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/SmallString.h" 19 #include "llvm/ADT/SmallVector.h" 20 #include "llvm/ADT/StringExtras.h" 21 #include "llvm/ADT/StringMap.h" 22 #include "llvm/ADT/StringRef.h" 23 #include "llvm/ADT/Twine.h" 24 #include "llvm/BinaryFormat/Dwarf.h" 25 #include "llvm/DebugInfo/CodeView/SymbolRecord.h" 26 #include "llvm/MC/MCAsmInfo.h" 27 #include "llvm/MC/MCCodeView.h" 28 #include "llvm/MC/MCContext.h" 29 #include "llvm/MC/MCDirectives.h" 30 #include "llvm/MC/MCDwarf.h" 31 #include "llvm/MC/MCExpr.h" 32 #include "llvm/MC/MCInstPrinter.h" 33 #include "llvm/MC/MCInstrDesc.h" 34 #include "llvm/MC/MCInstrInfo.h" 35 #include "llvm/MC/MCObjectFileInfo.h" 36 #include "llvm/MC/MCParser/AsmCond.h" 37 #include "llvm/MC/MCParser/AsmLexer.h" 38 #include "llvm/MC/MCParser/MCAsmLexer.h" 39 #include "llvm/MC/MCParser/MCAsmParser.h" 40 #include "llvm/MC/MCParser/MCAsmParserExtension.h" 41 #include "llvm/MC/MCParser/MCAsmParserUtils.h" 42 #include "llvm/MC/MCParser/MCParsedAsmOperand.h" 43 #include "llvm/MC/MCParser/MCTargetAsmParser.h" 44 #include "llvm/MC/MCRegisterInfo.h" 45 #include "llvm/MC/MCSection.h" 46 #include "llvm/MC/MCStreamer.h" 47 #include "llvm/MC/MCSymbol.h" 48 #include "llvm/MC/MCTargetOptions.h" 49 #include "llvm/MC/MCValue.h" 50 #include "llvm/Support/Casting.h" 51 #include "llvm/Support/CommandLine.h" 52 #include "llvm/Support/ErrorHandling.h" 53 #include "llvm/Support/MD5.h" 54 #include "llvm/Support/MathExtras.h" 55 #include "llvm/Support/MemoryBuffer.h" 56 #include "llvm/Support/SMLoc.h" 57 #include "llvm/Support/SourceMgr.h" 58 #include "llvm/Support/raw_ostream.h" 59 #include <algorithm> 60 #include <cassert> 61 #include <cctype> 62 #include <climits> 63 #include <cstddef> 64 #include <cstdint> 65 #include <deque> 66 #include <memory> 67 #include <sstream> 68 #include <string> 69 #include <tuple> 70 #include <utility> 71 #include <vector> 72 73 using namespace llvm; 74 75 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() = default; 76 77 extern cl::opt<unsigned> AsmMacroMaxNestingDepth; 78 79 namespace { 80 81 /// Helper types for tracking macro definitions. 82 typedef std::vector<AsmToken> MCAsmMacroArgument; 83 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 84 85 /// Helper class for storing information about an active macro 86 /// instantiation. 87 struct MacroInstantiation { 88 /// The location of the instantiation. 89 SMLoc InstantiationLoc; 90 91 /// The buffer where parsing should resume upon instantiation completion. 92 unsigned ExitBuffer; 93 94 /// The location where parsing should resume upon instantiation completion. 95 SMLoc ExitLoc; 96 97 /// The depth of TheCondStack at the start of the instantiation. 98 size_t CondStackDepth; 99 }; 100 101 struct ParseStatementInfo { 102 /// The parsed operands from the last parsed statement. 103 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 104 105 /// The opcode from the last parsed instruction. 106 unsigned Opcode = ~0U; 107 108 /// Was there an error parsing the inline assembly? 109 bool ParseError = false; 110 111 SmallVectorImpl<AsmRewrite> *AsmRewrites = nullptr; 112 113 ParseStatementInfo() = delete; 114 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 115 : AsmRewrites(rewrites) {} 116 }; 117 118 /// The concrete assembly parser instance. 119 class AsmParser : public MCAsmParser { 120 private: 121 AsmLexer Lexer; 122 MCContext &Ctx; 123 MCStreamer &Out; 124 const MCAsmInfo &MAI; 125 SourceMgr &SrcMgr; 126 SourceMgr::DiagHandlerTy SavedDiagHandler; 127 void *SavedDiagContext; 128 std::unique_ptr<MCAsmParserExtension> PlatformParser; 129 SMLoc StartTokLoc; 130 131 /// This is the current buffer index we're lexing from as managed by the 132 /// SourceMgr object. 133 unsigned CurBuffer; 134 135 AsmCond TheCondState; 136 std::vector<AsmCond> TheCondStack; 137 138 /// maps directive names to handler methods in parser 139 /// extensions. Extensions register themselves in this map by calling 140 /// addDirectiveHandler. 141 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 142 143 /// Stack of active macro instantiations. 144 std::vector<MacroInstantiation*> ActiveMacros; 145 146 /// List of bodies of anonymous macros. 147 std::deque<MCAsmMacro> MacroLikeBodies; 148 149 /// Boolean tracking whether macro substitution is enabled. 150 unsigned MacrosEnabledFlag : 1; 151 152 /// Keeps track of how many .macro's have been instantiated. 153 unsigned NumOfMacroInstantiations; 154 155 /// The values from the last parsed cpp hash file line comment if any. 156 struct CppHashInfoTy { 157 StringRef Filename; 158 int64_t LineNumber; 159 SMLoc Loc; 160 unsigned Buf; 161 CppHashInfoTy() : Filename(), LineNumber(0), Loc(), Buf(0) {} 162 }; 163 CppHashInfoTy CppHashInfo; 164 165 /// The filename from the first cpp hash file line comment, if any. 166 StringRef FirstCppHashFilename; 167 168 /// List of forward directional labels for diagnosis at the end. 169 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 170 171 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 172 unsigned AssemblerDialect = ~0U; 173 174 /// is Darwin compatibility enabled? 175 bool IsDarwin = false; 176 177 /// Are we parsing ms-style inline assembly? 178 bool ParsingMSInlineAsm = false; 179 180 /// Did we already inform the user about inconsistent MD5 usage? 181 bool ReportedInconsistentMD5 = false; 182 183 // Is alt macro mode enabled. 184 bool AltMacroMode = false; 185 186 public: 187 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 188 const MCAsmInfo &MAI, unsigned CB); 189 AsmParser(const AsmParser &) = delete; 190 AsmParser &operator=(const AsmParser &) = delete; 191 ~AsmParser() override; 192 193 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 194 195 void addDirectiveHandler(StringRef Directive, 196 ExtensionDirectiveHandler Handler) override { 197 ExtensionDirectiveMap[Directive] = Handler; 198 } 199 200 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 201 DirectiveKindMap[Directive.lower()] = DirectiveKindMap[Alias.lower()]; 202 } 203 204 /// @name MCAsmParser Interface 205 /// { 206 207 SourceMgr &getSourceManager() override { return SrcMgr; } 208 MCAsmLexer &getLexer() override { return Lexer; } 209 MCContext &getContext() override { return Ctx; } 210 MCStreamer &getStreamer() override { return Out; } 211 212 CodeViewContext &getCVContext() { return Ctx.getCVContext(); } 213 214 unsigned getAssemblerDialect() override { 215 if (AssemblerDialect == ~0U) 216 return MAI.getAssemblerDialect(); 217 else 218 return AssemblerDialect; 219 } 220 void setAssemblerDialect(unsigned i) override { 221 AssemblerDialect = i; 222 } 223 224 void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override; 225 bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override; 226 bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override; 227 228 const AsmToken &Lex() override; 229 230 void setParsingMSInlineAsm(bool V) override { 231 ParsingMSInlineAsm = V; 232 // When parsing MS inline asm, we must lex 0b1101 and 0ABCH as binary and 233 // hex integer literals. 234 Lexer.setLexMasmIntegers(V); 235 } 236 bool isParsingMSInlineAsm() override { return ParsingMSInlineAsm; } 237 238 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 239 unsigned &NumOutputs, unsigned &NumInputs, 240 SmallVectorImpl<std::pair<void *,bool>> &OpDecls, 241 SmallVectorImpl<std::string> &Constraints, 242 SmallVectorImpl<std::string> &Clobbers, 243 const MCInstrInfo *MII, const MCInstPrinter *IP, 244 MCAsmParserSemaCallback &SI) override; 245 246 bool parseExpression(const MCExpr *&Res); 247 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 248 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc, 249 AsmTypeInfo *TypeInfo) override; 250 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 251 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 252 SMLoc &EndLoc) override; 253 bool parseAbsoluteExpression(int64_t &Res) override; 254 255 /// Parse a floating point expression using the float \p Semantics 256 /// and set \p Res to the value. 257 bool parseRealValue(const fltSemantics &Semantics, APInt &Res); 258 259 /// Parse an identifier or string (as a quoted identifier) 260 /// and set \p Res to the identifier contents. 261 bool parseIdentifier(StringRef &Res) override; 262 void eatToEndOfStatement() override; 263 264 bool checkForValidSection() override; 265 266 /// } 267 268 private: 269 bool parseStatement(ParseStatementInfo &Info, 270 MCAsmParserSemaCallback *SI); 271 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 272 bool parseCppHashLineFilenameComment(SMLoc L, bool SaveLocInfo = true); 273 274 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 275 ArrayRef<MCAsmMacroParameter> Parameters); 276 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 277 ArrayRef<MCAsmMacroParameter> Parameters, 278 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 279 SMLoc L); 280 281 /// Are macros enabled in the parser? 282 bool areMacrosEnabled() {return MacrosEnabledFlag;} 283 284 /// Control a flag in the parser that enables or disables macros. 285 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 286 287 /// Are we inside a macro instantiation? 288 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 289 290 /// Handle entry to macro instantiation. 291 /// 292 /// \param M The macro. 293 /// \param NameLoc Instantiation location. 294 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 295 296 /// Handle exit from macro instantiation. 297 void handleMacroExit(); 298 299 /// Extract AsmTokens for a macro argument. 300 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 301 302 /// Parse all macro arguments for a given macro. 303 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 304 305 void printMacroInstantiations(); 306 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 307 SMRange Range = None) const { 308 ArrayRef<SMRange> Ranges(Range); 309 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 310 } 311 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 312 313 /// Should we emit DWARF describing this assembler source? (Returns false if 314 /// the source has .file directives, which means we don't want to generate 315 /// info describing the assembler source itself.) 316 bool enabledGenDwarfForAssembly(); 317 318 /// Enter the specified file. This returns true on failure. 319 bool enterIncludeFile(const std::string &Filename); 320 321 /// Process the specified file for the .incbin directive. 322 /// This returns true on failure. 323 bool processIncbinFile(const std::string &Filename, int64_t Skip = 0, 324 const MCExpr *Count = nullptr, SMLoc Loc = SMLoc()); 325 326 /// Reset the current lexer position to that given by \p Loc. The 327 /// current token is not set; clients should ensure Lex() is called 328 /// subsequently. 329 /// 330 /// \param InBuffer If not 0, should be the known buffer id that contains the 331 /// location. 332 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 333 334 /// Parse up to the end of statement and a return the contents from the 335 /// current token until the end of the statement; the current token on exit 336 /// will be either the EndOfStatement or EOF. 337 StringRef parseStringToEndOfStatement() override; 338 339 /// Parse until the end of a statement or a comma is encountered, 340 /// return the contents from the current token up to the end or comma. 341 StringRef parseStringToComma(); 342 343 bool parseAssignment(StringRef Name, bool allow_redef, 344 bool NoDeadStrip = false); 345 346 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 347 MCBinaryExpr::Opcode &Kind); 348 349 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 350 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 351 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 352 353 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 354 355 bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName); 356 bool parseCVFileId(int64_t &FileId, StringRef DirectiveName); 357 358 // Generic (target and platform independent) directive parsing. 359 enum DirectiveKind { 360 DK_NO_DIRECTIVE, // Placeholder 361 DK_SET, 362 DK_EQU, 363 DK_EQUIV, 364 DK_ASCII, 365 DK_ASCIZ, 366 DK_STRING, 367 DK_BYTE, 368 DK_SHORT, 369 DK_RELOC, 370 DK_VALUE, 371 DK_2BYTE, 372 DK_LONG, 373 DK_INT, 374 DK_4BYTE, 375 DK_QUAD, 376 DK_8BYTE, 377 DK_OCTA, 378 DK_DC, 379 DK_DC_A, 380 DK_DC_B, 381 DK_DC_D, 382 DK_DC_L, 383 DK_DC_S, 384 DK_DC_W, 385 DK_DC_X, 386 DK_DCB, 387 DK_DCB_B, 388 DK_DCB_D, 389 DK_DCB_L, 390 DK_DCB_S, 391 DK_DCB_W, 392 DK_DCB_X, 393 DK_DS, 394 DK_DS_B, 395 DK_DS_D, 396 DK_DS_L, 397 DK_DS_P, 398 DK_DS_S, 399 DK_DS_W, 400 DK_DS_X, 401 DK_SINGLE, 402 DK_FLOAT, 403 DK_DOUBLE, 404 DK_ALIGN, 405 DK_ALIGN32, 406 DK_BALIGN, 407 DK_BALIGNW, 408 DK_BALIGNL, 409 DK_P2ALIGN, 410 DK_P2ALIGNW, 411 DK_P2ALIGNL, 412 DK_ORG, 413 DK_FILL, 414 DK_ENDR, 415 DK_BUNDLE_ALIGN_MODE, 416 DK_BUNDLE_LOCK, 417 DK_BUNDLE_UNLOCK, 418 DK_ZERO, 419 DK_EXTERN, 420 DK_GLOBL, 421 DK_GLOBAL, 422 DK_LAZY_REFERENCE, 423 DK_NO_DEAD_STRIP, 424 DK_SYMBOL_RESOLVER, 425 DK_PRIVATE_EXTERN, 426 DK_REFERENCE, 427 DK_WEAK_DEFINITION, 428 DK_WEAK_REFERENCE, 429 DK_WEAK_DEF_CAN_BE_HIDDEN, 430 DK_COLD, 431 DK_COMM, 432 DK_COMMON, 433 DK_LCOMM, 434 DK_ABORT, 435 DK_INCLUDE, 436 DK_INCBIN, 437 DK_CODE16, 438 DK_CODE16GCC, 439 DK_REPT, 440 DK_IRP, 441 DK_IRPC, 442 DK_IF, 443 DK_IFEQ, 444 DK_IFGE, 445 DK_IFGT, 446 DK_IFLE, 447 DK_IFLT, 448 DK_IFNE, 449 DK_IFB, 450 DK_IFNB, 451 DK_IFC, 452 DK_IFEQS, 453 DK_IFNC, 454 DK_IFNES, 455 DK_IFDEF, 456 DK_IFNDEF, 457 DK_IFNOTDEF, 458 DK_ELSEIF, 459 DK_ELSE, 460 DK_ENDIF, 461 DK_SPACE, 462 DK_SKIP, 463 DK_FILE, 464 DK_LINE, 465 DK_LOC, 466 DK_STABS, 467 DK_CV_FILE, 468 DK_CV_FUNC_ID, 469 DK_CV_INLINE_SITE_ID, 470 DK_CV_LOC, 471 DK_CV_LINETABLE, 472 DK_CV_INLINE_LINETABLE, 473 DK_CV_DEF_RANGE, 474 DK_CV_STRINGTABLE, 475 DK_CV_STRING, 476 DK_CV_FILECHECKSUMS, 477 DK_CV_FILECHECKSUM_OFFSET, 478 DK_CV_FPO_DATA, 479 DK_CFI_SECTIONS, 480 DK_CFI_STARTPROC, 481 DK_CFI_ENDPROC, 482 DK_CFI_DEF_CFA, 483 DK_CFI_DEF_CFA_OFFSET, 484 DK_CFI_ADJUST_CFA_OFFSET, 485 DK_CFI_DEF_CFA_REGISTER, 486 DK_CFI_OFFSET, 487 DK_CFI_REL_OFFSET, 488 DK_CFI_PERSONALITY, 489 DK_CFI_LSDA, 490 DK_CFI_REMEMBER_STATE, 491 DK_CFI_RESTORE_STATE, 492 DK_CFI_SAME_VALUE, 493 DK_CFI_RESTORE, 494 DK_CFI_ESCAPE, 495 DK_CFI_RETURN_COLUMN, 496 DK_CFI_SIGNAL_FRAME, 497 DK_CFI_UNDEFINED, 498 DK_CFI_REGISTER, 499 DK_CFI_WINDOW_SAVE, 500 DK_CFI_B_KEY_FRAME, 501 DK_MACROS_ON, 502 DK_MACROS_OFF, 503 DK_ALTMACRO, 504 DK_NOALTMACRO, 505 DK_MACRO, 506 DK_EXITM, 507 DK_ENDM, 508 DK_ENDMACRO, 509 DK_PURGEM, 510 DK_SLEB128, 511 DK_ULEB128, 512 DK_ERR, 513 DK_ERROR, 514 DK_WARNING, 515 DK_PRINT, 516 DK_ADDRSIG, 517 DK_ADDRSIG_SYM, 518 DK_END 519 }; 520 521 /// Maps directive name --> DirectiveKind enum, for 522 /// directives parsed by this class. 523 StringMap<DirectiveKind> DirectiveKindMap; 524 525 // Codeview def_range type parsing. 526 enum CVDefRangeType { 527 CVDR_DEFRANGE = 0, // Placeholder 528 CVDR_DEFRANGE_REGISTER, 529 CVDR_DEFRANGE_FRAMEPOINTER_REL, 530 CVDR_DEFRANGE_SUBFIELD_REGISTER, 531 CVDR_DEFRANGE_REGISTER_REL 532 }; 533 534 /// Maps Codeview def_range types --> CVDefRangeType enum, for 535 /// Codeview def_range types parsed by this class. 536 StringMap<CVDefRangeType> CVDefRangeTypeMap; 537 538 // ".ascii", ".asciz", ".string" 539 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 540 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 541 bool parseDirectiveValue(StringRef IDVal, 542 unsigned Size); // ".byte", ".long", ... 543 bool parseDirectiveOctaValue(StringRef IDVal); // ".octa", ... 544 bool parseDirectiveRealValue(StringRef IDVal, 545 const fltSemantics &); // ".single", ... 546 bool parseDirectiveFill(); // ".fill" 547 bool parseDirectiveZero(); // ".zero" 548 // ".set", ".equ", ".equiv" 549 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 550 bool parseDirectiveOrg(); // ".org" 551 // ".align{,32}", ".p2align{,w,l}" 552 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 553 554 // ".file", ".line", ".loc", ".stabs" 555 bool parseDirectiveFile(SMLoc DirectiveLoc); 556 bool parseDirectiveLine(); 557 bool parseDirectiveLoc(); 558 bool parseDirectiveStabs(); 559 560 // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable", 561 // ".cv_inline_linetable", ".cv_def_range", ".cv_string" 562 bool parseDirectiveCVFile(); 563 bool parseDirectiveCVFuncId(); 564 bool parseDirectiveCVInlineSiteId(); 565 bool parseDirectiveCVLoc(); 566 bool parseDirectiveCVLinetable(); 567 bool parseDirectiveCVInlineLinetable(); 568 bool parseDirectiveCVDefRange(); 569 bool parseDirectiveCVString(); 570 bool parseDirectiveCVStringTable(); 571 bool parseDirectiveCVFileChecksums(); 572 bool parseDirectiveCVFileChecksumOffset(); 573 bool parseDirectiveCVFPOData(); 574 575 // .cfi directives 576 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 577 bool parseDirectiveCFIWindowSave(); 578 bool parseDirectiveCFISections(); 579 bool parseDirectiveCFIStartProc(); 580 bool parseDirectiveCFIEndProc(); 581 bool parseDirectiveCFIDefCfaOffset(); 582 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 583 bool parseDirectiveCFIAdjustCfaOffset(); 584 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 585 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 586 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 587 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 588 bool parseDirectiveCFIRememberState(); 589 bool parseDirectiveCFIRestoreState(); 590 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 591 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 592 bool parseDirectiveCFIEscape(); 593 bool parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc); 594 bool parseDirectiveCFISignalFrame(); 595 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 596 597 // macro directives 598 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 599 bool parseDirectiveExitMacro(StringRef Directive); 600 bool parseDirectiveEndMacro(StringRef Directive); 601 bool parseDirectiveMacro(SMLoc DirectiveLoc); 602 bool parseDirectiveMacrosOnOff(StringRef Directive); 603 // alternate macro mode directives 604 bool parseDirectiveAltmacro(StringRef Directive); 605 // ".bundle_align_mode" 606 bool parseDirectiveBundleAlignMode(); 607 // ".bundle_lock" 608 bool parseDirectiveBundleLock(); 609 // ".bundle_unlock" 610 bool parseDirectiveBundleUnlock(); 611 612 // ".space", ".skip" 613 bool parseDirectiveSpace(StringRef IDVal); 614 615 // ".dcb" 616 bool parseDirectiveDCB(StringRef IDVal, unsigned Size); 617 bool parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &); 618 // ".ds" 619 bool parseDirectiveDS(StringRef IDVal, unsigned Size); 620 621 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 622 bool parseDirectiveLEB128(bool Signed); 623 624 /// Parse a directive like ".globl" which 625 /// accepts a single symbol (which should be a label or an external). 626 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 627 628 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 629 630 bool parseDirectiveAbort(); // ".abort" 631 bool parseDirectiveInclude(); // ".include" 632 bool parseDirectiveIncbin(); // ".incbin" 633 634 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 635 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 636 // ".ifb" or ".ifnb", depending on ExpectBlank. 637 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 638 // ".ifc" or ".ifnc", depending on ExpectEqual. 639 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 640 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 641 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 642 // ".ifdef" or ".ifndef", depending on expect_defined 643 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 644 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 645 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 646 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 647 bool parseEscapedString(std::string &Data) override; 648 bool parseAngleBracketString(std::string &Data) override; 649 650 const MCExpr *applyModifierToExpr(const MCExpr *E, 651 MCSymbolRefExpr::VariantKind Variant); 652 653 // Macro-like directives 654 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 655 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 656 raw_svector_ostream &OS); 657 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 658 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 659 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 660 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 661 662 // "_emit" or "__emit" 663 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 664 size_t Len); 665 666 // "align" 667 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 668 669 // "end" 670 bool parseDirectiveEnd(SMLoc DirectiveLoc); 671 672 // ".err" or ".error" 673 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 674 675 // ".warning" 676 bool parseDirectiveWarning(SMLoc DirectiveLoc); 677 678 // .print <double-quotes-string> 679 bool parseDirectivePrint(SMLoc DirectiveLoc); 680 681 // Directives to support address-significance tables. 682 bool parseDirectiveAddrsig(); 683 bool parseDirectiveAddrsigSym(); 684 685 void initializeDirectiveKindMap(); 686 void initializeCVDefRangeTypeMap(); 687 }; 688 689 } // end anonymous namespace 690 691 namespace llvm { 692 693 extern MCAsmParserExtension *createDarwinAsmParser(); 694 extern MCAsmParserExtension *createELFAsmParser(); 695 extern MCAsmParserExtension *createCOFFAsmParser(); 696 extern MCAsmParserExtension *createWasmAsmParser(); 697 698 } // end namespace llvm 699 700 enum { DEFAULT_ADDRSPACE = 0 }; 701 702 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 703 const MCAsmInfo &MAI, unsigned CB = 0) 704 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 705 CurBuffer(CB ? CB : SM.getMainFileID()), MacrosEnabledFlag(true) { 706 HadError = false; 707 // Save the old handler. 708 SavedDiagHandler = SrcMgr.getDiagHandler(); 709 SavedDiagContext = SrcMgr.getDiagContext(); 710 // Set our own handler which calls the saved handler. 711 SrcMgr.setDiagHandler(DiagHandler, this); 712 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 713 // Make MCStreamer aware of the StartTokLoc for locations in diagnostics. 714 Out.setStartTokLocPtr(&StartTokLoc); 715 716 // Initialize the platform / file format parser. 717 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 718 case MCObjectFileInfo::IsCOFF: 719 PlatformParser.reset(createCOFFAsmParser()); 720 break; 721 case MCObjectFileInfo::IsMachO: 722 PlatformParser.reset(createDarwinAsmParser()); 723 IsDarwin = true; 724 break; 725 case MCObjectFileInfo::IsELF: 726 PlatformParser.reset(createELFAsmParser()); 727 break; 728 case MCObjectFileInfo::IsWasm: 729 PlatformParser.reset(createWasmAsmParser()); 730 break; 731 case MCObjectFileInfo::IsXCOFF: 732 report_fatal_error( 733 "Need to implement createXCOFFAsmParser for XCOFF format."); 734 break; 735 } 736 737 PlatformParser->Initialize(*this); 738 initializeDirectiveKindMap(); 739 initializeCVDefRangeTypeMap(); 740 741 NumOfMacroInstantiations = 0; 742 } 743 744 AsmParser::~AsmParser() { 745 assert((HadError || ActiveMacros.empty()) && 746 "Unexpected active macro instantiation!"); 747 748 // Remove MCStreamer's reference to the parser SMLoc. 749 Out.setStartTokLocPtr(nullptr); 750 // Restore the saved diagnostics handler and context for use during 751 // finalization. 752 SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext); 753 } 754 755 void AsmParser::printMacroInstantiations() { 756 // Print the active macro instantiation stack. 757 for (std::vector<MacroInstantiation *>::const_reverse_iterator 758 it = ActiveMacros.rbegin(), 759 ie = ActiveMacros.rend(); 760 it != ie; ++it) 761 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 762 "while in macro instantiation"); 763 } 764 765 void AsmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) { 766 printPendingErrors(); 767 printMessage(L, SourceMgr::DK_Note, Msg, Range); 768 printMacroInstantiations(); 769 } 770 771 bool AsmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) { 772 if(getTargetParser().getTargetOptions().MCNoWarn) 773 return false; 774 if (getTargetParser().getTargetOptions().MCFatalWarnings) 775 return Error(L, Msg, Range); 776 printMessage(L, SourceMgr::DK_Warning, Msg, Range); 777 printMacroInstantiations(); 778 return false; 779 } 780 781 bool AsmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) { 782 HadError = true; 783 printMessage(L, SourceMgr::DK_Error, Msg, Range); 784 printMacroInstantiations(); 785 return true; 786 } 787 788 bool AsmParser::enterIncludeFile(const std::string &Filename) { 789 std::string IncludedFile; 790 unsigned NewBuf = 791 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 792 if (!NewBuf) 793 return true; 794 795 CurBuffer = NewBuf; 796 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 797 return false; 798 } 799 800 /// Process the specified .incbin file by searching for it in the include paths 801 /// then just emitting the byte contents of the file to the streamer. This 802 /// returns true on failure. 