IDLoc, false); 1824 case DK_IFDEF: 1825 return parseDirectiveIfdef(IDLoc, true); 1826 case DK_IFNDEF: 1827 case DK_IFNOTDEF: 1828 return parseDirectiveIfdef(IDLoc, false); 1829 case DK_ELSEIF: 1830 return parseDirectiveElseIf(IDLoc); 1831 case DK_ELSE: 1832 return parseDirectiveElse(IDLoc); 1833 case DK_ENDIF: 1834 return parseDirectiveEndIf(IDLoc); 1835 } 1836 1837 // Ignore the statement if in the middle of inactive conditional 1838 // (e.g. ".if 0"). 1839 if (TheCondState.Ignore) { 1840 eatToEndOfStatement(); 1841 return false; 1842 } 1843 1844 // FIXME: Recurse on local labels? 1845 1846 // See what kind of statement we have. 1847 switch (Lexer.getKind()) { 1848 case AsmToken::Colon: { 1849 if (!getTargetParser().isLabel(ID)) 1850 break; 1851 if (checkForValidSection()) 1852 return true; 1853 1854 // identifier ':' -> Label. 1855 Lex(); 1856 1857 // Diagnose attempt to use '.' as a label. 1858 if (IDVal == ".") 1859 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1860 1861 // Diagnose attempt to use a variable as a label. 1862 // 1863 // FIXME: Diagnostics. Note the location of the definition as a label. 1864 // FIXME: This doesn't diagnose assignment to a symbol which has been 1865 // implicitly marked as external. 1866 MCSymbol *Sym; 1867 if (LocalLabelVal == -1) { 1868 if (ParsingMSInlineAsm && SI) { 1869 StringRef RewrittenLabel = 1870 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1871 assert(!RewrittenLabel.empty() && 1872 "We should have an internal name here."); 1873 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1874 RewrittenLabel); 1875 IDVal = RewrittenLabel; 1876 } 1877 Sym = getContext().getOrCreateSymbol(IDVal); 1878 } else 1879 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1880 // End of Labels should be treated as end of line for lexing 1881 // purposes but that information is not available to the Lexer who 1882 // does not understand Labels. This may cause us to see a Hash 1883 // here instead of a preprocessor line comment. 1884 if (getTok().is(AsmToken::Hash)) { 1885 StringRef CommentStr = parseStringToEndOfStatement(); 1886 Lexer.Lex(); 1887 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1888 } 1889 1890 // Consume any end of statement token, if present, to avoid spurious 1891 // AddBlankLine calls(). 1892 if (getTok().is(AsmToken::EndOfStatement)) { 1893 Lex(); 1894 } 1895 1896 if (discardLTOSymbol(IDVal)) 1897 return false; 1898 1899 getTargetParser().doBeforeLabelEmit(Sym); 1900 1901 // Emit the label. 1902 if (!getTargetParser().isParsingMSInlineAsm()) 1903 Out.emitLabel(Sym, IDLoc); 1904 1905 // If we are generating dwarf for assembly source files then gather the 1906 // info to make a dwarf label entry for this label if needed. 1907 if (enabledGenDwarfForAssembly()) 1908 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1909 IDLoc); 1910 1911 getTargetParser().onLabelParsed(Sym); 1912 1913 return false; 1914 } 1915 1916 case AsmToken::Equal: 1917 if (!getTargetParser().equalIsAsmAssignment()) 1918 break; 1919 // identifier '=' ... -> assignment statement 1920 Lex(); 1921 1922 return parseAssignment(IDVal, true); 1923 1924 default: // Normal instruction or directive. 1925 break; 1926 } 1927 1928 // If macros are enabled, check to see if this is a macro instantiation. 1929 if (areMacrosEnabled()) 1930 if (const MCAsmMacro *M = getContext().lookupMacro(IDVal)) { 1931 return handleMacroEntry(M, IDLoc); 1932 } 1933 1934 // Otherwise, we have a normal instruction or directive. 1935 1936 // Directives start with "." 1937 if (IDVal.startswith(".") && IDVal != ".") { 1938 // There are several entities interested in parsing directives: 1939 // 1940 // 1. The target-specific assembly parser. Some directives are target 1941 // specific or may potentially behave differently on certain targets. 1942 // 2. Asm parser extensions. For example, platform-specific parsers 1943 // (like the ELF parser) register themselves as extensions. 1944 // 3. The generic directive parser implemented by this class. These are 1945 // all the directives that behave in a target and platform independent 1946 // manner, or at least have a default behavior that's shared between 1947 // all targets and platforms. 1948 1949 getTargetParser().flushPendingInstructions(getStreamer()); 1950 1951 SMLoc StartTokLoc = getTok().getLoc(); 1952 bool TPDirectiveReturn = getTargetParser().ParseDirective(ID); 1953 1954 if (hasPendingError()) 1955 return true; 1956 // Currently the return value should be true if we are 1957 // uninterested but as this is at odds with the standard parsing 1958 // convention (return true = error) we have instances of a parsed 1959 // directive that fails returning true as an error. Catch these 1960 // cases as best as possible errors here. 1961 if (TPDirectiveReturn && StartTokLoc != getTok().getLoc()) 1962 return true; 1963 // Return if we did some parsing or believe we succeeded. 1964 if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc()) 1965 return false; 1966 1967 // Next, check the extension directive map to see if any extension has 1968 // registered itself to parse this directive. 1969 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1970 ExtensionDirectiveMap.lookup(IDVal); 1971 if (Handler.first) 1972 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1973 1974 // Finally, if no one else is interested in this directive, it must be 1975 // generic and familiar to this class. 1976 switch (DirKind) { 1977 default: 1978 break; 1979 case DK_SET: 1980 case DK_EQU: 1981 return parseDirectiveSet(IDVal, true); 1982 case DK_EQUIV: 1983 return parseDirectiveSet(IDVal, false); 1984 case DK_ASCII: 1985 return parseDirectiveAscii(IDVal, false); 1986 case DK_ASCIZ: 1987 case DK_STRING: 1988 return parseDirectiveAscii(IDVal, true); 1989 case DK_BYTE: 1990 case DK_DC_B: 1991 return parseDirectiveValue(IDVal, 1); 1992 case DK_DC: 1993 case DK_DC_W: 1994 case DK_SHORT: 1995 case DK_VALUE: 1996 case DK_2BYTE: 1997 return parseDirectiveValue(IDVal, 2); 1998 case DK_LONG: 1999 case DK_INT: 2000 case DK_4BYTE: 2001 case DK_DC_L: 2002 return parseDirectiveValue(IDVal, 4); 2003 case DK_QUAD: 2004 case DK_8BYTE: 2005 return parseDirectiveValue(IDVal, 8); 2006 case DK_DC_A: 2007 return parseDirectiveValue( 2008 IDVal, getContext().getAsmInfo()->getCodePointerSize()); 2009 case DK_OCTA: 2010 return parseDirectiveOctaValue(IDVal); 2011 case DK_SINGLE: 2012 case DK_FLOAT: 2013 case DK_DC_S: 2014 return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle()); 2015 case DK_DOUBLE: 2016 case DK_DC_D: 2017 return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble()); 2018 case DK_ALIGN: { 2019 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 2020 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 2021 } 2022 case DK_ALIGN32: { 2023 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 2024 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 2025 } 2026 case DK_BALIGN: 2027 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 2028 case DK_BALIGNW: 2029 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 2030 case DK_BALIGNL: 2031 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 2032 case DK_P2ALIGN: 2033 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 2034 case DK_P2ALIGNW: 2035 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 2036 case DK_P2ALIGNL: 2037 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 2038 case DK_ORG: 2039 return parseDirectiveOrg(); 2040 case DK_FILL: 2041 return parseDirectiveFill(); 2042 case DK_ZERO: 2043 return parseDirectiveZero(); 2044 case DK_EXTERN: 2045 eatToEndOfStatement(); // .extern is the default, ignore it. 2046 return false; 2047 case DK_GLOBL: 2048 case DK_GLOBAL: 2049 return parseDirectiveSymbolAttribute(MCSA_Global); 2050 case DK_LAZY_REFERENCE: 2051 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 2052 case DK_NO_DEAD_STRIP: 2053 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 2054 case DK_SYMBOL_RESOLVER: 2055 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 2056 case DK_PRIVATE_EXTERN: 2057 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 2058 case DK_REFERENCE: 2059 return parseDirectiveSymbolAttribute(MCSA_Reference); 2060 case DK_WEAK_DEFINITION: 2061 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 2062 case DK_WEAK_REFERENCE: 2063 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 2064 case DK_WEAK_DEF_CAN_BE_HIDDEN: 2065 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 2066 case DK_COLD: 2067 return parseDirectiveSymbolAttribute(MCSA_Cold); 2068 case DK_COMM: 2069 case DK_COMMON: 2070 return parseDirectiveComm(/*IsLocal=*/false); 2071 case DK_LCOMM: 2072 return parseDirectiveComm(/*IsLocal=*/true); 2073 case DK_ABORT: 2074 return parseDirectiveAbort(); 2075 case DK_INCLUDE: 2076 return parseDirectiveInclude(); 2077 case DK_INCBIN: 2078 return parseDirectiveIncbin(); 2079 case DK_CODE16: 2080 case DK_CODE16GCC: 2081 return TokError(Twine(IDVal) + 2082 " not currently supported for this target"); 2083 case DK_REPT: 2084 return parseDirectiveRept(IDLoc, IDVal); 2085 case DK_IRP: 2086 return parseDirectiveIrp(IDLoc); 2087 case DK_IRPC: 2088 return parseDirectiveIrpc(IDLoc); 2089 case DK_ENDR: 2090 return parseDirectiveEndr(IDLoc); 2091 case DK_BUNDLE_ALIGN_MODE: 2092 return parseDirectiveBundleAlignMode(); 2093 case DK_BUNDLE_LOCK: 2094 return parseDirectiveBundleLock(); 2095 case DK_BUNDLE_UNLOCK: 2096 return parseDirectiveBundleUnlock(); 2097 case DK_SLEB128: 2098 return parseDirectiveLEB128(true); 2099 case DK_ULEB128: 2100 return parseDirectiveLEB128(false); 2101 case DK_SPACE: 2102 case DK_SKIP: 2103 return parseDirectiveSpace(IDVal); 2104 case DK_FILE: 2105 return parseDirectiveFile(IDLoc); 2106 case DK_LINE: 2107 return parseDirectiveLine(); 2108 case DK_LOC: 2109 return parseDirectiveLoc(); 2110 case DK_STABS: 2111 return parseDirectiveStabs(); 2112 case DK_CV_FILE: 2113 return parseDirectiveCVFile(); 2114 case DK_CV_FUNC_ID: 2115 return parseDirectiveCVFuncId(); 2116 case DK_CV_INLINE_SITE_ID: 2117 return parseDirectiveCVInlineSiteId(); 2118 case DK_CV_LOC: 2119 return parseDirectiveCVLoc(); 2120 case DK_CV_LINETABLE: 2121 return parseDirectiveCVLinetable(); 2122 case DK_CV_INLINE_LINETABLE: 2123 return parseDirectiveCVInlineLinetable(); 2124 case DK_CV_DEF_RANGE: 2125 return parseDirectiveCVDefRange(); 2126 case DK_CV_STRING: 2127 return parseDirectiveCVString(); 2128 case DK_CV_STRINGTABLE: 2129 return parseDirectiveCVStringTable(); 2130 case DK_CV_FILECHECKSUMS: 2131 return parseDirectiveCVFileChecksums(); 2132 case DK_CV_FILECHECKSUM_OFFSET: 2133 return parseDirectiveCVFileChecksumOffset(); 2134 case DK_CV_FPO_DATA: 2135 return parseDirectiveCVFPOData(); 2136 case DK_CFI_SECTIONS: 2137 return parseDirectiveCFISections(); 2138 case DK_CFI_STARTPROC: 2139 return parseDirectiveCFIStartProc(); 2140 case DK_CFI_ENDPROC: 2141 return parseDirectiveCFIEndProc(); 2142 case DK_CFI_DEF_CFA: 2143 return parseDirectiveCFIDefCfa(IDLoc); 2144 case DK_CFI_DEF_CFA_OFFSET: 2145 return parseDirectiveCFIDefCfaOffset(); 2146 case DK_CFI_ADJUST_CFA_OFFSET: 2147 return parseDirectiveCFIAdjustCfaOffset(); 2148 case DK_CFI_DEF_CFA_REGISTER: 2149 return parseDirectiveCFIDefCfaRegister(IDLoc); 2150 case DK_CFI_OFFSET: 2151 return parseDirectiveCFIOffset(IDLoc); 2152 case DK_CFI_REL_OFFSET: 2153 return parseDirectiveCFIRelOffset(IDLoc); 2154 case DK_CFI_PERSONALITY: 2155 return parseDirectiveCFIPersonalityOrLsda(true); 2156 case DK_CFI_LSDA: 2157 return parseDirectiveCFIPersonalityOrLsda(false); 2158 case DK_CFI_REMEMBER_STATE: 2159 return parseDirectiveCFIRememberState(); 2160 case DK_CFI_RESTORE_STATE: 2161 return parseDirectiveCFIRestoreState(); 2162 case DK_CFI_SAME_VALUE: 2163 return parseDirectiveCFISameValue(IDLoc); 2164 case DK_CFI_RESTORE: 2165 return parseDirectiveCFIRestore(IDLoc); 2166 case DK_CFI_ESCAPE: 2167 return parseDirectiveCFIEscape(); 2168 case DK_CFI_RETURN_COLUMN: 2169 return parseDirectiveCFIReturnColumn(IDLoc); 2170 case DK_CFI_SIGNAL_FRAME: 2171 return parseDirectiveCFISignalFrame(); 2172 case DK_CFI_UNDEFINED: 2173 return parseDirectiveCFIUndefined(IDLoc); 2174 case DK_CFI_REGISTER: 2175 return parseDirectiveCFIRegister(IDLoc); 2176 case DK_CFI_WINDOW_SAVE: 2177 return parseDirectiveCFIWindowSave(); 2178 case DK_MACROS_ON: 2179 case DK_MACROS_OFF: 2180 return parseDirectiveMacrosOnOff(IDVal); 2181 case DK_MACRO: 2182 return parseDirectiveMacro(IDLoc); 2183 case DK_ALTMACRO: 2184 case DK_NOALTMACRO: 2185 return parseDirectiveAltmacro(IDVal); 2186 case DK_EXITM: 2187 return parseDirectiveExitMacro(IDVal); 2188 case DK_ENDM: 2189 case DK_ENDMACRO: 2190 return parseDirectiveEndMacro(IDVal); 2191 case DK_PURGEM: 2192 return parseDirectivePurgeMacro(IDLoc); 2193 case DK_END: 2194 return parseDirectiveEnd(IDLoc); 2195 case DK_ERR: 2196 return parseDirectiveError(IDLoc, false); 2197 case DK_ERROR: 2198 return parseDirectiveError(IDLoc, true); 2199 case DK_WARNING: 2200 return parseDirectiveWarning(IDLoc); 2201 case DK_RELOC: 2202 return parseDirectiveReloc(IDLoc); 2203 case DK_DCB: 2204 case DK_DCB_W: 2205 return parseDirectiveDCB(IDVal, 2); 2206 case DK_DCB_B: 2207 return parseDirectiveDCB(IDVal, 1); 2208 case DK_DCB_D: 2209 return parseDirectiveRealDCB(IDVal, APFloat::IEEEdouble()); 2210 case DK_DCB_L: 2211 return parseDirectiveDCB(IDVal, 4); 2212 case DK_DCB_S: 2213 return parseDirectiveRealDCB(IDVal, APFloat::IEEEsingle()); 2214 case DK_DC_X: 2215 case DK_DCB_X: 2216 return TokError(Twine(IDVal) + 2217 " not currently supported for this target"); 2218 case DK_DS: 2219 case DK_DS_W: 2220 return parseDirectiveDS(IDVal, 2); 2221 case DK_DS_B: 2222 return parseDirectiveDS(IDVal, 1); 2223 case DK_DS_D: 2224 return parseDirectiveDS(IDVal, 8); 2225 case DK_DS_L: 2226 case DK_DS_S: 2227 return parseDirectiveDS(IDVal, 4); 2228 case DK_DS_P: 2229 case DK_DS_X: 2230 return parseDirectiveDS(IDVal, 12); 2231 case DK_PRINT: 2232 return parseDirectivePrint(IDLoc); 2233 case DK_ADDRSIG: 2234 return parseDirectiveAddrsig(); 2235 case DK_ADDRSIG_SYM: 2236 return parseDirectiveAddrsigSym(); 2237 case DK_PSEUDO_PROBE: 2238 return parseDirectivePseudoProbe(); 2239 case DK_LTO_DISCARD: 2240 return parseDirectiveLTODiscard(); 2241 } 2242 2243 return Error(IDLoc, "unknown directive"); 2244 } 2245 2246 // __asm _emit or __asm __emit 2247 if (ParsingMSInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 2248 IDVal == "_EMIT" || IDVal == "__EMIT")) 2249 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 2250 2251 // __asm align 2252 if (ParsingMSInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 2253 return parseDirectiveMSAlign(IDLoc, Info); 2254 2255 if (ParsingMSInlineAsm && (IDVal == "even" || IDVal == "EVEN")) 2256 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 2257 if (checkForValidSection()) 2258 return true; 2259 2260 // Canonicalize the opcode to lower case. 2261 std::string OpcodeStr = IDVal.lower(); 2262 ParseInstructionInfo IInfo(Info.AsmRewrites); 2263 bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 2264 Info.ParsedOperands); 2265 Info.ParseError = ParseHadError; 2266 2267 // Dump the parsed representation, if requested. 2268 if (getShowParsedOperands()) { 2269 SmallString<256> Str; 2270 raw_svector_ostream OS(Str); 2271 OS << "parsed instruction: ["; 2272 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 2273 if (i != 0) 2274 OS << ", "; 2275 Info.ParsedOperands[i]->print(OS); 2276 } 2277 OS << "]"; 2278 2279 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 2280 } 2281 2282 // Fail even if ParseInstruction erroneously returns false. 2283 if (hasPendingError() || ParseHadError) 2284 return true; 2285 2286 // If we are generating dwarf for the current section then generate a .loc 2287 // directive for the instruction. 2288 if (!ParseHadError && enabledGenDwarfForAssembly() && 2289 getContext().getGenDwarfSectionSyms().count( 2290 getStreamer().getCurrentSectionOnly())) { 2291 unsigned Line; 2292 if (ActiveMacros.empty()) 2293 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 2294 else 2295 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 2296 ActiveMacros.front()->ExitBuffer); 2297 2298 // If we previously parsed a cpp hash file line comment then make sure the 2299 // current Dwarf File is for the CppHashFilename if not then emit the 2300 // Dwarf File table for it and adjust the line number for the .loc. 2301 if (!CppHashInfo.Filename.empty()) { 2302 unsigned FileNumber = getStreamer().emitDwarfFileDirective( 2303 0, StringRef(), CppHashInfo.Filename); 2304 getContext().setGenDwarfFileNumber(FileNumber); 2305 2306 unsigned CppHashLocLineNo = 2307 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 2308 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 2309 } 2310 2311 getStreamer().emitDwarfLocDirective( 2312 getContext().getGenDwarfFileNumber(), Line, 0, 2313 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 2314 StringRef()); 2315 } 2316 2317 // If parsing succeeded, match the instruction. 2318 if (!ParseHadError) { 2319 uint64_t ErrorInfo; 2320 if (getTargetParser().MatchAndEmitInstruction( 2321 IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo, 2322 getTargetParser().isParsingMSInlineAsm())) 2323 return true; 2324 } 2325 return false; 2326 } 2327 2328 // Parse and erase curly braces marking block start/end 2329 bool 2330 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 2331 // Identify curly brace marking block start/end 2332 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 2333 return false; 2334 2335 SMLoc StartLoc = Lexer.getLoc(); 2336 Lex(); // Eat the brace 2337 if (Lexer.is(AsmToken::EndOfStatement)) 2338 Lex(); // Eat EndOfStatement following the brace 2339 2340 // Erase the block start/end brace from the output asm string 2341 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 2342 StartLoc.getPointer()); 2343 return true; 2344 } 2345 2346 /// parseCppHashLineFilenameComment as this: 2347 /// ::= # number "filename" 2348 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L, bool SaveLocInfo) { 2349 Lex(); // Eat the hash token. 