803 bool AsmParser::processIncbinFile(const std::string &Filename, int64_t Skip, 804 const MCExpr *Count, SMLoc Loc) { 805 std::string IncludedFile; 806 unsigned NewBuf = 807 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 808 if (!NewBuf) 809 return true; 810 811 // Pick up the bytes from the file and emit them. 812 StringRef Bytes = SrcMgr.getMemoryBuffer(NewBuf)->getBuffer(); 813 Bytes = Bytes.drop_front(Skip); 814 if (Count) { 815 int64_t Res; 816 if (!Count->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 817 return Error(Loc, "expected absolute expression"); 818 if (Res < 0) 819 return Warning(Loc, "negative count has no effect"); 820 Bytes = Bytes.take_front(Res); 821 } 822 getStreamer().emitBytes(Bytes); 823 return false; 824 } 825 826 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 827 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 828 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 829 Loc.getPointer()); 830 } 831 832 const AsmToken &AsmParser::Lex() { 833 if (Lexer.getTok().is(AsmToken::Error)) 834 Error(Lexer.getErrLoc(), Lexer.getErr()); 835 836 // if it's a end of statement with a comment in it 837 if (getTok().is(AsmToken::EndOfStatement)) { 838 // if this is a line comment output it. 839 if (!getTok().getString().empty() && getTok().getString().front() != '\n' && 840 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 841 Out.addExplicitComment(Twine(getTok().getString())); 842 } 843 844 const AsmToken *tok = &Lexer.Lex(); 845 846 // Parse comments here to be deferred until end of next statement. 847 while (tok->is(AsmToken::Comment)) { 848 if (MAI.preserveAsmComments()) 849 Out.addExplicitComment(Twine(tok->getString())); 850 tok = &Lexer.Lex(); 851 } 852 853 if (tok->is(AsmToken::Eof)) { 854 // If this is the end of an included file, pop the parent file off the 855 // include stack. 856 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 857 if (ParentIncludeLoc != SMLoc()) { 858 jumpToLoc(ParentIncludeLoc); 859 return Lex(); 860 } 861 } 862 863 return *tok; 864 } 865 866 bool AsmParser::enabledGenDwarfForAssembly() { 867 // Check whether the user specified -g. 868 if (!getContext().getGenDwarfForAssembly()) 869 return false; 870 // If we haven't encountered any .file directives (which would imply that 871 // the assembler source was produced with debug info already) then emit one 872 // describing the assembler source file itself. 873 if (getContext().getGenDwarfFileNumber() == 0) { 874 // Use the first #line directive for this, if any. It's preprocessed, so 875 // there is no checksum, and of course no source directive. 876 if (!FirstCppHashFilename.empty()) 877 getContext().setMCLineTableRootFile(/*CUID=*/0, 878 getContext().getCompilationDir(), 879 FirstCppHashFilename, 880 /*Cksum=*/None, /*Source=*/None); 881 const MCDwarfFile &RootFile = 882 getContext().getMCDwarfLineTable(/*CUID=*/0).getRootFile(); 883 getContext().setGenDwarfFileNumber(getStreamer().emitDwarfFileDirective( 884 /*CUID=*/0, getContext().getCompilationDir(), RootFile.Name, 885 RootFile.Checksum, RootFile.Source)); 886 } 887 return true; 888 } 889 890 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 891 // Create the initial section, if requested. 892 if (!NoInitialTextSection) 893 Out.InitSections(false); 894 895 // Prime the lexer. 896 Lex(); 897 898 HadError = false; 899 AsmCond StartingCondState = TheCondState; 900 SmallVector<AsmRewrite, 4> AsmStrRewrites; 901 902 // If we are generating dwarf for assembly source files save the initial text 903 // section. (Don't use enabledGenDwarfForAssembly() here, as we aren't 904 // emitting any actual debug info yet and haven't had a chance to parse any 905 // embedded .file directives.) 906 if (getContext().getGenDwarfForAssembly()) { 907 MCSection *Sec = getStreamer().getCurrentSectionOnly(); 908 if (!Sec->getBeginSymbol()) { 909 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 910 getStreamer().emitLabel(SectionStartSym); 911 Sec->setBeginSymbol(SectionStartSym); 912 } 913 bool InsertResult = getContext().addGenDwarfSection(Sec); 914 assert(InsertResult && ".text section should not have debug info yet"); 915 (void)InsertResult; 916 } 917 918 // While we have input, parse each statement. 919 while (Lexer.isNot(AsmToken::Eof)) { 920 ParseStatementInfo Info(&AsmStrRewrites); 921 bool Parsed = parseStatement(Info, nullptr); 922 923 // If we have a Lexer Error we are on an Error Token. Load in Lexer Error 924 // for printing ErrMsg via Lex() only if no (presumably better) parser error 925 // exists. 926 if (Parsed && !hasPendingError() && Lexer.getTok().is(AsmToken::Error)) { 927 Lex(); 928 } 929 930 // parseStatement returned true so may need to emit an error. 931 printPendingErrors(); 932 933 // Skipping to the next line if needed. 934 if (Parsed && !getLexer().isAtStartOfStatement()) 935 eatToEndOfStatement(); 936 } 937 938 getTargetParser().onEndOfFile(); 939 printPendingErrors(); 940 941 // All errors should have been emitted. 942 assert(!hasPendingError() && "unexpected error from parseStatement"); 943 944 getTargetParser().flushPendingInstructions(getStreamer()); 945 946 if (TheCondState.TheCond != StartingCondState.TheCond || 947 TheCondState.Ignore != StartingCondState.Ignore) 948 printError(getTok().getLoc(), "unmatched .ifs or .elses"); 949 // Check to see there are no empty DwarfFile slots. 950 const auto &LineTables = getContext().getMCDwarfLineTables(); 951 if (!LineTables.empty()) { 952 unsigned Index = 0; 953 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 954 if (File.Name.empty() && Index != 0) 955 printError(getTok().getLoc(), "unassigned file number: " + 956 Twine(Index) + 957 " for .file directives"); 958 ++Index; 959 } 960 } 961 962 // Check to see that all assembler local symbols were actually defined. 963 // Targets that don't do subsections via symbols may not want this, though, 964 // so conservatively exclude them. Only do this if we're finalizing, though, 965 // as otherwise we won't necessarilly have seen everything yet. 966 if (!NoFinalize) { 967 if (MAI.hasSubsectionsViaSymbols()) { 968 for (const auto &TableEntry : getContext().getSymbols()) { 969 MCSymbol *Sym = TableEntry.getValue(); 970 // Variable symbols may not be marked as defined, so check those 971 // explicitly. If we know it's a variable, we have a definition for 972 // the purposes of this check. 973 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 974 // FIXME: We would really like to refer back to where the symbol was 975 // first referenced for a source location. We need to add something 976 // to track that. Currently, we just point to the end of the file. 977 printError(getTok().getLoc(), "assembler local symbol '" + 978 Sym->getName() + "' not defined"); 979 } 980 } 981 982 // Temporary symbols like the ones for directional jumps don't go in the 983 // symbol table. They also need to be diagnosed in all (final) cases. 984 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 985 if (std::get<2>(LocSym)->isUndefined()) { 986 // Reset the state of any "# line file" directives we've seen to the 987 // context as it was at the diagnostic site. 988 CppHashInfo = std::get<1>(LocSym); 989 printError(std::get<0>(LocSym), "directional label undefined"); 990 } 991 } 992 } 993 994 // Finalize the output stream if there are no errors and if the client wants 995 // us to. 996 if (!HadError && !NoFinalize) 997 Out.Finish(Lexer.getLoc()); 998 999 return HadError || getContext().hadError(); 1000 } 1001 1002 bool AsmParser::checkForValidSection() { 1003 if (!ParsingMSInlineAsm && !getStreamer().getCurrentSectionOnly()) { 1004 Out.InitSections(false); 1005 return Error(getTok().getLoc(), 1006 "expected section directive before assembly directive"); 1007 } 1008 return false; 1009 } 1010 1011 /// Throw away the rest of the line for testing purposes. 1012 void AsmParser::eatToEndOfStatement() { 1013 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1014 Lexer.Lex(); 1015 1016 // Eat EOL. 1017 if (Lexer.is(AsmToken::EndOfStatement)) 1018 Lexer.Lex(); 1019 } 1020 1021 StringRef AsmParser::parseStringToEndOfStatement() { 1022 const char *Start = getTok().getLoc().getPointer(); 1023 1024 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1025 Lexer.Lex(); 1026 1027 const char *End = getTok().getLoc().getPointer(); 1028 return StringRef(Start, End - Start); 1029 } 1030 1031 StringRef AsmParser::parseStringToComma() { 1032 const char *Start = getTok().getLoc().getPointer(); 1033 1034 while (Lexer.isNot(AsmToken::EndOfStatement) && 1035 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 1036 Lexer.Lex(); 1037 1038 const char *End = getTok().getLoc().getPointer(); 1039 return StringRef(Start, End - Start); 1040 } 1041 1042 /// Parse a paren expression and return it. 1043 /// NOTE: This assumes the leading '(' has already been consumed. 1044 /// 1045 /// parenexpr ::= expr) 1046 /// 1047 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1048 if (parseExpression(Res)) 1049 return true; 1050 if (Lexer.isNot(AsmToken::RParen)) 1051 return TokError("expected ')' in parentheses expression"); 1052 EndLoc = Lexer.getTok().getEndLoc(); 1053 Lex(); 1054 return false; 1055 } 1056 1057 /// Parse a bracket expression and return it. 1058 /// NOTE: This assumes the leading '[' has already been consumed. 1059 /// 1060 /// bracketexpr ::= expr] 1061 /// 1062 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1063 if (parseExpression(Res)) 1064 return true; 1065 EndLoc = getTok().getEndLoc(); 1066 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 1067 return true; 1068 return false; 1069 } 1070 1071 /// Parse a primary expression and return it. 1072 /// primaryexpr ::= (parenexpr 1073 /// primaryexpr ::= symbol 1074 /// primaryexpr ::= number 1075 /// primaryexpr ::= '.' 1076 /// primaryexpr ::= ~,+,- primaryexpr 1077 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc, 1078 AsmTypeInfo *TypeInfo) { 1079 SMLoc FirstTokenLoc = getLexer().getLoc(); 1080 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 1081 switch (FirstTokenKind) { 1082 default: 1083 return TokError("unknown token in expression"); 1084 // If we have an error assume that we've already handled it. 1085 case AsmToken::Error: 1086 return true; 1087 case AsmToken::Exclaim: 1088 Lex(); // Eat the operator. 1089 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1090 return true; 1091 Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc); 1092 return false; 1093 case AsmToken::Dollar: 1094 case AsmToken::At: 1095 case AsmToken::String: 1096 case AsmToken::Identifier: { 1097 StringRef Identifier; 1098 if (parseIdentifier(Identifier)) { 1099 // We may have failed but $ may be a valid token. 1100 if (getTok().is(AsmToken::Dollar)) { 1101 if (Lexer.getMAI().getDollarIsPC()) { 1102 Lex(); 1103 // This is a '$' reference, which references the current PC. Emit a 1104 // temporary label to the streamer and refer to it. 1105 MCSymbol *Sym = Ctx.createTempSymbol(); 1106 Out.emitLabel(Sym); 1107 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 1108 getContext()); 1109 EndLoc = FirstTokenLoc; 1110 return false; 1111 } 1112 return Error(FirstTokenLoc, "invalid token in expression"); 1113 } 1114 } 1115 // Parse symbol variant 1116 std::pair<StringRef, StringRef> Split; 1117 if (!MAI.useParensForSymbolVariant()) { 1118 if (FirstTokenKind == AsmToken::String) { 1119 if (Lexer.is(AsmToken::At)) { 1120 Lex(); // eat @ 1121 SMLoc AtLoc = getLexer().getLoc(); 1122 StringRef VName; 1123 if (parseIdentifier(VName)) 1124 return Error(AtLoc, "expected symbol variant after '@'"); 1125 1126 Split = std::make_pair(Identifier, VName); 1127 } 1128 } else { 1129 Split = Identifier.split('@'); 1130 } 1131 } else if (Lexer.is(AsmToken::LParen)) { 1132 Lex(); // eat '('. 1133 StringRef VName; 1134 parseIdentifier(VName); 1135 // eat ')'. 1136 if (parseToken(AsmToken::RParen, 1137 "unexpected token in variant, expected ')'")) 1138 return true; 1139 Split = std::make_pair(Identifier, VName); 1140 } 1141 1142 EndLoc = SMLoc::getFromPointer(Identifier.end()); 1143 1144 // This is a symbol reference. 1145 StringRef SymbolName = Identifier; 1146 if (SymbolName.empty()) 1147 return Error(getLexer().getLoc(), "expected a symbol reference"); 1148 1149 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1150 1151 // Lookup the symbol variant if used. 1152 if (!Split.second.empty()) { 1153 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1154 if (Variant != MCSymbolRefExpr::VK_Invalid) { 1155 SymbolName = Split.first; 1156 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 1157 Variant = MCSymbolRefExpr::VK_None; 1158 } else { 1159 return Error(SMLoc::getFromPointer(Split.second.begin()), 1160 "invalid variant '" + Split.second + "'"); 1161 } 1162 } 1163 1164 MCSymbol *Sym = getContext().getInlineAsmLabel(SymbolName); 1165 if (!Sym) 1166 Sym = getContext().getOrCreateSymbol(SymbolName); 1167 1168 // If this is an absolute variable reference, substitute it now to preserve 1169 // semantics in the face of reassignment. 1170 if (Sym->isVariable()) { 1171 auto V = Sym->getVariableValue(/*SetUsed*/ false); 1172 bool DoInline = isa<MCConstantExpr>(V) && !Variant; 1173 if (auto TV = dyn_cast<MCTargetExpr>(V)) 1174 DoInline = TV->inlineAssignedExpr(); 1175 if (DoInline) { 1176 if (Variant) 1177 return Error(EndLoc, "unexpected modifier on variable reference"); 1178 Res = Sym->getVariableValue(/*SetUsed*/ false); 1179 return false; 1180 } 1181 } 1182 1183 // Otherwise create a symbol ref. 1184 Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc); 1185 return false; 1186 } 1187 case AsmToken::BigNum: 1188 return TokError("literal value out of range for directive"); 1189 case AsmToken::Integer: { 1190 SMLoc Loc = getTok().getLoc(); 1191 int64_t IntVal = getTok().getIntVal(); 1192 Res = MCConstantExpr::create(IntVal, getContext()); 1193 EndLoc = Lexer.getTok().getEndLoc(); 1194 Lex(); // Eat token. 1195 // Look for 'b' or 'f' following an Integer as a directional label 1196 if (Lexer.getKind() == AsmToken::Identifier) { 1197 StringRef IDVal = getTok().getString(); 1198 // Lookup the symbol variant if used. 1199 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 1200 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1201 if (Split.first.size() != IDVal.size()) { 1202 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1203 if (Variant == MCSymbolRefExpr::VK_Invalid) 1204 return TokError("invalid variant '" + Split.second + "'"); 1205 IDVal = Split.first; 1206 } 1207 if (IDVal == "f" || IDVal == "b") { 1208 MCSymbol *Sym = 1209 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 1210 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 1211 if (IDVal == "b" && Sym->isUndefined()) 1212 return Error(Loc, "directional label undefined"); 1213 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 1214 EndLoc = Lexer.getTok().getEndLoc(); 1215 Lex(); // Eat identifier. 1216 } 1217 } 1218 return false; 1219 } 1220 case AsmToken::Real: { 1221 APFloat RealVal(APFloat::IEEEdouble(), getTok().getString()); 1222 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1223 Res = MCConstantExpr::create(IntVal, getContext()); 1224 EndLoc = Lexer.getTok().getEndLoc(); 1225 Lex(); // Eat token. 1226 return false; 1227 } 1228 case AsmToken::Dot: { 1229 // This is a '.' reference, which references the current PC. Emit a 1230 // temporary label to the streamer and refer to it. 1231 MCSymbol *Sym = Ctx.createTempSymbol(); 1232 Out.emitLabel(Sym); 1233 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1234 EndLoc = Lexer.getTok().getEndLoc(); 1235 Lex(); // Eat identifier. 1236 return false; 1237 } 1238 case AsmToken::LParen: 1239 Lex(); // Eat the '('. 1240 return parseParenExpr(Res, EndLoc); 1241 case AsmToken::LBrac: 1242 if (!PlatformParser->HasBracketExpressions()) 1243 return TokError("brackets expression not supported on this target"); 1244 Lex(); // Eat the '['. 1245 return parseBracketExpr(Res, EndLoc); 1246 case AsmToken::Minus: 1247 Lex(); // Eat the operator. 1248 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1249 return true; 1250 Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc); 1251 return false; 1252 case AsmToken::Plus: 1253 Lex(); // Eat the operator. 1254 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1255 return true; 1256 Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc); 1257 return false; 1258 case AsmToken::Tilde: 1259 Lex(); // Eat the operator. 1260 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1261 return true; 1262 Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc); 1263 return false; 1264 // MIPS unary expression operators. The lexer won't generate these tokens if 1265 // MCAsmInfo::HasMipsExpressions is false for the target. 1266 case AsmToken::PercentCall16: 1267 case AsmToken::PercentCall_Hi: 1268 case AsmToken::PercentCall_Lo: 1269 case AsmToken::PercentDtprel_Hi: 1270 case AsmToken::PercentDtprel_Lo: 1271 case AsmToken::PercentGot: 1272 case AsmToken::PercentGot_Disp: 1273 case AsmToken::PercentGot_Hi: 1274 case AsmToken::PercentGot_Lo: 1275 case AsmToken::PercentGot_Ofst: 1276 case AsmToken::PercentGot_Page: 1277 case AsmToken::PercentGottprel: 1278 case AsmToken::PercentGp_Rel: 1279 case AsmToken::PercentHi: 1280 case AsmToken::PercentHigher: 1281 case AsmToken::PercentHighest: 1282 case AsmToken::PercentLo: 1283 case AsmToken::PercentNeg: 1284 case AsmToken::PercentPcrel_Hi: 1285 case AsmToken::PercentPcrel_Lo: 1286 case AsmToken::PercentTlsgd: 1287 case AsmToken::PercentTlsldm: 1288 case AsmToken::PercentTprel_Hi: 1289 case AsmToken::PercentTprel_Lo: 1290 Lex(); // Eat the operator. 1291 if (Lexer.isNot(AsmToken::LParen)) 1292 return TokError("expected '(' after operator"); 1293 Lex(); // Eat the operator. 1294 if (parseExpression(Res, EndLoc)) 1295 return true; 1296 if (Lexer.isNot(AsmToken::RParen)) 1297 return TokError("expected ')'"); 1298 Lex(); // Eat the operator. 1299 Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx); 1300 return !Res; 1301 } 1302 } 1303 1304 bool AsmParser::parseExpression(const MCExpr *&Res) { 1305 SMLoc EndLoc; 1306 return parseExpression(Res, EndLoc); 1307 } 1308 1309 const MCExpr * 1310 AsmParser::applyModifierToExpr(const MCExpr *E, 1311 MCSymbolRefExpr::VariantKind Variant) { 1312 // Ask the target implementation about this expression first. 1313 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 1314 if (NewE) 1315 return NewE; 1316 // Recurse over the given expression, rebuilding it to apply the given variant 1317 // if there is exactly one symbol. 1318 switch (E->getKind()) { 1319 case MCExpr::Target: 1320 case MCExpr::Constant: 1321 return nullptr; 1322 1323 case MCExpr::SymbolRef: { 1324 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 1325 1326 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 1327 TokError("invalid variant on expression '" + getTok().getIdentifier() + 1328 "' (already modified)"); 1329 return E; 1330 } 1331 1332 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1333 } 1334 1335 case MCExpr::Unary: { 1336 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1337 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1338 if (!Sub) 1339 return nullptr; 1340 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1341 } 1342 1343 case MCExpr::Binary: { 1344 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1345 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1346 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1347 1348 if (!LHS && !RHS) 1349 return nullptr; 1350 1351 if (!LHS) 1352 LHS = BE->getLHS(); 1353 if (!RHS) 1354 RHS = BE->getRHS(); 1355 1356 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1357 } 1358 } 1359 1360 llvm_unreachable("Invalid expression kind!"); 1361 } 1362 1363 /// This function checks if the next token is <string> type or arithmetic. 1364 /// string that begin with character '<' must end with character '>'. 1365 /// otherwise it is arithmetics. 1366 /// If the function returns a 'true' value, 1367 /// the End argument will be filled with the last location pointed to the '>' 1368 /// character. 1369 1370 /// There is a gap between the AltMacro's documentation and the single quote 1371 /// implementation. GCC does not fully support this feature and so we will not 1372 /// support it. 1373 /// TODO: Adding single quote as a string. 1374 static bool isAngleBracketString(SMLoc &StrLoc, SMLoc &EndLoc) { 1375 assert((StrLoc.getPointer() != nullptr) && 1376 "Argument to the function cannot be a NULL value"); 1377 const char *CharPtr = StrLoc.getPointer(); 1378 while ((*CharPtr != '>') && (*CharPtr != '\n') && (*CharPtr != '\r') && 1379 (*CharPtr != '\0')) { 1380 if (*CharPtr == '!') 1381 CharPtr++; 1382 CharPtr++; 1383 } 1384 if (*CharPtr == '>') { 1385 EndLoc = StrLoc.getFromPointer(CharPtr + 1); 1386 return true; 1387 } 1388 return false; 1389 } 1390 1391 /// creating a string without the escape characters '!'. 1392 static std::string angleBracketString(StringRef AltMacroStr) { 1393 std::string Res; 1394 for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) { 1395 if (AltMacroStr[Pos] == '!') 1396 Pos++; 1397 Res += AltMacroStr[Pos]; 1398 } 1399 return Res; 1400 } 1401 1402 /// Parse an expression and return it. 1403 /// 1404 /// expr ::= expr &&,|| expr -> lowest. 1405 /// expr ::= expr |,^,&,! expr 1406 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1407 /// expr ::= expr <<,>> expr 1408 /// expr ::= expr +,- expr 1409 /// expr ::= expr *,/,% expr -> highest. 1410 /// expr ::= primaryexpr 1411 /// 1412 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1413 // Parse the expression. 1414 Res = nullptr; 1415 if (getTargetParser().parsePrimaryExpr(Res, EndLoc) || 1416 parseBinOpRHS(1, Res, EndLoc)) 1417 return true; 1418 1419 // As a special case, we support 'a op b @ modifier' by rewriting the 1420 // expression to include the modifier. This is inefficient, but in general we 1421 // expect users to use 'a@modifier op b'. 1422 if (Lexer.getKind() == AsmToken::At) { 1423 Lex(); 1424 1425 if (Lexer.isNot(AsmToken::Identifier)) 1426 return TokError("unexpected symbol modifier following '@'"); 1427 1428 MCSymbolRefExpr::VariantKind Variant = 1429 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1430 if (Variant == MCSymbolRefExpr::VK_Invalid) 1431 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1432 1433 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1434 if (!ModifiedRes) { 1435 return TokError("invalid modifier '" + getTok().getIdentifier() + 1436 "' (no symbols present)"); 1437 } 1438 1439 Res = ModifiedRes; 1440 Lex(); 1441 } 1442 1443 // Try to constant fold it up front, if possible. Do not exploit 1444 // assembler here. 1445 int64_t Value; 1446 if (Res->evaluateAsAbsolute(Value)) 1447 Res = MCConstantExpr::create(Value, getContext()); 1448 1449 return false; 1450 } 1451 1452 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1453 Res = nullptr; 1454 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1455 } 1456 1457 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1458 SMLoc &EndLoc) { 1459 if (parseParenExpr(Res, EndLoc)) 1460 return true; 1461 1462 for (; ParenDepth > 0; --ParenDepth) { 1463 if (parseBinOpRHS(1, Res, EndLoc)) 1464 return true; 1465 1466 // We don't Lex() the last RParen. 1467 // This is the same behavior as parseParenExpression(). 1468 if (ParenDepth - 1 > 0) { 1469 EndLoc = getTok().getEndLoc(); 1470 if (parseToken(AsmToken::RParen, 1471 "expected ')' in parentheses expression")) 1472 return true; 1473 } 1474 } 1475 return false; 1476 } 1477 1478 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1479 const MCExpr *Expr; 1480 1481 SMLoc StartLoc = Lexer.getLoc(); 1482 if (parseExpression(Expr)) 1483 return true; 1484 1485 if (!Expr->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 1486 return Error(StartLoc, "expected absolute expression"); 1487 1488 return false; 1489 } 1490 1491 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1492 MCBinaryExpr::Opcode &Kind, 1493 bool ShouldUseLogicalShr) { 1494 switch (K) { 1495 default: 1496 return 0; // not a binop. 1497 1498 // Lowest Precedence: &&, || 1499 case AsmToken::AmpAmp: 1500 Kind = MCBinaryExpr::LAnd; 1501 return 1; 1502 case AsmToken::PipePipe: 1503 Kind = MCBinaryExpr::LOr; 1504 return 1; 1505 1506 // Low Precedence: |, &, ^ 1507 case AsmToken::Pipe: 1508 Kind = MCBinaryExpr::Or; 1509 return 2; 1510 case AsmToken::Caret: 1511 Kind = MCBinaryExpr::Xor; 1512 return 2; 1513 case AsmToken::Amp: 1514 Kind = MCBinaryExpr::And; 1515 return 2; 1516 1517 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1518 case AsmToken::EqualEqual: 1519 Kind = MCBinaryExpr::EQ; 1520 return 3; 1521 case AsmToken::ExclaimEqual: 1522 case AsmToken::LessGreater: 1523 Kind = MCBinaryExpr::NE; 1524 return 3; 1525 case AsmToken::Less: 1526 Kind = MCBinaryExpr::LT; 1527 return 3; 1528 case AsmToken::LessEqual: 1529 Kind = MCBinaryExpr::LTE; 1530 return 3; 1531 case AsmToken::Greater: 1532 Kind = MCBinaryExpr::GT; 1533 return 3; 1534 case AsmToken::GreaterEqual: 1535 Kind = MCBinaryExpr::GTE; 1536 return 3; 1537 1538 // Intermediate Precedence: <<, >> 1539 case AsmToken::LessLess: 1540 Kind = MCBinaryExpr::Shl; 1541 return 4; 1542 case AsmToken::GreaterGreater: 1543 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1544 return 4; 1545 1546 // High Intermediate Precedence: +, - 1547 case AsmToken::Plus: 1548 Kind = MCBinaryExpr::Add; 1549 return 5; 1550 case AsmToken::Minus: 1551 Kind = MCBinaryExpr::Sub; 1552 return 5; 1553 1554 // Highest Precedence: *, /, % 1555 case AsmToken::Star: 1556 Kind = MCBinaryExpr::Mul; 1557 return 6; 1558 case AsmToken::Slash: 1559 Kind = MCBinaryExpr::Div; 1560 return 6; 1561 case AsmToken::Percent: 1562 Kind = MCBinaryExpr::Mod; 1563 return 6; 1564 } 1565 } 1566 1567 static unsigned getGNUBinOpPrecedence(const MCAsmInfo &MAI, 1568 AsmToken::TokenKind K, 1569 MCBinaryExpr::Opcode &Kind, 1570 bool ShouldUseLogicalShr) { 1571 switch (K) { 1572 default: 1573 return 0; // not a binop. 1574 1575 // Lowest Precedence: &&, || 1576 case AsmToken::AmpAmp: 1577 Kind = MCBinaryExpr::LAnd; 1578 return 2; 1579 case AsmToken::PipePipe: 1580 Kind = MCBinaryExpr::LOr; 1581 return 1; 1582 1583 // Low Precedence: ==, !=, <>, <, <=, >, >= 1584 case AsmToken::EqualEqual: 1585 Kind = MCBinaryExpr::EQ; 1586 return 3; 1587 case AsmToken::ExclaimEqual: 1588 case AsmToken::LessGreater: 1589 Kind = MCBinaryExpr::NE; 1590 return 3; 1591 case AsmToken::Less: 1592 Kind = MCBinaryExpr::LT; 1593 return 3; 1594 case AsmToken::LessEqual: 1595 Kind = MCBinaryExpr::LTE; 1596 return 3; 1597 case AsmToken::Greater: 1598 Kind = MCBinaryExpr::GT; 1599 return 3; 1600 case AsmToken::GreaterEqual: 1601 Kind = MCBinaryExpr::GTE; 1602 return 3; 1603 1604 // Low Intermediate Precedence: +, - 1605 case AsmToken::Plus: 1606 Kind = MCBinaryExpr::Add; 1607 return 4; 1608 case AsmToken::Minus: 1609 Kind = MCBinaryExpr::Sub; 1610 return 4; 1611 1612 // High Intermediate Precedence: |, !, &, ^ 1613 // 1614 case AsmToken::Pipe: 1615 Kind = MCBinaryExpr::Or; 1616 return 5; 1617 case AsmToken::Exclaim: 1618 // Hack to support ARM compatible aliases (implied 'sp' operand in 'srs*' 1619 // instructions like 'srsda #31!') and not parse ! as an infix operator. 1620 if (MAI.getCommentString() == "@") 1621 return 0; 1622 Kind = MCBinaryExpr::OrNot; 1623 return 5; 1624 case AsmToken::Caret: 1625 Kind = MCBinaryExpr::Xor; 1626 return 5; 1627 case AsmToken::Amp: 1628 Kind = MCBinaryExpr::And; 1629 return 5; 1630 1631 // Highest Precedence: *, /, %, <<, >> 1632 case AsmToken::Star: 1633 Kind = MCBinaryExpr::Mul; 1634 return 6; 1635 case AsmToken::Slash: 1636 Kind = MCBinaryExpr::Div; 1637 return 6; 1638 case AsmToken::Percent: 1639 Kind = MCBinaryExpr::Mod; 1640 return 6; 1641 case AsmToken::LessLess: 1642 Kind = MCBinaryExpr::Shl; 1643 return 6; 1644 case AsmToken::GreaterGreater: 1645 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1646 return 6; 1647 } 1648 } 1649 1650 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1651 MCBinaryExpr::Opcode &Kind) { 1652 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1653 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1654 : getGNUBinOpPrecedence(MAI, K, Kind, ShouldUseLogicalShr); 1655 } 1656 1657 /// Parse all binary operators with precedence >= 'Precedence'. 1658 /// Res contains the LHS of the expression on input. 1659 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1660 SMLoc &EndLoc) { 1661 SMLoc StartLoc = Lexer.getLoc(); 1662 while (true) { 1663 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1664 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1665 1666 // If the next token is lower precedence than we are allowed to eat, return 1667 // successfully with what we ate already. 1668 if (TokPrec < Precedence) 1669 return false; 1670 1671 Lex(); 1672 1673 // Eat the next primary expression. 1674 const MCExpr *RHS; 1675 if (getTargetParser().parsePrimaryExpr(RHS, EndLoc)) 1676 return true; 1677 1678 // If BinOp binds less tightly with RHS than the operator after RHS, let 1679 // the pending operator take RHS as its LHS. 1680 MCBinaryExpr::Opcode Dummy; 1681 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1682 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1683 return true; 1684 1685 // Merge LHS and RHS according to operator. 1686 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc); 1687 } 1688 } 1689 1690 /// ParseStatement: 1691 /// ::= EndOfStatement 1692 /// ::= Label* Directive ...Operands... EndOfStatement 1693 /// ::= Label* Identifier OperandList* EndOfStatement 1694 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1695 MCAsmParserSemaCallback *SI) { 1696 assert(!hasPendingError() && "parseStatement started with pending error"); 1697 // Eat initial spaces and comments 1698 while (Lexer.is(AsmToken::Space)) 1699 Lex(); 1700 if (Lexer.is(AsmToken::EndOfStatement)) { 1701 // if this is a line comment we can drop it safely 1702 if (getTok().getString().empty() || getTok().getString().front() == '\r' || 1703 getTok().getString().front() == '\n') 1704 Out.AddBlankLine(); 1705 Lex(); 1706 return false; 1707 } 1708 // Statements always start with an identifier. 1709 AsmToken ID = getTok(); 1710 SMLoc IDLoc = ID.getLoc(); 1711 StringRef IDVal; 1712 int64_t LocalLabelVal = -1; 1713 StartTokLoc = ID.getLoc(); 1714 if (Lexer.is(AsmToken::HashDirective)) 1715 return parseCppHashLineFilenameComment(IDLoc, 1716 !isInsideMacroInstantiation()); 1717 1718 // Allow an integer followed by a ':' as a directional local label. 1719 if (Lexer.is(AsmToken::Integer)) { 1720 LocalLabelVal = getTok().getIntVal(); 1721 if (LocalLabelVal < 0) { 1722 if (!TheCondState.Ignore) { 1723 Lex(); // always eat a token 1724 return Error(IDLoc, "unexpected token at start of statement"); 1725 } 1726 IDVal = ""; 1727 } else { 1728 IDVal = getTok().getString(); 1729 Lex(); // Consume the integer token to be used as an identifier token. 1730 if (Lexer.getKind() != AsmToken::Colon) { 1731 if (!TheCondState.Ignore) { 1732 Lex(); // always eat a token 1733 return Error(IDLoc, "unexpected token at start of statement"); 1734 } 1735 } 1736 } 1737 } else if (Lexer.is(AsmToken::Dot)) { 1738 // Treat '.' as a valid identifier in this context. 1739 Lex(); 1740 IDVal = "."; 1741 } else if (Lexer.is(AsmToken::LCurly)) { 1742 // Treat '{' as a valid identifier in this context. 1743 Lex(); 1744 IDVal = "{"; 1745 1746 } else if (Lexer.is(AsmToken::RCurly)) { 1747 // Treat '}' as a valid identifier in this context. 1748 Lex(); 1749 IDVal = "}"; 1750 } else if (Lexer.is(AsmToken::Star) && 1751 getTargetParser().starIsStartOfStatement()) { 1752 // Accept '*' as a valid start of statement. 1753 Lex(); 1754 IDVal = "*"; 1755 } else if (parseIdentifier(IDVal)) { 1756 if (!TheCondState.Ignore) { 1757 Lex(); // always eat a token 1758 return Error(IDLoc, "unexpected token at start of statement"); 1759 } 1760 IDVal = ""; 1761 } 1762 1763 // Handle conditional assembly here before checking for skipping. We 1764 // have to do this so that .endif isn't skipped in a ".if 0" block for 1765 // example. 1766 StringMap<DirectiveKind>::const_iterator DirKindIt = 1767 DirectiveKindMap.find(IDVal.lower()); 1768 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1769 1770 ? DK_NO_DIRECTIVE 1771 : DirKindIt->getValue(); 1772 switch (DirKind) { 1773 default: 1774 break; 1775 case DK_IF: 1776 case DK_IFEQ: 1777 case DK_IFGE: 1778 case DK_IFGT: 1779 case DK_IFLE: 1780 case DK_IFLT: 1781 case DK_IFNE: 1782 return parseDirectiveIf(IDLoc, DirKind); 1783 case DK_IFB: 1784 return parseDirectiveIfb(IDLoc, true); 1785 case DK_IFNB: 1786 return parseDirectiveIfb(IDLoc, false); 1787 case DK_IFC: 1788 return parseDirectiveIfc(IDLoc, true); 1789 case DK_IFEQS: 1790 return parseDirectiveIfeqs(IDLoc, true); 1791 case DK_IFNC: 1792 return parseDirectiveIfc(IDLoc, false); 1793 case DK_IFNES: 1794 return parseDirectiveIfeqs(IDLoc, false); 1795 case DK_IFDEF: 1796 return parseDirectiveIfdef(IDLoc, true); 1797 case DK_IFNDEF: 1798 case DK_IFNOTDEF: 1799 return parseDirectiveIfdef(IDLoc, false); 1800 case DK_ELSEIF: 1801 return parseDirectiveElseIf(IDLoc); 1802 case DK_ELSE: 1803 return parseDirectiveElse(IDLoc); 1804 case DK_ENDIF: 1805 return parseDirectiveEndIf(IDLoc); 1806 } 1807 1808 // Ignore the statement if in the middle of inactive conditional 1809 // (e.g. ".if 0"). 1810 if (TheCondState.Ignore) { 1811 eatToEndOfStatement(); 1812 return false; 1813 } 1814 1815 // FIXME: Recurse on local labels? 1816 1817 // See what kind of statement we have. 1818 switch (Lexer.getKind()) { 1819 case AsmToken::Colon: { 1820 if (!getTargetParser().isLabel(ID)) 1821 break; 1822 if (checkForValidSection()) 1823 return true; 1824 1825 // identifier ':' -> Label. 1826 Lex(); 1827 1828 // Diagnose attempt to use '.' as a label. 1829 if (IDVal == ".") 1830 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1831 1832 // Diagnose attempt to use a variable as a label. 1833 // 1834 // FIXME: Diagnostics. Note the location of the definition as a label. 1835 // FIXME: This doesn't diagnose assignment to a symbol which has been 1836 // implicitly marked as external. 1837 MCSymbol *Sym; 1838 if (LocalLabelVal == -1) { 1839 if (ParsingMSInlineAsm && SI) { 1840 StringRef RewrittenLabel = 1841 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1842 assert(!RewrittenLabel.empty() && 1843 "We should have an internal name here."); 1844 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1845 RewrittenLabel); 1846 IDVal = RewrittenLabel; 1847 } 1848 Sym = getContext().getOrCreateSymbol(IDVal); 1849 } else 1850 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1851 // End of Labels should be treated as end of line for lexing 1852 // purposes but that information is not available to the Lexer who 1853 // does not understand Labels. This may cause us to see a Hash 1854 // here instead of a preprocessor line comment. 1855 if (getTok().is(AsmToken::Hash)) { 1856 StringRef CommentStr = parseStringToEndOfStatement(); 1857 Lexer.Lex(); 1858 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1859 } 1860 1861 // Consume any end of statement token, if present, to avoid spurious 1862 // AddBlankLine calls(). 1863 if (getTok().is(AsmToken::EndOfStatement)) { 1864 Lex(); 1865 } 1866 1867 getTargetParser().doBeforeLabelEmit(Sym); 1868 1869 // Emit the label. 1870 if (!getTargetParser().isParsingMSInlineAsm()) 1871 Out.emitLabel(Sym, IDLoc); 1872 1873 // If we are generating dwarf for assembly source files then gather the 1874 // info to make a dwarf label entry for this label if needed. 1875 if (enabledGenDwarfForAssembly()) 1876 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1877 IDLoc); 1878 1879 getTargetParser().onLabelParsed(Sym); 1880 1881 return false; 1882 } 1883 1884 case AsmToken::Equal: 1885 if (!getTargetParser().equalIsAsmAssignment()) 1886 break; 1887 // identifier '=' ... -> assignment statement 1888 Lex(); 1889 1890 return parseAssignment(IDVal, true); 1891 1892 default: // Normal instruction or directive. 1893 break; 1894 } 1895 1896 // If macros are enabled, check to see if this is a macro instantiation. 1897 if (areMacrosEnabled()) 1898 if (const MCAsmMacro *M = getContext().lookupMacro(IDVal)) { 1899 return handleMacroEntry(M, IDLoc); 1900 } 1901 1902 // Otherwise, we have a normal instruction or directive. 1903 1904 // Directives start with "." 1905 if (IDVal.startswith(".") && IDVal != ".") { 1906 // There are several entities interested in parsing directives: 1907 // 1908 // 1. The target-specific assembly parser. Some directives are target 1909 // specific or may potentially behave differently on certain targets. 1910 // 2. Asm parser extensions. For example, platform-specific parsers 1911 // (like the ELF parser) register themselves as extensions. 1912 // 3. The generic directive parser implemented by this class. These are 1913 // all the directives that behave in a target and platform independent 1914 // manner, or at least have a default behavior that's shared between 1915 // all targets and platforms. 1916 1917 getTargetParser().flushPendingInstructions(getStreamer()); 1918 1919 SMLoc StartTokLoc = getTok().getLoc(); 1920 bool TPDirectiveReturn = getTargetParser().ParseDirective(ID); 1921 1922 if (hasPendingError()) 1923 return true; 1924 // Currently the return value should be true if we are 1925 // uninterested but as this is at odds with the standard parsing 1926 // convention (return true = error) we have instances of a parsed 1927 // directive that fails returning true as an error. Catch these 1928 // cases as best as possible errors here. 1929 if (TPDirectiveReturn && StartTokLoc != getTok().getLoc()) 1930 return true; 1931 // Return if we did some parsing or believe we succeeded. 1932 if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc()) 1933 return false; 1934 1935 // Next, check the extension directive map to see if any extension has 1936 // registered itself to parse this directive. 1937 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1938 ExtensionDirectiveMap.lookup(IDVal); 1939 if (Handler.first) 1940 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1941 1942 // Finally, if no one else is interested in this directive, it must be 1943 // generic and familiar to this class. 1944 switch (DirKind) { 1945 default: 1946 break; 1947 case DK_SET: 1948 case DK_EQU: 1949 return parseDirectiveSet(IDVal, true); 1950 case DK_EQUIV: 1951 return parseDirectiveSet(IDVal, false); 1952 case DK_ASCII: 1953 return parseDirectiveAscii(IDVal, false); 1954 case DK_ASCIZ: 1955 case DK_STRING: 1956 return parseDirectiveAscii(IDVal, true); 1957 case DK_BYTE: 1958 case DK_DC_B: 1959 return parseDirectiveValue(IDVal, 1); 1960 case DK_DC: 1961 case DK_DC_W: 1962 case DK_SHORT: 1963 case DK_VALUE: 1964 case DK_2BYTE: 1965 return parseDirectiveValue(IDVal, 2); 1966 case DK_LONG: 1967 case DK_INT: 1968 case DK_4BYTE: 1969 case DK_DC_L: 1970 return parseDirectiveValue(IDVal, 4); 1971 case DK_QUAD: 1972 case DK_8BYTE: 1973 return parseDirectiveValue(IDVal, 8); 1974 case DK_DC_A: 1975 return parseDirectiveValue( 1976 IDVal, getContext().getAsmInfo()->getCodePointerSize()); 1977 case DK_OCTA: 1978 return parseDirectiveOctaValue(IDVal); 1979 case DK_SINGLE: 1980 case DK_FLOAT: 1981 case DK_DC_S: 1982 return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle()); 1983 case DK_DOUBLE: 1984 case DK_DC_D: 1985 return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble()); 1986 case DK_ALIGN: { 1987 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1988 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1989 } 1990 case DK_ALIGN32: { 1991 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1992 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1993 } 1994 case DK_BALIGN: 1995 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1996 case DK_BALIGNW: 1997 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1998 case DK_BALIGNL: 1999 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 2000 case DK_P2ALIGN: 2001 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 2002 case DK_P2ALIGNW: 2003 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 2004 case DK_P2ALIGNL: 2005 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 2006 case DK_ORG: 2007 return parseDirectiveOrg(); 2008 case DK_FILL: 2009 return parseDirectiveFill(); 2010 case DK_ZERO: 2011 return parseDirectiveZero(); 2012 case DK_EXTERN: 2013 eatToEndOfStatement(); // .extern is the default, ignore it. 2014 return false; 2015 case DK_GLOBL: 2016 case DK_GLOBAL: 2017 return parseDirectiveSymbolAttribute(MCSA_Global); 2018 case DK_LAZY_REFERENCE: 2019 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 2020 case DK_NO_DEAD_STRIP: 2021 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 2022 case DK_SYMBOL_RESOLVER: 2023 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 2024 case DK_PRIVATE_EXTERN: 2025 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 2026 case DK_REFERENCE: 2027 return parseDirectiveSymbolAttribute(MCSA_Reference); 2028 case DK_WEAK_DEFINITION: 2029 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 2030 case DK_WEAK_REFERENCE: 2031 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 2032 case DK_WEAK_DEF_CAN_BE_HIDDEN: 2033 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 2034 case DK_COLD: 2035 return parseDirectiveSymbolAttribute(MCSA_Cold); 2036 case DK_COMM: 2037 case DK_COMMON: 2038 return parseDirectiveComm(/*IsLocal=*/false); 2039 case DK_LCOMM: 2040 return parseDirectiveComm(/*IsLocal=*/true); 2041 case DK_ABORT: 2042 return parseDirectiveAbort(); 2043 case DK_INCLUDE: 2044 return parseDirectiveInclude(); 2045 case DK_INCBIN: 2046 return parseDirectiveIncbin(); 2047 case DK_CODE16: 2048 case DK_CODE16GCC: 2049 return TokError(Twine(IDVal) + 2050 " not currently supported for this target"); 2051 case DK_REPT: 2052 return parseDirectiveRept(IDLoc, IDVal); 2053 case DK_IRP: 2054 return parseDirectiveIrp(IDLoc); 2055 case DK_IRPC: 2056 return parseDirectiveIrpc(IDLoc); 2057 case DK_ENDR: 2058 return parseDirectiveEndr(IDLoc); 2059 case DK_BUNDLE_ALIGN_MODE: 2060 return parseDirectiveBundleAlignMode(); 2061 case DK_BUNDLE_LOCK: 2062 return parseDirectiveBundleLock(); 2063 case DK_BUNDLE_UNLOCK: 2064 return parseDirectiveBundleUnlock(); 2065 case DK_SLEB128: 2066 return parseDirectiveLEB128(true); 2067 case DK_ULEB128: 2068 return parseDirectiveLEB128(false); 2069 case DK_SPACE: 2070 case DK_SKIP: 2071 return parseDirectiveSpace(IDVal); 2072 case DK_FILE: 2073 return parseDirectiveFile(IDLoc); 2074 case DK_LINE: 2075 return parseDirectiveLine(); 2076 case DK_LOC: 2077 return parseDirectiveLoc(); 2078 case DK_STABS: 2079 return parseDirectiveStabs(); 2080 case DK_CV_FILE: 2081 return parseDirectiveCVFile(); 2082 case DK_CV_FUNC_ID: 2083 return parseDirectiveCVFuncId(); 2084 case DK_CV_INLINE_SITE_ID: 2085 return parseDirectiveCVInlineSiteId(); 2086 case DK_CV_LOC: 2087 return parseDirectiveCVLoc(); 2088 case DK_CV_LINETABLE: 2089 return parseDirectiveCVLinetable(); 2090 case DK_CV_INLINE_LINETABLE: 2091 return parseDirectiveCVInlineLinetable(); 2092 case DK_CV_DEF_RANGE: 2093 return parseDirectiveCVDefRange(); 2094 case DK_CV_STRING: 2095 return parseDirectiveCVString(); 2096 case DK_CV_STRINGTABLE: 2097 return parseDirectiveCVStringTable(); 2098 case DK_CV_FILECHECKSUMS: 2099 return parseDirectiveCVFileChecksums(); 2100 case DK_CV_FILECHECKSUM_OFFSET: 2101 return parseDirectiveCVFileChecksumOffset(); 2102 case DK_CV_FPO_DATA: 2103 return parseDirectiveCVFPOData(); 2104 case DK_CFI_SECTIONS: 2105 return parseDirectiveCFISections(); 2106 case DK_CFI_STARTPROC: 2107 return parseDirectiveCFIStartProc(); 2108 case DK_CFI_ENDPROC: 2109 return parseDirectiveCFIEndProc(); 2110 case DK_CFI_DEF_CFA: 2111 return parseDirectiveCFIDefCfa(IDLoc); 2112 case DK_CFI_DEF_CFA_OFFSET: 2113 return parseDirectiveCFIDefCfaOffset(); 2114 case DK_CFI_ADJUST_CFA_OFFSET: 2115 return parseDirectiveCFIAdjustCfaOffset(); 2116 case DK_CFI_DEF_CFA_REGISTER: 2117 return parseDirectiveCFIDefCfaRegister(IDLoc); 2118 case DK_CFI_OFFSET: 2119 return parseDirectiveCFIOffset(IDLoc); 2120 case DK_CFI_REL_OFFSET: 2121 return parseDirectiveCFIRelOffset(IDLoc); 2122 case DK_CFI_PERSONALITY: 2123 return parseDirectiveCFIPersonalityOrLsda(true); 2124 case DK_CFI_LSDA: 2125 return parseDirectiveCFIPersonalityOrLsda(false); 2126 case DK_CFI_REMEMBER_STATE: 2127 return parseDirectiveCFIRememberState(); 2128 case DK_CFI_RESTORE_STATE: 2129 return parseDirectiveCFIRestoreState(); 2130 case DK_CFI_SAME_VALUE: 2131 return parseDirectiveCFISameValue(IDLoc); 2132 case DK_CFI_RESTORE: 2133 return parseDirectiveCFIRestore(IDLoc); 2134 case DK_CFI_ESCAPE: 2135 return parseDirectiveCFIEscape(); 2136 case DK_CFI_RETURN_COLUMN: 2137 return parseDirectiveCFIReturnColumn(IDLoc); 2138 case DK_CFI_SIGNAL_FRAME: 2139 return parseDirectiveCFISignalFrame(); 2140 case DK_CFI_UNDEFINED: 2141 return parseDirectiveCFIUndefined(IDLoc); 2142 case DK_CFI_REGISTER: 2143 return parseDirectiveCFIRegister(IDLoc); 2144 case DK_CFI_WINDOW_SAVE: 2145 return parseDirectiveCFIWindowSave(); 2146 case DK_MACROS_ON: 2147 case DK_MACROS_OFF: 2148 return parseDirectiveMacrosOnOff(IDVal); 2149 case DK_MACRO: 2150 return parseDirectiveMacro(IDLoc); 2151 case DK_ALTMACRO: 2152 case DK_NOALTMACRO: 2153 return parseDirectiveAltmacro(IDVal); 2154 case DK_EXITM: 2155 return parseDirectiveExitMacro(IDVal); 2156 case DK_ENDM: 2157 case DK_ENDMACRO: 2158 return parseDirectiveEndMacro(IDVal); 2159 case DK_PURGEM: 2160 return parseDirectivePurgeMacro(IDLoc); 2161 case DK_END: 2162 return parseDirectiveEnd(IDLoc); 2163 case DK_ERR: 2164 return parseDirectiveError(IDLoc, false); 2165 case DK_ERROR: 2166 return parseDirectiveError(IDLoc, true); 2167 case DK_WARNING: 2168 return parseDirectiveWarning(IDLoc); 2169 case DK_RELOC: 2170 return parseDirectiveReloc(IDLoc); 2171 case DK_DCB: 2172 case DK_DCB_W: 2173 return parseDirectiveDCB(IDVal, 2); 2174 case DK_DCB_B: 2175 return parseDirectiveDCB(IDVal, 1); 2176 case DK_DCB_D: 2177 return parseDirectiveRealDCB(IDVal, APFloat::IEEEdouble()); 2178 case DK_DCB_L: 2179 return parseDirectiveDCB(IDVal, 4); 2180 case DK_DCB_S: 2181 return parseDirectiveRealDCB(IDVal, APFloat::IEEEsingle()); 2182 case DK_DC_X: 2183 case DK_DCB_X: 2184 return TokError(Twine(IDVal) + 2185 " not currently supported for this target"); 2186 case DK_DS: 2187 case DK_DS_W: 2188 return parseDirectiveDS(IDVal, 2); 2189 case DK_DS_B: 2190 return parseDirectiveDS(IDVal, 1); 2191 case DK_DS_D: 2192 return parseDirectiveDS(IDVal, 8); 2193 case DK_DS_L: 2194 case DK_DS_S: 2195 return parseDirectiveDS(IDVal, 4); 2196 case DK_DS_P: 2197 case DK_DS_X: 2198 return parseDirectiveDS(IDVal, 12); 2199 case DK_PRINT: 2200 return parseDirectivePrint(IDLoc); 2201 case DK_ADDRSIG: 2202 return parseDirectiveAddrsig(); 2203 case DK_ADDRSIG_SYM: 2204 return parseDirectiveAddrsigSym(); 2205 } 2206 2207 return Error(IDLoc, "unknown directive"); 2208 } 2209 2210 // __asm _emit or __asm __emit 2211 if (ParsingMSInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 2212 IDVal == "_EMIT" || IDVal == "__EMIT")) 2213 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 2214 2215 // __asm align 2216 if (ParsingMSInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 2217 return parseDirectiveMSAlign(IDLoc, Info); 2218 2219 if (ParsingMSInlineAsm && (IDVal == "even" || IDVal == "EVEN")) 2220 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 2221 if (checkForValidSection()) 2222 return true; 2223 2224 // Canonicalize the opcode to lower case. 2225 std::string OpcodeStr = IDVal.lower(); 2226 ParseInstructionInfo IInfo(Info.AsmRewrites); 2227 bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 2228 Info.ParsedOperands); 2229 Info.ParseError = ParseHadError; 2230 2231 // Dump the parsed representation, if requested. 