2350 // Lexer only ever emits HashDirective if it fully formed if it's 2351 // done the checking already so this is an internal error. 2352 assert(getTok().is(AsmToken::Integer) && 2353 "Lexing Cpp line comment: Expected Integer"); 2354 int64_t LineNumber = getTok().getIntVal(); 2355 Lex(); 2356 assert(getTok().is(AsmToken::String) && 2357 "Lexing Cpp line comment: Expected String"); 2358 StringRef Filename = getTok().getString(); 2359 Lex(); 2360 2361 if (!SaveLocInfo) 2362 return false; 2363 2364 // Get rid of the enclosing quotes. 2365 Filename = Filename.substr(1, Filename.size() - 2); 2366 2367 // Save the SMLoc, Filename and LineNumber for later use by diagnostics 2368 // and possibly DWARF file info. 2369 CppHashInfo.Loc = L; 2370 CppHashInfo.Filename = Filename; 2371 CppHashInfo.LineNumber = LineNumber; 2372 CppHashInfo.Buf = CurBuffer; 2373 if (FirstCppHashFilename.empty()) 2374 FirstCppHashFilename = Filename; 2375 return false; 2376 } 2377 2378 /// will use the last parsed cpp hash line filename comment 2379 /// for the Filename and LineNo if any in the diagnostic. 2380 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 2381 auto *Parser = static_cast<AsmParser *>(Context); 2382 raw_ostream &OS = errs(); 2383 2384 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 2385 SMLoc DiagLoc = Diag.getLoc(); 2386 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2387 unsigned CppHashBuf = 2388 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 2389 2390 // Like SourceMgr::printMessage() we need to print the include stack if any 2391 // before printing the message. 2392 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2393 if (!Parser->SavedDiagHandler && DiagCurBuffer && 2394 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 2395 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 2396 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 2397 } 2398 2399 // If we have not parsed a cpp hash line filename comment or the source 2400 // manager changed or buffer changed (like in a nested include) then just 2401 // print the normal diagnostic using its Filename and LineNo. 2402 if (!Parser->CppHashInfo.LineNumber || DiagBuf != CppHashBuf) { 2403 if (Parser->SavedDiagHandler) 2404 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2405 else 2406 Parser->getContext().diagnose(Diag); 2407 return; 2408 } 2409 2410 // Use the CppHashFilename and calculate a line number based on the 2411 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 2412 // for the diagnostic. 2413 const std::string &Filename = std::string(Parser->CppHashInfo.Filename); 2414 2415 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 2416 int CppHashLocLineNo = 2417 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 2418 int LineNo = 2419 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 2420 2421 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 2422 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 2423 Diag.getLineContents(), Diag.getRanges()); 2424 2425 if (Parser->SavedDiagHandler) 2426 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2427 else 2428 Parser->getContext().diagnose(NewDiag); 2429 } 2430 2431 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2432 // difference being that that function accepts '@' as part of identifiers and 2433 // we can't do that. AsmLexer.cpp should probably be changed to handle 2434 // '@' as a special case when needed. 2435 static bool isIdentifierChar(char c) { 2436 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2437 c == '.'; 2438 } 2439 2440 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2441 ArrayRef<MCAsmMacroParameter> Parameters, 2442 ArrayRef<MCAsmMacroArgument> A, 2443 bool EnableAtPseudoVariable, SMLoc L) { 2444 unsigned NParameters = Parameters.size(); 2445 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2446 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2447 return Error(L, "Wrong number of arguments"); 2448 2449 // A macro without parameters is handled differently on Darwin: 2450 // gas accepts no arguments and does no substitutions 2451 while (!Body.empty()) { 2452 // Scan for the next substitution. 2453 std::size_t End = Body.size(), Pos = 0; 2454 for (; Pos != End; ++Pos) { 2455 // Check for a substitution or escape. 2456 if (IsDarwin && !NParameters) { 2457 // This macro has no parameters, look for $0, $1, etc. 2458 if (Body[Pos] != '$' || Pos + 1 == End) 2459 continue; 2460 2461 char Next = Body[Pos + 1]; 2462 if (Next == '$' || Next == 'n' || 2463 isdigit(static_cast<unsigned char>(Next))) 2464 break; 2465 } else { 2466 // This macro has parameters, look for \foo, \bar, etc. 2467 if (Body[Pos] == '\\' && Pos + 1 != End) 2468 break; 2469 } 2470 } 2471 2472 // Add the prefix. 2473 OS << Body.slice(0, Pos); 2474 2475 // Check if we reached the end. 2476 if (Pos == End) 2477 break; 2478 2479 if (IsDarwin && !NParameters) { 2480 switch (Body[Pos + 1]) { 2481 // $$ => $ 2482 case '$': 2483 OS << '$'; 2484 break; 2485 2486 // $n => number of arguments 2487 case 'n': 2488 OS << A.size(); 2489 break; 2490 2491 // $[0-9] => argument 2492 default: { 2493 // Missing arguments are ignored. 2494 unsigned Index = Body[Pos + 1] - '0'; 2495 if (Index >= A.size()) 2496 break; 2497 2498 // Otherwise substitute with the token values, with spaces eliminated. 2499 for (const AsmToken &Token : A[Index]) 2500 OS << Token.getString(); 2501 break; 2502 } 2503 } 2504 Pos += 2; 2505 } else { 2506 unsigned I = Pos + 1; 2507 2508 // Check for the \@ pseudo-variable. 2509 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2510 ++I; 2511 else 2512 while (isIdentifierChar(Body[I]) && I + 1 != End) 2513 ++I; 2514 2515 const char *Begin = Body.data() + Pos + 1; 2516 StringRef Argument(Begin, I - (Pos + 1)); 2517 unsigned Index = 0; 2518 2519 if (Argument == "@") { 2520 OS << NumOfMacroInstantiations; 2521 Pos += 2; 2522 } else { 2523 for (; Index < NParameters; ++Index) 2524 if (Parameters[Index].Name == Argument) 2525 break; 2526 2527 if (Index == NParameters) { 2528 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2529 Pos += 3; 2530 else { 2531 OS << '\\' << Argument; 2532 Pos = I; 2533 } 2534 } else { 2535 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2536 for (const AsmToken &Token : A[Index]) 2537 // For altmacro mode, you can write '%expr'. 2538 // The prefix '%' evaluates the expression 'expr' 2539 // and uses the result as a string (e.g. replace %(1+2) with the 2540 // string "3"). 2541 // Here, we identify the integer token which is the result of the 2542 // absolute expression evaluation and replace it with its string 2543 // representation. 2544 if (AltMacroMode && Token.getString().front() == '%' && 2545 Token.is(AsmToken::Integer)) 2546 // Emit an integer value to the buffer. 2547 OS << Token.getIntVal(); 2548 // Only Token that was validated as a string and begins with '<' 2549 // is considered altMacroString!!! 2550 else if (AltMacroMode && Token.getString().front() == '<' && 2551 Token.is(AsmToken::String)) { 2552 OS << angleBracketString(Token.getStringContents()); 2553 } 2554 // We expect no quotes around the string's contents when 2555 // parsing for varargs. 2556 else if (Token.isNot(AsmToken::String) || VarargParameter) 2557 OS << Token.getString(); 2558 else 2559 OS << Token.getStringContents(); 2560 2561 Pos += 1 + Argument.size(); 2562 } 2563 } 2564 } 2565 // Update the scan point. 2566 Body = Body.substr(Pos); 2567 } 2568 2569 return false; 2570 } 2571 2572 static bool isOperator(AsmToken::TokenKind kind) { 2573 switch (kind) { 2574 default: 2575 return false; 2576 case AsmToken::Plus: 2577 case AsmToken::Minus: 2578 case AsmToken::Tilde: 2579 case AsmToken::Slash: 2580 case AsmToken::Star: 2581 case AsmToken::Dot: 2582 case AsmToken::Equal: 2583 case AsmToken::EqualEqual: 2584 case AsmToken::Pipe: 2585 case AsmToken::PipePipe: 2586 case AsmToken::Caret: 2587 case AsmToken::Amp: 2588 case AsmToken::AmpAmp: 2589 case AsmToken::Exclaim: 2590 case AsmToken::ExclaimEqual: 2591 case AsmToken::Less: 2592 case AsmToken::LessEqual: 2593 case AsmToken::LessLess: 2594 case AsmToken::LessGreater: 2595 case AsmToken::Greater: 2596 case AsmToken::GreaterEqual: 2597 case AsmToken::GreaterGreater: 2598 return true; 2599 } 2600 } 2601 2602 namespace { 2603 2604 class AsmLexerSkipSpaceRAII { 2605 public: 2606 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2607 Lexer.setSkipSpace(SkipSpace); 2608 } 2609 2610 ~AsmLexerSkipSpaceRAII() { 2611 Lexer.setSkipSpace(true); 2612 } 2613 2614 private: 2615 AsmLexer &Lexer; 2616 }; 2617 2618 } // end anonymous namespace 2619 2620 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2621 2622 if (Vararg) { 2623 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2624 StringRef Str = parseStringToEndOfStatement(); 2625 MA.emplace_back(AsmToken::String, Str); 2626 } 2627 return false; 2628 } 2629 2630 unsigned ParenLevel = 0; 2631 2632 // Darwin doesn't use spaces to delmit arguments. 2633 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2634 2635 bool SpaceEaten; 2636 2637 while (true) { 2638 SpaceEaten = false; 2639 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2640 return TokError("unexpected token in macro instantiation"); 2641 2642 if (ParenLevel == 0) { 2643 2644 if (Lexer.is(AsmToken::Comma)) 2645 break; 2646 2647 if (Lexer.is(AsmToken::Space)) { 2648 SpaceEaten = true; 2649 Lexer.Lex(); // Eat spaces 2650 } 2651 2652 // Spaces can delimit parameters, but could also be part an expression. 2653 // If the token after a space is an operator, add the token and the next 2654 // one into this argument 2655 if (!IsDarwin) { 2656 if (isOperator(Lexer.getKind())) { 2657 MA.push_back(getTok()); 2658 Lexer.Lex(); 2659 2660 // Whitespace after an operator can be ignored. 2661 if (Lexer.is(AsmToken::Space)) 2662 Lexer.Lex(); 2663 2664 continue; 2665 } 2666 } 2667 if (SpaceEaten) 2668 break; 2669 } 2670 2671 // handleMacroEntry relies on not advancing the lexer here 2672 // to be able to fill in the remaining default parameter values 2673 if (Lexer.is(AsmToken::EndOfStatement)) 2674 break; 2675 2676 // Adjust the current parentheses level. 2677 if (Lexer.is(AsmToken::LParen)) 2678 ++ParenLevel; 2679 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2680 --ParenLevel; 2681 2682 // Append the token to the current argument list. 2683 MA.push_back(getTok()); 2684 Lexer.Lex(); 2685 } 2686 2687 if (ParenLevel != 0) 2688 return TokError("unbalanced parentheses in macro argument"); 2689 return false; 2690 } 2691 2692 // Parse the macro instantiation arguments. 2693 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2694 MCAsmMacroArguments &A) { 2695 const unsigned NParameters = M ? M->Parameters.size() : 0; 2696 bool NamedParametersFound = false; 2697 SmallVector<SMLoc, 4> FALocs; 2698 2699 A.resize(NParameters); 2700 FALocs.resize(NParameters); 2701 2702 // Parse two kinds of macro invocations: 2703 // - macros defined without any parameters accept an arbitrary number of them 2704 // - macros defined with parameters accept at most that many of them 2705 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2706 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2707 ++Parameter) { 2708 SMLoc IDLoc = Lexer.getLoc(); 2709 MCAsmMacroParameter FA; 2710 2711 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2712 if (parseIdentifier(FA.Name)) 2713 return Error(IDLoc, "invalid argument identifier for formal argument"); 2714 2715 if (Lexer.isNot(AsmToken::Equal)) 2716 return TokError("expected '=' after formal parameter identifier"); 2717 2718 Lex(); 2719 2720 NamedParametersFound = true; 2721 } 2722 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2723 2724 if (NamedParametersFound && FA.Name.empty()) 2725 return Error(IDLoc, "cannot mix positional and keyword arguments"); 2726 2727 SMLoc StrLoc = Lexer.getLoc(); 2728 SMLoc EndLoc; 2729 if (AltMacroMode && Lexer.is(AsmToken::Percent)) { 2730 const MCExpr *AbsoluteExp; 2731 int64_t Value; 2732 /// Eat '%' 2733 Lex(); 2734 if (parseExpression(AbsoluteExp, EndLoc)) 2735 return false; 2736 if (!AbsoluteExp->evaluateAsAbsolute(Value, 2737 getStreamer().getAssemblerPtr())) 2738 return Error(StrLoc, "expected absolute expression"); 2739 const char *StrChar = StrLoc.getPointer(); 2740 const char *EndChar = EndLoc.getPointer(); 2741 AsmToken newToken(AsmToken::Integer, 2742 StringRef(StrChar, EndChar - StrChar), Value); 2743 FA.Value.push_back(newToken); 2744 } else if (AltMacroMode && Lexer.is(AsmToken::Less) && 2745 isAngleBracketString(StrLoc, EndLoc)) { 2746 const char *StrChar = StrLoc.getPointer(); 2747 const char *EndChar = EndLoc.getPointer(); 2748 jumpToLoc(EndLoc, CurBuffer); 2749 /// Eat from '<' to '>' 2750 Lex(); 2751 AsmToken newToken(AsmToken::String, 2752 StringRef(StrChar, EndChar - StrChar)); 2753 FA.Value.push_back(newToken); 2754 } else if(parseMacroArgument(FA.Value, Vararg)) 2755 return true; 2756 2757 unsigned PI = Parameter; 2758 if (!FA.Name.empty()) { 2759 unsigned FAI = 0; 2760 for (FAI = 0; FAI < NParameters; ++FAI) 2761 if (M->Parameters[FAI].Name == FA.Name) 2762 break; 2763 2764 if (FAI >= NParameters) { 2765 assert(M && "expected macro to be defined"); 2766 return Error(IDLoc, "parameter named '" + FA.Name + 2767 "' does not exist for macro '" + M->Name + "'"); 2768 } 2769 PI = FAI; 2770 } 2771 2772 if (!FA.Value.empty()) { 2773 if (A.size() <= PI) 2774 A.resize(PI + 1); 2775 A[PI] = FA.Value; 2776 2777 if (FALocs.size() <= PI) 2778 FALocs.resize(PI + 1); 2779 2780 FALocs[PI] = Lexer.getLoc(); 2781 } 2782 2783 // At the end of the statement, fill in remaining arguments that have 2784 // default values. If there aren't any, then the next argument is 2785 // required but missing 2786 if (Lexer.is(AsmToken::EndOfStatement)) { 2787 bool Failure = false; 2788 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2789 if (A[FAI].empty()) { 2790 if (M->Parameters[FAI].Required) { 2791 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2792 "missing value for required parameter " 2793 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2794 Failure = true; 2795 } 2796 2797 if (!M->Parameters[FAI].Value.empty()) 2798 A[FAI] = M->Parameters[FAI].Value; 2799 } 2800 } 2801 return Failure; 2802 } 2803 2804 if (Lexer.is(AsmToken::Comma)) 2805 Lex(); 2806 } 2807 2808 return TokError("too many positional arguments"); 2809 } 2810 2811 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2812 // Arbitrarily limit macro nesting depth (default matches 'as'). We can 2813 // eliminate this, although we should protect against infinite loops. 2814 unsigned MaxNestingDepth = AsmMacroMaxNestingDepth; 2815 if (ActiveMacros.size() == MaxNestingDepth) { 2816 std::ostringstream MaxNestingDepthError; 2817 MaxNestingDepthError << "macros cannot be nested more than " 2818 << MaxNestingDepth << " levels deep." 2819 << " Use -asm-macro-max-nesting-depth to increase " 2820 "this limit."; 2821 return TokError(MaxNestingDepthError.str()); 2822 } 2823 2824 MCAsmMacroArguments A; 2825 if (parseMacroArguments(M, A)) 2826 return true; 2827 2828 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2829 // to hold the macro body with substitutions. 2830 SmallString<256> Buf; 2831 StringRef Body = M->Body; 2832 raw_svector_ostream OS(Buf); 2833 2834 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2835 return true; 2836 2837 // We include the .endmacro in the buffer as our cue to exit the macro 2838 // instantiation. 2839 OS << ".endmacro\n"; 2840 2841 std::unique_ptr<MemoryBuffer> Instantiation = 2842 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2843 2844 // Create the macro instantiation object and add to the current macro 2845 // instantiation stack. 2846 MacroInstantiation *MI = new MacroInstantiation{ 2847 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 2848 ActiveMacros.push_back(MI); 2849 2850 ++NumOfMacroInstantiations; 2851 2852 // Jump to the macro instantiation and prime the lexer. 2853 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2854 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2855 Lex(); 2856 2857 return false; 2858 } 2859 2860 void AsmParser::handleMacroExit() { 2861 // Jump to the EndOfStatement we should return to, and consume it. 2862 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2863 Lex(); 2864 2865 // Pop the instantiation entry. 2866 delete ActiveMacros.back(); 2867 ActiveMacros.pop_back(); 2868 } 2869 2870 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2871 bool NoDeadStrip) { 2872 MCSymbol *Sym; 2873 const MCExpr *Value; 2874 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2875 Value)) 2876 return true; 2877 2878 if (!Sym) { 2879 // In the case where we parse an expression starting with a '.', we will 2880 // not generate an error, nor will we create a symbol. In this case we 2881 // should just return out. 2882 return false; 2883 } 2884 2885 if (discardLTOSymbol(Name)) 2886 return false; 2887 2888 // Do the assignment. 2889 Out.emitAssignment(Sym, Value); 2890 if (NoDeadStrip) 2891 Out.emitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2892 2893 return false; 2894 } 2895 2896 /// parseIdentifier: 2897 /// ::= identifier 2898 /// ::= string 2899 bool AsmParser::parseIdentifier(StringRef &Res) { 2900 // The assembler has relaxed rules for accepting identifiers, in particular we 2901 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2902 // separate tokens. At this level, we have already lexed so we cannot (currently) 2903 // handle this as a context dependent token, instead we detect adjacent tokens 2904 // and return the combined identifier. 2905 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2906 SMLoc PrefixLoc = getLexer().getLoc(); 2907 2908 // Consume the prefix character, and check for a following identifier. 2909 2910 AsmToken Buf[1]; 2911 Lexer.peekTokens(Buf, false); 2912 2913 if (Buf[0].isNot(AsmToken::Identifier) && Buf[0].isNot(AsmToken::Integer)) 2914 return true; 2915 2916 // We have a '$' or '@' followed by an identifier or integer token, make 2917 // sure they are adjacent. 2918 if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer()) 2919 return true; 2920 2921 // eat $ or @ 2922 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2923 // Construct the joined identifier and consume the token. 2924 Res = StringRef(PrefixLoc.getPointer(), getTok().getString().size() + 1); 2925 Lex(); // Parser Lex to maintain invariants. 2926 return false; 2927 } 2928 2929 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2930 return true; 2931 2932 Res = getTok().getIdentifier(); 2933 2934 Lex(); // Consume the identifier token. 2935 2936 return false; 2937 } 2938 2939 /// parseDirectiveSet: 2940 /// ::= .equ identifier ',' expression 2941 /// ::= .equiv identifier ',' expression 2942 /// ::= .set identifier ',' expression 2943 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2944 StringRef Name; 2945 if (check(parseIdentifier(Name), "expected identifier") || parseComma() || 2946 parseAssignment(Name, allow_redef, true)) 2947 return true; 2948 return false; 2949 } 2950 2951 bool AsmParser::parseEscapedString(std::string &Data) { 2952 if (check(getTok().isNot(AsmToken::String), "expected string")) 2953 return true; 2954 2955 Data = ""; 2956 StringRef Str = getTok().getStringContents(); 2957 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2958 if (Str[i] != '\\') { 2959 Data += Str[i]; 2960 continue; 2961 } 2962 2963 // Recognize escaped characters. Note that this escape semantics currently 2964 // loosely follows Darwin 'as'. 2965 ++i; 2966 if (i == e) 2967 return TokError("unexpected backslash at end of string"); 2968 2969 // Recognize hex sequences similarly to GNU 'as'. 2970 if (Str[i] == 'x' || Str[i] == 'X') { 2971 size_t length = Str.size(); 2972 if (i + 1 >= length || !isHexDigit(Str[i + 1])) 2973 return TokError("invalid hexadecimal escape sequence"); 2974 2975 // Consume hex characters. GNU 'as' reads all hexadecimal characters and 2976 // then truncates to the lower 16 bits. Seems reasonable. 2977 unsigned Value = 0; 2978 while (i + 1 < length && isHexDigit(Str[i + 1])) 2979 Value = Value * 16 + hexDigitValue(Str[++i]); 2980 2981 Data += (unsigned char)(Value & 0xFF); 2982 continue; 2983 } 2984 2985 // Recognize octal sequences. 2986 if ((unsigned)(Str[i] - '0') <= 7) { 2987 // Consume up to three octal characters. 2988 unsigned Value = Str[i] - '0'; 2989 2990 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2991 ++i; 2992 Value = Value * 8 + (Str[i] - '0'); 2993 2994 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2995 ++i; 2996 Value = Value * 8 + (Str[i] - '0'); 2997 } 2998 } 2999 3000 if (Value > 255) 3001 return TokError("invalid octal escape sequence (out of range)"); 3002 3003 Data += (unsigned char)Value; 3004 continue; 3005 } 3006 3007 // Otherwise recognize individual escapes. 3008 switch (Str[i]) { 3009 default: 3010 // Just reject invalid escape sequences for now. 3011 return TokError("invalid escape sequence (unrecognized character)"); 3012 3013 case 'b': Data += '\b'; break; 3014 case 'f': Data += '\f'; break; 3015 case 'n': Data += '\n'; break; 3016 case 'r': Data += '\r'; break; 3017 case 't': Data += '\t'; break; 3018 case '"': Data += '"'; break; 3019 case '\\': Data += '\\'; break; 3020 } 3021 } 3022 3023 Lex(); 3024 return false; 3025 } 3026 3027 bool AsmParser::parseAngleBracketString(std::string &Data) { 3028 SMLoc EndLoc, StartLoc = getTok().getLoc(); 3029 if (isAngleBracketString(StartLoc, EndLoc)) { 3030 const char *StartChar = StartLoc.getPointer() + 1; 3031 const char *EndChar = EndLoc.getPointer() - 1; 3032 jumpToLoc(EndLoc, CurBuffer); 3033 /// Eat from '<' to '>' 3034 Lex(); 3035 3036 Data = angleBracketString(StringRef(StartChar, EndChar - StartChar)); 3037 return false; 3038 } 3039 return true; 3040 } 3041 3042 /// parseDirectiveAscii: 3043 // ::= .ascii [ "string"+ ( , "string"+ )* ] 3044 /// ::= ( .asciz | .string ) [ "string" ( , "string" )* ] 3045 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 3046 auto parseOp = [&]() -> bool { 3047 std::string Data; 3048 if (checkForValidSection()) 3049 return true; 3050 // Only support spaces as separators for .ascii directive for now. See the 3051 // discusssion at https://reviews.llvm.org/D91460 for more details. 3052 do { 3053 if (parseEscapedString(Data)) 3054 return true; 3055 getStreamer().emitBytes(Data); 3056 } while (!ZeroTerminated && getTok().is(AsmToken::String)); 3057 if (ZeroTerminated) 3058 getStreamer().emitBytes(StringRef("\0", 1)); 3059 return false; 3060 }; 3061 3062 return parseMany(parseOp); 3063 } 3064 3065 /// parseDirectiveReloc 3066 /// ::= .reloc expression , identifier [ , expression ] 3067 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 3068 const MCExpr *Offset; 3069 const MCExpr *Expr = nullptr; 3070 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 3071 3072 if (parseExpression(Offset)) 3073 return true; 3074 if (parseComma() || 3075 check(getTok().isNot(AsmToken::Identifier), "expected relocation name")) 3076 return true; 3077 3078 SMLoc NameLoc = Lexer.getTok().getLoc(); 3079 StringRef Name = Lexer.getTok().getIdentifier(); 3080 Lex(); 3081 3082 if (Lexer.is(AsmToken::Comma)) { 3083 Lex(); 3084 SMLoc ExprLoc = Lexer.getLoc(); 3085 if (parseExpression(Expr)) 3086 return true; 3087 3088 MCValue Value; 3089 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 3090 return Error(ExprLoc, "expression must be relocatable"); 3091 } 3092 3093 if (parseEOL()) 3094 return true; 3095 3096 const MCTargetAsmParser &MCT = getTargetParser(); 3097 const MCSubtargetInfo &STI = MCT.getSTI(); 3098 if (Optional<std::pair<bool, std::string>> Err = 3099 getStreamer().emitRelocDirective(*Offset, Name, Expr, DirectiveLoc, 3100 STI)) 3101 return Error(Err->first ? NameLoc : OffsetLoc, Err->second); 3102 3103 return false; 3104 } 3105 3106 /// parseDirectiveValue 3107 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 3108 bool AsmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) { 3109 auto parseOp = [&]() -> bool { 3110 const MCExpr *Value; 3111 SMLoc ExprLoc = getLexer().getLoc(); 3112 if (checkForValidSection() || parseExpression(Value)) 3113 return true; 3114 // Special case constant expressions to match code generator. 3115 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3116 assert(Size <= 8 && "Invalid size"); 3117 uint64_t IntValue = MCE->getValue(); 3118 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 3119 return Error(ExprLoc, "out of range literal value"); 3120 getStreamer().emitIntValue(IntValue, Size); 3121 } else 3122 getStreamer().emitValue(Value, Size, ExprLoc); 3123 return false; 3124 }; 3125 3126 return parseMany(parseOp); 3127 } 3128 3129 static bool parseHexOcta(AsmParser &Asm, uint64_t &hi, uint64_t &lo) { 3130 if (Asm.getTok().isNot(AsmToken::Integer) && 3131 Asm.getTok().isNot(AsmToken::BigNum)) 3132 return Asm.TokError("unknown token in expression"); 3133 SMLoc ExprLoc = Asm.getTok().getLoc(); 3134 APInt IntValue = Asm.getTok().getAPIntVal(); 3135 Asm.Lex(); 3136 if (!IntValue.isIntN(128)) 3137 return Asm.Error(ExprLoc, "out of range literal value"); 3138 if (!IntValue.isIntN(64)) { 3139 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 3140 lo = IntValue.getLoBits(64).getZExtValue(); 3141 } else { 3142 hi = 0; 3143 lo = IntValue.getZExtValue(); 3144 } 3145 return false; 3146 } 3147 3148 /// ParseDirectiveOctaValue 3149 /// ::= .octa [ hexconstant (, hexconstant)* ] 3150 3151 bool AsmParser::parseDirectiveOctaValue(StringRef IDVal) { 3152 auto parseOp = [&]() -> bool { 3153 if (checkForValidSection()) 3154 return true; 3155 uint64_t hi, lo; 3156 if (parseHexOcta(*this, hi, lo)) 3157 return true; 3158 if (MAI.isLittleEndian()) { 3159 getStreamer().emitInt64(lo); 3160 getStreamer().emitInt64(hi); 3161 } else { 3162 getStreamer().emitInt64(hi); 3163 getStreamer().emitInt64(lo); 3164 } 3165 return false; 3166 }; 3167 3168 return parseMany(parseOp); 3169 } 3170 3171 bool AsmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) { 3172 // We don't truly support arithmetic on floating point expressions, so we 3173 // have to manually parse unary prefixes. 3174 bool IsNeg = false; 3175 if (getLexer().is(AsmToken::Minus)) { 3176 Lexer.Lex(); 3177 IsNeg = true; 3178 } else if (getLexer().is(AsmToken::Plus)) 3179 Lexer.Lex(); 3180 3181 if (Lexer.is(AsmToken::Error)) 3182 return TokError(Lexer.getErr()); 3183 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 3184 Lexer.isNot(AsmToken::Identifier)) 3185 return TokError("unexpected token in directive"); 3186 3187 // Convert to an APFloat. 3188 APFloat Value(Semantics); 3189 StringRef IDVal = getTok().getString(); 3190 if (getLexer().is(AsmToken::Identifier)) { 3191 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 3192 Value = APFloat::getInf(Semantics); 3193 else if (!IDVal.compare_lower("nan")) 3194 Value = APFloat::getNaN(Semantics, false, ~0); 3195 else 3196 return TokError("invalid floating point literal"); 3197 } else if (errorToBool( 3198 Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) 3199 .takeError())) 3200 return TokError("invalid floating point literal"); 3201 if (IsNeg) 3202 Value.changeSign(); 3203 3204 // Consume the numeric token. 3205 Lex(); 3206 3207 Res = Value.bitcastToAPInt(); 3208 3209 return false; 3210 } 3211 3212 /// parseDirectiveRealValue 3213 /// ::= (.single | .double) [ expression (, expression)* ] 3214 bool AsmParser::parseDirectiveRealValue(StringRef IDVal, 3215 const fltSemantics &Semantics) { 3216 auto parseOp = [&]() -> bool { 3217 APInt AsInt; 3218 if (checkForValidSection() || parseRealValue(Semantics, AsInt)) 3219 return true; 3220 getStreamer().emitIntValue(AsInt.getLimitedValue(), 3221 AsInt.getBitWidth() / 8); 3222 return false; 3223 }; 3224 3225 return parseMany(parseOp); 3226 } 3227 3228 /// parseDirectiveZero 3229 /// ::= .zero expression 3230 bool AsmParser::parseDirectiveZero() { 3231 SMLoc NumBytesLoc = Lexer.getLoc(); 3232 const MCExpr *NumBytes; 3233 if (checkForValidSection() || parseExpression(NumBytes)) 3234 return true; 3235 3236 int64_t Val = 0; 3237 if (getLexer().is(AsmToken::Comma)) { 3238 Lex(); 3239 if (parseAbsoluteExpression(Val)) 3240 return true; 3241 } 3242 3243 if (parseEOL()) 3244 return true; 3245 getStreamer().emitFill(*NumBytes, Val, NumBytesLoc); 3246 3247 return false; 3248 } 3249 3250 /// parseDirectiveFill 3251 /// ::= .fill expression [ , expression [ , expression ] ] 3252 bool AsmParser::parseDirectiveFill() { 3253 SMLoc NumValuesLoc = Lexer.getLoc(); 3254 const MCExpr *NumValues; 3255 if (checkForValidSection() || parseExpression(NumValues)) 3256 return true; 3257 3258 int64_t FillSize = 1; 3259 int64_t FillExpr = 0; 3260 3261 SMLoc SizeLoc, ExprLoc; 3262 3263 if (parseOptionalToken(AsmToken::Comma)) { 3264 SizeLoc = getTok().getLoc(); 3265 if (parseAbsoluteExpression(FillSize)) 3266 return true; 3267 if (parseOptionalToken(AsmToken::Comma)) { 3268 ExprLoc = getTok().getLoc(); 3269 if (parseAbsoluteExpression(FillExpr)) 3270 return true; 3271 } 3272 } 3273 if (parseEOL()) 3274 return true; 3275 3276 if (FillSize < 0) { 3277 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 3278 return false; 3279 } 3280 if (FillSize > 8) { 3281 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 3282 FillSize = 8; 3283 } 3284 3285 if (!isUInt<32>(FillExpr) && FillSize > 4) 3286 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 3287 3288 getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc); 3289 3290 return false; 3291 } 3292 3293 /// parseDirectiveOrg 3294 /// ::= .org expression [ , expression ] 3295 bool AsmParser::parseDirectiveOrg() { 3296 const MCExpr *Offset; 3297 SMLoc OffsetLoc = Lexer.getLoc(); 3298 if (checkForValidSection() || parseExpression(Offset)) 3299 return true; 3300 3301 // Parse optional fill expression. 3302 int64_t FillExpr = 0; 3303 if (parseOptionalToken(AsmToken::Comma)) 3304 if (parseAbsoluteExpression(FillExpr)) 3305 return true; 3306 if (parseEOL()) 3307 return true; 3308 3309 getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc); 3310 return false; 3311 } 3312 3313 /// parseDirectiveAlign 3314 /// ::= {.align, ...} expression [ , expression [ , expression ]] 3315 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 3316 SMLoc AlignmentLoc = getLexer().getLoc(); 3317 int64_t Alignment; 3318 SMLoc MaxBytesLoc; 3319 bool HasFillExpr = false; 3320 int64_t FillExpr = 0; 3321 int64_t MaxBytesToFill = 0; 3322 3323 auto parseAlign = [&]() -> bool { 3324 if (parseAbsoluteExpression(Alignment)) 3325 return true; 3326 if (parseOptionalToken(AsmToken::Comma)) { 3327 // The fill expression can be omitted while specifying a maximum number of 3328 // alignment bytes, e.g: 3329 // .align 3,,4 3330 if (getTok().isNot(AsmToken::Comma)) { 3331 HasFillExpr = true; 3332 if (parseAbsoluteExpression(FillExpr)) 3333 return true; 3334 } 3335 if (parseOptionalToken(AsmToken::Comma)) 3336 if (parseTokenLoc(MaxBytesLoc) || 3337 parseAbsoluteExpression(MaxBytesToFill)) 3338 return true; 3339 } 3340 return parseEOL(); 3341 }; 3342 3343 if (checkForValidSection()) 3344 return true; 3345 // Ignore empty '.p2align' directives for GNU-as compatibility 3346 if (IsPow2 && (ValueSize == 1) && getTok().is(AsmToken::EndOfStatement)) { 3347 Warning(AlignmentLoc, "p2align directive with no operand(s) is ignored"); 3348 return parseEOL(); 3349 } 3350 if (parseAlign()) 3351 return true; 3352 3353 // Always emit an alignment here even if we thrown an error. 3354 bool ReturnVal = false; 3355 3356 // Compute alignment in bytes. 3357 if (IsPow2) { 3358 // FIXME: Diagnose overflow. 3359 if (Alignment >= 32) { 3360 ReturnVal |= Error(AlignmentLoc, "invalid alignment value"); 3361 Alignment = 31; 3362 } 3363 3364 Alignment = 1ULL << Alignment; 3365 } else { 3366 // Reject alignments that aren't either a power of two or zero, 3367 // for gas compatibility. Alignment of zero is silently rounded 3368 // up to one. 3369 if (Alignment == 0) 3370 Alignment = 1; 3371 if (!isPowerOf2_64(Alignment)) 3372 ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2"); 3373 if (!isUInt<32>(Alignment)) 3374 ReturnVal |= Error(AlignmentLoc, "alignment must be smaller than 2**32"); 3375 } 3376 3377 // Diagnose non-sensical max bytes to align. 3378 if (MaxBytesLoc.isValid()) { 3379 if (MaxBytesToFill < 1) { 3380 ReturnVal |= Error(MaxBytesLoc, 3381 "alignment directive can never be satisfied in this " 3382 "many bytes, ignoring maximum bytes expression"); 3383 MaxBytesToFill = 0; 3384 } 3385 3386 if (MaxBytesToFill >= Alignment) { 3387 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 3388 "has no effect"); 3389 MaxBytesToFill = 0; 3390 } 3391 } 3392 3393 // Check whether we should use optimal code alignment for this .align 3394 // directive. 3395 const MCSection *Section = getStreamer().getCurrentSectionOnly(); 3396 assert(Section && "must have section to emit alignment"); 3397 bool UseCodeAlign = Section->UseCodeAlign(); 3398 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 3399 ValueSize == 1 && UseCodeAlign) { 3400 getStreamer().emitCodeAlignment(Alignment, MaxBytesToFill); 3401 } else { 3402 // FIXME: Target specific behavior about how the "extra" bytes are filled. 3403 getStreamer().emitValueToAlignment(Alignment, FillExpr, ValueSize, 3404 MaxBytesToFill); 3405 } 3406 3407 return ReturnVal; 3408 } 3409 3410 /// parseDirectiveFile 3411 /// ::= .file filename 3412 /// ::= .file number [directory] filename [md5 checksum] [source source-text] 3413 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 3414 // FIXME: I'm not sure what this is. 3415 int64_t FileNumber = -1; 3416 if (getLexer().is(AsmToken::Integer)) { 3417 FileNumber = getTok().getIntVal(); 3418 Lex(); 3419 3420 if (FileNumber < 0) 3421 return TokError("negative file number"); 3422 } 3423 3424 std::string Path; 3425 3426 // Usually the directory and filename together, otherwise just the directory. 3427 // Allow the strings to have escaped octal character sequence. 3428 if (parseEscapedString(Path)) 3429 return true; 3430 3431 StringRef Directory; 3432 StringRef Filename; 3433 std::string FilenameData; 3434 if (getLexer().is(AsmToken::String)) { 3435 if (check(FileNumber == -1, 3436 "explicit path specified, but no file number") || 3437 parseEscapedString(FilenameData)) 3438 return true; 3439 Filename = FilenameData; 3440 Directory = Path; 3441 } else { 3442 Filename = Path; 3443 } 3444 3445 uint64_t MD5Hi, MD5Lo; 3446 bool HasMD5 = false; 3447 3448 Optional<StringRef> Source; 3449 bool HasSource = false; 3450 std::string SourceString; 3451 3452 while (!parseOptionalToken(AsmToken::EndOfStatement)) { 3453 StringRef Keyword; 3454 if (check(getTok().isNot(AsmToken::Identifier), 3455 "unexpected token in '.file' directive") || 3456 parseIdentifier(Keyword)) 3457 return true; 3458 if (Keyword == "md5") { 3459 HasMD5 = true; 3460 if (check(FileNumber == -1, 3461 "MD5 checksum specified, but no file number") || 3462 parseHexOcta(*this, MD5Hi, MD5Lo)) 3463 return true; 3464 } else if (Keyword == "source") { 3465 HasSource = true; 3466 if (check(FileNumber == -1, 3467 "source specified, but no file number") || 3468 check(getTok().isNot(AsmToken::String), 3469 "unexpected token in '.file' directive") || 3470 parseEscapedString(SourceString)) 3471 return true; 3472 } else { 3473 return TokError("unexpected token in '.file' directive"); 3474 } 3475 } 3476 3477 if (FileNumber == -1) { 3478 // Ignore the directive if there is no number and the target doesn't support 3479 // numberless .file directives. This allows some portability of assembler 3480 // between different object file formats. 3481 if (getContext().getAsmInfo()->hasSingleParameterDotFile()) 3482 getStreamer().emitFileDirective(Filename); 3483 } else { 3484 // In case there is a -g option as well as debug info from directive .file, 3485 // we turn off the -g option, directly use the existing debug info instead. 3486 // Throw away any implicit file table for the assembler source. 