2232 if (getShowParsedOperands()) { 2233 SmallString<256> Str; 2234 raw_svector_ostream OS(Str); 2235 OS << "parsed instruction: ["; 2236 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 2237 if (i != 0) 2238 OS << ", "; 2239 Info.ParsedOperands[i]->print(OS); 2240 } 2241 OS << "]"; 2242 2243 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 2244 } 2245 2246 // Fail even if ParseInstruction erroneously returns false. 2247 if (hasPendingError() || ParseHadError) 2248 return true; 2249 2250 // If we are generating dwarf for the current section then generate a .loc 2251 // directive for the instruction. 2252 if (!ParseHadError && enabledGenDwarfForAssembly() && 2253 getContext().getGenDwarfSectionSyms().count( 2254 getStreamer().getCurrentSectionOnly())) { 2255 unsigned Line; 2256 if (ActiveMacros.empty()) 2257 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 2258 else 2259 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 2260 ActiveMacros.front()->ExitBuffer); 2261 2262 // If we previously parsed a cpp hash file line comment then make sure the 2263 // current Dwarf File is for the CppHashFilename if not then emit the 2264 // Dwarf File table for it and adjust the line number for the .loc. 2265 if (!CppHashInfo.Filename.empty()) { 2266 unsigned FileNumber = getStreamer().emitDwarfFileDirective( 2267 0, StringRef(), CppHashInfo.Filename); 2268 getContext().setGenDwarfFileNumber(FileNumber); 2269 2270 unsigned CppHashLocLineNo = 2271 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 2272 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 2273 } 2274 2275 getStreamer().emitDwarfLocDirective( 2276 getContext().getGenDwarfFileNumber(), Line, 0, 2277 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 2278 StringRef()); 2279 } 2280 2281 // If parsing succeeded, match the instruction. 2282 if (!ParseHadError) { 2283 uint64_t ErrorInfo; 2284 if (getTargetParser().MatchAndEmitInstruction( 2285 IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo, 2286 getTargetParser().isParsingMSInlineAsm())) 2287 return true; 2288 } 2289 return false; 2290 } 2291 2292 // Parse and erase curly braces marking block start/end 2293 bool 2294 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 2295 // Identify curly brace marking block start/end 2296 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 2297 return false; 2298 2299 SMLoc StartLoc = Lexer.getLoc(); 2300 Lex(); // Eat the brace 2301 if (Lexer.is(AsmToken::EndOfStatement)) 2302 Lex(); // Eat EndOfStatement following the brace 2303 2304 // Erase the block start/end brace from the output asm string 2305 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 2306 StartLoc.getPointer()); 2307 return true; 2308 } 2309 2310 /// parseCppHashLineFilenameComment as this: 2311 /// ::= # number "filename" 2312 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L, bool SaveLocInfo) { 2313 Lex(); // Eat the hash token. 2314 // Lexer only ever emits HashDirective if it fully formed if it's 2315 // done the checking already so this is an internal error. 2316 assert(getTok().is(AsmToken::Integer) && 2317 "Lexing Cpp line comment: Expected Integer"); 2318 int64_t LineNumber = getTok().getIntVal(); 2319 Lex(); 2320 assert(getTok().is(AsmToken::String) && 2321 "Lexing Cpp line comment: Expected String"); 2322 StringRef Filename = getTok().getString(); 2323 Lex(); 2324 2325 if (!SaveLocInfo) 2326 return false; 2327 2328 // Get rid of the enclosing quotes. 2329 Filename = Filename.substr(1, Filename.size() - 2); 2330 2331 // Save the SMLoc, Filename and LineNumber for later use by diagnostics 2332 // and possibly DWARF file info. 2333 CppHashInfo.Loc = L; 2334 CppHashInfo.Filename = Filename; 2335 CppHashInfo.LineNumber = LineNumber; 2336 CppHashInfo.Buf = CurBuffer; 2337 if (FirstCppHashFilename.empty()) 2338 FirstCppHashFilename = Filename; 2339 return false; 2340 } 2341 2342 /// will use the last parsed cpp hash line filename comment 2343 /// for the Filename and LineNo if any in the diagnostic. 2344 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 2345 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 2346 raw_ostream &OS = errs(); 2347 2348 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 2349 SMLoc DiagLoc = Diag.getLoc(); 2350 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2351 unsigned CppHashBuf = 2352 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 2353 2354 // Like SourceMgr::printMessage() we need to print the include stack if any 2355 // before printing the message. 2356 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2357 if (!Parser->SavedDiagHandler && DiagCurBuffer && 2358 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 2359 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 2360 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 2361 } 2362 2363 // If we have not parsed a cpp hash line filename comment or the source 2364 // manager changed or buffer changed (like in a nested include) then just 2365 // print the normal diagnostic using its Filename and LineNo. 2366 if (!Parser->CppHashInfo.LineNumber || &DiagSrcMgr != &Parser->SrcMgr || 2367 DiagBuf != CppHashBuf) { 2368 if (Parser->SavedDiagHandler) 2369 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2370 else 2371 Diag.print(nullptr, OS); 2372 return; 2373 } 2374 2375 // Use the CppHashFilename and calculate a line number based on the 2376 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 2377 // for the diagnostic. 2378 const std::string &Filename = std::string(Parser->CppHashInfo.Filename); 2379 2380 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 2381 int CppHashLocLineNo = 2382 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 2383 int LineNo = 2384 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 2385 2386 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 2387 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 2388 Diag.getLineContents(), Diag.getRanges()); 2389 2390 if (Parser->SavedDiagHandler) 2391 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 2392 else 2393 NewDiag.print(nullptr, OS); 2394 } 2395 2396 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2397 // difference being that that function accepts '@' as part of identifiers and 2398 // we can't do that. AsmLexer.cpp should probably be changed to handle 2399 // '@' as a special case when needed. 2400 static bool isIdentifierChar(char c) { 2401 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2402 c == '.'; 2403 } 2404 2405 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2406 ArrayRef<MCAsmMacroParameter> Parameters, 2407 ArrayRef<MCAsmMacroArgument> A, 2408 bool EnableAtPseudoVariable, SMLoc L) { 2409 unsigned NParameters = Parameters.size(); 2410 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2411 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2412 return Error(L, "Wrong number of arguments"); 2413 2414 // A macro without parameters is handled differently on Darwin: 2415 // gas accepts no arguments and does no substitutions 2416 while (!Body.empty()) { 2417 // Scan for the next substitution. 2418 std::size_t End = Body.size(), Pos = 0; 2419 for (; Pos != End; ++Pos) { 2420 // Check for a substitution or escape. 2421 if (IsDarwin && !NParameters) { 2422 // This macro has no parameters, look for $0, $1, etc. 2423 if (Body[Pos] != '$' || Pos + 1 == End) 2424 continue; 2425 2426 char Next = Body[Pos + 1]; 2427 if (Next == '$' || Next == 'n' || 2428 isdigit(static_cast<unsigned char>(Next))) 2429 break; 2430 } else { 2431 // This macro has parameters, look for \foo, \bar, etc. 2432 if (Body[Pos] == '\\' && Pos + 1 != End) 2433 break; 2434 } 2435 } 2436 2437 // Add the prefix. 2438 OS << Body.slice(0, Pos); 2439 2440 // Check if we reached the end. 2441 if (Pos == End) 2442 break; 2443 2444 if (IsDarwin && !NParameters) { 2445 switch (Body[Pos + 1]) { 2446 // $$ => $ 2447 case '$': 2448 OS << '$'; 2449 break; 2450 2451 // $n => number of arguments 2452 case 'n': 2453 OS << A.size(); 2454 break; 2455 2456 // $[0-9] => argument 2457 default: { 2458 // Missing arguments are ignored. 2459 unsigned Index = Body[Pos + 1] - '0'; 2460 if (Index >= A.size()) 2461 break; 2462 2463 // Otherwise substitute with the token values, with spaces eliminated. 2464 for (const AsmToken &Token : A[Index]) 2465 OS << Token.getString(); 2466 break; 2467 } 2468 } 2469 Pos += 2; 2470 } else { 2471 unsigned I = Pos + 1; 2472 2473 // Check for the \@ pseudo-variable. 2474 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2475 ++I; 2476 else 2477 while (isIdentifierChar(Body[I]) && I + 1 != End) 2478 ++I; 2479 2480 const char *Begin = Body.data() + Pos + 1; 2481 StringRef Argument(Begin, I - (Pos + 1)); 2482 unsigned Index = 0; 2483 2484 if (Argument == "@") { 2485 OS << NumOfMacroInstantiations; 2486 Pos += 2; 2487 } else { 2488 for (; Index < NParameters; ++Index) 2489 if (Parameters[Index].Name == Argument) 2490 break; 2491 2492 if (Index == NParameters) { 2493 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2494 Pos += 3; 2495 else { 2496 OS << '\\' << Argument; 2497 Pos = I; 2498 } 2499 } else { 2500 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2501 for (const AsmToken &Token : A[Index]) 2502 // For altmacro mode, you can write '%expr'. 2503 // The prefix '%' evaluates the expression 'expr' 2504 // and uses the result as a string (e.g. replace %(1+2) with the 2505 // string "3"). 2506 // Here, we identify the integer token which is the result of the 2507 // absolute expression evaluation and replace it with its string 2508 // representation. 2509 if (AltMacroMode && Token.getString().front() == '%' && 2510 Token.is(AsmToken::Integer)) 2511 // Emit an integer value to the buffer. 2512 OS << Token.getIntVal(); 2513 // Only Token that was validated as a string and begins with '<' 2514 // is considered altMacroString!!! 2515 else if (AltMacroMode && Token.getString().front() == '<' && 2516 Token.is(AsmToken::String)) { 2517 OS << angleBracketString(Token.getStringContents()); 2518 } 2519 // We expect no quotes around the string's contents when 2520 // parsing for varargs. 2521 else if (Token.isNot(AsmToken::String) || VarargParameter) 2522 OS << Token.getString(); 2523 else 2524 OS << Token.getStringContents(); 2525 2526 Pos += 1 + Argument.size(); 2527 } 2528 } 2529 } 2530 // Update the scan point. 2531 Body = Body.substr(Pos); 2532 } 2533 2534 return false; 2535 } 2536 2537 static bool isOperator(AsmToken::TokenKind kind) { 2538 switch (kind) { 2539 default: 2540 return false; 2541 case AsmToken::Plus: 2542 case AsmToken::Minus: 2543 case AsmToken::Tilde: 2544 case AsmToken::Slash: 2545 case AsmToken::Star: 2546 case AsmToken::Dot: 2547 case AsmToken::Equal: 2548 case AsmToken::EqualEqual: 2549 case AsmToken::Pipe: 2550 case AsmToken::PipePipe: 2551 case AsmToken::Caret: 2552 case AsmToken::Amp: 2553 case AsmToken::AmpAmp: 2554 case AsmToken::Exclaim: 2555 case AsmToken::ExclaimEqual: 2556 case AsmToken::Less: 2557 case AsmToken::LessEqual: 2558 case AsmToken::LessLess: 2559 case AsmToken::LessGreater: 2560 case AsmToken::Greater: 2561 case AsmToken::GreaterEqual: 2562 case AsmToken::GreaterGreater: 2563 return true; 2564 } 2565 } 2566 2567 namespace { 2568 2569 class AsmLexerSkipSpaceRAII { 2570 public: 2571 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2572 Lexer.setSkipSpace(SkipSpace); 2573 } 2574 2575 ~AsmLexerSkipSpaceRAII() { 2576 Lexer.setSkipSpace(true); 2577 } 2578 2579 private: 2580 AsmLexer &Lexer; 2581 }; 2582 2583 } // end anonymous namespace 2584 2585 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2586 2587 if (Vararg) { 2588 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2589 StringRef Str = parseStringToEndOfStatement(); 2590 MA.emplace_back(AsmToken::String, Str); 2591 } 2592 return false; 2593 } 2594 2595 unsigned ParenLevel = 0; 2596 2597 // Darwin doesn't use spaces to delmit arguments. 2598 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2599 2600 bool SpaceEaten; 2601 2602 while (true) { 2603 SpaceEaten = false; 2604 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2605 return TokError("unexpected token in macro instantiation"); 2606 2607 if (ParenLevel == 0) { 2608 2609 if (Lexer.is(AsmToken::Comma)) 2610 break; 2611 2612 if (Lexer.is(AsmToken::Space)) { 2613 SpaceEaten = true; 2614 Lexer.Lex(); // Eat spaces 2615 } 2616 2617 // Spaces can delimit parameters, but could also be part an expression. 2618 // If the token after a space is an operator, add the token and the next 2619 // one into this argument 2620 if (!IsDarwin) { 2621 if (isOperator(Lexer.getKind())) { 2622 MA.push_back(getTok()); 2623 Lexer.Lex(); 2624 2625 // Whitespace after an operator can be ignored. 2626 if (Lexer.is(AsmToken::Space)) 2627 Lexer.Lex(); 2628 2629 continue; 2630 } 2631 } 2632 if (SpaceEaten) 2633 break; 2634 } 2635 2636 // handleMacroEntry relies on not advancing the lexer here 2637 // to be able to fill in the remaining default parameter values 2638 if (Lexer.is(AsmToken::EndOfStatement)) 2639 break; 2640 2641 // Adjust the current parentheses level. 2642 if (Lexer.is(AsmToken::LParen)) 2643 ++ParenLevel; 2644 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2645 --ParenLevel; 2646 2647 // Append the token to the current argument list. 2648 MA.push_back(getTok()); 2649 Lexer.Lex(); 2650 } 2651 2652 if (ParenLevel != 0) 2653 return TokError("unbalanced parentheses in macro argument"); 2654 return false; 2655 } 2656 2657 // Parse the macro instantiation arguments. 2658 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2659 MCAsmMacroArguments &A) { 2660 const unsigned NParameters = M ? M->Parameters.size() : 0; 2661 bool NamedParametersFound = false; 2662 SmallVector<SMLoc, 4> FALocs; 2663 2664 A.resize(NParameters); 2665 FALocs.resize(NParameters); 2666 2667 // Parse two kinds of macro invocations: 2668 // - macros defined without any parameters accept an arbitrary number of them 2669 // - macros defined with parameters accept at most that many of them 2670 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2671 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2672 ++Parameter) { 2673 SMLoc IDLoc = Lexer.getLoc(); 2674 MCAsmMacroParameter FA; 2675 2676 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2677 if (parseIdentifier(FA.Name)) 2678 return Error(IDLoc, "invalid argument identifier for formal argument"); 2679 2680 if (Lexer.isNot(AsmToken::Equal)) 2681 return TokError("expected '=' after formal parameter identifier"); 2682 2683 Lex(); 2684 2685 NamedParametersFound = true; 2686 } 2687 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2688 2689 if (NamedParametersFound && FA.Name.empty()) 2690 return Error(IDLoc, "cannot mix positional and keyword arguments"); 2691 2692 SMLoc StrLoc = Lexer.getLoc(); 2693 SMLoc EndLoc; 2694 if (AltMacroMode && Lexer.is(AsmToken::Percent)) { 2695 const MCExpr *AbsoluteExp; 2696 int64_t Value; 2697 /// Eat '%' 2698 Lex(); 2699 if (parseExpression(AbsoluteExp, EndLoc)) 2700 return false; 2701 if (!AbsoluteExp->evaluateAsAbsolute(Value, 2702 getStreamer().getAssemblerPtr())) 2703 return Error(StrLoc, "expected absolute expression"); 2704 const char *StrChar = StrLoc.getPointer(); 2705 const char *EndChar = EndLoc.getPointer(); 2706 AsmToken newToken(AsmToken::Integer, 2707 StringRef(StrChar, EndChar - StrChar), Value); 2708 FA.Value.push_back(newToken); 2709 } else if (AltMacroMode && Lexer.is(AsmToken::Less) && 2710 isAngleBracketString(StrLoc, EndLoc)) { 2711 const char *StrChar = StrLoc.getPointer(); 2712 const char *EndChar = EndLoc.getPointer(); 2713 jumpToLoc(EndLoc, CurBuffer); 2714 /// Eat from '<' to '>' 2715 Lex(); 2716 AsmToken newToken(AsmToken::String, 2717 StringRef(StrChar, EndChar - StrChar)); 2718 FA.Value.push_back(newToken); 2719 } else if(parseMacroArgument(FA.Value, Vararg)) 2720 return true; 2721 2722 unsigned PI = Parameter; 2723 if (!FA.Name.empty()) { 2724 unsigned FAI = 0; 2725 for (FAI = 0; FAI < NParameters; ++FAI) 2726 if (M->Parameters[FAI].Name == FA.Name) 2727 break; 2728 2729 if (FAI >= NParameters) { 2730 assert(M && "expected macro to be defined"); 2731 return Error(IDLoc, "parameter named '" + FA.Name + 2732 "' does not exist for macro '" + M->Name + "'"); 2733 } 2734 PI = FAI; 2735 } 2736 2737 if (!FA.Value.empty()) { 2738 if (A.size() <= PI) 2739 A.resize(PI + 1); 2740 A[PI] = FA.Value; 2741 2742 if (FALocs.size() <= PI) 2743 FALocs.resize(PI + 1); 2744 2745 FALocs[PI] = Lexer.getLoc(); 2746 } 2747 2748 // At the end of the statement, fill in remaining arguments that have 2749 // default values. If there aren't any, then the next argument is 2750 // required but missing 2751 if (Lexer.is(AsmToken::EndOfStatement)) { 2752 bool Failure = false; 2753 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2754 if (A[FAI].empty()) { 2755 if (M->Parameters[FAI].Required) { 2756 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2757 "missing value for required parameter " 2758 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2759 Failure = true; 2760 } 2761 2762 if (!M->Parameters[FAI].Value.empty()) 2763 A[FAI] = M->Parameters[FAI].Value; 2764 } 2765 } 2766 return Failure; 2767 } 2768 2769 if (Lexer.is(AsmToken::Comma)) 2770 Lex(); 2771 } 2772 2773 return TokError("too many positional arguments"); 2774 } 2775 2776 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2777 // Arbitrarily limit macro nesting depth (default matches 'as'). We can 2778 // eliminate this, although we should protect against infinite loops. 2779 unsigned MaxNestingDepth = AsmMacroMaxNestingDepth; 2780 if (ActiveMacros.size() == MaxNestingDepth) { 2781 std::ostringstream MaxNestingDepthError; 2782 MaxNestingDepthError << "macros cannot be nested more than " 2783 << MaxNestingDepth << " levels deep." 2784 << " Use -asm-macro-max-nesting-depth to increase " 2785 "this limit."; 2786 return TokError(MaxNestingDepthError.str()); 2787 } 2788 2789 MCAsmMacroArguments A; 2790 if (parseMacroArguments(M, A)) 2791 return true; 2792 2793 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2794 // to hold the macro body with substitutions. 2795 SmallString<256> Buf; 2796 StringRef Body = M->Body; 2797 raw_svector_ostream OS(Buf); 2798 2799 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2800 return true; 2801 2802 // We include the .endmacro in the buffer as our cue to exit the macro 2803 // instantiation. 2804 OS << ".endmacro\n"; 2805 2806 std::unique_ptr<MemoryBuffer> Instantiation = 2807 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2808 2809 // Create the macro instantiation object and add to the current macro 2810 // instantiation stack. 2811 MacroInstantiation *MI = new MacroInstantiation{ 2812 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 2813 ActiveMacros.push_back(MI); 2814 2815 ++NumOfMacroInstantiations; 2816 2817 // Jump to the macro instantiation and prime the lexer. 2818 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2819 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2820 Lex(); 2821 2822 return false; 2823 } 2824 2825 void AsmParser::handleMacroExit() { 2826 // Jump to the EndOfStatement we should return to, and consume it. 2827 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2828 Lex(); 2829 2830 // Pop the instantiation entry. 2831 delete ActiveMacros.back(); 2832 ActiveMacros.pop_back(); 2833 } 2834 2835 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2836 bool NoDeadStrip) { 2837 MCSymbol *Sym; 2838 const MCExpr *Value; 2839 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2840 Value)) 2841 return true; 2842 2843 if (!Sym) { 2844 // In the case where we parse an expression starting with a '.', we will 2845 // not generate an error, nor will we create a symbol. In this case we 2846 // should just return out. 2847 return false; 2848 } 2849 2850 // Do the assignment. 2851 Out.emitAssignment(Sym, Value); 2852 if (NoDeadStrip) 2853 Out.emitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2854 2855 return false; 2856 } 2857 2858 /// parseIdentifier: 2859 /// ::= identifier 2860 /// ::= string 2861 bool AsmParser::parseIdentifier(StringRef &Res) { 2862 // The assembler has relaxed rules for accepting identifiers, in particular we 2863 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2864 // separate tokens. At this level, we have already lexed so we cannot (currently) 2865 // handle this as a context dependent token, instead we detect adjacent tokens 2866 // and return the combined identifier. 2867 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2868 SMLoc PrefixLoc = getLexer().getLoc(); 2869 2870 // Consume the prefix character, and check for a following identifier. 2871 2872 AsmToken Buf[1]; 2873 Lexer.peekTokens(Buf, false); 2874 2875 if (Buf[0].isNot(AsmToken::Identifier) && Buf[0].isNot(AsmToken::Integer)) 2876 return true; 2877 2878 // We have a '$' or '@' followed by an identifier or integer token, make 2879 // sure they are adjacent. 2880 if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer()) 2881 return true; 2882 2883 // eat $ or @ 2884 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2885 // Construct the joined identifier and consume the token. 2886 Res = StringRef(PrefixLoc.getPointer(), getTok().getString().size() + 1); 2887 Lex(); // Parser Lex to maintain invariants. 2888 return false; 2889 } 2890 2891 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2892 return true; 2893 2894 Res = getTok().getIdentifier(); 2895 2896 Lex(); // Consume the identifier token. 2897 2898 return false; 2899 } 2900 2901 /// parseDirectiveSet: 2902 /// ::= .equ identifier ',' expression 2903 /// ::= .equiv identifier ',' expression 2904 /// ::= .set identifier ',' expression 2905 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2906 StringRef Name; 2907 if (check(parseIdentifier(Name), "expected identifier") || 2908 parseToken(AsmToken::Comma) || parseAssignment(Name, allow_redef, true)) 2909 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2910 return false; 2911 } 2912 2913 bool AsmParser::parseEscapedString(std::string &Data) { 2914 if (check(getTok().isNot(AsmToken::String), "expected string")) 2915 return true; 2916 2917 Data = ""; 2918 StringRef Str = getTok().getStringContents(); 2919 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2920 if (Str[i] != '\\') { 2921 Data += Str[i]; 2922 continue; 2923 } 2924 2925 // Recognize escaped characters. Note that this escape semantics currently 2926 // loosely follows Darwin 'as'. 2927 ++i; 2928 if (i == e) 2929 return TokError("unexpected backslash at end of string"); 2930 2931 // Recognize hex sequences similarly to GNU 'as'. 2932 if (Str[i] == 'x' || Str[i] == 'X') { 2933 size_t length = Str.size(); 2934 if (i + 1 >= length || !isHexDigit(Str[i + 1])) 2935 return TokError("invalid hexadecimal escape sequence"); 2936 2937 // Consume hex characters. GNU 'as' reads all hexadecimal characters and 2938 // then truncates to the lower 16 bits. Seems reasonable. 2939 unsigned Value = 0; 2940 while (i + 1 < length && isHexDigit(Str[i + 1])) 2941 Value = Value * 16 + hexDigitValue(Str[++i]); 2942 2943 Data += (unsigned char)(Value & 0xFF); 2944 continue; 2945 } 2946 2947 // Recognize octal sequences. 2948 if ((unsigned)(Str[i] - '0') <= 7) { 2949 // Consume up to three octal characters. 2950 unsigned Value = Str[i] - '0'; 2951 2952 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2953 ++i; 2954 Value = Value * 8 + (Str[i] - '0'); 2955 2956 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2957 ++i; 2958 Value = Value * 8 + (Str[i] - '0'); 2959 } 2960 } 2961 2962 if (Value > 255) 2963 return TokError("invalid octal escape sequence (out of range)"); 2964 2965 Data += (unsigned char)Value; 2966 continue; 2967 } 2968 2969 // Otherwise recognize individual escapes. 