3487 if (Ctx.getGenDwarfForAssembly()) { 3488 Ctx.getMCDwarfLineTable(0).resetFileTable(); 3489 Ctx.setGenDwarfForAssembly(false); 3490 } 3491 3492 Optional<MD5::MD5Result> CKMem; 3493 if (HasMD5) { 3494 MD5::MD5Result Sum; 3495 for (unsigned i = 0; i != 8; ++i) { 3496 Sum.Bytes[i] = uint8_t(MD5Hi >> ((7 - i) * 8)); 3497 Sum.Bytes[i + 8] = uint8_t(MD5Lo >> ((7 - i) * 8)); 3498 } 3499 CKMem = Sum; 3500 } 3501 if (HasSource) { 3502 char *SourceBuf = static_cast<char *>(Ctx.allocate(SourceString.size())); 3503 memcpy(SourceBuf, SourceString.data(), SourceString.size()); 3504 Source = StringRef(SourceBuf, SourceString.size()); 3505 } 3506 if (FileNumber == 0) { 3507 // Upgrade to Version 5 for assembly actions like clang -c a.s. 3508 if (Ctx.getDwarfVersion() < 5) 3509 Ctx.setDwarfVersion(5); 3510 getStreamer().emitDwarfFile0Directive(Directory, Filename, CKMem, Source); 3511 } else { 3512 Expected<unsigned> FileNumOrErr = getStreamer().tryEmitDwarfFileDirective( 3513 FileNumber, Directory, Filename, CKMem, Source); 3514 if (!FileNumOrErr) 3515 return Error(DirectiveLoc, toString(FileNumOrErr.takeError())); 3516 } 3517 // Alert the user if there are some .file directives with MD5 and some not. 3518 // But only do that once. 3519 if (!ReportedInconsistentMD5 && !Ctx.isDwarfMD5UsageConsistent(0)) { 3520 ReportedInconsistentMD5 = true; 3521 return Warning(DirectiveLoc, "inconsistent use of MD5 checksums"); 3522 } 3523 } 3524 3525 return false; 3526 } 3527 3528 /// parseDirectiveLine 3529 /// ::= .line [number] 3530 bool AsmParser::parseDirectiveLine() { 3531 int64_t LineNumber; 3532 if (getLexer().is(AsmToken::Integer)) { 3533 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3534 return true; 3535 (void)LineNumber; 3536 // FIXME: Do something with the .line. 3537 } 3538 return parseEOL(); 3539 } 3540 3541 /// parseDirectiveLoc 3542 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3543 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3544 /// The first number is a file number, must have been previously assigned with 3545 /// a .file directive, the second number is the line number and optionally the 3546 /// third number is a column position (zero if not specified). The remaining 3547 /// optional items are .loc sub-directives. 3548 bool AsmParser::parseDirectiveLoc() { 3549 int64_t FileNumber = 0, LineNumber = 0; 3550 SMLoc Loc = getTok().getLoc(); 3551 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3552 check(FileNumber < 1 && Ctx.getDwarfVersion() < 5, Loc, 3553 "file number less than one in '.loc' directive") || 3554 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3555 "unassigned file number in '.loc' directive")) 3556 return true; 3557 3558 // optional 3559 if (getLexer().is(AsmToken::Integer)) { 3560 LineNumber = getTok().getIntVal(); 3561 if (LineNumber < 0) 3562 return TokError("line number less than zero in '.loc' directive"); 3563 Lex(); 3564 } 3565 3566 int64_t ColumnPos = 0; 3567 if (getLexer().is(AsmToken::Integer)) { 3568 ColumnPos = getTok().getIntVal(); 3569 if (ColumnPos < 0) 3570 return TokError("column position less than zero in '.loc' directive"); 3571 Lex(); 3572 } 3573 3574 auto PrevFlags = getContext().getCurrentDwarfLoc().getFlags(); 3575 unsigned Flags = PrevFlags & DWARF2_FLAG_IS_STMT; 3576 unsigned Isa = 0; 3577 int64_t Discriminator = 0; 3578 3579 auto parseLocOp = [&]() -> bool { 3580 StringRef Name; 3581 SMLoc Loc = getTok().getLoc(); 3582 if (parseIdentifier(Name)) 3583 return TokError("unexpected token in '.loc' directive"); 3584 3585 if (Name == "basic_block") 3586 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3587 else if (Name == "prologue_end") 3588 Flags |= DWARF2_FLAG_PROLOGUE_END; 3589 else if (Name == "epilogue_begin") 3590 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3591 else if (Name == "is_stmt") { 3592 Loc = getTok().getLoc(); 3593 const MCExpr *Value; 3594 if (parseExpression(Value)) 3595 return true; 3596 // The expression must be the constant 0 or 1. 3597 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3598 int Value = MCE->getValue(); 3599 if (Value == 0) 3600 Flags &= ~DWARF2_FLAG_IS_STMT; 3601 else if (Value == 1) 3602 Flags |= DWARF2_FLAG_IS_STMT; 3603 else 3604 return Error(Loc, "is_stmt value not 0 or 1"); 3605 } else { 3606 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3607 } 3608 } else if (Name == "isa") { 3609 Loc = getTok().getLoc(); 3610 const MCExpr *Value; 3611 if (parseExpression(Value)) 3612 return true; 3613 // The expression must be a constant greater or equal to 0. 3614 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3615 int Value = MCE->getValue(); 3616 if (Value < 0) 3617 return Error(Loc, "isa number less than zero"); 3618 Isa = Value; 3619 } else { 3620 return Error(Loc, "isa number not a constant value"); 3621 } 3622 } else if (Name == "discriminator") { 3623 if (parseAbsoluteExpression(Discriminator)) 3624 return true; 3625 } else { 3626 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3627 } 3628 return false; 3629 }; 3630 3631 if (parseMany(parseLocOp, false /*hasComma*/)) 3632 return true; 3633 3634 getStreamer().emitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3635 Isa, Discriminator, StringRef()); 3636 3637 return false; 3638 } 3639 3640 /// parseDirectiveStabs 3641 /// ::= .stabs string, number, number, number 3642 bool AsmParser::parseDirectiveStabs() { 3643 return TokError("unsupported directive '.stabs'"); 3644 } 3645 3646 /// parseDirectiveCVFile 3647 /// ::= .cv_file number filename [checksum] [checksumkind] 3648 bool AsmParser::parseDirectiveCVFile() { 3649 SMLoc FileNumberLoc = getTok().getLoc(); 3650 int64_t FileNumber; 3651 std::string Filename; 3652 std::string Checksum; 3653 int64_t ChecksumKind = 0; 3654 3655 if (parseIntToken(FileNumber, 3656 "expected file number in '.cv_file' directive") || 3657 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3658 check(getTok().isNot(AsmToken::String), 3659 "unexpected token in '.cv_file' directive") || 3660 parseEscapedString(Filename)) 3661 return true; 3662 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3663 if (check(getTok().isNot(AsmToken::String), 3664 "unexpected token in '.cv_file' directive") || 3665 parseEscapedString(Checksum) || 3666 parseIntToken(ChecksumKind, 3667 "expected checksum kind in '.cv_file' directive") || 3668 parseToken(AsmToken::EndOfStatement, 3669 "unexpected token in '.cv_file' directive")) 3670 return true; 3671 } 3672 3673 Checksum = fromHex(Checksum); 3674 void *CKMem = Ctx.allocate(Checksum.size(), 1); 3675 memcpy(CKMem, Checksum.data(), Checksum.size()); 3676 ArrayRef<uint8_t> ChecksumAsBytes(reinterpret_cast<const uint8_t *>(CKMem), 3677 Checksum.size()); 3678 3679 if (!getStreamer().EmitCVFileDirective(FileNumber, Filename, ChecksumAsBytes, 3680 static_cast<uint8_t>(ChecksumKind))) 3681 return Error(FileNumberLoc, "file number already allocated"); 3682 3683 return false; 3684 } 3685 3686 bool AsmParser::parseCVFunctionId(int64_t &FunctionId, 3687 StringRef DirectiveName) { 3688 SMLoc Loc; 3689 return parseTokenLoc(Loc) || 3690 parseIntToken(FunctionId, "expected function id in '" + DirectiveName + 3691 "' directive") || 3692 check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc, 3693 "expected function id within range [0, UINT_MAX)"); 3694 } 3695 3696 bool AsmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) { 3697 SMLoc Loc; 3698 return parseTokenLoc(Loc) || 3699 parseIntToken(FileNumber, "expected integer in '" + DirectiveName + 3700 "' directive") || 3701 check(FileNumber < 1, Loc, "file number less than one in '" + 3702 DirectiveName + "' directive") || 3703 check(!getCVContext().isValidFileNumber(FileNumber), Loc, 3704 "unassigned file number in '" + DirectiveName + "' directive"); 3705 } 3706 3707 /// parseDirectiveCVFuncId 3708 /// ::= .cv_func_id FunctionId 3709 /// 3710 /// Introduces a function ID that can be used with .cv_loc. 3711 bool AsmParser::parseDirectiveCVFuncId() { 3712 SMLoc FunctionIdLoc = getTok().getLoc(); 3713 int64_t FunctionId; 3714 3715 if (parseCVFunctionId(FunctionId, ".cv_func_id") || 3716 parseToken(AsmToken::EndOfStatement, 3717 "unexpected token in '.cv_func_id' directive")) 3718 return true; 3719 3720 if (!getStreamer().EmitCVFuncIdDirective(FunctionId)) 3721 return Error(FunctionIdLoc, "function id already allocated"); 3722 3723 return false; 3724 } 3725 3726 /// parseDirectiveCVInlineSiteId 3727 /// ::= .cv_inline_site_id FunctionId 3728 /// "within" IAFunc 3729 /// "inlined_at" IAFile IALine [IACol] 3730 /// 3731 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined 3732 /// at" source location information for use in the line table of the caller, 3733 /// whether the caller is a real function or another inlined call site. 3734 bool AsmParser::parseDirectiveCVInlineSiteId() { 3735 SMLoc FunctionIdLoc = getTok().getLoc(); 3736 int64_t FunctionId; 3737 int64_t IAFunc; 3738 int64_t IAFile; 3739 int64_t IALine; 3740 int64_t IACol = 0; 3741 3742 // FunctionId 3743 if (parseCVFunctionId(FunctionId, ".cv_inline_site_id")) 3744 return true; 3745 3746 // "within" 3747 if (check((getLexer().isNot(AsmToken::Identifier) || 3748 getTok().getIdentifier() != "within"), 3749 "expected 'within' identifier in '.cv_inline_site_id' directive")) 3750 return true; 3751 Lex(); 3752 3753 // IAFunc 3754 if (parseCVFunctionId(IAFunc, ".cv_inline_site_id")) 3755 return true; 3756 3757 // "inlined_at" 3758 if (check((getLexer().isNot(AsmToken::Identifier) || 3759 getTok().getIdentifier() != "inlined_at"), 3760 "expected 'inlined_at' identifier in '.cv_inline_site_id' " 3761 "directive") ) 3762 return true; 3763 Lex(); 3764 3765 // IAFile IALine 3766 if (parseCVFileId(IAFile, ".cv_inline_site_id") || 3767 parseIntToken(IALine, "expected line number after 'inlined_at'")) 3768 return true; 3769 3770 // [IACol] 3771 if (getLexer().is(AsmToken::Integer)) { 3772 IACol = getTok().getIntVal(); 3773 Lex(); 3774 } 3775 3776 if (parseToken(AsmToken::EndOfStatement, 3777 "unexpected token in '.cv_inline_site_id' directive")) 3778 return true; 3779 3780 if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile, 3781 IALine, IACol, FunctionIdLoc)) 3782 return Error(FunctionIdLoc, "function id already allocated"); 3783 3784 return false; 3785 } 3786 3787 /// parseDirectiveCVLoc 3788 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3789 /// [is_stmt VALUE] 3790 /// The first number is a file number, must have been previously assigned with 3791 /// a .file directive, the second number is the line number and optionally the 3792 /// third number is a column position (zero if not specified). The remaining 3793 /// optional items are .loc sub-directives. 3794 bool AsmParser::parseDirectiveCVLoc() { 3795 SMLoc DirectiveLoc = getTok().getLoc(); 3796 int64_t FunctionId, FileNumber; 3797 if (parseCVFunctionId(FunctionId, ".cv_loc") || 3798 parseCVFileId(FileNumber, ".cv_loc")) 3799 return true; 3800 3801 int64_t LineNumber = 0; 3802 if (getLexer().is(AsmToken::Integer)) { 3803 LineNumber = getTok().getIntVal(); 3804 if (LineNumber < 0) 3805 return TokError("line number less than zero in '.cv_loc' directive"); 3806 Lex(); 3807 } 3808 3809 int64_t ColumnPos = 0; 3810 if (getLexer().is(AsmToken::Integer)) { 3811 ColumnPos = getTok().getIntVal(); 3812 if (ColumnPos < 0) 3813 return TokError("column position less than zero in '.cv_loc' directive"); 3814 Lex(); 3815 } 3816 3817 bool PrologueEnd = false; 3818 uint64_t IsStmt = 0; 3819 3820 auto parseOp = [&]() -> bool { 3821 StringRef Name; 3822 SMLoc Loc = getTok().getLoc(); 3823 if (parseIdentifier(Name)) 3824 return TokError("unexpected token in '.cv_loc' directive"); 3825 if (Name == "prologue_end") 3826 PrologueEnd = true; 3827 else if (Name == "is_stmt") { 3828 Loc = getTok().getLoc(); 3829 const MCExpr *Value; 3830 if (parseExpression(Value)) 3831 return true; 3832 // The expression must be the constant 0 or 1. 3833 IsStmt = ~0ULL; 3834 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3835 IsStmt = MCE->getValue(); 3836 3837 if (IsStmt > 1) 3838 return Error(Loc, "is_stmt value not 0 or 1"); 3839 } else { 3840 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3841 } 3842 return false; 3843 }; 3844 3845 if (parseMany(parseOp, false /*hasComma*/)) 3846 return true; 3847 3848 getStreamer().emitCVLocDirective(FunctionId, FileNumber, LineNumber, 3849 ColumnPos, PrologueEnd, IsStmt, StringRef(), 3850 DirectiveLoc); 3851 return false; 3852 } 3853 3854 /// parseDirectiveCVLinetable 3855 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3856 bool AsmParser::parseDirectiveCVLinetable() { 3857 int64_t FunctionId; 3858 StringRef FnStartName, FnEndName; 3859 SMLoc Loc = getTok().getLoc(); 3860 if (parseCVFunctionId(FunctionId, ".cv_linetable") || parseComma() || 3861 parseTokenLoc(Loc) || 3862 check(parseIdentifier(FnStartName), Loc, 3863 "expected identifier in directive") || 3864 parseComma() || parseTokenLoc(Loc) || 3865 check(parseIdentifier(FnEndName), Loc, 3866 "expected identifier in directive")) 3867 return true; 3868 3869 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3870 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3871 3872 getStreamer().emitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3873 return false; 3874 } 3875 3876 /// parseDirectiveCVInlineLinetable 3877 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3878 bool AsmParser::parseDirectiveCVInlineLinetable() { 3879 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3880 StringRef FnStartName, FnEndName; 3881 SMLoc Loc = getTok().getLoc(); 3882 if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") || 3883 parseTokenLoc(Loc) || 3884 parseIntToken( 3885 SourceFileId, 3886 "expected SourceField in '.cv_inline_linetable' directive") || 3887 check(SourceFileId <= 0, Loc, 3888 "File id less than zero in '.cv_inline_linetable' directive") || 3889 parseTokenLoc(Loc) || 3890 parseIntToken( 3891 SourceLineNum, 3892 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3893 check(SourceLineNum < 0, Loc, 3894 "Line number less than zero in '.cv_inline_linetable' directive") || 3895 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3896 "expected identifier in directive") || 3897 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3898 "expected identifier in directive")) 3899 return true; 3900 3901 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3902 return true; 3903 3904 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3905 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3906 getStreamer().emitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3907 SourceLineNum, FnStartSym, 3908 FnEndSym); 3909 return false; 3910 } 3911 3912 void AsmParser::initializeCVDefRangeTypeMap() { 3913 CVDefRangeTypeMap["reg"] = CVDR_DEFRANGE_REGISTER; 3914 CVDefRangeTypeMap["frame_ptr_rel"] = CVDR_DEFRANGE_FRAMEPOINTER_REL; 3915 CVDefRangeTypeMap["subfield_reg"] = CVDR_DEFRANGE_SUBFIELD_REGISTER; 3916 CVDefRangeTypeMap["reg_rel"] = CVDR_DEFRANGE_REGISTER_REL; 3917 } 3918 3919 /// parseDirectiveCVDefRange 3920 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3921 bool AsmParser::parseDirectiveCVDefRange() { 3922 SMLoc Loc; 3923 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3924 while (getLexer().is(AsmToken::Identifier)) { 3925 Loc = getLexer().getLoc(); 3926 StringRef GapStartName; 3927 if (parseIdentifier(GapStartName)) 3928 return Error(Loc, "expected identifier in directive"); 3929 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3930 3931 Loc = getLexer().getLoc(); 3932 StringRef GapEndName; 3933 if (parseIdentifier(GapEndName)) 3934 return Error(Loc, "expected identifier in directive"); 3935 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3936 3937 Ranges.push_back({GapStartSym, GapEndSym}); 3938 } 3939 3940 StringRef CVDefRangeTypeStr; 3941 if (parseToken( 3942 AsmToken::Comma, 3943 "expected comma before def_range type in .cv_def_range directive") || 3944 parseIdentifier(CVDefRangeTypeStr)) 3945 return Error(Loc, "expected def_range type in directive"); 3946 3947 StringMap<CVDefRangeType>::const_iterator CVTypeIt = 3948 CVDefRangeTypeMap.find(CVDefRangeTypeStr); 3949 CVDefRangeType CVDRType = (CVTypeIt == CVDefRangeTypeMap.end()) 3950 ? CVDR_DEFRANGE 3951 : CVTypeIt->getValue(); 3952 switch (CVDRType) { 3953 case CVDR_DEFRANGE_REGISTER: { 3954 int64_t DRRegister; 3955 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3956 ".cv_def_range directive") || 3957 parseAbsoluteExpression(DRRegister)) 3958 return Error(Loc, "expected register number"); 3959 3960 codeview::DefRangeRegisterHeader DRHdr; 3961 DRHdr.Register = DRRegister; 3962 DRHdr.MayHaveNoName = 0; 3963 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3964 break; 3965 } 3966 case CVDR_DEFRANGE_FRAMEPOINTER_REL: { 3967 int64_t DROffset; 3968 if (parseToken(AsmToken::Comma, 3969 "expected comma before offset in .cv_def_range directive") || 3970 parseAbsoluteExpression(DROffset)) 3971 return Error(Loc, "expected offset value"); 3972 3973 codeview::DefRangeFramePointerRelHeader DRHdr; 3974 DRHdr.Offset = DROffset; 3975 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3976 break; 3977 } 3978 case CVDR_DEFRANGE_SUBFIELD_REGISTER: { 3979 int64_t DRRegister; 3980 int64_t DROffsetInParent; 3981 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3982 ".cv_def_range directive") || 3983 parseAbsoluteExpression(DRRegister)) 3984 return Error(Loc, "expected register number"); 3985 if (parseToken(AsmToken::Comma, 3986 "expected comma before offset in .cv_def_range directive") || 3987 parseAbsoluteExpression(DROffsetInParent)) 3988 return Error(Loc, "expected offset value"); 3989 3990 codeview::DefRangeSubfieldRegisterHeader DRHdr; 3991 DRHdr.Register = DRRegister; 3992 DRHdr.MayHaveNoName = 0; 3993 DRHdr.OffsetInParent = DROffsetInParent; 3994 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3995 break; 3996 } 3997 case CVDR_DEFRANGE_REGISTER_REL: { 3998 int64_t DRRegister; 3999 int64_t DRFlags; 4000 int64_t DRBasePointerOffset; 4001 if (parseToken(AsmToken::Comma, "expected comma before register number in " 4002 ".cv_def_range directive") || 4003 parseAbsoluteExpression(DRRegister)) 4004 return Error(Loc, "expected register value"); 4005 if (parseToken( 4006 AsmToken::Comma, 4007 "expected comma before flag value in .cv_def_range directive") || 4008 parseAbsoluteExpression(DRFlags)) 4009 return Error(Loc, "expected flag value"); 4010 if (parseToken(AsmToken::Comma, "expected comma before base pointer offset " 4011 "in .cv_def_range directive") || 4012 parseAbsoluteExpression(DRBasePointerOffset)) 4013 return Error(Loc, "expected base pointer offset value"); 4014 4015 codeview::DefRangeRegisterRelHeader DRHdr; 4016 DRHdr.Register = DRRegister; 4017 DRHdr.Flags = DRFlags; 4018 DRHdr.BasePointerOffset = DRBasePointerOffset; 4019 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 4020 break; 4021 } 4022 default: 4023 return Error(Loc, "unexpected def_range type in .cv_def_range directive"); 4024 } 4025 return true; 4026 } 4027 4028 /// parseDirectiveCVString 4029 /// ::= .cv_stringtable "string" 4030 bool AsmParser::parseDirectiveCVString() { 4031 std::string Data; 4032 if (checkForValidSection() || parseEscapedString(Data)) 4033 return true; 4034 4035 // Put the string in the table and emit the offset. 4036 std::pair<StringRef, unsigned> Insertion = 4037 getCVContext().addToStringTable(Data); 4038 getStreamer().emitInt32(Insertion.second); 4039 return false; 4040 } 4041 4042 /// parseDirectiveCVStringTable 4043 /// ::= .cv_stringtable 4044 bool AsmParser::parseDirectiveCVStringTable() { 4045 getStreamer().emitCVStringTableDirective(); 4046 return false; 4047 } 4048 4049 /// parseDirectiveCVFileChecksums 4050 /// ::= .cv_filechecksums 4051 bool AsmParser::parseDirectiveCVFileChecksums() { 4052 getStreamer().emitCVFileChecksumsDirective(); 4053 return false; 4054 } 4055 4056 /// parseDirectiveCVFileChecksumOffset 4057 /// ::= .cv_filechecksumoffset fileno 4058 bool AsmParser::parseDirectiveCVFileChecksumOffset() { 4059 int64_t FileNo; 4060 if (parseIntToken(FileNo, "expected identifier in directive")) 4061 return true; 4062 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 4063 return true; 4064 getStreamer().emitCVFileChecksumOffsetDirective(FileNo); 4065 return false; 4066 } 4067 4068 /// parseDirectiveCVFPOData 4069 /// ::= .cv_fpo_data procsym 4070 bool AsmParser::parseDirectiveCVFPOData() { 4071 SMLoc DirLoc = getLexer().getLoc(); 4072 StringRef ProcName; 4073 if (parseIdentifier(ProcName)) 4074 return TokError("expected symbol name"); 4075 if (parseEOL()) 4076 return true; 4077 MCSymbol *ProcSym = getContext().getOrCreateSymbol(ProcName); 4078 getStreamer().EmitCVFPOData(ProcSym, DirLoc); 4079 return false; 4080 } 4081 4082 /// parseDirectiveCFISections 4083 /// ::= .cfi_sections section [, section] 4084 bool AsmParser::parseDirectiveCFISections() { 4085 StringRef Name; 4086 bool EH = false; 4087 bool Debug = false; 4088 4089 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4090 for (;;) { 4091 if (parseIdentifier(Name)) 4092 return TokError("expected .eh_frame or .debug_frame"); 4093 if (Name == ".eh_frame") 4094 EH = true; 4095 else if (Name == ".debug_frame") 4096 Debug = true; 4097 if (parseOptionalToken(AsmToken::EndOfStatement)) 4098 break; 4099 if (parseComma()) 4100 return true; 4101 } 4102 } 4103 getStreamer().emitCFISections(EH, Debug); 4104 return false; 4105 } 4106 4107 /// parseDirectiveCFIStartProc 4108 /// ::= .cfi_startproc [simple] 4109 bool AsmParser::parseDirectiveCFIStartProc() { 4110 StringRef Simple; 4111 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4112 if (check(parseIdentifier(Simple) || Simple != "simple", 4113 "unexpected token") || 4114 parseEOL()) 4115 return true; 4116 } 4117 4118 // TODO(kristina): Deal with a corner case of incorrect diagnostic context 4119 // being produced if this directive is emitted as part of preprocessor macro 4120 // expansion which can *ONLY* happen if Clang's cc1as is the API consumer. 4121 // Tools like llvm-mc on the other hand are not affected by it, and report 4122 // correct context information. 4123 getStreamer().emitCFIStartProc(!Simple.empty(), Lexer.getLoc()); 4124 return false; 4125 } 4126 4127 /// parseDirectiveCFIEndProc 4128 /// ::= .cfi_endproc 4129 bool AsmParser::parseDirectiveCFIEndProc() { 4130 if (parseEOL()) 4131 return true; 4132 getStreamer().emitCFIEndProc(); 4133 return false; 4134 } 4135 4136 /// parse register name or number. 4137 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 4138 SMLoc DirectiveLoc) { 4139 unsigned RegNo; 4140 4141 if (getLexer().isNot(AsmToken::Integer)) { 4142 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 4143 return true; 4144 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 4145 } else 4146 return parseAbsoluteExpression(Register); 4147 4148 return false; 4149 } 4150 4151 /// parseDirectiveCFIDefCfa 4152 /// ::= .cfi_def_cfa register, offset 4153 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 4154 int64_t Register = 0, Offset = 0; 4155 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseComma() || 4156 parseAbsoluteExpression(Offset) || parseEOL()) 4157 return true; 4158 4159 getStreamer().emitCFIDefCfa(Register, Offset); 4160 return false; 4161 } 4162 4163 /// parseDirectiveCFIDefCfaOffset 4164 /// ::= .cfi_def_cfa_offset offset 4165 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 4166 int64_t Offset = 0; 4167 if (parseAbsoluteExpression(Offset) || parseEOL()) 4168 return true; 4169 4170 getStreamer().emitCFIDefCfaOffset(Offset); 4171 return false; 4172 } 4173 4174 /// parseDirectiveCFIRegister 4175 /// ::= .cfi_register register, register 4176 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 4177 int64_t Register1 = 0, Register2 = 0; 4178 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || parseComma() || 4179 parseRegisterOrRegisterNumber(Register2, DirectiveLoc) || parseEOL()) 4180 return true; 4181 4182 getStreamer().emitCFIRegister(Register1, Register2); 4183 return false; 4184 } 4185 4186 /// parseDirectiveCFIWindowSave 4187 /// ::= .cfi_window_save 4188 bool AsmParser::parseDirectiveCFIWindowSave() { 4189 if (parseEOL()) 4190 return true; 4191 getStreamer().emitCFIWindowSave(); 4192 return false; 4193 } 4194 4195 /// parseDirectiveCFIAdjustCfaOffset 4196 /// ::= .cfi_adjust_cfa_offset adjustment 4197 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 4198 int64_t Adjustment = 0; 4199 if (parseAbsoluteExpression(Adjustment) || parseEOL()) 4200 return true; 4201 4202 getStreamer().emitCFIAdjustCfaOffset(Adjustment); 4203 return false; 4204 } 4205 4206 /// parseDirectiveCFIDefCfaRegister 4207 /// ::= .cfi_def_cfa_register register 4208 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 4209 int64_t Register = 0; 4210 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4211 return true; 4212 4213 getStreamer().emitCFIDefCfaRegister(Register); 4214 return false; 4215 } 4216 4217 /// parseDirectiveCFIOffset 4218 /// ::= .cfi_offset register, offset 4219 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 4220 int64_t Register = 0; 4221 int64_t Offset = 0; 4222 4223 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseComma() || 4224 parseAbsoluteExpression(Offset) || parseEOL()) 4225 return true; 4226 4227 getStreamer().emitCFIOffset(Register, Offset); 4228 return false; 4229 } 4230 4231 /// parseDirectiveCFIRelOffset 4232 /// ::= .cfi_rel_offset register, offset 4233 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 4234 int64_t Register = 0, Offset = 0; 4235 4236 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseComma() || 4237 parseAbsoluteExpression(Offset) || parseEOL()) 4238 return true; 4239 4240 getStreamer().emitCFIRelOffset(Register, Offset); 4241 return false; 4242 } 4243 4244 static bool isValidEncoding(int64_t Encoding) { 4245 if (Encoding & ~0xff) 4246 return false; 4247 4248 if (Encoding == dwarf::DW_EH_PE_omit) 4249 return true; 4250 4251 const unsigned Format = Encoding & 0xf; 4252 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 4253 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 4254 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 4255 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 4256 return false; 4257 4258 const unsigned Application = Encoding & 0x70; 4259 if (Application != dwarf::DW_EH_PE_absptr && 4260 Application != dwarf::DW_EH_PE_pcrel) 4261 return false; 4262 4263 return true; 4264 } 4265 4266 /// parseDirectiveCFIPersonalityOrLsda 4267 /// IsPersonality true for cfi_personality, false for cfi_lsda 4268 /// ::= .cfi_personality encoding, [symbol_name] 4269 /// ::= .cfi_lsda encoding, [symbol_name] 4270 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 4271 int64_t Encoding = 0; 4272 if (parseAbsoluteExpression(Encoding)) 4273 return true; 4274 if (Encoding == dwarf::DW_EH_PE_omit) 4275 return false; 4276 4277 StringRef Name; 4278 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 4279 parseComma() || 4280 check(parseIdentifier(Name), "expected identifier in directive") || 4281 parseEOL()) 4282 return true; 4283 4284 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4285 4286 if (IsPersonality) 4287 getStreamer().emitCFIPersonality(Sym, Encoding); 4288 else 4289 getStreamer().emitCFILsda(Sym, Encoding); 4290 return false; 4291 } 4292 4293 /// parseDirectiveCFIRememberState 4294 /// ::= .cfi_remember_state 4295 bool AsmParser::parseDirectiveCFIRememberState() { 4296 if (parseEOL()) 4297 return true; 4298 getStreamer().emitCFIRememberState(); 4299 return false; 4300 } 4301 4302 /// parseDirectiveCFIRestoreState 4303 /// ::= .cfi_remember_state 4304 bool AsmParser::parseDirectiveCFIRestoreState() { 4305 if (parseEOL()) 4306 return true; 4307 getStreamer().emitCFIRestoreState(); 4308 return false; 4309 } 4310 4311 /// parseDirectiveCFISameValue 4312 /// ::= .cfi_same_value register 4313 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 4314 int64_t Register = 0; 4315 4316 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4317 return true; 4318 4319 getStreamer().emitCFISameValue(Register); 4320 return false; 4321 } 4322 4323 /// parseDirectiveCFIRestore 4324 /// ::= .cfi_restore register 4325 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 4326 int64_t Register = 0; 4327 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4328 return true; 4329 4330 getStreamer().emitCFIRestore(Register); 4331 return false; 4332 } 4333 4334 /// parseDirectiveCFIEscape 4335 /// ::= .cfi_escape expression[,...] 4336 bool AsmParser::parseDirectiveCFIEscape() { 4337 std::string Values; 4338 int64_t CurrValue; 4339 if (parseAbsoluteExpression(CurrValue)) 4340 return true; 4341 4342 Values.push_back((uint8_t)CurrValue); 4343 4344 while (getLexer().is(AsmToken::Comma)) { 4345 Lex(); 4346 4347 if (parseAbsoluteExpression(CurrValue)) 4348 return true; 4349 4350 Values.push_back((uint8_t)CurrValue); 4351 } 4352 4353 getStreamer().emitCFIEscape(Values); 4354 return false; 4355 } 4356 4357 /// parseDirectiveCFIReturnColumn 4358 /// ::= .cfi_return_column register 4359 bool AsmParser::parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc) { 4360 int64_t Register = 0; 4361 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4362 return true; 4363 getStreamer().emitCFIReturnColumn(Register); 4364 return false; 4365 } 4366 4367 /// parseDirectiveCFISignalFrame 4368 /// ::= .cfi_signal_frame 4369 bool AsmParser::parseDirectiveCFISignalFrame() { 4370 if (parseEOL()) 4371 return true; 4372 4373 getStreamer().emitCFISignalFrame(); 4374 return false; 4375 } 4376 4377 /// parseDirectiveCFIUndefined 4378 /// ::= .cfi_undefined register 4379 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 4380 int64_t Register = 0; 4381 4382 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4383 return true; 4384 4385 getStreamer().emitCFIUndefined(Register); 4386 return false; 4387 } 4388 4389 /// parseDirectiveAltmacro 4390 /// ::= .altmacro 4391 /// ::= .noaltmacro 4392 bool AsmParser::parseDirectiveAltmacro(StringRef Directive) { 4393 if (parseEOL()) 4394 return true; 4395 AltMacroMode = (Directive == ".altmacro"); 4396 return false; 4397 } 4398 4399 /// parseDirectiveMacrosOnOff 4400 /// ::= .macros_on 4401 /// ::= .macros_off 4402 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 4403 if (parseEOL()) 4404 return true; 4405 setMacrosEnabled(Directive == ".macros_on"); 4406 return false; 4407 } 4408 4409 /// parseDirectiveMacro 4410 /// ::= .macro name[,] [parameters] 4411 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 4412 StringRef Name; 4413 if (parseIdentifier(Name)) 4414 return TokError("expected identifier in '.macro' directive"); 4415 4416 if (getLexer().is(AsmToken::Comma)) 4417 Lex(); 4418 4419 MCAsmMacroParameters Parameters; 4420 while (getLexer().isNot(AsmToken::EndOfStatement)) { 4421 4422 if (!Parameters.empty() && Parameters.back().Vararg) 4423 return Error(Lexer.getLoc(), "vararg parameter '" + 4424 Parameters.back().Name + 4425 "' should be the last parameter"); 4426 4427 MCAsmMacroParameter Parameter; 4428 if (parseIdentifier(Parameter.Name)) 4429 return TokError("expected identifier in '.macro' directive"); 4430 4431 // Emit an error if two (or more) named parameters share the same name 4432 for (const MCAsmMacroParameter& CurrParam : Parameters) 4433 if (CurrParam.Name.equals(Parameter.Name)) 4434 return TokError("macro '" + Name + "' has multiple parameters" 4435 " named '" + Parameter.Name + "'"); 4436 4437 if (Lexer.is(AsmToken::Colon)) { 4438 Lex(); // consume ':' 4439 4440 SMLoc QualLoc; 4441 StringRef Qualifier; 4442 4443 QualLoc = Lexer.getLoc(); 4444 if (parseIdentifier(Qualifier)) 4445 return Error(QualLoc, "missing parameter qualifier for " 4446 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4447 4448 if (Qualifier == "req") 4449 Parameter.Required = true; 4450 else if (Qualifier == "vararg") 4451 Parameter.Vararg = true; 4452 else 4453 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 4454 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 4455 } 4456 4457 if (getLexer().is(AsmToken::Equal)) { 4458 Lex(); 4459 4460 SMLoc ParamLoc; 4461 4462 ParamLoc = Lexer.getLoc(); 4463 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 4464 return true; 4465 4466 if (Parameter.Required) 4467 Warning(ParamLoc, "pointless default value for required parameter " 4468 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4469 } 4470 4471 Parameters.push_back(std::move(Parameter)); 4472 4473 if (getLexer().is(AsmToken::Comma)) 4474 Lex(); 4475 } 4476 4477 // Eat just the end of statement. 4478 Lexer.Lex(); 4479 4480 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 4481 AsmToken EndToken, StartToken = getTok(); 4482 unsigned MacroDepth = 0; 4483 // Lex the macro definition. 4484 while (true) { 4485 // Ignore Lexing errors in macros. 4486 while (Lexer.is(AsmToken::Error)) { 4487 Lexer.Lex(); 4488 } 4489 4490 // Check whether we have reached the end of the file. 4491 if (getLexer().is(AsmToken::Eof)) 4492 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 4493 4494 // Otherwise, check whether we have reach the .endmacro or the start of a 4495 // preprocessor line marker. 4496 if (getLexer().is(AsmToken::Identifier)) { 4497 if (getTok().getIdentifier() == ".endm" || 4498 getTok().getIdentifier() == ".endmacro") { 4499 if (MacroDepth == 0) { // Outermost macro. 4500 EndToken = getTok(); 4501 Lexer.Lex(); 4502 if (getLexer().isNot(AsmToken::EndOfStatement)) 4503 return TokError("unexpected token in '" + EndToken.getIdentifier() + 4504 "' directive"); 4505 break; 4506 } else { 4507 // Otherwise we just found the end of an inner macro. 4508 --MacroDepth; 4509 } 4510 } else if (getTok().getIdentifier() == ".macro") { 4511 // We allow nested macros. Those aren't instantiated until the outermost 4512 // macro is expanded so just ignore them for now. 4513 ++MacroDepth; 4514 } 4515 } else if (Lexer.is(AsmToken::HashDirective)) { 4516 (void)parseCppHashLineFilenameComment(getLexer().getLoc()); 4517 } 4518 4519 // Otherwise, scan til the end of the statement. 4520 eatToEndOfStatement(); 4521 } 4522 4523 if (getContext().lookupMacro(Name)) { 4524 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 4525 } 4526 4527 const char *BodyStart = StartToken.getLoc().getPointer(); 4528 const char *BodyEnd = EndToken.getLoc().getPointer(); 4529 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4530 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 4531 MCAsmMacro Macro(Name, Body, std::move(Parameters)); 4532 DEBUG_WITH_TYPE("asm-macros", dbgs() << "Defining new macro:\n"; 4533 Macro.dump()); 4534 getContext().defineMacro(Name, std::move(Macro)); 4535 return false; 4536 } 4537 4538 /// checkForBadMacro 4539 /// 4540 /// With the support added for named parameters there may be code out there that 4541 /// is transitioning from positional parameters. In versions of gas that did 4542 /// not support named parameters they would be ignored on the macro definition. 4543 /// But to support both styles of parameters this is not possible so if a macro 4544 /// definition has named parameters but does not use them and has what appears 4545 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 4546 /// warning that the positional parameter found in body which have no effect. 4547 /// Hoping the developer will either remove the named parameters from the macro 4548 /// definition so the positional parameters get used if that was what was 4549 /// intended or change the macro to use the named parameters. It is possible 4550 /// this warning will trigger when the none of the named parameters are used 4551 /// and the strings like $1 are infact to simply to be passed trough unchanged. 4552 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 4553 StringRef Body, 4554 ArrayRef<MCAsmMacroParameter> Parameters) { 4555 // If this macro is not defined with named parameters the warning we are 4556 // checking for here doesn't apply. 