2970 switch (Str[i]) { 2971 default: 2972 // Just reject invalid escape sequences for now. 2973 return TokError("invalid escape sequence (unrecognized character)"); 2974 2975 case 'b': Data += '\b'; break; 2976 case 'f': Data += '\f'; break; 2977 case 'n': Data += '\n'; break; 2978 case 'r': Data += '\r'; break; 2979 case 't': Data += '\t'; break; 2980 case '"': Data += '"'; break; 2981 case '\\': Data += '\\'; break; 2982 } 2983 } 2984 2985 Lex(); 2986 return false; 2987 } 2988 2989 bool AsmParser::parseAngleBracketString(std::string &Data) { 2990 SMLoc EndLoc, StartLoc = getTok().getLoc(); 2991 if (isAngleBracketString(StartLoc, EndLoc)) { 2992 const char *StartChar = StartLoc.getPointer() + 1; 2993 const char *EndChar = EndLoc.getPointer() - 1; 2994 jumpToLoc(EndLoc, CurBuffer); 2995 /// Eat from '<' to '>' 2996 Lex(); 2997 2998 Data = angleBracketString(StringRef(StartChar, EndChar - StartChar)); 2999 return false; 3000 } 3001 return true; 3002 } 3003 3004 /// parseDirectiveAscii: 3005 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 3006 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 3007 auto parseOp = [&]() -> bool { 3008 std::string Data; 3009 if (checkForValidSection() || parseEscapedString(Data)) 3010 return true; 3011 getStreamer().emitBytes(Data); 3012 if (ZeroTerminated) 3013 getStreamer().emitBytes(StringRef("\0", 1)); 3014 return false; 3015 }; 3016 3017 if (parseMany(parseOp)) 3018 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3019 return false; 3020 } 3021 3022 /// parseDirectiveReloc 3023 /// ::= .reloc expression , identifier [ , expression ] 3024 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 3025 const MCExpr *Offset; 3026 const MCExpr *Expr = nullptr; 3027 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 3028 3029 if (parseExpression(Offset)) 3030 return true; 3031 if (parseToken(AsmToken::Comma, "expected comma") || 3032 check(getTok().isNot(AsmToken::Identifier), "expected relocation name")) 3033 return true; 3034 3035 SMLoc NameLoc = Lexer.getTok().getLoc(); 3036 StringRef Name = Lexer.getTok().getIdentifier(); 3037 Lex(); 3038 3039 if (Lexer.is(AsmToken::Comma)) { 3040 Lex(); 3041 SMLoc ExprLoc = Lexer.getLoc(); 3042 if (parseExpression(Expr)) 3043 return true; 3044 3045 MCValue Value; 3046 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 3047 return Error(ExprLoc, "expression must be relocatable"); 3048 } 3049 3050 if (parseToken(AsmToken::EndOfStatement, 3051 "unexpected token in .reloc directive")) 3052 return true; 3053 3054 const MCTargetAsmParser &MCT = getTargetParser(); 3055 const MCSubtargetInfo &STI = MCT.getSTI(); 3056 if (Optional<std::pair<bool, std::string>> Err = 3057 getStreamer().emitRelocDirective(*Offset, Name, Expr, DirectiveLoc, 3058 STI)) 3059 return Error(Err->first ? NameLoc : OffsetLoc, Err->second); 3060 3061 return false; 3062 } 3063 3064 /// parseDirectiveValue 3065 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 3066 bool AsmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) { 3067 auto parseOp = [&]() -> bool { 3068 const MCExpr *Value; 3069 SMLoc ExprLoc = getLexer().getLoc(); 3070 if (checkForValidSection() || parseExpression(Value)) 3071 return true; 3072 // Special case constant expressions to match code generator. 3073 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3074 assert(Size <= 8 && "Invalid size"); 3075 uint64_t IntValue = MCE->getValue(); 3076 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 3077 return Error(ExprLoc, "out of range literal value"); 3078 getStreamer().emitIntValue(IntValue, Size); 3079 } else 3080 getStreamer().emitValue(Value, Size, ExprLoc); 3081 return false; 3082 }; 3083 3084 if (parseMany(parseOp)) 3085 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3086 return false; 3087 } 3088 3089 static bool parseHexOcta(AsmParser &Asm, uint64_t &hi, uint64_t &lo) { 3090 if (Asm.getTok().isNot(AsmToken::Integer) && 3091 Asm.getTok().isNot(AsmToken::BigNum)) 3092 return Asm.TokError("unknown token in expression"); 3093 SMLoc ExprLoc = Asm.getTok().getLoc(); 3094 APInt IntValue = Asm.getTok().getAPIntVal(); 3095 Asm.Lex(); 3096 if (!IntValue.isIntN(128)) 3097 return Asm.Error(ExprLoc, "out of range literal value"); 3098 if (!IntValue.isIntN(64)) { 3099 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 3100 lo = IntValue.getLoBits(64).getZExtValue(); 3101 } else { 3102 hi = 0; 3103 lo = IntValue.getZExtValue(); 3104 } 3105 return false; 3106 } 3107 3108 /// ParseDirectiveOctaValue 3109 /// ::= .octa [ hexconstant (, hexconstant)* ] 3110 3111 bool AsmParser::parseDirectiveOctaValue(StringRef IDVal) { 3112 auto parseOp = [&]() -> bool { 3113 if (checkForValidSection()) 3114 return true; 3115 uint64_t hi, lo; 3116 if (parseHexOcta(*this, hi, lo)) 3117 return true; 3118 if (MAI.isLittleEndian()) { 3119 getStreamer().emitInt64(lo); 3120 getStreamer().emitInt64(hi); 3121 } else { 3122 getStreamer().emitInt64(hi); 3123 getStreamer().emitInt64(lo); 3124 } 3125 return false; 3126 }; 3127 3128 if (parseMany(parseOp)) 3129 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3130 return false; 3131 } 3132 3133 bool AsmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) { 3134 // We don't truly support arithmetic on floating point expressions, so we 3135 // have to manually parse unary prefixes. 3136 bool IsNeg = false; 3137 if (getLexer().is(AsmToken::Minus)) { 3138 Lexer.Lex(); 3139 IsNeg = true; 3140 } else if (getLexer().is(AsmToken::Plus)) 3141 Lexer.Lex(); 3142 3143 if (Lexer.is(AsmToken::Error)) 3144 return TokError(Lexer.getErr()); 3145 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 3146 Lexer.isNot(AsmToken::Identifier)) 3147 return TokError("unexpected token in directive"); 3148 3149 // Convert to an APFloat. 3150 APFloat Value(Semantics); 3151 StringRef IDVal = getTok().getString(); 3152 if (getLexer().is(AsmToken::Identifier)) { 3153 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 3154 Value = APFloat::getInf(Semantics); 3155 else if (!IDVal.compare_lower("nan")) 3156 Value = APFloat::getNaN(Semantics, false, ~0); 3157 else 3158 return TokError("invalid floating point literal"); 3159 } else if (errorToBool( 3160 Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) 3161 .takeError())) 3162 return TokError("invalid floating point literal"); 3163 if (IsNeg) 3164 Value.changeSign(); 3165 3166 // Consume the numeric token. 3167 Lex(); 3168 3169 Res = Value.bitcastToAPInt(); 3170 3171 return false; 3172 } 3173 3174 /// parseDirectiveRealValue 3175 /// ::= (.single | .double) [ expression (, expression)* ] 3176 bool AsmParser::parseDirectiveRealValue(StringRef IDVal, 3177 const fltSemantics &Semantics) { 3178 auto parseOp = [&]() -> bool { 3179 APInt AsInt; 3180 if (checkForValidSection() || parseRealValue(Semantics, AsInt)) 3181 return true; 3182 getStreamer().emitIntValue(AsInt.getLimitedValue(), 3183 AsInt.getBitWidth() / 8); 3184 return false; 3185 }; 3186 3187 if (parseMany(parseOp)) 3188 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3189 return false; 3190 } 3191 3192 /// parseDirectiveZero 3193 /// ::= .zero expression 3194 bool AsmParser::parseDirectiveZero() { 3195 SMLoc NumBytesLoc = Lexer.getLoc(); 3196 const MCExpr *NumBytes; 3197 if (checkForValidSection() || parseExpression(NumBytes)) 3198 return true; 3199 3200 int64_t Val = 0; 3201 if (getLexer().is(AsmToken::Comma)) { 3202 Lex(); 3203 if (parseAbsoluteExpression(Val)) 3204 return true; 3205 } 3206 3207 if (parseToken(AsmToken::EndOfStatement, 3208 "unexpected token in '.zero' directive")) 3209 return true; 3210 getStreamer().emitFill(*NumBytes, Val, NumBytesLoc); 3211 3212 return false; 3213 } 3214 3215 /// parseDirectiveFill 3216 /// ::= .fill expression [ , expression [ , expression ] ] 3217 bool AsmParser::parseDirectiveFill() { 3218 SMLoc NumValuesLoc = Lexer.getLoc(); 3219 const MCExpr *NumValues; 3220 if (checkForValidSection() || parseExpression(NumValues)) 3221 return true; 3222 3223 int64_t FillSize = 1; 3224 int64_t FillExpr = 0; 3225 3226 SMLoc SizeLoc, ExprLoc; 3227 3228 if (parseOptionalToken(AsmToken::Comma)) { 3229 SizeLoc = getTok().getLoc(); 3230 if (parseAbsoluteExpression(FillSize)) 3231 return true; 3232 if (parseOptionalToken(AsmToken::Comma)) { 3233 ExprLoc = getTok().getLoc(); 3234 if (parseAbsoluteExpression(FillExpr)) 3235 return true; 3236 } 3237 } 3238 if (parseToken(AsmToken::EndOfStatement, 3239 "unexpected token in '.fill' directive")) 3240 return true; 3241 3242 if (FillSize < 0) { 3243 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 3244 return false; 3245 } 3246 if (FillSize > 8) { 3247 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 3248 FillSize = 8; 3249 } 3250 3251 if (!isUInt<32>(FillExpr) && FillSize > 4) 3252 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 3253 3254 getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc); 3255 3256 return false; 3257 } 3258 3259 /// parseDirectiveOrg 3260 /// ::= .org expression [ , expression ] 3261 bool AsmParser::parseDirectiveOrg() { 3262 const MCExpr *Offset; 3263 SMLoc OffsetLoc = Lexer.getLoc(); 3264 if (checkForValidSection() || parseExpression(Offset)) 3265 return true; 3266 3267 // Parse optional fill expression. 3268 int64_t FillExpr = 0; 3269 if (parseOptionalToken(AsmToken::Comma)) 3270 if (parseAbsoluteExpression(FillExpr)) 3271 return addErrorSuffix(" in '.org' directive"); 3272 if (parseToken(AsmToken::EndOfStatement)) 3273 return addErrorSuffix(" in '.org' directive"); 3274 3275 getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc); 3276 return false; 3277 } 3278 3279 /// parseDirectiveAlign 3280 /// ::= {.align, ...} expression [ , expression [ , expression ]] 3281 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 3282 SMLoc AlignmentLoc = getLexer().getLoc(); 3283 int64_t Alignment; 3284 SMLoc MaxBytesLoc; 3285 bool HasFillExpr = false; 3286 int64_t FillExpr = 0; 3287 int64_t MaxBytesToFill = 0; 3288 3289 auto parseAlign = [&]() -> bool { 3290 if (parseAbsoluteExpression(Alignment)) 3291 return true; 3292 if (parseOptionalToken(AsmToken::Comma)) { 3293 // The fill expression can be omitted while specifying a maximum number of 3294 // alignment bytes, e.g: 3295 // .align 3,,4 3296 if (getTok().isNot(AsmToken::Comma)) { 3297 HasFillExpr = true; 3298 if (parseAbsoluteExpression(FillExpr)) 3299 return true; 3300 } 3301 if (parseOptionalToken(AsmToken::Comma)) 3302 if (parseTokenLoc(MaxBytesLoc) || 3303 parseAbsoluteExpression(MaxBytesToFill)) 3304 return true; 3305 } 3306 return parseToken(AsmToken::EndOfStatement); 3307 }; 3308 3309 if (checkForValidSection()) 3310 return addErrorSuffix(" in directive"); 3311 // Ignore empty '.p2align' directives for GNU-as compatibility 3312 if (IsPow2 && (ValueSize == 1) && getTok().is(AsmToken::EndOfStatement)) { 3313 Warning(AlignmentLoc, "p2align directive with no operand(s) is ignored"); 3314 return parseToken(AsmToken::EndOfStatement); 3315 } 3316 if (parseAlign()) 3317 return addErrorSuffix(" in directive"); 3318 3319 // Always emit an alignment here even if we thrown an error. 3320 bool ReturnVal = false; 3321 3322 // Compute alignment in bytes. 3323 if (IsPow2) { 3324 // FIXME: Diagnose overflow. 3325 if (Alignment >= 32) { 3326 ReturnVal |= Error(AlignmentLoc, "invalid alignment value"); 3327 Alignment = 31; 3328 } 3329 3330 Alignment = 1ULL << Alignment; 3331 } else { 3332 // Reject alignments that aren't either a power of two or zero, 3333 // for gas compatibility. Alignment of zero is silently rounded 3334 // up to one. 3335 if (Alignment == 0) 3336 Alignment = 1; 3337 if (!isPowerOf2_64(Alignment)) 3338 ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2"); 3339 } 3340 3341 // Diagnose non-sensical max bytes to align. 3342 if (MaxBytesLoc.isValid()) { 3343 if (MaxBytesToFill < 1) { 3344 ReturnVal |= Error(MaxBytesLoc, 3345 "alignment directive can never be satisfied in this " 3346 "many bytes, ignoring maximum bytes expression"); 3347 MaxBytesToFill = 0; 3348 } 3349 3350 if (MaxBytesToFill >= Alignment) { 3351 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 3352 "has no effect"); 3353 MaxBytesToFill = 0; 3354 } 3355 } 3356 3357 // Check whether we should use optimal code alignment for this .align 3358 // directive. 3359 const MCSection *Section = getStreamer().getCurrentSectionOnly(); 3360 assert(Section && "must have section to emit alignment"); 3361 bool UseCodeAlign = Section->UseCodeAlign(); 3362 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 3363 ValueSize == 1 && UseCodeAlign) { 3364 getStreamer().emitCodeAlignment(Alignment, MaxBytesToFill); 3365 } else { 3366 // FIXME: Target specific behavior about how the "extra" bytes are filled. 3367 getStreamer().emitValueToAlignment(Alignment, FillExpr, ValueSize, 3368 MaxBytesToFill); 3369 } 3370 3371 return ReturnVal; 3372 } 3373 3374 /// parseDirectiveFile 3375 /// ::= .file filename 3376 /// ::= .file number [directory] filename [md5 checksum] [source source-text] 3377 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 3378 // FIXME: I'm not sure what this is. 3379 int64_t FileNumber = -1; 3380 if (getLexer().is(AsmToken::Integer)) { 3381 FileNumber = getTok().getIntVal(); 3382 Lex(); 3383 3384 if (FileNumber < 0) 3385 return TokError("negative file number"); 3386 } 3387 3388 std::string Path; 3389 3390 // Usually the directory and filename together, otherwise just the directory. 3391 // Allow the strings to have escaped octal character sequence. 3392 if (check(getTok().isNot(AsmToken::String), 3393 "unexpected token in '.file' directive") || 3394 parseEscapedString(Path)) 3395 return true; 3396 3397 StringRef Directory; 3398 StringRef Filename; 3399 std::string FilenameData; 3400 if (getLexer().is(AsmToken::String)) { 3401 if (check(FileNumber == -1, 3402 "explicit path specified, but no file number") || 3403 parseEscapedString(FilenameData)) 3404 return true; 3405 Filename = FilenameData; 3406 Directory = Path; 3407 } else { 3408 Filename = Path; 3409 } 3410 3411 uint64_t MD5Hi, MD5Lo; 3412 bool HasMD5 = false; 3413 3414 Optional<StringRef> Source; 3415 bool HasSource = false; 3416 std::string SourceString; 3417 3418 while (!parseOptionalToken(AsmToken::EndOfStatement)) { 3419 StringRef Keyword; 3420 if (check(getTok().isNot(AsmToken::Identifier), 3421 "unexpected token in '.file' directive") || 3422 parseIdentifier(Keyword)) 3423 return true; 3424 if (Keyword == "md5") { 3425 HasMD5 = true; 3426 if (check(FileNumber == -1, 3427 "MD5 checksum specified, but no file number") || 3428 parseHexOcta(*this, MD5Hi, MD5Lo)) 3429 return true; 3430 } else if (Keyword == "source") { 3431 HasSource = true; 3432 if (check(FileNumber == -1, 3433 "source specified, but no file number") || 3434 check(getTok().isNot(AsmToken::String), 3435 "unexpected token in '.file' directive") || 3436 parseEscapedString(SourceString)) 3437 return true; 3438 } else { 3439 return TokError("unexpected token in '.file' directive"); 3440 } 3441 } 3442 3443 if (FileNumber == -1) { 3444 // Ignore the directive if there is no number and the target doesn't support 3445 // numberless .file directives. This allows some portability of assembler 3446 // between different object file formats. 3447 if (getContext().getAsmInfo()->hasSingleParameterDotFile()) 3448 getStreamer().emitFileDirective(Filename); 3449 } else { 3450 // In case there is a -g option as well as debug info from directive .file, 3451 // we turn off the -g option, directly use the existing debug info instead. 3452 // Throw away any implicit file table for the assembler source. 3453 if (Ctx.getGenDwarfForAssembly()) { 3454 Ctx.getMCDwarfLineTable(0).resetFileTable(); 3455 Ctx.setGenDwarfForAssembly(false); 3456 } 3457 3458 Optional<MD5::MD5Result> CKMem; 3459 if (HasMD5) { 3460 MD5::MD5Result Sum; 3461 for (unsigned i = 0; i != 8; ++i) { 3462 Sum.Bytes[i] = uint8_t(MD5Hi >> ((7 - i) * 8)); 3463 Sum.Bytes[i + 8] = uint8_t(MD5Lo >> ((7 - i) * 8)); 3464 } 3465 CKMem = Sum; 3466 } 3467 if (HasSource) { 3468 char *SourceBuf = static_cast<char *>(Ctx.allocate(SourceString.size())); 3469 memcpy(SourceBuf, SourceString.data(), SourceString.size()); 3470 Source = StringRef(SourceBuf, SourceString.size()); 3471 } 3472 if (FileNumber == 0) { 3473 if (Ctx.getDwarfVersion() < 5) 3474 return Warning(DirectiveLoc, "file 0 not supported prior to DWARF-5"); 3475 getStreamer().emitDwarfFile0Directive(Directory, Filename, CKMem, Source); 3476 } else { 3477 Expected<unsigned> FileNumOrErr = getStreamer().tryEmitDwarfFileDirective( 3478 FileNumber, Directory, Filename, CKMem, Source); 3479 if (!FileNumOrErr) 3480 return Error(DirectiveLoc, toString(FileNumOrErr.takeError())); 3481 } 3482 // Alert the user if there are some .file directives with MD5 and some not. 3483 // But only do that once. 3484 if (!ReportedInconsistentMD5 && !Ctx.isDwarfMD5UsageConsistent(0)) { 3485 ReportedInconsistentMD5 = true; 3486 return Warning(DirectiveLoc, "inconsistent use of MD5 checksums"); 3487 } 3488 } 3489 3490 return false; 3491 } 3492 3493 /// parseDirectiveLine 3494 /// ::= .line [number] 3495 bool AsmParser::parseDirectiveLine() { 3496 int64_t LineNumber; 3497 if (getLexer().is(AsmToken::Integer)) { 3498 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3499 return true; 3500 (void)LineNumber; 3501 // FIXME: Do something with the .line. 3502 } 3503 if (parseToken(AsmToken::EndOfStatement, 3504 "unexpected token in '.line' directive")) 3505 return true; 3506 3507 return false; 3508 } 3509 3510 /// parseDirectiveLoc 3511 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3512 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3513 /// The first number is a file number, must have been previously assigned with 3514 /// a .file directive, the second number is the line number and optionally the 3515 /// third number is a column position (zero if not specified). The remaining 3516 /// optional items are .loc sub-directives. 3517 bool AsmParser::parseDirectiveLoc() { 3518 int64_t FileNumber = 0, LineNumber = 0; 3519 SMLoc Loc = getTok().getLoc(); 3520 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3521 check(FileNumber < 1 && Ctx.getDwarfVersion() < 5, Loc, 3522 "file number less than one in '.loc' directive") || 3523 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3524 "unassigned file number in '.loc' directive")) 3525 return true; 3526 3527 // optional 3528 if (getLexer().is(AsmToken::Integer)) { 3529 LineNumber = getTok().getIntVal(); 3530 if (LineNumber < 0) 3531 return TokError("line number less than zero in '.loc' directive"); 3532 Lex(); 3533 } 3534 3535 int64_t ColumnPos = 0; 3536 if (getLexer().is(AsmToken::Integer)) { 3537 ColumnPos = getTok().getIntVal(); 3538 if (ColumnPos < 0) 3539 return TokError("column position less than zero in '.loc' directive"); 3540 Lex(); 3541 } 3542 3543 auto PrevFlags = getContext().getCurrentDwarfLoc().getFlags(); 3544 unsigned Flags = PrevFlags & DWARF2_FLAG_IS_STMT; 3545 unsigned Isa = 0; 3546 int64_t Discriminator = 0; 3547 3548 auto parseLocOp = [&]() -> bool { 3549 StringRef Name; 3550 SMLoc Loc = getTok().getLoc(); 3551 if (parseIdentifier(Name)) 3552 return TokError("unexpected token in '.loc' directive"); 3553 3554 if (Name == "basic_block") 3555 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3556 else if (Name == "prologue_end") 3557 Flags |= DWARF2_FLAG_PROLOGUE_END; 3558 else if (Name == "epilogue_begin") 3559 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3560 else if (Name == "is_stmt") { 3561 Loc = getTok().getLoc(); 3562 const MCExpr *Value; 3563 if (parseExpression(Value)) 3564 return true; 3565 // The expression must be the constant 0 or 1. 3566 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3567 int Value = MCE->getValue(); 3568 if (Value == 0) 3569 Flags &= ~DWARF2_FLAG_IS_STMT; 3570 else if (Value == 1) 3571 Flags |= DWARF2_FLAG_IS_STMT; 3572 else 3573 return Error(Loc, "is_stmt value not 0 or 1"); 3574 } else { 3575 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3576 } 3577 } else if (Name == "isa") { 3578 Loc = getTok().getLoc(); 3579 const MCExpr *Value; 3580 if (parseExpression(Value)) 3581 return true; 3582 // The expression must be a constant greater or equal to 0. 3583 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3584 int Value = MCE->getValue(); 3585 if (Value < 0) 3586 return Error(Loc, "isa number less than zero"); 3587 Isa = Value; 3588 } else { 3589 return Error(Loc, "isa number not a constant value"); 3590 } 3591 } else if (Name == "discriminator") { 3592 if (parseAbsoluteExpression(Discriminator)) 3593 return true; 3594 } else { 3595 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3596 } 3597 return false; 3598 }; 3599 3600 if (parseMany(parseLocOp, false /*hasComma*/)) 3601 return true; 3602 3603 getStreamer().emitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3604 Isa, Discriminator, StringRef()); 3605 3606 return false; 3607 } 3608 3609 /// parseDirectiveStabs 3610 /// ::= .stabs string, number, number, number 3611 bool AsmParser::parseDirectiveStabs() { 3612 return TokError("unsupported directive '.stabs'"); 3613 } 3614 3615 /// parseDirectiveCVFile 3616 /// ::= .cv_file number filename [checksum] [checksumkind] 3617 bool AsmParser::parseDirectiveCVFile() { 3618 SMLoc FileNumberLoc = getTok().getLoc(); 3619 int64_t FileNumber; 3620 std::string Filename; 3621 std::string Checksum; 3622 int64_t ChecksumKind = 0; 3623 3624 if (parseIntToken(FileNumber, 3625 "expected file number in '.cv_file' directive") || 3626 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3627 check(getTok().isNot(AsmToken::String), 3628 "unexpected token in '.cv_file' directive") || 3629 parseEscapedString(Filename)) 3630 return true; 3631 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3632 if (check(getTok().isNot(AsmToken::String), 3633 "unexpected token in '.cv_file' directive") || 3634 parseEscapedString(Checksum) || 3635 parseIntToken(ChecksumKind, 3636 "expected checksum kind in '.cv_file' directive") || 3637 parseToken(AsmToken::EndOfStatement, 3638 "unexpected token in '.cv_file' directive")) 3639 return true; 3640 } 3641 3642 Checksum = fromHex(Checksum); 3643 void *CKMem = Ctx.allocate(Checksum.size(), 1); 3644 memcpy(CKMem, Checksum.data(), Checksum.size()); 3645 ArrayRef<uint8_t> ChecksumAsBytes(reinterpret_cast<const uint8_t *>(CKMem), 3646 Checksum.size()); 3647 3648 if (!getStreamer().EmitCVFileDirective(FileNumber, Filename, ChecksumAsBytes, 3649 static_cast<uint8_t>(ChecksumKind))) 3650 return Error(FileNumberLoc, "file number already allocated"); 3651 3652 return false; 3653 } 3654 3655 bool AsmParser::parseCVFunctionId(int64_t &FunctionId, 3656 StringRef DirectiveName) { 3657 SMLoc Loc; 3658 return parseTokenLoc(Loc) || 3659 parseIntToken(FunctionId, "expected function id in '" + DirectiveName + 3660 "' directive") || 3661 check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc, 3662 "expected function id within range [0, UINT_MAX)"); 3663 } 3664 3665 bool AsmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) { 3666 SMLoc Loc; 3667 return parseTokenLoc(Loc) || 3668 parseIntToken(FileNumber, "expected integer in '" + DirectiveName + 3669 "' directive") || 3670 check(FileNumber < 1, Loc, "file number less than one in '" + 3671 DirectiveName + "' directive") || 3672 check(!getCVContext().isValidFileNumber(FileNumber), Loc, 3673 "unassigned file number in '" + DirectiveName + "' directive"); 3674 } 3675 3676 /// parseDirectiveCVFuncId 3677 /// ::= .cv_func_id FunctionId 3678 /// 3679 /// Introduces a function ID that can be used with .cv_loc. 3680 bool AsmParser::parseDirectiveCVFuncId() { 3681 SMLoc FunctionIdLoc = getTok().getLoc(); 3682 int64_t FunctionId; 3683 3684 if (parseCVFunctionId(FunctionId, ".cv_func_id") || 3685 parseToken(AsmToken::EndOfStatement, 3686 "unexpected token in '.cv_func_id' directive")) 3687 return true; 3688 3689 if (!getStreamer().EmitCVFuncIdDirective(FunctionId)) 3690 return Error(FunctionIdLoc, "function id already allocated"); 3691 3692 return false; 3693 } 3694 3695 /// parseDirectiveCVInlineSiteId 3696 /// ::= .cv_inline_site_id FunctionId 3697 /// "within" IAFunc 3698 /// "inlined_at" IAFile IALine [IACol] 3699 /// 3700 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined 3701 /// at" source location information for use in the line table of the caller, 3702 /// whether the caller is a real function or another inlined call site. 3703 bool AsmParser::parseDirectiveCVInlineSiteId() { 3704 SMLoc FunctionIdLoc = getTok().getLoc(); 3705 int64_t FunctionId; 3706 int64_t IAFunc; 3707 int64_t IAFile; 3708 int64_t IALine; 3709 int64_t IACol = 0; 3710 3711 // FunctionId 3712 if (parseCVFunctionId(FunctionId, ".cv_inline_site_id")) 3713 return true; 3714 3715 // "within" 3716 if (check((getLexer().isNot(AsmToken::Identifier) || 3717 getTok().getIdentifier() != "within"), 3718 "expected 'within' identifier in '.cv_inline_site_id' directive")) 3719 return true; 3720 Lex(); 3721 3722 // IAFunc 3723 if (parseCVFunctionId(IAFunc, ".cv_inline_site_id")) 3724 return true; 3725 3726 // "inlined_at" 3727 if (check((getLexer().isNot(AsmToken::Identifier) || 3728 getTok().getIdentifier() != "inlined_at"), 3729 "expected 'inlined_at' identifier in '.cv_inline_site_id' " 3730 "directive") ) 3731 return true; 3732 Lex(); 3733 3734 // IAFile IALine 3735 if (parseCVFileId(IAFile, ".cv_inline_site_id") || 3736 parseIntToken(IALine, "expected line number after 'inlined_at'")) 3737 return true; 3738 3739 // [IACol] 3740 if (getLexer().is(AsmToken::Integer)) { 3741 IACol = getTok().getIntVal(); 3742 Lex(); 3743 } 3744 3745 if (parseToken(AsmToken::EndOfStatement, 3746 "unexpected token in '.cv_inline_site_id' directive")) 3747 return true; 3748 3749 if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile, 3750 IALine, IACol, FunctionIdLoc)) 3751 return Error(FunctionIdLoc, "function id already allocated"); 3752 3753 return false; 3754 } 3755 3756 /// parseDirectiveCVLoc 3757 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3758 /// [is_stmt VALUE] 3759 /// The first number is a file number, must have been previously assigned with 3760 /// a .file directive, the second number is the line number and optionally the 3761 /// third number is a column position (zero if not specified). The remaining 3762 /// optional items are .loc sub-directives. 