4557 unsigned NParameters = Parameters.size(); 4558 if (NParameters == 0) 4559 return; 4560 4561 bool NamedParametersFound = false; 4562 bool PositionalParametersFound = false; 4563 4564 // Look at the body of the macro for use of both the named parameters and what 4565 // are likely to be positional parameters. This is what expandMacro() is 4566 // doing when it finds the parameters in the body. 4567 while (!Body.empty()) { 4568 // Scan for the next possible parameter. 4569 std::size_t End = Body.size(), Pos = 0; 4570 for (; Pos != End; ++Pos) { 4571 // Check for a substitution or escape. 4572 // This macro is defined with parameters, look for \foo, \bar, etc. 4573 if (Body[Pos] == '\\' && Pos + 1 != End) 4574 break; 4575 4576 // This macro should have parameters, but look for $0, $1, ..., $n too. 4577 if (Body[Pos] != '$' || Pos + 1 == End) 4578 continue; 4579 char Next = Body[Pos + 1]; 4580 if (Next == '$' || Next == 'n' || 4581 isdigit(static_cast<unsigned char>(Next))) 4582 break; 4583 } 4584 4585 // Check if we reached the end. 4586 if (Pos == End) 4587 break; 4588 4589 if (Body[Pos] == '$') { 4590 switch (Body[Pos + 1]) { 4591 // $$ => $ 4592 case '$': 4593 break; 4594 4595 // $n => number of arguments 4596 case 'n': 4597 PositionalParametersFound = true; 4598 break; 4599 4600 // $[0-9] => argument 4601 default: { 4602 PositionalParametersFound = true; 4603 break; 4604 } 4605 } 4606 Pos += 2; 4607 } else { 4608 unsigned I = Pos + 1; 4609 while (isIdentifierChar(Body[I]) && I + 1 != End) 4610 ++I; 4611 4612 const char *Begin = Body.data() + Pos + 1; 4613 StringRef Argument(Begin, I - (Pos + 1)); 4614 unsigned Index = 0; 4615 for (; Index < NParameters; ++Index) 4616 if (Parameters[Index].Name == Argument) 4617 break; 4618 4619 if (Index == NParameters) { 4620 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 4621 Pos += 3; 4622 else { 4623 Pos = I; 4624 } 4625 } else { 4626 NamedParametersFound = true; 4627 Pos += 1 + Argument.size(); 4628 } 4629 } 4630 // Update the scan point. 4631 Body = Body.substr(Pos); 4632 } 4633 4634 if (!NamedParametersFound && PositionalParametersFound) 4635 Warning(DirectiveLoc, "macro defined with named parameters which are not " 4636 "used in macro body, possible positional parameter " 4637 "found in body which will have no effect"); 4638 } 4639 4640 /// parseDirectiveExitMacro 4641 /// ::= .exitm 4642 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 4643 if (parseEOL()) 4644 return true; 4645 4646 if (!isInsideMacroInstantiation()) 4647 return TokError("unexpected '" + Directive + "' in file, " 4648 "no current macro definition"); 4649 4650 // Exit all conditionals that are active in the current macro. 4651 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 4652 TheCondState = TheCondStack.back(); 4653 TheCondStack.pop_back(); 4654 } 4655 4656 handleMacroExit(); 4657 return false; 4658 } 4659 4660 /// parseDirectiveEndMacro 4661 /// ::= .endm 4662 /// ::= .endmacro 4663 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 4664 if (getLexer().isNot(AsmToken::EndOfStatement)) 4665 return TokError("unexpected token in '" + Directive + "' directive"); 4666 4667 // If we are inside a macro instantiation, terminate the current 4668 // instantiation. 4669 if (isInsideMacroInstantiation()) { 4670 handleMacroExit(); 4671 return false; 4672 } 4673 4674 // Otherwise, this .endmacro is a stray entry in the file; well formed 4675 // .endmacro directives are handled during the macro definition parsing. 4676 return TokError("unexpected '" + Directive + "' in file, " 4677 "no current macro definition"); 4678 } 4679 4680 /// parseDirectivePurgeMacro 4681 /// ::= .purgem name 4682 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 4683 StringRef Name; 4684 SMLoc Loc; 4685 if (parseTokenLoc(Loc) || 4686 check(parseIdentifier(Name), Loc, 4687 "expected identifier in '.purgem' directive") || 4688 parseEOL()) 4689 return true; 4690 4691 if (!getContext().lookupMacro(Name)) 4692 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 4693 4694 getContext().undefineMacro(Name); 4695 DEBUG_WITH_TYPE("asm-macros", dbgs() 4696 << "Un-defining macro: " << Name << "\n"); 4697 return false; 4698 } 4699 4700 /// parseDirectiveBundleAlignMode 4701 /// ::= {.bundle_align_mode} expression 4702 bool AsmParser::parseDirectiveBundleAlignMode() { 4703 // Expect a single argument: an expression that evaluates to a constant 4704 // in the inclusive range 0-30. 4705 SMLoc ExprLoc = getLexer().getLoc(); 4706 int64_t AlignSizePow2; 4707 if (checkForValidSection() || parseAbsoluteExpression(AlignSizePow2) || 4708 parseEOL() || 4709 check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc, 4710 "invalid bundle alignment size (expected between 0 and 30)")) 4711 return true; 4712 4713 // Because of AlignSizePow2's verified range we can safely truncate it to 4714 // unsigned. 4715 getStreamer().emitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 4716 return false; 4717 } 4718 4719 /// parseDirectiveBundleLock 4720 /// ::= {.bundle_lock} [align_to_end] 4721 bool AsmParser::parseDirectiveBundleLock() { 4722 if (checkForValidSection()) 4723 return true; 4724 bool AlignToEnd = false; 4725 4726 StringRef Option; 4727 SMLoc Loc = getTok().getLoc(); 4728 const char *kInvalidOptionError = 4729 "invalid option for '.bundle_lock' directive"; 4730 4731 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4732 if (check(parseIdentifier(Option), Loc, kInvalidOptionError) || 4733 check(Option != "align_to_end", Loc, kInvalidOptionError) || parseEOL()) 4734 return true; 4735 AlignToEnd = true; 4736 } 4737 4738 getStreamer().emitBundleLock(AlignToEnd); 4739 return false; 4740 } 4741 4742 /// parseDirectiveBundleLock 4743 /// ::= {.bundle_lock} 4744 bool AsmParser::parseDirectiveBundleUnlock() { 4745 if (checkForValidSection() || parseEOL()) 4746 return true; 4747 4748 getStreamer().emitBundleUnlock(); 4749 return false; 4750 } 4751 4752 /// parseDirectiveSpace 4753 /// ::= (.skip | .space) expression [ , expression ] 4754 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4755 SMLoc NumBytesLoc = Lexer.getLoc(); 4756 const MCExpr *NumBytes; 4757 if (checkForValidSection() || parseExpression(NumBytes)) 4758 return true; 4759 4760 int64_t FillExpr = 0; 4761 if (parseOptionalToken(AsmToken::Comma)) 4762 if (parseAbsoluteExpression(FillExpr)) 4763 return true; 4764 if (parseEOL()) 4765 return true; 4766 4767 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4768 getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc); 4769 4770 return false; 4771 } 4772 4773 /// parseDirectiveDCB 4774 /// ::= .dcb.{b, l, w} expression, expression 4775 bool AsmParser::parseDirectiveDCB(StringRef IDVal, unsigned Size) { 4776 SMLoc NumValuesLoc = Lexer.getLoc(); 4777 int64_t NumValues; 4778 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4779 return true; 4780 4781 if (NumValues < 0) { 4782 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4783 return false; 4784 } 4785 4786 if (parseComma()) 4787 return true; 4788 4789 const MCExpr *Value; 4790 SMLoc ExprLoc = getLexer().getLoc(); 4791 if (parseExpression(Value)) 4792 return true; 4793 4794 // Special case constant expressions to match code generator. 4795 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 4796 assert(Size <= 8 && "Invalid size"); 4797 uint64_t IntValue = MCE->getValue(); 4798 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 4799 return Error(ExprLoc, "literal value out of range for directive"); 4800 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4801 getStreamer().emitIntValue(IntValue, Size); 4802 } else { 4803 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4804 getStreamer().emitValue(Value, Size, ExprLoc); 4805 } 4806 4807 return parseEOL(); 4808 } 4809 4810 /// parseDirectiveRealDCB 4811 /// ::= .dcb.{d, s} expression, expression 4812 bool AsmParser::parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &Semantics) { 4813 SMLoc NumValuesLoc = Lexer.getLoc(); 4814 int64_t NumValues; 4815 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4816 return true; 4817 4818 if (NumValues < 0) { 4819 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4820 return false; 4821 } 4822 4823 if (parseComma()) 4824 return true; 4825 4826 APInt AsInt; 4827 if (parseRealValue(Semantics, AsInt) || parseEOL()) 4828 return true; 4829 4830 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4831 getStreamer().emitIntValue(AsInt.getLimitedValue(), 4832 AsInt.getBitWidth() / 8); 4833 4834 return false; 4835 } 4836 4837 /// parseDirectiveDS 4838 /// ::= .ds.{b, d, l, p, s, w, x} expression 4839 bool AsmParser::parseDirectiveDS(StringRef IDVal, unsigned Size) { 4840 SMLoc NumValuesLoc = Lexer.getLoc(); 4841 int64_t NumValues; 4842 if (checkForValidSection() || parseAbsoluteExpression(NumValues) || 4843 parseEOL()) 4844 return true; 4845 4846 if (NumValues < 0) { 4847 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4848 return false; 4849 } 4850 4851 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4852 getStreamer().emitFill(Size, 0); 4853 4854 return false; 4855 } 4856 4857 /// parseDirectiveLEB128 4858 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4859 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4860 if (checkForValidSection()) 4861 return true; 4862 4863 auto parseOp = [&]() -> bool { 4864 const MCExpr *Value; 4865 if (parseExpression(Value)) 4866 return true; 4867 if (Signed) 4868 getStreamer().emitSLEB128Value(Value); 4869 else 4870 getStreamer().emitULEB128Value(Value); 4871 return false; 4872 }; 4873 4874 return parseMany(parseOp); 4875 } 4876 4877 /// parseDirectiveSymbolAttribute 4878 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4879 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4880 auto parseOp = [&]() -> bool { 4881 StringRef Name; 4882 SMLoc Loc = getTok().getLoc(); 4883 if (parseIdentifier(Name)) 4884 return Error(Loc, "expected identifier"); 4885 4886 if (discardLTOSymbol(Name)) 4887 return false; 4888 4889 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4890 4891 // Assembler local symbols don't make any sense here. Complain loudly. 4892 if (Sym->isTemporary()) 4893 return Error(Loc, "non-local symbol required"); 4894 4895 if (!getStreamer().emitSymbolAttribute(Sym, Attr)) 4896 return Error(Loc, "unable to emit symbol attribute"); 4897 return false; 4898 }; 4899 4900 return parseMany(parseOp); 4901 } 4902 4903 /// parseDirectiveComm 4904 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4905 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4906 if (checkForValidSection()) 4907 return true; 4908 4909 SMLoc IDLoc = getLexer().getLoc(); 4910 StringRef Name; 4911 if (parseIdentifier(Name)) 4912 return TokError("expected identifier in directive"); 4913 4914 // Handle the identifier as the key symbol. 4915 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4916 4917 if (parseComma()) 4918 return true; 4919 4920 int64_t Size; 4921 SMLoc SizeLoc = getLexer().getLoc(); 4922 if (parseAbsoluteExpression(Size)) 4923 return true; 4924 4925 int64_t Pow2Alignment = 0; 4926 SMLoc Pow2AlignmentLoc; 4927 if (getLexer().is(AsmToken::Comma)) { 4928 Lex(); 4929 Pow2AlignmentLoc = getLexer().getLoc(); 4930 if (parseAbsoluteExpression(Pow2Alignment)) 4931 return true; 4932 4933 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4934 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4935 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4936 4937 // If this target takes alignments in bytes (not log) validate and convert. 4938 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4939 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4940 if (!isPowerOf2_64(Pow2Alignment)) 4941 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4942 Pow2Alignment = Log2_64(Pow2Alignment); 4943 } 4944 } 4945 4946 if (parseEOL()) 4947 return true; 4948 4949 // NOTE: a size of zero for a .comm should create a undefined symbol 4950 // but a size of .lcomm creates a bss symbol of size zero. 4951 if (Size < 0) 4952 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4953 "be less than zero"); 4954 4955 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4956 // may internally end up wanting an alignment in bytes. 4957 // FIXME: Diagnose overflow. 4958 if (Pow2Alignment < 0) 4959 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4960 "alignment, can't be less than zero"); 4961 4962 Sym->redefineIfPossible(); 4963 if (!Sym->isUndefined()) 4964 return Error(IDLoc, "invalid symbol redefinition"); 4965 4966 // Create the Symbol as a common or local common with Size and Pow2Alignment 4967 if (IsLocal) { 4968 getStreamer().emitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4969 return false; 4970 } 4971 4972 getStreamer().emitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4973 return false; 4974 } 4975 4976 /// parseDirectiveAbort 4977 /// ::= .abort [... message ...] 4978 bool AsmParser::parseDirectiveAbort() { 4979 // FIXME: Use loc from directive. 4980 SMLoc Loc = getLexer().getLoc(); 4981 4982 StringRef Str = parseStringToEndOfStatement(); 4983 if (parseEOL()) 4984 return true; 4985 4986 if (Str.empty()) 4987 return Error(Loc, ".abort detected. Assembly stopping."); 4988 else 4989 return Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 4990 // FIXME: Actually abort assembly here. 4991 4992 return false; 4993 } 4994 4995 /// parseDirectiveInclude 4996 /// ::= .include "filename" 4997 bool AsmParser::parseDirectiveInclude() { 4998 // Allow the strings to have escaped octal character sequence. 4999 std::string Filename; 5000 SMLoc IncludeLoc = getTok().getLoc(); 5001 5002 if (check(getTok().isNot(AsmToken::String), 5003 "expected string in '.include' directive") || 5004 parseEscapedString(Filename) || 5005 check(getTok().isNot(AsmToken::EndOfStatement), 5006 "unexpected token in '.include' directive") || 5007 // Attempt to switch the lexer to the included file before consuming the 5008 // end of statement to avoid losing it when we switch. 5009 check(enterIncludeFile(Filename), IncludeLoc, 5010 "Could not find include file '" + Filename + "'")) 5011 return true; 5012 5013 return false; 5014 } 5015 5016 /// parseDirectiveIncbin 5017 /// ::= .incbin "filename" [ , skip [ , count ] ] 5018 bool AsmParser::parseDirectiveIncbin() { 5019 // Allow the strings to have escaped octal character sequence. 5020 std::string Filename; 5021 SMLoc IncbinLoc = getTok().getLoc(); 5022 if (check(getTok().isNot(AsmToken::String), 5023 "expected string in '.incbin' directive") || 5024 parseEscapedString(Filename)) 5025 return true; 5026 5027 int64_t Skip = 0; 5028 const MCExpr *Count = nullptr; 5029 SMLoc SkipLoc, CountLoc; 5030 if (parseOptionalToken(AsmToken::Comma)) { 5031 // The skip expression can be omitted while specifying the count, e.g: 5032 // .incbin "filename",,4 5033 if (getTok().isNot(AsmToken::Comma)) { 5034 if (parseTokenLoc(SkipLoc) || parseAbsoluteExpression(Skip)) 5035 return true; 5036 } 5037 if (parseOptionalToken(AsmToken::Comma)) { 5038 CountLoc = getTok().getLoc(); 5039 if (parseExpression(Count)) 5040 return true; 5041 } 5042 } 5043 5044 if (parseEOL()) 5045 return true; 5046 5047 if (check(Skip < 0, SkipLoc, "skip is negative")) 5048 return true; 5049 5050 // Attempt to process the included file. 5051 if (processIncbinFile(Filename, Skip, Count, CountLoc)) 5052 return Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 5053 return false; 5054 } 5055 5056 /// parseDirectiveIf 5057 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 5058 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 5059 TheCondStack.push_back(TheCondState); 5060 TheCondState.TheCond = AsmCond::IfCond; 5061 if (TheCondState.Ignore) { 5062 eatToEndOfStatement(); 5063 } else { 5064 int64_t ExprValue; 5065 if (parseAbsoluteExpression(ExprValue) || parseEOL()) 5066 return true; 5067 5068 switch (DirKind) { 5069 default: 5070 llvm_unreachable("unsupported directive"); 5071 case DK_IF: 5072 case DK_IFNE: 5073 break; 5074 case DK_IFEQ: 5075 ExprValue = ExprValue == 0; 5076 break; 5077 case DK_IFGE: 5078 ExprValue = ExprValue >= 0; 5079 break; 5080 case DK_IFGT: 5081 ExprValue = ExprValue > 0; 5082 break; 5083 case DK_IFLE: 5084 ExprValue = ExprValue <= 0; 5085 break; 5086 case DK_IFLT: 5087 ExprValue = ExprValue < 0; 5088 break; 5089 } 5090 5091 TheCondState.CondMet = ExprValue; 5092 TheCondState.Ignore = !TheCondState.CondMet; 5093 } 5094 5095 return false; 5096 } 5097 5098 /// parseDirectiveIfb 5099 /// ::= .ifb string 5100 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 5101 TheCondStack.push_back(TheCondState); 5102 TheCondState.TheCond = AsmCond::IfCond; 5103 5104 if (TheCondState.Ignore) { 5105 eatToEndOfStatement(); 5106 } else { 5107 StringRef Str = parseStringToEndOfStatement(); 5108 5109 if (parseEOL()) 5110 return true; 5111 5112 TheCondState.CondMet = ExpectBlank == Str.empty(); 5113 TheCondState.Ignore = !TheCondState.CondMet; 5114 } 5115 5116 return false; 5117 } 5118 5119 /// parseDirectiveIfc 5120 /// ::= .ifc string1, string2 5121 /// ::= .ifnc string1, string2 5122 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 5123 TheCondStack.push_back(TheCondState); 5124 TheCondState.TheCond = AsmCond::IfCond; 5125 5126 if (TheCondState.Ignore) { 5127 eatToEndOfStatement(); 5128 } else { 5129 StringRef Str1 = parseStringToComma(); 5130 5131 if (parseComma()) 5132 return true; 5133 5134 StringRef Str2 = parseStringToEndOfStatement(); 5135 5136 if (parseEOL()) 5137 return true; 5138 5139 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 5140 TheCondState.Ignore = !TheCondState.CondMet; 5141 } 5142 5143 return false; 5144 } 5145 5146 /// parseDirectiveIfeqs 5147 /// ::= .ifeqs string1, string2 5148 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 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 String1 = getTok().getStringContents(); 5156 Lex(); 5157 5158 if (Lexer.isNot(AsmToken::Comma)) { 5159 if (ExpectEqual) 5160 return TokError( 5161 "expected comma after first string for '.ifeqs' directive"); 5162 return TokError("expected comma after first string for '.ifnes' directive"); 