3763 bool AsmParser::parseDirectiveCVLoc() { 3764 SMLoc DirectiveLoc = getTok().getLoc(); 3765 int64_t FunctionId, FileNumber; 3766 if (parseCVFunctionId(FunctionId, ".cv_loc") || 3767 parseCVFileId(FileNumber, ".cv_loc")) 3768 return true; 3769 3770 int64_t LineNumber = 0; 3771 if (getLexer().is(AsmToken::Integer)) { 3772 LineNumber = getTok().getIntVal(); 3773 if (LineNumber < 0) 3774 return TokError("line number less than zero in '.cv_loc' directive"); 3775 Lex(); 3776 } 3777 3778 int64_t ColumnPos = 0; 3779 if (getLexer().is(AsmToken::Integer)) { 3780 ColumnPos = getTok().getIntVal(); 3781 if (ColumnPos < 0) 3782 return TokError("column position less than zero in '.cv_loc' directive"); 3783 Lex(); 3784 } 3785 3786 bool PrologueEnd = false; 3787 uint64_t IsStmt = 0; 3788 3789 auto parseOp = [&]() -> bool { 3790 StringRef Name; 3791 SMLoc Loc = getTok().getLoc(); 3792 if (parseIdentifier(Name)) 3793 return TokError("unexpected token in '.cv_loc' directive"); 3794 if (Name == "prologue_end") 3795 PrologueEnd = true; 3796 else if (Name == "is_stmt") { 3797 Loc = getTok().getLoc(); 3798 const MCExpr *Value; 3799 if (parseExpression(Value)) 3800 return true; 3801 // The expression must be the constant 0 or 1. 3802 IsStmt = ~0ULL; 3803 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3804 IsStmt = MCE->getValue(); 3805 3806 if (IsStmt > 1) 3807 return Error(Loc, "is_stmt value not 0 or 1"); 3808 } else { 3809 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3810 } 3811 return false; 3812 }; 3813 3814 if (parseMany(parseOp, false /*hasComma*/)) 3815 return true; 3816 3817 getStreamer().emitCVLocDirective(FunctionId, FileNumber, LineNumber, 3818 ColumnPos, PrologueEnd, IsStmt, StringRef(), 3819 DirectiveLoc); 3820 return false; 3821 } 3822 3823 /// parseDirectiveCVLinetable 3824 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3825 bool AsmParser::parseDirectiveCVLinetable() { 3826 int64_t FunctionId; 3827 StringRef FnStartName, FnEndName; 3828 SMLoc Loc = getTok().getLoc(); 3829 if (parseCVFunctionId(FunctionId, ".cv_linetable") || 3830 parseToken(AsmToken::Comma, 3831 "unexpected token in '.cv_linetable' directive") || 3832 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3833 "expected identifier in directive") || 3834 parseToken(AsmToken::Comma, 3835 "unexpected token in '.cv_linetable' directive") || 3836 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3837 "expected identifier in directive")) 3838 return true; 3839 3840 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3841 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3842 3843 getStreamer().emitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3844 return false; 3845 } 3846 3847 /// parseDirectiveCVInlineLinetable 3848 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3849 bool AsmParser::parseDirectiveCVInlineLinetable() { 3850 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3851 StringRef FnStartName, FnEndName; 3852 SMLoc Loc = getTok().getLoc(); 3853 if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") || 3854 parseTokenLoc(Loc) || 3855 parseIntToken( 3856 SourceFileId, 3857 "expected SourceField in '.cv_inline_linetable' directive") || 3858 check(SourceFileId <= 0, Loc, 3859 "File id less than zero in '.cv_inline_linetable' directive") || 3860 parseTokenLoc(Loc) || 3861 parseIntToken( 3862 SourceLineNum, 3863 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3864 check(SourceLineNum < 0, Loc, 3865 "Line number less than zero in '.cv_inline_linetable' directive") || 3866 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3867 "expected identifier in directive") || 3868 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3869 "expected identifier in directive")) 3870 return true; 3871 3872 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3873 return true; 3874 3875 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3876 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3877 getStreamer().emitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3878 SourceLineNum, FnStartSym, 3879 FnEndSym); 3880 return false; 3881 } 3882 3883 void AsmParser::initializeCVDefRangeTypeMap() { 3884 CVDefRangeTypeMap["reg"] = CVDR_DEFRANGE_REGISTER; 3885 CVDefRangeTypeMap["frame_ptr_rel"] = CVDR_DEFRANGE_FRAMEPOINTER_REL; 3886 CVDefRangeTypeMap["subfield_reg"] = CVDR_DEFRANGE_SUBFIELD_REGISTER; 3887 CVDefRangeTypeMap["reg_rel"] = CVDR_DEFRANGE_REGISTER_REL; 3888 } 3889 3890 /// parseDirectiveCVDefRange 3891 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3892 bool AsmParser::parseDirectiveCVDefRange() { 3893 SMLoc Loc; 3894 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3895 while (getLexer().is(AsmToken::Identifier)) { 3896 Loc = getLexer().getLoc(); 3897 StringRef GapStartName; 3898 if (parseIdentifier(GapStartName)) 3899 return Error(Loc, "expected identifier in directive"); 3900 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3901 3902 Loc = getLexer().getLoc(); 3903 StringRef GapEndName; 3904 if (parseIdentifier(GapEndName)) 3905 return Error(Loc, "expected identifier in directive"); 3906 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3907 3908 Ranges.push_back({GapStartSym, GapEndSym}); 3909 } 3910 3911 StringRef CVDefRangeTypeStr; 3912 if (parseToken( 3913 AsmToken::Comma, 3914 "expected comma before def_range type in .cv_def_range directive") || 3915 parseIdentifier(CVDefRangeTypeStr)) 3916 return Error(Loc, "expected def_range type in directive"); 3917 3918 StringMap<CVDefRangeType>::const_iterator CVTypeIt = 3919 CVDefRangeTypeMap.find(CVDefRangeTypeStr); 3920 CVDefRangeType CVDRType = (CVTypeIt == CVDefRangeTypeMap.end()) 3921 ? CVDR_DEFRANGE 3922 : CVTypeIt->getValue(); 3923 switch (CVDRType) { 3924 case CVDR_DEFRANGE_REGISTER: { 3925 int64_t DRRegister; 3926 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3927 ".cv_def_range directive") || 3928 parseAbsoluteExpression(DRRegister)) 3929 return Error(Loc, "expected register number"); 3930 3931 codeview::DefRangeRegisterHeader DRHdr; 3932 DRHdr.Register = DRRegister; 3933 DRHdr.MayHaveNoName = 0; 3934 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3935 break; 3936 } 3937 case CVDR_DEFRANGE_FRAMEPOINTER_REL: { 3938 int64_t DROffset; 3939 if (parseToken(AsmToken::Comma, 3940 "expected comma before offset in .cv_def_range directive") || 3941 parseAbsoluteExpression(DROffset)) 3942 return Error(Loc, "expected offset value"); 3943 3944 codeview::DefRangeFramePointerRelHeader DRHdr; 3945 DRHdr.Offset = DROffset; 3946 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3947 break; 3948 } 3949 case CVDR_DEFRANGE_SUBFIELD_REGISTER: { 3950 int64_t DRRegister; 3951 int64_t DROffsetInParent; 3952 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3953 ".cv_def_range directive") || 3954 parseAbsoluteExpression(DRRegister)) 3955 return Error(Loc, "expected register number"); 3956 if (parseToken(AsmToken::Comma, 3957 "expected comma before offset in .cv_def_range directive") || 3958 parseAbsoluteExpression(DROffsetInParent)) 3959 return Error(Loc, "expected offset value"); 3960 3961 codeview::DefRangeSubfieldRegisterHeader DRHdr; 3962 DRHdr.Register = DRRegister; 3963 DRHdr.MayHaveNoName = 0; 3964 DRHdr.OffsetInParent = DROffsetInParent; 3965 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3966 break; 3967 } 3968 case CVDR_DEFRANGE_REGISTER_REL: { 3969 int64_t DRRegister; 3970 int64_t DRFlags; 3971 int64_t DRBasePointerOffset; 3972 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3973 ".cv_def_range directive") || 3974 parseAbsoluteExpression(DRRegister)) 3975 return Error(Loc, "expected register value"); 3976 if (parseToken( 3977 AsmToken::Comma, 3978 "expected comma before flag value in .cv_def_range directive") || 3979 parseAbsoluteExpression(DRFlags)) 3980 return Error(Loc, "expected flag value"); 3981 if (parseToken(AsmToken::Comma, "expected comma before base pointer offset " 3982 "in .cv_def_range directive") || 3983 parseAbsoluteExpression(DRBasePointerOffset)) 3984 return Error(Loc, "expected base pointer offset value"); 3985 3986 codeview::DefRangeRegisterRelHeader DRHdr; 3987 DRHdr.Register = DRRegister; 3988 DRHdr.Flags = DRFlags; 3989 DRHdr.BasePointerOffset = DRBasePointerOffset; 3990 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3991 break; 3992 } 3993 default: 3994 return Error(Loc, "unexpected def_range type in .cv_def_range directive"); 3995 } 3996 return true; 3997 } 3998 3999 /// parseDirectiveCVString 4000 /// ::= .cv_stringtable "string" 4001 bool AsmParser::parseDirectiveCVString() { 4002 std::string Data; 4003 if (checkForValidSection() || parseEscapedString(Data)) 4004 return addErrorSuffix(" in '.cv_string' directive"); 4005 4006 // Put the string in the table and emit the offset. 4007 std::pair<StringRef, unsigned> Insertion = 4008 getCVContext().addToStringTable(Data); 4009 getStreamer().emitInt32(Insertion.second); 4010 return false; 4011 } 4012 4013 /// parseDirectiveCVStringTable 4014 /// ::= .cv_stringtable 4015 bool AsmParser::parseDirectiveCVStringTable() { 4016 getStreamer().emitCVStringTableDirective(); 4017 return false; 4018 } 4019 4020 /// parseDirectiveCVFileChecksums 4021 /// ::= .cv_filechecksums 4022 bool AsmParser::parseDirectiveCVFileChecksums() { 4023 getStreamer().emitCVFileChecksumsDirective(); 4024 return false; 4025 } 4026 4027 /// parseDirectiveCVFileChecksumOffset 4028 /// ::= .cv_filechecksumoffset fileno 4029 bool AsmParser::parseDirectiveCVFileChecksumOffset() { 4030 int64_t FileNo; 4031 if (parseIntToken(FileNo, "expected identifier in directive")) 4032 return true; 4033 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 4034 return true; 4035 getStreamer().emitCVFileChecksumOffsetDirective(FileNo); 4036 return false; 4037 } 4038 4039 /// parseDirectiveCVFPOData 4040 /// ::= .cv_fpo_data procsym 4041 bool AsmParser::parseDirectiveCVFPOData() { 4042 SMLoc DirLoc = getLexer().getLoc(); 4043 StringRef ProcName; 4044 if (parseIdentifier(ProcName)) 4045 return TokError("expected symbol name"); 4046 if (parseEOL("unexpected tokens")) 4047 return addErrorSuffix(" in '.cv_fpo_data' directive"); 4048 MCSymbol *ProcSym = getContext().getOrCreateSymbol(ProcName); 4049 getStreamer().EmitCVFPOData(ProcSym, DirLoc); 4050 return false; 4051 } 4052 4053 /// parseDirectiveCFISections 4054 /// ::= .cfi_sections section [, section] 4055 bool AsmParser::parseDirectiveCFISections() { 4056 StringRef Name; 4057 bool EH = false; 4058 bool Debug = false; 4059 4060 if (parseIdentifier(Name)) 4061 return TokError("Expected an identifier"); 4062 4063 if (Name == ".eh_frame") 4064 EH = true; 4065 else if (Name == ".debug_frame") 4066 Debug = true; 4067 4068 if (getLexer().is(AsmToken::Comma)) { 4069 Lex(); 4070 4071 if (parseIdentifier(Name)) 4072 return TokError("Expected an identifier"); 4073 4074 if (Name == ".eh_frame") 4075 EH = true; 4076 else if (Name == ".debug_frame") 4077 Debug = true; 4078 } 4079 4080 getStreamer().emitCFISections(EH, Debug); 4081 return false; 4082 } 4083 4084 /// parseDirectiveCFIStartProc 4085 /// ::= .cfi_startproc [simple] 4086 bool AsmParser::parseDirectiveCFIStartProc() { 4087 StringRef Simple; 4088 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4089 if (check(parseIdentifier(Simple) || Simple != "simple", 4090 "unexpected token") || 4091 parseToken(AsmToken::EndOfStatement)) 4092 return addErrorSuffix(" in '.cfi_startproc' directive"); 4093 } 4094 4095 // TODO(kristina): Deal with a corner case of incorrect diagnostic context 4096 // being produced if this directive is emitted as part of preprocessor macro 4097 // expansion which can *ONLY* happen if Clang's cc1as is the API consumer. 4098 // Tools like llvm-mc on the other hand are not affected by it, and report 4099 // correct context information. 4100 getStreamer().emitCFIStartProc(!Simple.empty(), Lexer.getLoc()); 4101 return false; 4102 } 4103 4104 /// parseDirectiveCFIEndProc 4105 /// ::= .cfi_endproc 4106 bool AsmParser::parseDirectiveCFIEndProc() { 4107 getStreamer().emitCFIEndProc(); 4108 return false; 4109 } 4110 4111 /// parse register name or number. 4112 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 4113 SMLoc DirectiveLoc) { 4114 unsigned RegNo; 4115 4116 if (getLexer().isNot(AsmToken::Integer)) { 4117 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 4118 return true; 4119 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 4120 } else 4121 return parseAbsoluteExpression(Register); 4122 4123 return false; 4124 } 4125 4126 /// parseDirectiveCFIDefCfa 4127 /// ::= .cfi_def_cfa register, offset 4128 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 4129 int64_t Register = 0, Offset = 0; 4130 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4131 parseToken(AsmToken::Comma, "unexpected token in directive") || 4132 parseAbsoluteExpression(Offset)) 4133 return true; 4134 4135 getStreamer().emitCFIDefCfa(Register, Offset); 4136 return false; 4137 } 4138 4139 /// parseDirectiveCFIDefCfaOffset 4140 /// ::= .cfi_def_cfa_offset offset 4141 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 4142 int64_t Offset = 0; 4143 if (parseAbsoluteExpression(Offset)) 4144 return true; 4145 4146 getStreamer().emitCFIDefCfaOffset(Offset); 4147 return false; 4148 } 4149 4150 /// parseDirectiveCFIRegister 4151 /// ::= .cfi_register register, register 4152 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 4153 int64_t Register1 = 0, Register2 = 0; 4154 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || 4155 parseToken(AsmToken::Comma, "unexpected token in directive") || 4156 parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 4157 return true; 4158 4159 getStreamer().emitCFIRegister(Register1, Register2); 4160 return false; 4161 } 4162 4163 /// parseDirectiveCFIWindowSave 4164 /// ::= .cfi_window_save 4165 bool AsmParser::parseDirectiveCFIWindowSave() { 4166 getStreamer().emitCFIWindowSave(); 4167 return false; 4168 } 4169 4170 /// parseDirectiveCFIAdjustCfaOffset 4171 /// ::= .cfi_adjust_cfa_offset adjustment 4172 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 4173 int64_t Adjustment = 0; 4174 if (parseAbsoluteExpression(Adjustment)) 4175 return true; 4176 4177 getStreamer().emitCFIAdjustCfaOffset(Adjustment); 4178 return false; 4179 } 4180 4181 /// parseDirectiveCFIDefCfaRegister 4182 /// ::= .cfi_def_cfa_register register 4183 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 4184 int64_t Register = 0; 4185 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4186 return true; 4187 4188 getStreamer().emitCFIDefCfaRegister(Register); 4189 return false; 4190 } 4191 4192 /// parseDirectiveCFIOffset 4193 /// ::= .cfi_offset register, offset 4194 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 4195 int64_t Register = 0; 4196 int64_t Offset = 0; 4197 4198 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4199 parseToken(AsmToken::Comma, "unexpected token in directive") || 4200 parseAbsoluteExpression(Offset)) 4201 return true; 4202 4203 getStreamer().emitCFIOffset(Register, Offset); 4204 return false; 4205 } 4206 4207 /// parseDirectiveCFIRelOffset 4208 /// ::= .cfi_rel_offset register, offset 4209 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 4210 int64_t Register = 0, Offset = 0; 4211 4212 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4213 parseToken(AsmToken::Comma, "unexpected token in directive") || 4214 parseAbsoluteExpression(Offset)) 4215 return true; 4216 4217 getStreamer().emitCFIRelOffset(Register, Offset); 4218 return false; 4219 } 4220 4221 static bool isValidEncoding(int64_t Encoding) { 4222 if (Encoding & ~0xff) 4223 return false; 4224 4225 if (Encoding == dwarf::DW_EH_PE_omit) 4226 return true; 4227 4228 const unsigned Format = Encoding & 0xf; 4229 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 4230 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 4231 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 4232 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 4233 return false; 4234 4235 const unsigned Application = Encoding & 0x70; 4236 if (Application != dwarf::DW_EH_PE_absptr && 4237 Application != dwarf::DW_EH_PE_pcrel) 4238 return false; 4239 4240 return true; 4241 } 4242 4243 /// parseDirectiveCFIPersonalityOrLsda 4244 /// IsPersonality true for cfi_personality, false for cfi_lsda 4245 /// ::= .cfi_personality encoding, [symbol_name] 4246 /// ::= .cfi_lsda encoding, [symbol_name] 4247 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 4248 int64_t Encoding = 0; 4249 if (parseAbsoluteExpression(Encoding)) 4250 return true; 4251 if (Encoding == dwarf::DW_EH_PE_omit) 4252 return false; 4253 4254 StringRef Name; 4255 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 4256 parseToken(AsmToken::Comma, "unexpected token in directive") || 4257 check(parseIdentifier(Name), "expected identifier in directive")) 4258 return true; 4259 4260 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4261 4262 if (IsPersonality) 4263 getStreamer().emitCFIPersonality(Sym, Encoding); 4264 else 4265 getStreamer().emitCFILsda(Sym, Encoding); 4266 return false; 4267 } 4268 4269 /// parseDirectiveCFIRememberState 4270 /// ::= .cfi_remember_state 4271 bool AsmParser::parseDirectiveCFIRememberState() { 4272 getStreamer().emitCFIRememberState(); 4273 return false; 4274 } 4275 4276 /// parseDirectiveCFIRestoreState 4277 /// ::= .cfi_remember_state 4278 bool AsmParser::parseDirectiveCFIRestoreState() { 4279 getStreamer().emitCFIRestoreState(); 4280 return false; 4281 } 4282 4283 /// parseDirectiveCFISameValue 4284 /// ::= .cfi_same_value register 4285 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 4286 int64_t Register = 0; 4287 4288 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4289 return true; 4290 4291 getStreamer().emitCFISameValue(Register); 4292 return false; 4293 } 4294 4295 /// parseDirectiveCFIRestore 4296 /// ::= .cfi_restore register 4297 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 4298 int64_t Register = 0; 4299 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4300 return true; 4301 4302 getStreamer().emitCFIRestore(Register); 4303 return false; 4304 } 4305 4306 /// parseDirectiveCFIEscape 4307 /// ::= .cfi_escape expression[,...] 4308 bool AsmParser::parseDirectiveCFIEscape() { 4309 std::string Values; 4310 int64_t CurrValue; 4311 if (parseAbsoluteExpression(CurrValue)) 4312 return true; 4313 4314 Values.push_back((uint8_t)CurrValue); 4315 4316 while (getLexer().is(AsmToken::Comma)) { 4317 Lex(); 4318 4319 if (parseAbsoluteExpression(CurrValue)) 4320 return true; 4321 4322 Values.push_back((uint8_t)CurrValue); 4323 } 4324 4325 getStreamer().emitCFIEscape(Values); 4326 return false; 4327 } 4328 4329 /// parseDirectiveCFIReturnColumn 4330 /// ::= .cfi_return_column register 4331 bool AsmParser::parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc) { 4332 int64_t Register = 0; 4333 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4334 return true; 4335 getStreamer().emitCFIReturnColumn(Register); 4336 return false; 4337 } 4338 4339 /// parseDirectiveCFISignalFrame 4340 /// ::= .cfi_signal_frame 4341 bool AsmParser::parseDirectiveCFISignalFrame() { 4342 if (parseToken(AsmToken::EndOfStatement, 4343 "unexpected token in '.cfi_signal_frame'")) 4344 return true; 4345 4346 getStreamer().emitCFISignalFrame(); 4347 return false; 4348 } 4349 4350 /// parseDirectiveCFIUndefined 4351 /// ::= .cfi_undefined register 4352 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 4353 int64_t Register = 0; 4354 4355 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4356 return true; 4357 4358 getStreamer().emitCFIUndefined(Register); 4359 return false; 4360 } 4361 4362 /// parseDirectiveAltmacro 4363 /// ::= .altmacro 4364 /// ::= .noaltmacro 4365 bool AsmParser::parseDirectiveAltmacro(StringRef Directive) { 4366 if (getLexer().isNot(AsmToken::EndOfStatement)) 4367 return TokError("unexpected token in '" + Directive + "' directive"); 4368 AltMacroMode = (Directive == ".altmacro"); 4369 return false; 4370 } 4371 4372 /// parseDirectiveMacrosOnOff 4373 /// ::= .macros_on 4374 /// ::= .macros_off 4375 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 4376 if (parseToken(AsmToken::EndOfStatement, 4377 "unexpected token in '" + Directive + "' directive")) 4378 return true; 4379 4380 setMacrosEnabled(Directive == ".macros_on"); 4381 return false; 4382 } 4383 4384 /// parseDirectiveMacro 4385 /// ::= .macro name[,] [parameters] 4386 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 4387 StringRef Name; 4388 if (parseIdentifier(Name)) 4389 return TokError("expected identifier in '.macro' directive"); 4390 4391 if (getLexer().is(AsmToken::Comma)) 4392 Lex(); 4393 4394 MCAsmMacroParameters Parameters; 4395 while (getLexer().isNot(AsmToken::EndOfStatement)) { 4396 4397 if (!Parameters.empty() && Parameters.back().Vararg) 4398 return Error(Lexer.getLoc(), "vararg parameter '" + 4399 Parameters.back().Name + 4400 "' should be the last parameter"); 4401 4402 MCAsmMacroParameter Parameter; 4403 if (parseIdentifier(Parameter.Name)) 4404 return TokError("expected identifier in '.macro' directive"); 4405 4406 // Emit an error if two (or more) named parameters share the same name 4407 for (const MCAsmMacroParameter& CurrParam : Parameters) 4408 if (CurrParam.Name.equals(Parameter.Name)) 4409 return TokError("macro '" + Name + "' has multiple parameters" 4410 " named '" + Parameter.Name + "'"); 4411 4412 if (Lexer.is(AsmToken::Colon)) { 4413 Lex(); // consume ':' 4414 4415 SMLoc QualLoc; 4416 StringRef Qualifier; 4417 4418 QualLoc = Lexer.getLoc(); 4419 if (parseIdentifier(Qualifier)) 4420 return Error(QualLoc, "missing parameter qualifier for " 4421 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4422 4423 if (Qualifier == "req") 4424 Parameter.Required = true; 4425 else if (Qualifier == "vararg") 4426 Parameter.Vararg = true; 4427 else 4428 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 4429 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 4430 } 4431 4432 if (getLexer().is(AsmToken::Equal)) { 4433 Lex(); 4434 4435 SMLoc ParamLoc; 4436 4437 ParamLoc = Lexer.getLoc(); 4438 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 4439 return true; 4440 4441 if (Parameter.Required) 4442 Warning(ParamLoc, "pointless default value for required parameter " 4443 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4444 } 4445 4446 Parameters.push_back(std::move(Parameter)); 4447 4448 if (getLexer().is(AsmToken::Comma)) 4449 Lex(); 4450 } 4451 4452 // Eat just the end of statement. 4453 Lexer.Lex(); 4454 4455 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 4456 AsmToken EndToken, StartToken = getTok(); 4457 unsigned MacroDepth = 0; 4458 // Lex the macro definition. 4459 while (true) { 4460 // Ignore Lexing errors in macros. 4461 while (Lexer.is(AsmToken::Error)) { 4462 Lexer.Lex(); 4463 } 4464 4465 // Check whether we have reached the end of the file. 4466 if (getLexer().is(AsmToken::Eof)) 4467 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 4468 4469 // Otherwise, check whether we have reach the .endmacro or the start of a 4470 // preprocessor line marker. 4471 if (getLexer().is(AsmToken::Identifier)) { 4472 if (getTok().getIdentifier() == ".endm" || 4473 getTok().getIdentifier() == ".endmacro") { 4474 if (MacroDepth == 0) { // Outermost macro. 