5163 } 5164 5165 Lex(); 5166 5167 if (Lexer.isNot(AsmToken::String)) { 5168 if (ExpectEqual) 5169 return TokError("expected string parameter for '.ifeqs' directive"); 5170 return TokError("expected string parameter for '.ifnes' directive"); 5171 } 5172 5173 StringRef String2 = getTok().getStringContents(); 5174 Lex(); 5175 5176 TheCondStack.push_back(TheCondState); 5177 TheCondState.TheCond = AsmCond::IfCond; 5178 TheCondState.CondMet = ExpectEqual == (String1 == String2); 5179 TheCondState.Ignore = !TheCondState.CondMet; 5180 5181 return false; 5182 } 5183 5184 /// parseDirectiveIfdef 5185 /// ::= .ifdef symbol 5186 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 5187 StringRef Name; 5188 TheCondStack.push_back(TheCondState); 5189 TheCondState.TheCond = AsmCond::IfCond; 5190 5191 if (TheCondState.Ignore) { 5192 eatToEndOfStatement(); 5193 } else { 5194 if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") || 5195 parseEOL()) 5196 return true; 5197 5198 MCSymbol *Sym = getContext().lookupSymbol(Name); 5199 5200 if (expect_defined) 5201 TheCondState.CondMet = (Sym && !Sym->isUndefined(false)); 5202 else 5203 TheCondState.CondMet = (!Sym || Sym->isUndefined(false)); 5204 TheCondState.Ignore = !TheCondState.CondMet; 5205 } 5206 5207 return false; 5208 } 5209 5210 /// parseDirectiveElseIf 5211 /// ::= .elseif expression 5212 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 5213 if (TheCondState.TheCond != AsmCond::IfCond && 5214 TheCondState.TheCond != AsmCond::ElseIfCond) 5215 return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an" 5216 " .if or an .elseif"); 5217 TheCondState.TheCond = AsmCond::ElseIfCond; 5218 5219 bool LastIgnoreState = false; 5220 if (!TheCondStack.empty()) 5221 LastIgnoreState = TheCondStack.back().Ignore; 5222 if (LastIgnoreState || TheCondState.CondMet) { 5223 TheCondState.Ignore = true; 5224 eatToEndOfStatement(); 5225 } else { 5226 int64_t ExprValue; 5227 if (parseAbsoluteExpression(ExprValue)) 5228 return true; 5229 5230 if (parseEOL()) 5231 return true; 5232 5233 TheCondState.CondMet = ExprValue; 5234 TheCondState.Ignore = !TheCondState.CondMet; 5235 } 5236 5237 return false; 5238 } 5239 5240 /// parseDirectiveElse 5241 /// ::= .else 5242 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 5243 if (parseEOL()) 5244 return true; 5245 5246 if (TheCondState.TheCond != AsmCond::IfCond && 5247 TheCondState.TheCond != AsmCond::ElseIfCond) 5248 return Error(DirectiveLoc, "Encountered a .else that doesn't follow " 5249 " an .if or an .elseif"); 5250 TheCondState.TheCond = AsmCond::ElseCond; 5251 bool LastIgnoreState = false; 5252 if (!TheCondStack.empty()) 5253 LastIgnoreState = TheCondStack.back().Ignore; 5254 if (LastIgnoreState || TheCondState.CondMet) 5255 TheCondState.Ignore = true; 5256 else 5257 TheCondState.Ignore = false; 5258 5259 return false; 5260 } 5261 5262 /// parseDirectiveEnd 5263 /// ::= .end 5264 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 5265 if (parseEOL()) 5266 return true; 5267 5268 while (Lexer.isNot(AsmToken::Eof)) 5269 Lexer.Lex(); 5270 5271 return false; 5272 } 5273 5274 /// parseDirectiveError 5275 /// ::= .err 5276 /// ::= .error [string] 5277 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 5278 if (!TheCondStack.empty()) { 5279 if (TheCondStack.back().Ignore) { 5280 eatToEndOfStatement(); 5281 return false; 5282 } 5283 } 5284 5285 if (!WithMessage) 5286 return Error(L, ".err encountered"); 5287 5288 StringRef Message = ".error directive invoked in source file"; 5289 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5290 if (Lexer.isNot(AsmToken::String)) 5291 return TokError(".error argument must be a string"); 5292 5293 Message = getTok().getStringContents(); 5294 Lex(); 5295 } 5296 5297 return Error(L, Message); 5298 } 5299 5300 /// parseDirectiveWarning 5301 /// ::= .warning [string] 5302 bool AsmParser::parseDirectiveWarning(SMLoc L) { 5303 if (!TheCondStack.empty()) { 5304 if (TheCondStack.back().Ignore) { 5305 eatToEndOfStatement(); 5306 return false; 5307 } 5308 } 5309 5310 StringRef Message = ".warning directive invoked in source file"; 5311 5312 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 5313 if (Lexer.isNot(AsmToken::String)) 5314 return TokError(".warning argument must be a string"); 5315 5316 Message = getTok().getStringContents(); 5317 Lex(); 5318 if (parseEOL()) 5319 return true; 5320 } 5321 5322 return Warning(L, Message); 5323 } 5324 5325 /// parseDirectiveEndIf 5326 /// ::= .endif 5327 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 5328 if (parseEOL()) 5329 return true; 5330 5331 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 5332 return Error(DirectiveLoc, "Encountered a .endif that doesn't follow " 5333 "an .if or .else"); 5334 if (!TheCondStack.empty()) { 5335 TheCondState = TheCondStack.back(); 5336 TheCondStack.pop_back(); 5337 } 5338 5339 return false; 5340 } 5341 5342 void AsmParser::initializeDirectiveKindMap() { 5343 /* Lookup will be done with the directive 5344 * converted to lower case, so all these 5345 * keys should be lower case. 5346 * (target specific directives are handled 5347 * elsewhere) 5348 */ 5349 DirectiveKindMap[".set"] = DK_SET; 5350 DirectiveKindMap[".equ"] = DK_EQU; 5351 DirectiveKindMap[".equiv"] = DK_EQUIV; 5352 DirectiveKindMap[".ascii"] = DK_ASCII; 5353 DirectiveKindMap[".asciz"] = DK_ASCIZ; 5354 DirectiveKindMap[".string"] = DK_STRING; 5355 DirectiveKindMap[".byte"] = DK_BYTE; 5356 DirectiveKindMap[".short"] = DK_SHORT; 5357 DirectiveKindMap[".value"] = DK_VALUE; 5358 DirectiveKindMap[".2byte"] = DK_2BYTE; 5359 DirectiveKindMap[".long"] = DK_LONG; 5360 DirectiveKindMap[".int"] = DK_INT; 5361 DirectiveKindMap[".4byte"] = DK_4BYTE; 5362 DirectiveKindMap[".quad"] = DK_QUAD; 5363 DirectiveKindMap[".8byte"] = DK_8BYTE; 5364 DirectiveKindMap[".octa"] = DK_OCTA; 5365 DirectiveKindMap[".single"] = DK_SINGLE; 5366 DirectiveKindMap[".float"] = DK_FLOAT; 5367 DirectiveKindMap[".double"] = DK_DOUBLE; 5368 DirectiveKindMap[".align"] = DK_ALIGN; 5369 DirectiveKindMap[".align32"] = DK_ALIGN32; 5370 DirectiveKindMap[".balign"] = DK_BALIGN; 5371 DirectiveKindMap[".balignw"] = DK_BALIGNW; 5372 DirectiveKindMap[".balignl"] = DK_BALIGNL; 5373 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 5374 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 5375 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 5376 DirectiveKindMap[".org"] = DK_ORG; 5377 DirectiveKindMap[".fill"] = DK_FILL; 5378 DirectiveKindMap[".zero"] = DK_ZERO; 5379 DirectiveKindMap[".extern"] = DK_EXTERN; 5380 DirectiveKindMap[".globl"] = DK_GLOBL; 5381 DirectiveKindMap[".global"] = DK_GLOBAL; 5382 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 5383 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 5384 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 5385 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 5386 DirectiveKindMap[".reference"] = DK_REFERENCE; 5387 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 5388 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 5389 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 5390 DirectiveKindMap[".cold"] = DK_COLD; 5391 DirectiveKindMap[".comm"] = DK_COMM; 5392 DirectiveKindMap[".common"] = DK_COMMON; 5393 DirectiveKindMap[".lcomm"] = DK_LCOMM; 5394 DirectiveKindMap[".abort"] = DK_ABORT; 5395 DirectiveKindMap[".include"] = DK_INCLUDE; 5396 DirectiveKindMap[".incbin"] = DK_INCBIN; 5397 DirectiveKindMap[".code16"] = DK_CODE16; 5398 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 5399 DirectiveKindMap[".rept"] = DK_REPT; 5400 DirectiveKindMap[".rep"] = DK_REPT; 5401 DirectiveKindMap[".irp"] = DK_IRP; 5402 DirectiveKindMap[".irpc"] = DK_IRPC; 5403 DirectiveKindMap[".endr"] = DK_ENDR; 5404 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 5405 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 5406 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 5407 DirectiveKindMap[".if"] = DK_IF; 5408 DirectiveKindMap[".ifeq"] = DK_IFEQ; 5409 DirectiveKindMap[".ifge"] = DK_IFGE; 5410 DirectiveKindMap[".ifgt"] = DK_IFGT; 5411 DirectiveKindMap[".ifle"] = DK_IFLE; 5412 DirectiveKindMap[".iflt"] = DK_IFLT; 5413 DirectiveKindMap[".ifne"] = DK_IFNE; 5414 DirectiveKindMap[".ifb"] = DK_IFB; 5415 DirectiveKindMap[".ifnb"] = DK_IFNB; 5416 DirectiveKindMap[".ifc"] = DK_IFC; 5417 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 5418 DirectiveKindMap[".ifnc"] = DK_IFNC; 5419 DirectiveKindMap[".ifnes"] = DK_IFNES; 5420 DirectiveKindMap[".ifdef"] = DK_IFDEF; 5421 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 5422 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 5423 DirectiveKindMap[".elseif"] = DK_ELSEIF; 5424 DirectiveKindMap[".else"] = DK_ELSE; 5425 DirectiveKindMap[".end"] = DK_END; 5426 DirectiveKindMap[".endif"] = DK_ENDIF; 5427 DirectiveKindMap[".skip"] = DK_SKIP; 5428 DirectiveKindMap[".space"] = DK_SPACE; 5429 DirectiveKindMap[".file"] = DK_FILE; 5430 DirectiveKindMap[".line"] = DK_LINE; 5431 DirectiveKindMap[".loc"] = DK_LOC; 5432 DirectiveKindMap[".stabs"] = DK_STABS; 5433 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 5434 DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID; 5435 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 5436 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 5437 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 5438 DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID; 5439 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 5440 DirectiveKindMap[".cv_string"] = DK_CV_STRING; 5441 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 5442 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 5443 DirectiveKindMap[".cv_filechecksumoffset"] = DK_CV_FILECHECKSUM_OFFSET; 5444 DirectiveKindMap[".cv_fpo_data"] = DK_CV_FPO_DATA; 5445 DirectiveKindMap[".sleb128"] = DK_SLEB128; 5446 DirectiveKindMap[".uleb128"] = DK_ULEB128; 5447 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 5448 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 5449 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 5450 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 5451 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 5452 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 5453 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 5454 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 5455 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 5456 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 5457 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 5458 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 5459 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 5460 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 5461 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 5462 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 5463 DirectiveKindMap[".cfi_return_column"] = DK_CFI_RETURN_COLUMN; 5464 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 5465 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 5466 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 5467 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 5468 DirectiveKindMap[".cfi_b_key_frame"] = DK_CFI_B_KEY_FRAME; 5469 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 5470 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 5471 DirectiveKindMap[".macro"] = DK_MACRO; 5472 DirectiveKindMap[".exitm"] = DK_EXITM; 5473 DirectiveKindMap[".endm"] = DK_ENDM; 5474 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 5475 DirectiveKindMap[".purgem"] = DK_PURGEM; 5476 DirectiveKindMap[".err"] = DK_ERR; 5477 DirectiveKindMap[".error"] = DK_ERROR; 5478 DirectiveKindMap[".warning"] = DK_WARNING; 5479 DirectiveKindMap[".altmacro"] = DK_ALTMACRO; 5480 DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO; 5481 DirectiveKindMap[".reloc"] = DK_RELOC; 5482 DirectiveKindMap[".dc"] = DK_DC; 5483 DirectiveKindMap[".dc.a"] = DK_DC_A; 5484 DirectiveKindMap[".dc.b"] = DK_DC_B; 5485 DirectiveKindMap[".dc.d"] = DK_DC_D; 5486 DirectiveKindMap[".dc.l"] = DK_DC_L; 5487 DirectiveKindMap[".dc.s"] = DK_DC_S; 5488 DirectiveKindMap[".dc.w"] = DK_DC_W; 5489 DirectiveKindMap[".dc.x"] = DK_DC_X; 5490 DirectiveKindMap[".dcb"] = DK_DCB; 5491 DirectiveKindMap[".dcb.b"] = DK_DCB_B; 5492 DirectiveKindMap[".dcb.d"] = DK_DCB_D; 5493 DirectiveKindMap[".dcb.l"] = DK_DCB_L; 5494 DirectiveKindMap[".dcb.s"] = DK_DCB_S; 5495 DirectiveKindMap[".dcb.w"] = DK_DCB_W; 5496 DirectiveKindMap[".dcb.x"] = DK_DCB_X; 5497 DirectiveKindMap[".ds"] = DK_DS; 5498 DirectiveKindMap[".ds.b"] = DK_DS_B; 5499 DirectiveKindMap[".ds.d"] = DK_DS_D; 5500 DirectiveKindMap[".ds.l"] = DK_DS_L; 5501 DirectiveKindMap[".ds.p"] = DK_DS_P; 5502 DirectiveKindMap[".ds.s"] = DK_DS_S; 5503 DirectiveKindMap[".ds.w"] = DK_DS_W; 5504 DirectiveKindMap[".ds.x"] = DK_DS_X; 5505 DirectiveKindMap[".print"] = DK_PRINT; 5506 DirectiveKindMap[".addrsig"] = DK_ADDRSIG; 5507 DirectiveKindMap[".addrsig_sym"] = DK_ADDRSIG_SYM; 5508 DirectiveKindMap[".pseudoprobe"] = DK_PSEUDO_PROBE; 5509 DirectiveKindMap[".lto_discard"] = DK_LTO_DISCARD; 5510 } 5511 5512 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 5513 AsmToken EndToken, StartToken = getTok(); 5514 5515 unsigned NestLevel = 0; 5516 while (true) { 5517 // Check whether we have reached the end of the file. 5518 if (getLexer().is(AsmToken::Eof)) { 5519 printError(DirectiveLoc, "no matching '.endr' in definition"); 5520 return nullptr; 5521 } 5522 5523 if (Lexer.is(AsmToken::Identifier) && 5524 (getTok().getIdentifier() == ".rep" || 5525 getTok().getIdentifier() == ".rept" || 5526 getTok().getIdentifier() == ".irp" || 5527 getTok().getIdentifier() == ".irpc")) { 5528 ++NestLevel; 5529 } 5530 5531 // Otherwise, check whether we have reached the .endr. 5532 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 5533 if (NestLevel == 0) { 5534 EndToken = getTok(); 5535 Lex(); 5536 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5537 printError(getTok().getLoc(), 5538 "unexpected token in '.endr' directive"); 5539 return nullptr; 5540 } 5541 break; 5542 } 5543 --NestLevel; 5544 } 5545 5546 // Otherwise, scan till the end of the statement. 5547 eatToEndOfStatement(); 5548 } 5549 5550 const char *BodyStart = StartToken.getLoc().getPointer(); 5551 const char *BodyEnd = EndToken.getLoc().getPointer(); 5552 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 5553 5554 // We Are Anonymous. 5555 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 5556 return &MacroLikeBodies.back(); 5557 } 5558 5559 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 5560 raw_svector_ostream &OS) { 5561 OS << ".endr\n"; 5562 5563 std::unique_ptr<MemoryBuffer> Instantiation = 5564 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 5565 5566 // Create the macro instantiation object and add to the current macro 5567 // instantiation stack. 5568 MacroInstantiation *MI = new MacroInstantiation{ 5569 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 5570 ActiveMacros.push_back(MI); 5571 5572 // Jump to the macro instantiation and prime the lexer. 5573 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 5574 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 5575 Lex(); 5576 } 5577 5578 /// parseDirectiveRept 5579 /// ::= .rep | .rept count 5580 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 5581 const MCExpr *CountExpr; 5582 SMLoc CountLoc = getTok().getLoc(); 5583 if (parseExpression(CountExpr)) 5584 return true; 5585 5586 int64_t Count; 5587 if (!CountExpr->evaluateAsAbsolute(Count, getStreamer().getAssemblerPtr())) { 5588 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 5589 } 5590 5591 if (check(Count < 0, CountLoc, "Count is negative") || parseEOL()) 5592 return true; 5593 5594 // Lex the rept definition. 5595 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5596 if (!M) 5597 return true; 5598 5599 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5600 // to hold the macro body with substitutions. 5601 SmallString<256> Buf; 5602 raw_svector_ostream OS(Buf); 5603 while (Count--) { 5604 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 5605 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 5606 return true; 5607 } 5608 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5609 5610 return false; 5611 } 5612 5613 /// parseDirectiveIrp 5614 /// ::= .irp symbol,values 5615 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 5616 MCAsmMacroParameter Parameter; 5617 MCAsmMacroArguments A; 5618 if (check(parseIdentifier(Parameter.Name), 5619 "expected identifier in '.irp' directive") || 5620 parseComma() || parseMacroArguments(nullptr, A) || parseEOL()) 5621 return true; 5622 5623 // Lex the irp definition. 5624 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5625 if (!M) 5626 return true; 5627 5628 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5629 // to hold the macro body with substitutions. 5630 SmallString<256> Buf; 5631 raw_svector_ostream OS(Buf); 5632 5633 for (const MCAsmMacroArgument &Arg : A) { 5634 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 5635 // This is undocumented, but GAS seems to support it. 5636 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5637 return true; 5638 } 5639 5640 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5641 5642 return false; 5643 } 5644 5645 /// parseDirectiveIrpc 5646 /// ::= .irpc symbol,values 5647 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 5648 MCAsmMacroParameter Parameter; 5649 MCAsmMacroArguments A; 5650 5651 if (check(parseIdentifier(Parameter.Name), 5652 "expected identifier in '.irpc' directive") || 5653 parseComma() || parseMacroArguments(nullptr, A)) 5654 return true; 5655 5656 if (A.size() != 1 || A.front().size() != 1) 5657 return TokError("unexpected token in '.irpc' directive"); 5658 if (parseEOL()) 5659 return true; 5660 5661 // Lex the irpc definition. 5662 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5663 if (!M) 5664 return true; 5665 5666 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5667 // to hold the macro body with substitutions. 5668 SmallString<256> Buf; 5669 raw_svector_ostream OS(Buf); 5670 5671 StringRef Values = A.front().front().getString(); 5672 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 5673 MCAsmMacroArgument Arg; 5674 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 5675 5676 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 5677 // This is undocumented, but GAS seems to support it. 5678 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5679 return true; 5680 } 5681 5682 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5683 5684 return false; 5685 } 5686 5687 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 5688 if (ActiveMacros.empty()) 5689 return TokError("unmatched '.endr' directive"); 5690 5691 // The only .repl that should get here are the ones created by 5692 // instantiateMacroLikeBody. 5693 assert(getLexer().is(AsmToken::EndOfStatement)); 5694 5695 handleMacroExit(); 5696 return false; 5697 } 5698 5699 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 5700 size_t Len) { 5701 const MCExpr *Value; 5702 SMLoc ExprLoc = getLexer().getLoc(); 5703 if (parseExpression(Value)) 5704 return true; 5705 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5706 if (!MCE) 5707 return Error(ExprLoc, "unexpected expression in _emit"); 5708 uint64_t IntValue = MCE->getValue(); 5709 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 5710 return Error(ExprLoc, "literal value out of range for directive"); 5711 5712 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 5713 return false; 5714 } 5715 5716 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 5717 const MCExpr *Value; 5718 SMLoc ExprLoc = getLexer().getLoc(); 5719 if (parseExpression(Value)) 5720 return true; 5721 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5722 if (!MCE) 5723 return Error(ExprLoc, "unexpected expression in align"); 5724 uint64_t IntValue = MCE->getValue(); 5725 if (!isPowerOf2_64(IntValue)) 5726 return Error(ExprLoc, "literal value not a power of two greater then zero"); 5727 5728 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 5729 return false; 5730 } 5731 5732 bool AsmParser::parseDirectivePrint(SMLoc DirectiveLoc) { 5733 const AsmToken StrTok = getTok(); 5734 Lex(); 5735 if (StrTok.isNot(AsmToken::String) || StrTok.getString().front() != '"') 5736 return Error(DirectiveLoc, "expected double quoted string after .print"); 5737 if (parseEOL()) 5738 return true; 5739 llvm::outs() << StrTok.getStringContents() << '\n'; 5740 return false; 5741 } 5742 5743 bool AsmParser::parseDirectiveAddrsig() { 5744 if (parseEOL()) 5745 return true; 5746 getStreamer().emitAddrsig(); 5747 return false; 5748 } 5749 5750 bool AsmParser::parseDirectiveAddrsigSym() { 5751 StringRef Name; 5752 if (check(parseIdentifier(Name), "expected identifier") || parseEOL()) 5753 return true; 5754 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 5755 getStreamer().emitAddrsigSym(Sym); 5756 return false; 5757 } 5758 5759 bool AsmParser::parseDirectivePseudoProbe() { 5760 int64_t Guid; 5761 int64_t Index; 5762 int64_t Type; 5763 int64_t Attr; 5764 5765 if (getLexer().is(AsmToken::Integer)) { 5766 if (parseIntToken(Guid, "unexpected token in '.pseudoprobe' directive")) 5767 return true; 5768 } 5769 5770 if (getLexer().is(AsmToken::Integer)) { 5771 if (parseIntToken(Index, "unexpected token in '.pseudoprobe' directive")) 5772 return true; 5773 } 5774 5775 if (getLexer().is(AsmToken::Integer)) { 5776 if (parseIntToken(Type, "unexpected token in '.pseudoprobe' directive")) 5777 return true; 5778 } 5779 5780 if (getLexer().is(AsmToken::Integer)) { 5781 if (parseIntToken(Attr, "unexpected token in '.pseudoprobe' directive")) 5782 return true; 5783 } 5784 5785 // Parse inline stack like @ GUID:11:12 @ GUID:1:11 @ GUID:3:21 5786 MCPseudoProbeInlineStack InlineStack; 5787 5788 while (getLexer().is(AsmToken::At)) { 5789 // eat @ 5790 Lex(); 5791 5792 int64_t CallerGuid = 0; 5793 if (getLexer().is(AsmToken::Integer)) { 5794 if (parseIntToken(CallerGuid, 5795 "unexpected token in '.pseudoprobe' directive")) 5796 return true; 5797 } 5798 5799 // eat colon 5800 if (getLexer().is(AsmToken::Colon)) 5801 Lex(); 5802 5803 int64_t CallerProbeId = 0; 5804 if (getLexer().is(AsmToken::Integer)) { 5805 if (parseIntToken(CallerProbeId, 5806 "unexpected token in '.pseudoprobe' directive")) 5807 return true; 5808 } 5809 5810 InlineSite Site(CallerGuid, CallerProbeId); 5811 InlineStack.push_back(Site); 5812 } 5813 5814 if (parseEOL()) 5815 return true; 5816 5817 getStreamer().emitPseudoProbe(Guid, Index, Type, Attr, InlineStack); 5818 return false; 5819 } 5820 5821 /// parseDirectiveLTODiscard 5822 /// ::= ".lto_discard" [ identifier ( , identifier )* ] 5823 /// The LTO library emits this directive to discard non-prevailing symbols. 5824 /// We ignore symbol assignments and attribute changes for the specified 5825 /// symbols. 5826 bool AsmParser::parseDirectiveLTODiscard() { 5827 auto ParseOp = [&]() -> bool { 5828 StringRef Name; 5829 SMLoc Loc = getTok().getLoc(); 5830 if (parseIdentifier(Name)) 5831 return Error(Loc, "expected identifier"); 5832 LTODiscardSymbols.insert(Name); 5833 return false; 5834 }; 5835 5836 LTODiscardSymbols.clear(); 5837 return parseMany(ParseOp); 5838 } 5839 5840 // We are comparing pointers, but the pointers are relative to a single string. 5841 // Thus, this should always be deterministic. 5842 static int rewritesSort(const AsmRewrite *AsmRewriteA, 5843 const AsmRewrite *AsmRewriteB) { 5844 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 5845 return -1; 5846 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 5847 return 1; 5848 5849 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 5850 // rewrite to the same location. Make sure the SizeDirective rewrite is 5851 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 5852 // ensures the sort algorithm is stable. 5853 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 5854 AsmRewritePrecedence[AsmRewriteB->Kind]) 5855 return -1; 5856 5857 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 5858 AsmRewritePrecedence[AsmRewriteB->Kind]) 5859 return 1; 5860 llvm_unreachable("Unstable rewrite sort."); 5861 } 5862 5863 bool AsmParser::parseMSInlineAsm( 5864 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 5865 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls, 5866 SmallVectorImpl<std::string> &Constraints, 5867 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 5868 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 5869 SmallVector<void *, 4> InputDecls; 5870 SmallVector<void *, 4> OutputDecls; 5871 SmallVector<bool, 4> InputDeclsAddressOf; 5872 SmallVector<bool, 4> OutputDeclsAddressOf; 5873 SmallVector<std::string, 4> InputConstraints; 5874 SmallVector<std::string, 4> OutputConstraints; 5875 SmallVector<unsigned, 4> ClobberRegs; 5876 5877 SmallVector<AsmRewrite, 4> AsmStrRewrites; 5878 5879 // Prime the lexer. 5880 Lex(); 5881 5882 // While we have input, parse each statement. 5883 unsigned InputIdx = 0; 5884 unsigned OutputIdx = 0; 5885 while (getLexer().isNot(AsmToken::Eof)) { 5886 // Parse curly braces marking block start/end 5887 if (parseCurlyBlockScope(AsmStrRewrites)) 5888 continue; 5889 5890 ParseStatementInfo Info(&AsmStrRewrites); 5891 bool StatementErr = parseStatement(Info, &SI); 5892 5893 if (StatementErr || Info.ParseError) { 5894 // Emit pending errors if any exist. 5895 printPendingErrors(); 5896 return true; 5897 } 5898 5899 // No pending error should exist here. 5900 assert(!hasPendingError() && "unexpected error from parseStatement"); 5901 5902 if (Info.Opcode == ~0U) 5903 continue; 5904 5905 const MCInstrDesc &Desc = MII->get(Info.Opcode); 5906 5907 // Build the list of clobbers, outputs and inputs. 5908 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 5909 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 5910 5911 // Register operand. 5912 if (Operand.isReg() && !Operand.needAddressOf() && 5913 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5914 unsigned NumDefs = Desc.getNumDefs(); 5915 // Clobber. 5916 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5917 ClobberRegs.push_back(Operand.getReg()); 5918 continue; 5919 } 5920 5921 // Expr/Input or Output. 5922 StringRef SymName = Operand.getSymName(); 5923 if (SymName.empty()) 5924 continue; 5925 5926 void *OpDecl = Operand.getOpDecl(); 5927 if (!OpDecl) 5928 continue; 5929 5930 StringRef Constraint = Operand.getConstraint(); 5931 if (Operand.isImm()) { 5932 // Offset as immediate 5933 if (Operand.isOffsetOfLocal()) 5934 Constraint = "r"; 5935 else 5936 Constraint = "i"; 5937 } 5938 5939 bool isOutput = (i == 1) && Desc.mayStore(); 5940 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5941 if (isOutput) { 5942 ++InputIdx; 5943 OutputDecls.push_back(OpDecl); 5944 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5945 OutputConstraints.push_back(("=" + Constraint).str()); 5946 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5947 } else { 5948 InputDecls.push_back(OpDecl); 5949 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5950 InputConstraints.push_back(Constraint.str()); 5951 if (Desc.OpInfo[i - 1].isBranchTarget()) 5952 AsmStrRewrites.emplace_back(AOK_CallInput, Start, SymName.size()); 5953 else 5954 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5955 } 5956 } 5957 5958 // Consider implicit defs to be clobbers. Think of cpuid and push. 5959 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5960 Desc.getNumImplicitDefs()); 5961 llvm::append_range(ClobberRegs, ImpDefs); 5962 } 5963 5964 // Set the number of Outputs and Inputs. 5965 NumOutputs = OutputDecls.size(); 5966 NumInputs = InputDecls.size(); 5967 5968 // Set the unique clobbers. 5969 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5970 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5971 ClobberRegs.end()); 5972 Clobbers.assign(ClobberRegs.size(), std::string()); 5973 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5974 raw_string_ostream OS(Clobbers[I]); 5975 IP->printRegName(OS, ClobberRegs[I]); 5976 } 5977 5978 // Merge the various outputs and inputs. Output are expected first. 5979 if (NumOutputs || NumInputs) { 5980 unsigned NumExprs = NumOutputs + NumInputs; 5981 OpDecls.resize(NumExprs); 5982 Constraints.resize(NumExprs); 5983 for (unsigned i = 0; i < NumOutputs; ++i) { 5984 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5985 Constraints[i] = OutputConstraints[i]; 5986 } 5987 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5988 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5989 Constraints[j] = InputConstraints[i]; 5990 } 5991 } 5992 5993 // Build the IR assembly string. 5994 std::string AsmStringIR; 5995 raw_string_ostream OS(AsmStringIR); 5996 StringRef ASMString = 5997 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5998 const char *AsmStart = ASMString.begin(); 5999 const char *AsmEnd = ASMString.end(); 6000 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 6001 for (auto it = AsmStrRewrites.begin(); it != AsmStrRewrites.end(); ++it) { 6002 const AsmRewrite &AR = *it; 6003 // Check if this has already been covered by another rewrite... 6004 if (AR.Done) 6005 continue; 6006 AsmRewriteKind Kind = AR.Kind; 6007 6008 const char *Loc = AR.Loc.getPointer(); 6009 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 6010 6011 // Emit everything up to the immediate/expression. 6012 if (unsigned Len = Loc - AsmStart) 6013 OS << StringRef(AsmStart, Len); 6014 6015 // Skip the original expression. 6016 if (Kind == AOK_Skip) { 6017 AsmStart = Loc + AR.Len; 6018 continue; 6019 } 6020 6021 unsigned AdditionalSkip = 0; 6022 // Rewrite expressions in $N notation. 6023 switch (Kind) { 6024 default: 6025 break; 6026 case AOK_IntelExpr: 6027 assert(AR.IntelExp.isValid() && "cannot write invalid intel expression"); 6028 if (AR.IntelExp.NeedBracs) 6029 OS << "["; 6030 if (AR.IntelExp.hasBaseReg()) 6031 OS << AR.IntelExp.BaseReg; 6032 if (AR.IntelExp.hasIndexReg()) 6033 OS << (AR.IntelExp.hasBaseReg() ? " + " : "") 6034 << AR.IntelExp.IndexReg; 6035 if (AR.IntelExp.Scale > 1) 6036 OS << " * $$" << AR.IntelExp.Scale; 6037 if (AR.IntelExp.hasOffset()) { 6038 if (AR.IntelExp.hasRegs()) 6039 OS << " + "; 6040 // Fuse this rewrite with a rewrite of the offset name, if present. 6041 StringRef OffsetName = AR.IntelExp.OffsetName; 6042 SMLoc OffsetLoc = SMLoc::getFromPointer(AR.IntelExp.OffsetName.data()); 6043 size_t OffsetLen = OffsetName.size(); 6044 auto rewrite_it = std::find_if( 6045 it, AsmStrRewrites.end(), [&](const AsmRewrite &FusingAR) { 6046 return FusingAR.Loc == OffsetLoc && FusingAR.Len == OffsetLen && 6047 (FusingAR.Kind == AOK_Input || 6048 FusingAR.Kind == AOK_CallInput); 6049 }); 6050 if (rewrite_it == AsmStrRewrites.end()) { 6051 OS << "offset " << OffsetName; 6052 } else if (rewrite_it->Kind == AOK_CallInput) { 6053 OS << "${" << InputIdx++ << ":P}"; 6054 rewrite_it->Done = true; 6055 } else { 6056 OS << '$' << InputIdx++; 6057 rewrite_it->Done = true; 6058 } 6059 } 6060 if (AR.IntelExp.Imm || AR.IntelExp.emitImm()) 6061 OS << (AR.IntelExp.emitImm() ? "$$" : " + $$") << AR.IntelExp.Imm; 6062 if (AR.IntelExp.NeedBracs) 6063 OS << "]"; 6064 break; 6065 case AOK_Label: 6066 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 6067 break; 6068 case AOK_Input: 6069 OS << '$' << InputIdx++; 6070 break; 6071 case AOK_CallInput: 6072 OS << "${" << InputIdx++ << ":P}"; 6073 break; 6074 case AOK_Output: 6075 OS << '$' << OutputIdx++; 6076 break; 6077 case AOK_SizeDirective: 6078 switch (AR.Val) { 6079 default: break; 6080 case 8: OS << "byte ptr "; break; 6081 case 16: OS << "word ptr "; break; 6082 case 32: OS << "dword ptr "; break; 6083 case 64: OS << "qword ptr "; break; 6084 case 80: OS << "xword ptr "; break; 6085 case 128: OS << "xmmword ptr "; break; 6086 case 256: OS << "ymmword ptr "; break; 6087 } 6088 break; 6089 case AOK_Emit: 6090 OS << ".byte"; 6091 break; 6092 case AOK_Align: { 6093 // MS alignment directives are measured in bytes. If the native assembler 6094 // measures alignment in bytes, we can pass it straight through. 6095 OS << ".align"; 6096 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 6097 break; 6098 6099 // Alignment is in log2 form, so print that instead and skip the original 6100 // immediate. 6101 unsigned Val = AR.Val; 6102 OS << ' ' << Val; 6103 assert(Val < 10 && "Expected alignment less then 2^10."); 6104 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 6105 break; 6106 } 6107 case AOK_EVEN: 6108 OS << ".even"; 6109 break; 6110 case AOK_EndOfStatement: 6111 OS << "\n\t"; 6112 break; 6113 } 6114 6115 // Skip the original expression. 6116 AsmStart = Loc + AR.Len + AdditionalSkip; 6117 } 6118 6119 // Emit the remainder of the asm string. 6120 if (AsmStart != AsmEnd) 6121 OS << StringRef(AsmStart, AsmEnd - AsmStart); 6122 6123 AsmString = OS.str(); 6124 return false; 6125 } 6126 6127 namespace llvm { 6128 namespace MCParserUtils { 6129 6130 /// Returns whether the given symbol is used anywhere in the given expression, 6131 /// or subexpressions. 6132 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 6133 switch (Value->getKind()) { 6134 case MCExpr::Binary: { 6135 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 6136 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 6137 isSymbolUsedInExpression(Sym, BE->getRHS()); 6138 } 6139 case MCExpr::Target: 6140 case MCExpr::Constant: 6141 return false; 6142 case MCExpr::SymbolRef: { 6143 const MCSymbol &S = 6144 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 6145 if (S.isVariable()) 6146 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 6147 return &S == Sym; 6148 } 6149 case MCExpr::Unary: 6150 return isSymbolUsedInExpression( 6151 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 6152 } 6153 6154 llvm_unreachable("Unknown expr kind!"); 6155 } 6156 6157 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 6158 MCAsmParser &Parser, MCSymbol *&Sym, 6159 const MCExpr *&Value) { 6160 6161 // FIXME: Use better location, we should use proper tokens. 6162 SMLoc EqualLoc = Parser.getTok().getLoc(); 6163 if (Parser.parseExpression(Value)) 6164 return Parser.TokError("missing expression"); 6165 6166 // Note: we don't count b as used in "a = b". This is to allow 6167 // a = b 6168 // b = c 6169 6170 if (Parser.parseEOL()) 6171 return true; 6172 6173 // Validate that the LHS is allowed to be a variable (either it has not been 6174 // used as a symbol, or it is an absolute symbol). 6175 Sym = Parser.getContext().lookupSymbol(Name); 6176 if (Sym) { 6177 // Diagnose assignment to a label. 6178 // 6179 // FIXME: Diagnostics. Note the location of the definition as a label. 6180 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 6181 if (isSymbolUsedInExpression(Sym, Value)) 6182 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 6183 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 6184 !Sym->isVariable()) 6185 ; // Allow redefinitions of undefined symbols only used in directives. 6186 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 6187 ; // Allow redefinitions of variables that haven't yet been used. 6188 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 6189 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 6190 else if (!Sym->isVariable()) 6191 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 6192 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 6193 return Parser.Error(EqualLoc, 6194 "invalid reassignment of non-absolute variable '" + 6195 Name + "'"); 6196 } else if (Name == ".") { 6197 Parser.getStreamer().emitValueToOffset(Value, 0, EqualLoc); 6198 return false; 6199 } else 6200 Sym = Parser.getContext().getOrCreateSymbol(Name); 6201 6202 Sym->setRedefinable(allow_redef); 6203 6204 return false; 6205 } 6206 6207 } // end namespace MCParserUtils 6208 } // end namespace llvm 6209 6210 /// Create an MCAsmParser instance. 6211 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 6212 MCStreamer &Out, const MCAsmInfo &MAI, 6213 unsigned CB) { 6214 return new AsmParser(SM, C, Out, MAI, CB); 6215 } 6216