4475 EndToken = getTok(); 4476 Lexer.Lex(); 4477 if (getLexer().isNot(AsmToken::EndOfStatement)) 4478 return TokError("unexpected token in '" + EndToken.getIdentifier() + 4479 "' directive"); 4480 break; 4481 } else { 4482 // Otherwise we just found the end of an inner macro. 4483 --MacroDepth; 4484 } 4485 } else if (getTok().getIdentifier() == ".macro") { 4486 // We allow nested macros. Those aren't instantiated until the outermost 4487 // macro is expanded so just ignore them for now. 4488 ++MacroDepth; 4489 } 4490 } else if (Lexer.is(AsmToken::HashDirective)) { 4491 (void)parseCppHashLineFilenameComment(getLexer().getLoc()); 4492 } 4493 4494 // Otherwise, scan til the end of the statement. 4495 eatToEndOfStatement(); 4496 } 4497 4498 if (getContext().lookupMacro(Name)) { 4499 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 4500 } 4501 4502 const char *BodyStart = StartToken.getLoc().getPointer(); 4503 const char *BodyEnd = EndToken.getLoc().getPointer(); 4504 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4505 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 4506 MCAsmMacro Macro(Name, Body, std::move(Parameters)); 4507 DEBUG_WITH_TYPE("asm-macros", dbgs() << "Defining new macro:\n"; 4508 Macro.dump()); 4509 getContext().defineMacro(Name, std::move(Macro)); 4510 return false; 4511 } 4512 4513 /// checkForBadMacro 4514 /// 4515 /// With the support added for named parameters there may be code out there that 4516 /// is transitioning from positional parameters. In versions of gas that did 4517 /// not support named parameters they would be ignored on the macro definition. 4518 /// But to support both styles of parameters this is not possible so if a macro 4519 /// definition has named parameters but does not use them and has what appears 4520 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 4521 /// warning that the positional parameter found in body which have no effect. 4522 /// Hoping the developer will either remove the named parameters from the macro 4523 /// definition so the positional parameters get used if that was what was 4524 /// intended or change the macro to use the named parameters. It is possible 4525 /// this warning will trigger when the none of the named parameters are used 4526 /// and the strings like $1 are infact to simply to be passed trough unchanged. 4527 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 4528 StringRef Body, 4529 ArrayRef<MCAsmMacroParameter> Parameters) { 4530 // If this macro is not defined with named parameters the warning we are 4531 // checking for here doesn't apply. 4532 unsigned NParameters = Parameters.size(); 4533 if (NParameters == 0) 4534 return; 4535 4536 bool NamedParametersFound = false; 4537 bool PositionalParametersFound = false; 4538 4539 // Look at the body of the macro for use of both the named parameters and what 4540 // are likely to be positional parameters. This is what expandMacro() is 4541 // doing when it finds the parameters in the body. 4542 while (!Body.empty()) { 4543 // Scan for the next possible parameter. 4544 std::size_t End = Body.size(), Pos = 0; 4545 for (; Pos != End; ++Pos) { 4546 // Check for a substitution or escape. 4547 // This macro is defined with parameters, look for \foo, \bar, etc. 4548 if (Body[Pos] == '\\' && Pos + 1 != End) 4549 break; 4550 4551 // This macro should have parameters, but look for $0, $1, ..., $n too. 4552 if (Body[Pos] != '$' || Pos + 1 == End) 4553 continue; 4554 char Next = Body[Pos + 1]; 4555 if (Next == '$' || Next == 'n' || 4556 isdigit(static_cast<unsigned char>(Next))) 4557 break; 4558 } 4559 4560 // Check if we reached the end. 4561 if (Pos == End) 4562 break; 4563 4564 if (Body[Pos] == '$') { 4565 switch (Body[Pos + 1]) { 4566 // $$ => $ 4567 case '$': 4568 break; 4569 4570 // $n => number of arguments 4571 case 'n': 4572 PositionalParametersFound = true; 4573 break; 4574 4575 // $[0-9] => argument 4576 default: { 4577 PositionalParametersFound = true; 4578 break; 4579 } 4580 } 4581 Pos += 2; 4582 } else { 4583 unsigned I = Pos + 1; 4584 while (isIdentifierChar(Body[I]) && I + 1 != End) 4585 ++I; 4586 4587 const char *Begin = Body.data() + Pos + 1; 4588 StringRef Argument(Begin, I - (Pos + 1)); 4589 unsigned Index = 0; 4590 for (; Index < NParameters; ++Index) 4591 if (Parameters[Index].Name == Argument) 4592 break; 4593 4594 if (Index == NParameters) { 4595 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 4596 Pos += 3; 4597 else { 4598 Pos = I; 4599 } 4600 } else { 4601 NamedParametersFound = true; 4602 Pos += 1 + Argument.size(); 4603 } 4604 } 4605 // Update the scan point. 4606 Body = Body.substr(Pos); 4607 } 4608 4609 if (!NamedParametersFound && PositionalParametersFound) 4610 Warning(DirectiveLoc, "macro defined with named parameters which are not " 4611 "used in macro body, possible positional parameter " 4612 "found in body which will have no effect"); 4613 } 4614 4615 /// parseDirectiveExitMacro 4616 /// ::= .exitm 4617 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 4618 if (parseToken(AsmToken::EndOfStatement, 4619 "unexpected token in '" + Directive + "' directive")) 4620 return true; 4621 4622 if (!isInsideMacroInstantiation()) 4623 return TokError("unexpected '" + Directive + "' in file, " 4624 "no current macro definition"); 4625 4626 // Exit all conditionals that are active in the current macro. 4627 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 4628 TheCondState = TheCondStack.back(); 4629 TheCondStack.pop_back(); 4630 } 4631 4632 handleMacroExit(); 4633 return false; 4634 } 4635 4636 /// parseDirectiveEndMacro 4637 /// ::= .endm 4638 /// ::= .endmacro 4639 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 4640 if (getLexer().isNot(AsmToken::EndOfStatement)) 4641 return TokError("unexpected token in '" + Directive + "' directive"); 4642 4643 // If we are inside a macro instantiation, terminate the current 4644 // instantiation. 4645 if (isInsideMacroInstantiation()) { 4646 handleMacroExit(); 4647 return false; 4648 } 4649 4650 // Otherwise, this .endmacro is a stray entry in the file; well formed 4651 // .endmacro directives are handled during the macro definition parsing. 4652 return TokError("unexpected '" + Directive + "' in file, " 4653 "no current macro definition"); 4654 } 4655 4656 /// parseDirectivePurgeMacro 4657 /// ::= .purgem 4658 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 4659 StringRef Name; 4660 SMLoc Loc; 4661 if (parseTokenLoc(Loc) || 4662 check(parseIdentifier(Name), Loc, 4663 "expected identifier in '.purgem' directive") || 4664 parseToken(AsmToken::EndOfStatement, 4665 "unexpected token in '.purgem' directive")) 4666 return true; 4667 4668 if (!getContext().lookupMacro(Name)) 4669 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 4670 4671 getContext().undefineMacro(Name); 4672 DEBUG_WITH_TYPE("asm-macros", dbgs() 4673 << "Un-defining macro: " << Name << "\n"); 4674 return false; 4675 } 4676 4677 /// parseDirectiveBundleAlignMode 4678 /// ::= {.bundle_align_mode} expression 4679 bool AsmParser::parseDirectiveBundleAlignMode() { 4680 // Expect a single argument: an expression that evaluates to a constant 4681 // in the inclusive range 0-30. 4682 SMLoc ExprLoc = getLexer().getLoc(); 4683 int64_t AlignSizePow2; 4684 if (checkForValidSection() || parseAbsoluteExpression(AlignSizePow2) || 4685 parseToken(AsmToken::EndOfStatement, "unexpected token after expression " 4686 "in '.bundle_align_mode' " 4687 "directive") || 4688 check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc, 4689 "invalid bundle alignment size (expected between 0 and 30)")) 4690 return true; 4691 4692 // Because of AlignSizePow2's verified range we can safely truncate it to 4693 // unsigned. 4694 getStreamer().emitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 4695 return false; 4696 } 4697 4698 /// parseDirectiveBundleLock 4699 /// ::= {.bundle_lock} [align_to_end] 4700 bool AsmParser::parseDirectiveBundleLock() { 4701 if (checkForValidSection()) 4702 return true; 4703 bool AlignToEnd = false; 4704 4705 StringRef Option; 4706 SMLoc Loc = getTok().getLoc(); 4707 const char *kInvalidOptionError = 4708 "invalid option for '.bundle_lock' directive"; 4709 4710 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4711 if (check(parseIdentifier(Option), Loc, kInvalidOptionError) || 4712 check(Option != "align_to_end", Loc, kInvalidOptionError) || 4713 parseToken(AsmToken::EndOfStatement, 4714 "unexpected token after '.bundle_lock' directive option")) 4715 return true; 4716 AlignToEnd = true; 4717 } 4718 4719 getStreamer().emitBundleLock(AlignToEnd); 4720 return false; 4721 } 4722 4723 /// parseDirectiveBundleLock 4724 /// ::= {.bundle_lock} 4725 bool AsmParser::parseDirectiveBundleUnlock() { 4726 if (checkForValidSection() || 4727 parseToken(AsmToken::EndOfStatement, 4728 "unexpected token in '.bundle_unlock' directive")) 4729 return true; 4730 4731 getStreamer().emitBundleUnlock(); 4732 return false; 4733 } 4734 4735 /// parseDirectiveSpace 4736 /// ::= (.skip | .space) expression [ , expression ] 4737 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4738 SMLoc NumBytesLoc = Lexer.getLoc(); 4739 const MCExpr *NumBytes; 4740 if (checkForValidSection() || parseExpression(NumBytes)) 4741 return true; 4742 4743 int64_t FillExpr = 0; 4744 if (parseOptionalToken(AsmToken::Comma)) 4745 if (parseAbsoluteExpression(FillExpr)) 4746 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4747 if (parseToken(AsmToken::EndOfStatement)) 4748 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4749 4750 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4751 getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc); 4752 4753 return false; 4754 } 4755 4756 /// parseDirectiveDCB 4757 /// ::= .dcb.{b, l, w} expression, expression 4758 bool AsmParser::parseDirectiveDCB(StringRef IDVal, unsigned Size) { 4759 SMLoc NumValuesLoc = Lexer.getLoc(); 4760 int64_t NumValues; 4761 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4762 return true; 4763 4764 if (NumValues < 0) { 4765 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4766 return false; 4767 } 4768 4769 if (parseToken(AsmToken::Comma, 4770 "unexpected token in '" + Twine(IDVal) + "' directive")) 4771 return true; 4772 4773 const MCExpr *Value; 4774 SMLoc ExprLoc = getLexer().getLoc(); 4775 if (parseExpression(Value)) 4776 return true; 4777 4778 // Special case constant expressions to match code generator. 4779 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 4780 assert(Size <= 8 && "Invalid size"); 4781 uint64_t IntValue = MCE->getValue(); 4782 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 4783 return Error(ExprLoc, "literal value out of range for directive"); 4784 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4785 getStreamer().emitIntValue(IntValue, Size); 4786 } else { 4787 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4788 getStreamer().emitValue(Value, Size, ExprLoc); 4789 } 4790 4791 if (parseToken(AsmToken::EndOfStatement, 4792 "unexpected token in '" + Twine(IDVal) + "' directive")) 4793 return true; 4794 4795 return false; 4796 } 4797 4798 /// parseDirectiveRealDCB 4799 /// ::= .dcb.{d, s} expression, expression 4800 bool AsmParser::parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &Semantics) { 4801 SMLoc NumValuesLoc = Lexer.getLoc(); 4802 int64_t NumValues; 4803 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4804 return true; 4805 4806 if (NumValues < 0) { 4807 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4808 return false; 4809 } 4810 4811 if (parseToken(AsmToken::Comma, 4812 "unexpected token in '" + Twine(IDVal) + "' directive")) 4813 return true; 4814 4815 APInt AsInt; 4816 if (parseRealValue(Semantics, AsInt)) 4817 return true; 4818 4819 if (parseToken(AsmToken::EndOfStatement, 4820 "unexpected token in '" + Twine(IDVal) + "' directive")) 4821 return true; 4822 4823 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4824 getStreamer().emitIntValue(AsInt.getLimitedValue(), 4825 AsInt.getBitWidth() / 8); 4826 4827 return false; 4828 } 4829 4830 /// parseDirectiveDS 4831 /// ::= .ds.{b, d, l, p, s, w, x} expression 4832 bool AsmParser::parseDirectiveDS(StringRef IDVal, unsigned Size) { 4833 SMLoc NumValuesLoc = Lexer.getLoc(); 4834 int64_t NumValues; 4835 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4836 return true; 4837 4838 if (NumValues < 0) { 4839 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4840 return false; 4841 } 4842 4843 if (parseToken(AsmToken::EndOfStatement, 4844 "unexpected token in '" + Twine(IDVal) + "' directive")) 4845 return true; 4846 4847 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4848 getStreamer().emitFill(Size, 0); 4849 4850 return false; 4851 } 4852 4853 /// parseDirectiveLEB128 4854 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4855 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4856 if (checkForValidSection()) 4857 return true; 4858 4859 auto parseOp = [&]() -> bool { 4860 const MCExpr *Value; 4861 if (parseExpression(Value)) 4862 return true; 4863 if (Signed) 4864 getStreamer().emitSLEB128Value(Value); 4865 else 4866 getStreamer().emitULEB128Value(Value); 4867 return false; 4868 }; 4869 4870 if (parseMany(parseOp)) 4871 return addErrorSuffix(" in directive"); 4872 4873 return false; 4874 } 4875 4876 /// parseDirectiveSymbolAttribute 4877 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4878 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4879 auto parseOp = [&]() -> bool { 4880 StringRef Name; 4881 SMLoc Loc = getTok().getLoc(); 4882 if (parseIdentifier(Name)) 4883 return Error(Loc, "expected identifier"); 4884 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4885 4886 // Assembler local symbols don't make any sense here. Complain loudly. 4887 if (Sym->isTemporary()) 4888 return Error(Loc, "non-local symbol required"); 4889 4890 if (!getStreamer().emitSymbolAttribute(Sym, Attr)) 4891 return Error(Loc, "unable to emit symbol attribute"); 4892 return false; 4893 }; 4894 4895 if (parseMany(parseOp)) 4896 return addErrorSuffix(" in directive"); 4897 return false; 4898 } 4899 4900 /// parseDirectiveComm 4901 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4902 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4903 if (checkForValidSection()) 4904 return true; 4905 4906 SMLoc IDLoc = getLexer().getLoc(); 4907 StringRef Name; 4908 if (parseIdentifier(Name)) 4909 return TokError("expected identifier in directive"); 4910 4911 // Handle the identifier as the key symbol. 4912 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4913 4914 if (getLexer().isNot(AsmToken::Comma)) 4915 return TokError("unexpected token in directive"); 4916 Lex(); 4917 4918 int64_t Size; 4919 SMLoc SizeLoc = getLexer().getLoc(); 4920 if (parseAbsoluteExpression(Size)) 4921 return true; 4922 4923 int64_t Pow2Alignment = 0; 4924 SMLoc Pow2AlignmentLoc; 4925 if (getLexer().is(AsmToken::Comma)) { 4926 Lex(); 4927 Pow2AlignmentLoc = getLexer().getLoc(); 4928 if (parseAbsoluteExpression(Pow2Alignment)) 4929 return true; 4930 4931 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4932 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4933 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4934 4935 // If this target takes alignments in bytes (not log) validate and convert. 4936 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4937 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4938 if (!isPowerOf2_64(Pow2Alignment)) 4939 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4940 Pow2Alignment = Log2_64(Pow2Alignment); 4941 } 4942 } 4943 4944 if (parseToken(AsmToken::EndOfStatement, 4945 "unexpected token in '.comm' or '.lcomm' directive")) 4946 return true; 4947 4948 // NOTE: a size of zero for a .comm should create a undefined symbol 4949 // but a size of .lcomm creates a bss symbol of size zero. 4950 if (Size < 0) 4951 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4952 "be less than zero"); 4953 4954 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4955 // may internally end up wanting an alignment in bytes. 4956 // FIXME: Diagnose overflow. 4957 if (Pow2Alignment < 0) 4958 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4959 "alignment, can't be less than zero"); 4960 4961 Sym->redefineIfPossible(); 4962 if (!Sym->isUndefined()) 4963 return Error(IDLoc, "invalid symbol redefinition"); 4964 4965 // Create the Symbol as a common or local common with Size and Pow2Alignment 4966 if (IsLocal) { 4967 getStreamer().emitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4968 return false; 4969 } 4970 4971 getStreamer().emitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4972 return false; 4973 } 4974 4975 /// parseDirectiveAbort 4976 /// ::= .abort [... message ...] 4977 bool AsmParser::parseDirectiveAbort() { 4978 // FIXME: Use loc from directive. 4979 SMLoc Loc = getLexer().getLoc(); 4980 4981 StringRef Str = parseStringToEndOfStatement(); 4982 if (parseToken(AsmToken::EndOfStatement, 4983 "unexpected token in '.abort' directive")) 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 (parseToken(AsmToken::EndOfStatement, 5045 "unexpected token in '.incbin' directive")) 5046 return true; 5047 5048 if (check(Skip < 0, SkipLoc, "skip is negative")) 5049 return true; 5050 5051 // Attempt to process the included file. 5052 if (processIncbinFile(Filename, Skip, Count, CountLoc)) 5053 return Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 5054 return false; 5055 } 5056 5057 /// parseDirectiveIf 5058 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 5059 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 5060 TheCondStack.push_back(TheCondState); 5061 TheCondState.TheCond = AsmCond::IfCond; 5062 if (TheCondState.Ignore) { 5063 eatToEndOfStatement(); 5064 } else { 5065 int64_t ExprValue; 5066 if (parseAbsoluteExpression(ExprValue) || 5067 parseToken(AsmToken::EndOfStatement, 5068 "unexpected token in '.if' directive")) 5069 return true; 5070 5071 switch (DirKind) { 5072 default: 5073 llvm_unreachable("unsupported directive"); 5074 case DK_IF: 5075 case DK_IFNE: 5076 break; 5077 case DK_IFEQ: 5078 ExprValue = ExprValue == 0; 5079 break; 5080 case DK_IFGE: 5081 ExprValue = ExprValue >= 0; 5082 break; 5083 case DK_IFGT: 5084 ExprValue = ExprValue > 0; 5085 break; 5086 case DK_IFLE: 5087 ExprValue = ExprValue <= 0; 5088 break; 5089 case DK_IFLT: 5090 ExprValue = ExprValue < 0; 5091 break; 5092 } 5093 5094 TheCondState.CondMet = ExprValue; 5095 TheCondState.Ignore = !TheCondState.CondMet; 5096 } 5097 5098 return false; 5099 } 5100 5101 /// parseDirectiveIfb 5102 /// ::= .ifb string 5103 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 5104 TheCondStack.push_back(TheCondState); 5105 TheCondState.TheCond = AsmCond::IfCond; 5106 5107 if (TheCondState.Ignore) { 5108 eatToEndOfStatement(); 5109 } else { 5110 StringRef Str = parseStringToEndOfStatement(); 5111 5112 if (parseToken(AsmToken::EndOfStatement, 5113 "unexpected token in '.ifb' directive")) 5114 return true; 5115 5116 TheCondState.CondMet = ExpectBlank == Str.empty(); 5117 TheCondState.Ignore = !TheCondState.CondMet; 5118 } 5119 5120 return false; 5121 } 5122 5123 /// parseDirectiveIfc 5124 /// ::= .ifc string1, string2 5125 /// ::= .ifnc string1, string2 5126 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 5127 TheCondStack.push_back(TheCondState); 5128 TheCondState.TheCond = AsmCond::IfCond; 5129 5130 if (TheCondState.Ignore) { 5131 eatToEndOfStatement(); 5132 } else { 5133 StringRef Str1 = parseStringToComma(); 5134 5135 if (parseToken(AsmToken::Comma, "unexpected token in '.ifc' directive")) 5136 return true; 5137 5138 StringRef Str2 = parseStringToEndOfStatement(); 5139 5140 if (parseToken(AsmToken::EndOfStatement, 5141 "unexpected token in '.ifc' directive")) 5142 return true; 5143 5144 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 5145 TheCondState.Ignore = !TheCondState.CondMet; 5146 } 5147 5148 return false; 5149 } 5150 5151 /// parseDirectiveIfeqs 5152 /// ::= .ifeqs string1, string2 5153 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 5154 if (Lexer.isNot(AsmToken::String)) { 5155 if (ExpectEqual) 5156 return TokError("expected string parameter for '.ifeqs' directive"); 5157 return TokError("expected string parameter for '.ifnes' directive"); 5158 } 5159 5160 StringRef String1 = getTok().getStringContents(); 5161 Lex(); 5162 5163 if (Lexer.isNot(AsmToken::Comma)) { 5164 if (ExpectEqual) 5165 return TokError( 5166 "expected comma after first string for '.ifeqs' directive"); 5167 return TokError("expected comma after first string for '.ifnes' directive"); 5168 } 5169 5170 Lex(); 5171 5172 if (Lexer.isNot(AsmToken::String)) { 5173 if (ExpectEqual) 5174 return TokError("expected string parameter for '.ifeqs' directive"); 5175 return TokError("expected string parameter for '.ifnes' directive"); 5176 } 5177 5178 StringRef String2 = getTok().getStringContents(); 5179 Lex(); 5180 5181 TheCondStack.push_back(TheCondState); 5182 TheCondState.TheCond = AsmCond::IfCond; 5183 TheCondState.CondMet = ExpectEqual == (String1 == String2); 5184 TheCondState.Ignore = !TheCondState.CondMet; 5185 5186 return false; 5187 } 5188 5189 /// parseDirectiveIfdef 5190 /// ::= .ifdef symbol 5191 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 5192 StringRef Name; 5193 TheCondStack.push_back(TheCondState); 5194 TheCondState.TheCond = AsmCond::IfCond; 5195 5196 if (TheCondState.Ignore) { 5197 eatToEndOfStatement(); 5198 } else { 5199 if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") || 5200 parseToken(AsmToken::EndOfStatement, "unexpected token in '.ifdef'")) 5201 return true; 5202 5203 MCSymbol *Sym = getContext().lookupSymbol(Name); 5204 5205 if (expect_defined) 5206 TheCondState.CondMet = (Sym && !Sym->isUndefined(false)); 5207 else 5208 TheCondState.CondMet = (!Sym || Sym->isUndefined(false)); 5209 TheCondState.Ignore = !TheCondState.CondMet; 5210 } 5211 5212 return false; 5213 } 5214 5215 /// parseDirectiveElseIf 5216 /// ::= .elseif expression 5217 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 5218 if (TheCondState.TheCond != AsmCond::IfCond && 5219 TheCondState.TheCond != AsmCond::ElseIfCond) 5220 return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an" 5221 " .if or an .elseif"); 5222 TheCondState.TheCond = AsmCond::ElseIfCond; 5223 5224 bool LastIgnoreState = false; 5225 if (!TheCondStack.empty()) 5226 LastIgnoreState = TheCondStack.back().Ignore; 5227 if (LastIgnoreState || TheCondState.CondMet) { 5228 TheCondState.Ignore = true; 5229 eatToEndOfStatement(); 5230 } else { 5231 int64_t ExprValue; 5232 if (parseAbsoluteExpression(ExprValue)) 5233 return true; 5234 5235 if (parseToken(AsmToken::EndOfStatement, 5236 "unexpected token in '.elseif' directive")) 5237 return true; 5238 5239 TheCondState.CondMet = ExprValue; 5240 TheCondState.Ignore = !TheCondState.CondMet; 5241 } 5242 5243 return false; 5244 } 5245 5246 /// parseDirectiveElse 5247 /// ::= .else 5248 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 5249 if (parseToken(AsmToken::EndOfStatement, 5250 "unexpected token in '.else' directive")) 5251 return true; 5252 5253 if (TheCondState.TheCond != AsmCond::IfCond && 5254 TheCondState.TheCond != AsmCond::ElseIfCond) 5255 return Error(DirectiveLoc, "Encountered a .else that doesn't follow " 5256 " an .if or an .elseif"); 5257 TheCondState.TheCond = AsmCond::ElseCond; 5258 bool LastIgnoreState = false; 5259 if (!TheCondStack.empty()) 5260 LastIgnoreState = TheCondStack.back().Ignore; 5261 if (LastIgnoreState || TheCondState.CondMet) 5262 TheCondState.Ignore = true; 5263 else 5264 TheCondState.Ignore = false; 5265 5266 return false; 5267 } 5268 5269 /// parseDirectiveEnd 5270 /// ::= .end 5271 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 5272 if (parseToken(AsmToken::EndOfStatement, 5273 "unexpected token in '.end' directive")) 5274 return true; 5275 5276 while (Lexer.isNot(AsmToken::Eof)) 5277 Lexer.Lex(); 5278 5279 return false; 5280 } 5281 5282 /// parseDirectiveError 5283 /// ::= .err 5284 /// ::= .error [string] 5285 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 5286 if (!TheCondStack.empty()) { 5287 if (TheCondStack.back().Ignore) { 5288 eatToEndOfStatement(); 5289 return false; 5290 } 5291 } 5292 5293 if (!WithMessage) 5294 return Error(L, ".err encountered"); 5295 5296 StringRef Message = ".error directive invoked in source file"; 5297 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5298 if (Lexer.isNot(AsmToken::String)) 5299 return TokError(".error argument must be a string"); 5300 5301 Message = getTok().getStringContents(); 5302 Lex(); 5303 } 5304 5305 return Error(L, Message); 5306 } 5307 5308 /// parseDirectiveWarning 5309 /// ::= .warning [string] 5310 bool AsmParser::parseDirectiveWarning(SMLoc L) { 5311 if (!TheCondStack.empty()) { 5312 if (TheCondStack.back().Ignore) { 5313 eatToEndOfStatement(); 5314 return false; 5315 } 5316 } 5317 5318 StringRef Message = ".warning directive invoked in source file"; 5319 5320 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 5321 if (Lexer.isNot(AsmToken::String)) 5322 return TokError(".warning argument must be a string"); 5323 5324 Message = getTok().getStringContents(); 5325 Lex(); 5326 if (parseToken(AsmToken::EndOfStatement, 5327 "expected end of statement in '.warning' directive")) 5328 return true; 5329 } 5330 5331 return Warning(L, Message); 5332 } 5333 5334 /// parseDirectiveEndIf 5335 /// ::= .endif 5336 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 5337 if (parseToken(AsmToken::EndOfStatement, 5338 "unexpected token in '.endif' directive")) 5339 return true; 5340 5341 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 5342 return Error(DirectiveLoc, "Encountered a .endif that doesn't follow " 5343 "an .if or .else"); 5344 if (!TheCondStack.empty()) { 5345 TheCondState = TheCondStack.back(); 5346 TheCondStack.pop_back(); 5347 } 5348 5349 return false; 5350 } 5351 5352 void AsmParser::initializeDirectiveKindMap() { 5353 /* Lookup will be done with the directive 5354 * converted to lower case, so all these 5355 * keys should be lower case. 5356 * (target specific directives are handled 5357 * elsewhere) 5358 */ 5359 DirectiveKindMap[".set"] = DK_SET; 5360 DirectiveKindMap[".equ"] = DK_EQU; 5361 DirectiveKindMap[".equiv"] = DK_EQUIV; 5362 DirectiveKindMap[".ascii"] = DK_ASCII; 5363 DirectiveKindMap[".asciz"] = DK_ASCIZ; 5364 DirectiveKindMap[".string"] = DK_STRING; 5365 DirectiveKindMap[".byte"] = DK_BYTE; 5366 DirectiveKindMap[".short"] = DK_SHORT; 5367 DirectiveKindMap[".value"] = DK_VALUE; 5368 DirectiveKindMap[".2byte"] = DK_2BYTE; 5369 DirectiveKindMap[".long"] = DK_LONG; 5370 DirectiveKindMap[".int"] = DK_INT; 5371 DirectiveKindMap[".4byte"] = DK_4BYTE; 5372 DirectiveKindMap[".quad"] = DK_QUAD; 5373 DirectiveKindMap[".8byte"] = DK_8BYTE; 5374 DirectiveKindMap[".octa"] = DK_OCTA; 5375 DirectiveKindMap[".single"] = DK_SINGLE; 5376 DirectiveKindMap[".float"] = DK_FLOAT; 5377 DirectiveKindMap[".double"] = DK_DOUBLE; 5378 DirectiveKindMap[".align"] = DK_ALIGN; 5379 DirectiveKindMap[".align32"] = DK_ALIGN32; 5380 DirectiveKindMap[".balign"] = DK_BALIGN; 5381 DirectiveKindMap[".balignw"] = DK_BALIGNW; 5382 DirectiveKindMap[".balignl"] = DK_BALIGNL; 5383 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 5384 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 5385 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 5386 DirectiveKindMap[".org"] = DK_ORG; 5387 DirectiveKindMap[".fill"] = DK_FILL; 5388 DirectiveKindMap[".zero"] = DK_ZERO; 5389 DirectiveKindMap[".extern"] = DK_EXTERN; 5390 DirectiveKindMap[".globl"] = DK_GLOBL; 5391 DirectiveKindMap[".global"] = DK_GLOBAL; 5392 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 5393 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 5394 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 5395 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 5396 DirectiveKindMap[".reference"] = DK_REFERENCE; 5397 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 5398 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 5399 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 5400 DirectiveKindMap[".cold"] = DK_COLD; 5401 DirectiveKindMap[".comm"] = DK_COMM; 5402 DirectiveKindMap[".common"] = DK_COMMON; 5403 DirectiveKindMap[".lcomm"] = DK_LCOMM; 5404 DirectiveKindMap[".abort"] = DK_ABORT; 5405 DirectiveKindMap[".include"] = DK_INCLUDE; 5406 DirectiveKindMap[".incbin"] = DK_INCBIN; 5407 DirectiveKindMap[".code16"] = DK_CODE16; 5408 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 5409 DirectiveKindMap[".rept"] = DK_REPT; 5410 DirectiveKindMap[".rep"] = DK_REPT; 5411 DirectiveKindMap[".irp"] = DK_IRP; 5412 DirectiveKindMap[".irpc"] = DK_IRPC; 5413 DirectiveKindMap[".endr"] = DK_ENDR; 5414 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 5415 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 5416 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 5417 DirectiveKindMap[".if"] = DK_IF; 5418 DirectiveKindMap[".ifeq"] = DK_IFEQ; 5419 DirectiveKindMap[".ifge"] = DK_IFGE; 5420 DirectiveKindMap[".ifgt"] = DK_IFGT; 5421 DirectiveKindMap[".ifle"] = DK_IFLE; 5422 DirectiveKindMap[".iflt"] = DK_IFLT; 5423 DirectiveKindMap[".ifne"] = DK_IFNE; 5424 DirectiveKindMap[".ifb"] = DK_IFB; 5425 DirectiveKindMap[".ifnb"] = DK_IFNB; 5426 DirectiveKindMap[".ifc"] = DK_IFC; 5427 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 5428 DirectiveKindMap[".ifnc"] = DK_IFNC; 5429 DirectiveKindMap[".ifnes"] = DK_IFNES; 5430 DirectiveKindMap[".ifdef"] = DK_IFDEF; 5431 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 5432 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 5433 DirectiveKindMap[".elseif"] = DK_ELSEIF; 5434 DirectiveKindMap[".else"] = DK_ELSE; 5435 DirectiveKindMap[".end"] = DK_END; 5436 DirectiveKindMap[".endif"] = DK_ENDIF; 5437 DirectiveKindMap[".skip"] = DK_SKIP; 5438 DirectiveKindMap[".space"] = DK_SPACE; 5439 DirectiveKindMap[".file"] = DK_FILE; 5440 DirectiveKindMap[".line"] = DK_LINE; 5441 DirectiveKindMap[".loc"] = DK_LOC; 5442 DirectiveKindMap[".stabs"] = DK_STABS; 5443 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 5444 DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID; 5445 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 5446 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 5447 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 5448 DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID; 5449 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 5450 DirectiveKindMap[".cv_string"] = DK_CV_STRING; 5451 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 5452 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 5453 DirectiveKindMap[".cv_filechecksumoffset"] = DK_CV_FILECHECKSUM_OFFSET; 5454 DirectiveKindMap[".cv_fpo_data"] = DK_CV_FPO_DATA; 5455 DirectiveKindMap[".sleb128"] = DK_SLEB128; 5456 DirectiveKindMap[".uleb128"] = DK_ULEB128; 5457 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 5458 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 5459 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 5460 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 5461 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 5462 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 5463 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 5464 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 5465 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 5466 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 5467 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 5468 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 5469 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 5470 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 5471 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 5472 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 5473 DirectiveKindMap[".cfi_return_column"] = DK_CFI_RETURN_COLUMN; 5474 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 5475 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 5476 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 5477 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 5478 DirectiveKindMap[".cfi_b_key_frame"] = DK_CFI_B_KEY_FRAME; 5479 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 5480 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 5481 DirectiveKindMap[".macro"] = DK_MACRO; 5482 DirectiveKindMap[".exitm"] = DK_EXITM; 5483 DirectiveKindMap[".endm"] = DK_ENDM; 5484 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 5485 DirectiveKindMap[".purgem"] = DK_PURGEM; 5486 DirectiveKindMap[".err"] = DK_ERR; 5487 DirectiveKindMap[".error"] = DK_ERROR; 5488 DirectiveKindMap[".warning"] = DK_WARNING; 5489 DirectiveKindMap[".altmacro"] = DK_ALTMACRO; 5490 DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO; 5491 DirectiveKindMap[".reloc"] = DK_RELOC; 5492 DirectiveKindMap[".dc"] = DK_DC; 5493 DirectiveKindMap[".dc.a"] = DK_DC_A; 5494 DirectiveKindMap[".dc.b"] = DK_DC_B; 5495 DirectiveKindMap[".dc.d"] = DK_DC_D; 5496 DirectiveKindMap[".dc.l"] = DK_DC_L; 5497 DirectiveKindMap[".dc.s"] = DK_DC_S; 5498 DirectiveKindMap[".dc.w"] = DK_DC_W; 5499 DirectiveKindMap[".dc.x"] = DK_DC_X; 5500 DirectiveKindMap[".dcb"] = DK_DCB; 5501 DirectiveKindMap[".dcb.b"] = DK_DCB_B; 5502 DirectiveKindMap[".dcb.d"] = DK_DCB_D; 5503 DirectiveKindMap[".dcb.l"] = DK_DCB_L; 5504 DirectiveKindMap[".dcb.s"] = DK_DCB_S; 5505 DirectiveKindMap[".dcb.w"] = DK_DCB_W; 5506 DirectiveKindMap[".dcb.x"] = DK_DCB_X; 5507 DirectiveKindMap[".ds"] = DK_DS; 5508 DirectiveKindMap[".ds.b"] = DK_DS_B; 5509 DirectiveKindMap[".ds.d"] = DK_DS_D; 5510 DirectiveKindMap[".ds.l"] = DK_DS_L; 5511 DirectiveKindMap[".ds.p"] = DK_DS_P; 5512 DirectiveKindMap[".ds.s"] = DK_DS_S; 5513 DirectiveKindMap[".ds.w"] = DK_DS_W; 5514 DirectiveKindMap[".ds.x"] = DK_DS_X; 5515 DirectiveKindMap[".print"] = DK_PRINT; 5516 DirectiveKindMap[".addrsig"] = DK_ADDRSIG; 5517 DirectiveKindMap[".addrsig_sym"] = DK_ADDRSIG_SYM; 5518 } 5519 5520 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 5521 AsmToken EndToken, StartToken = getTok(); 5522 5523 unsigned NestLevel = 0; 5524 while (true) { 5525 // Check whether we have reached the end of the file. 5526 if (getLexer().is(AsmToken::Eof)) { 5527 printError(DirectiveLoc, "no matching '.endr' in definition"); 5528 return nullptr; 5529 } 5530 5531 if (Lexer.is(AsmToken::Identifier) && 5532 (getTok().getIdentifier() == ".rep" || 5533 getTok().getIdentifier() == ".rept" || 5534 getTok().getIdentifier() == ".irp" || 5535 getTok().getIdentifier() == ".irpc")) { 5536 ++NestLevel; 5537 } 5538 5539 // Otherwise, check whether we have reached the .endr. 5540 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 5541 if (NestLevel == 0) { 5542 EndToken = getTok(); 5543 Lex(); 5544 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5545 printError(getTok().getLoc(), 5546 "unexpected token in '.endr' directive"); 5547 return nullptr; 5548 } 5549 break; 5550 } 5551 --NestLevel; 5552 } 5553 5554 // Otherwise, scan till the end of the statement. 5555 eatToEndOfStatement(); 5556 } 5557 5558 const char *BodyStart = StartToken.getLoc().getPointer(); 5559 const char *BodyEnd = EndToken.getLoc().getPointer(); 5560 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 5561 5562 // We Are Anonymous. 5563 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 5564 return &MacroLikeBodies.back(); 5565 } 5566 5567 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 5568 raw_svector_ostream &OS) { 5569 OS << ".endr\n"; 5570 5571 std::unique_ptr<MemoryBuffer> Instantiation = 5572 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 5573 5574 // Create the macro instantiation object and add to the current macro 5575 // instantiation stack. 5576 MacroInstantiation *MI = new MacroInstantiation{ 5577 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 5578 ActiveMacros.push_back(MI); 5579 5580 // Jump to the macro instantiation and prime the lexer. 5581 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 5582 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 5583 Lex(); 5584 } 5585 5586 /// parseDirectiveRept 5587 /// ::= .rep | .rept count 5588 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 5589 const MCExpr *CountExpr; 5590 SMLoc CountLoc = getTok().getLoc(); 5591 if (parseExpression(CountExpr)) 5592 return true; 5593 5594 int64_t Count; 5595 if (!CountExpr->evaluateAsAbsolute(Count, getStreamer().getAssemblerPtr())) { 5596 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 5597 } 5598 5599 if (check(Count < 0, CountLoc, "Count is negative") || 5600 parseToken(AsmToken::EndOfStatement, 5601 "unexpected token in '" + Dir + "' directive")) 5602 return true; 5603 5604 // Lex the rept definition. 5605 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5606 if (!M) 5607 return true; 5608 5609 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5610 // to hold the macro body with substitutions. 5611 SmallString<256> Buf; 5612 raw_svector_ostream OS(Buf); 5613 while (Count--) { 5614 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 5615 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 5616 return true; 5617 } 5618 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5619 5620 return false; 5621 } 5622 5623 /// parseDirectiveIrp 5624 /// ::= .irp symbol,values 5625 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 5626 MCAsmMacroParameter Parameter; 5627 MCAsmMacroArguments A; 5628 if (check(parseIdentifier(Parameter.Name), 5629 "expected identifier in '.irp' directive") || 5630 parseToken(AsmToken::Comma, "expected comma in '.irp' directive") || 5631 parseMacroArguments(nullptr, A) || 5632 parseToken(AsmToken::EndOfStatement, "expected End of Statement")) 5633 return true; 5634 5635 // Lex the irp definition. 5636 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5637 if (!M) 5638 return true; 5639 5640 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5641 // to hold the macro body with substitutions. 5642 SmallString<256> Buf; 5643 raw_svector_ostream OS(Buf); 5644 5645 for (const MCAsmMacroArgument &Arg : A) { 5646 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 5647 // This is undocumented, but GAS seems to support it. 5648 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5649 return true; 5650 } 5651 5652 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5653 5654 return false; 5655 } 5656 5657 /// parseDirectiveIrpc 5658 /// ::= .irpc symbol,values 5659 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 5660 MCAsmMacroParameter Parameter; 5661 MCAsmMacroArguments A; 5662 5663 if (check(parseIdentifier(Parameter.Name), 5664 "expected identifier in '.irpc' directive") || 5665 parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") || 5666 parseMacroArguments(nullptr, A)) 5667 return true; 5668 5669 if (A.size() != 1 || A.front().size() != 1) 5670 return TokError("unexpected token in '.irpc' directive"); 5671 5672 // Eat the end of statement. 5673 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5674 return true; 5675 5676 // Lex the irpc definition. 5677 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5678 if (!M) 5679 return true; 5680 5681 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5682 // to hold the macro body with substitutions. 5683 SmallString<256> Buf; 5684 raw_svector_ostream OS(Buf); 5685 5686 StringRef Values = A.front().front().getString(); 5687 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 5688 MCAsmMacroArgument Arg; 5689 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 5690 5691 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 5692 // This is undocumented, but GAS seems to support it. 5693 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5694 return true; 5695 } 5696 5697 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5698 5699 return false; 5700 } 5701 5702 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 5703 if (ActiveMacros.empty()) 5704 return TokError("unmatched '.endr' directive"); 5705 5706 // The only .repl that should get here are the ones created by 5707 // instantiateMacroLikeBody. 5708 assert(getLexer().is(AsmToken::EndOfStatement)); 5709 5710 handleMacroExit(); 5711 return false; 5712 } 5713 5714 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 5715 size_t Len) { 5716 const MCExpr *Value; 5717 SMLoc ExprLoc = getLexer().getLoc(); 5718 if (parseExpression(Value)) 5719 return true; 5720 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5721 if (!MCE) 5722 return Error(ExprLoc, "unexpected expression in _emit"); 5723 uint64_t IntValue = MCE->getValue(); 5724 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 5725 return Error(ExprLoc, "literal value out of range for directive"); 5726 5727 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 5728 return false; 5729 } 5730 5731 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 5732 const MCExpr *Value; 5733 SMLoc ExprLoc = getLexer().getLoc(); 5734 if (parseExpression(Value)) 5735 return true; 5736 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5737 if (!MCE) 5738 return Error(ExprLoc, "unexpected expression in align"); 5739 uint64_t IntValue = MCE->getValue(); 5740 if (!isPowerOf2_64(IntValue)) 5741 return Error(ExprLoc, "literal value not a power of two greater then zero"); 5742 5743 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 5744 return false; 5745 } 5746 5747 bool AsmParser::parseDirectivePrint(SMLoc DirectiveLoc) { 5748 const AsmToken StrTok = getTok(); 5749 Lex(); 5750 if (StrTok.isNot(AsmToken::String) || StrTok.getString().front() != '"') 5751 return Error(DirectiveLoc, "expected double quoted string after .print"); 5752 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5753 return true; 5754 llvm::outs() << StrTok.getStringContents() << '\n'; 5755 return false; 5756 } 5757 5758 bool AsmParser::parseDirectiveAddrsig() { 5759 getStreamer().emitAddrsig(); 5760 return false; 5761 } 5762 5763 bool AsmParser::parseDirectiveAddrsigSym() { 5764 StringRef Name; 5765 if (check(parseIdentifier(Name), 5766 "expected identifier in '.addrsig_sym' directive")) 5767 return true; 5768 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 5769 getStreamer().emitAddrsigSym(Sym); 5770 return false; 5771 } 5772 5773 // We are comparing pointers, but the pointers are relative to a single string. 5774 // Thus, this should always be deterministic. 5775 static int rewritesSort(const AsmRewrite *AsmRewriteA, 5776 const AsmRewrite *AsmRewriteB) { 5777 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 5778 return -1; 5779 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 5780 return 1; 5781 5782 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 5783 // rewrite to the same location. Make sure the SizeDirective rewrite is 5784 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 5785 // ensures the sort algorithm is stable. 5786 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 5787 AsmRewritePrecedence[AsmRewriteB->Kind]) 5788 return -1; 5789 5790 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 5791 AsmRewritePrecedence[AsmRewriteB->Kind]) 5792 return 1; 5793 llvm_unreachable("Unstable rewrite sort."); 5794 } 5795 5796 bool AsmParser::parseMSInlineAsm( 5797 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 5798 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls, 5799 SmallVectorImpl<std::string> &Constraints, 5800 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 5801 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 5802 SmallVector<void *, 4> InputDecls; 5803 SmallVector<void *, 4> OutputDecls; 5804 SmallVector<bool, 4> InputDeclsAddressOf; 5805 SmallVector<bool, 4> OutputDeclsAddressOf; 5806 SmallVector<std::string, 4> InputConstraints; 5807 SmallVector<std::string, 4> OutputConstraints; 5808 SmallVector<unsigned, 4> ClobberRegs; 5809 5810 SmallVector<AsmRewrite, 4> AsmStrRewrites; 5811 5812 // Prime the lexer. 5813 Lex(); 5814 5815 // While we have input, parse each statement. 5816 unsigned InputIdx = 0; 5817 unsigned OutputIdx = 0; 5818 while (getLexer().isNot(AsmToken::Eof)) { 5819 // Parse curly braces marking block start/end 5820 if (parseCurlyBlockScope(AsmStrRewrites)) 5821 continue; 5822 5823 ParseStatementInfo Info(&AsmStrRewrites); 5824 bool StatementErr = parseStatement(Info, &SI); 5825 5826 if (StatementErr || Info.ParseError) { 5827 // Emit pending errors if any exist. 5828 printPendingErrors(); 5829 return true; 5830 } 5831 5832 // No pending error should exist here. 5833 assert(!hasPendingError() && "unexpected error from parseStatement"); 5834 5835 if (Info.Opcode == ~0U) 5836 continue; 5837 5838 const MCInstrDesc &Desc = MII->get(Info.Opcode); 5839 5840 // Build the list of clobbers, outputs and inputs. 5841 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 5842 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 5843 5844 // Register operand. 5845 if (Operand.isReg() && !Operand.needAddressOf() && 5846 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5847 unsigned NumDefs = Desc.getNumDefs(); 5848 // Clobber. 5849 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5850 ClobberRegs.push_back(Operand.getReg()); 5851 continue; 5852 } 5853 5854 // Expr/Input or Output. 5855 StringRef SymName = Operand.getSymName(); 5856 if (SymName.empty()) 5857 continue; 5858 5859 void *OpDecl = Operand.getOpDecl(); 5860 if (!OpDecl) 5861 continue; 5862 5863 StringRef Constraint = Operand.getConstraint(); 5864 if (Operand.isImm()) { 5865 // Offset as immediate 5866 if (Operand.isOffsetOfLocal()) 5867 Constraint = "r"; 5868 else 5869 Constraint = "i"; 5870 } 5871 5872 bool isOutput = (i == 1) && Desc.mayStore(); 5873 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5874 if (isOutput) { 5875 ++InputIdx; 5876 OutputDecls.push_back(OpDecl); 5877 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5878 OutputConstraints.push_back(("=" + Constraint).str()); 5879 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5880 } else { 5881 InputDecls.push_back(OpDecl); 5882 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5883 InputConstraints.push_back(Constraint.str()); 5884 if (Desc.OpInfo[i - 1].isBranchTarget()) 5885 AsmStrRewrites.emplace_back(AOK_CallInput, Start, SymName.size()); 5886 else 5887 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5888 } 5889 } 5890 5891 // Consider implicit defs to be clobbers. Think of cpuid and push. 5892 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5893 Desc.getNumImplicitDefs()); 5894 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 5895 } 5896 5897 // Set the number of Outputs and Inputs. 5898 NumOutputs = OutputDecls.size(); 5899 NumInputs = InputDecls.size(); 5900 5901 // Set the unique clobbers. 5902 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5903 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5904 ClobberRegs.end()); 5905 Clobbers.assign(ClobberRegs.size(), std::string()); 5906 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5907 raw_string_ostream OS(Clobbers[I]); 5908 IP->printRegName(OS, ClobberRegs[I]); 5909 } 5910 5911 // Merge the various outputs and inputs. Output are expected first. 5912 if (NumOutputs || NumInputs) { 5913 unsigned NumExprs = NumOutputs + NumInputs; 5914 OpDecls.resize(NumExprs); 5915 Constraints.resize(NumExprs); 5916 for (unsigned i = 0; i < NumOutputs; ++i) { 5917 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5918 Constraints[i] = OutputConstraints[i]; 5919 } 5920 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5921 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5922 Constraints[j] = InputConstraints[i]; 5923 } 5924 } 5925 5926 // Build the IR assembly string. 5927 std::string AsmStringIR; 5928 raw_string_ostream OS(AsmStringIR); 5929 StringRef ASMString = 5930 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5931 const char *AsmStart = ASMString.begin(); 5932 const char *AsmEnd = ASMString.end(); 5933 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 5934 for (auto it = AsmStrRewrites.begin(); it != AsmStrRewrites.end(); ++it) { 5935 const AsmRewrite &AR = *it; 5936 // Check if this has already been covered by another rewrite... 5937 if (AR.Done) 5938 continue; 5939 AsmRewriteKind Kind = AR.Kind; 5940 5941 const char *Loc = AR.Loc.getPointer(); 5942 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 5943 5944 // Emit everything up to the immediate/expression. 5945 if (unsigned Len = Loc - AsmStart) 5946 OS << StringRef(AsmStart, Len); 5947 5948 // Skip the original expression. 5949 if (Kind == AOK_Skip) { 5950 AsmStart = Loc + AR.Len; 5951 continue; 5952 } 5953 5954 unsigned AdditionalSkip = 0; 5955 // Rewrite expressions in $N notation. 5956 switch (Kind) { 5957 default: 5958 break; 5959 case AOK_IntelExpr: 5960 assert(AR.IntelExp.isValid() && "cannot write invalid intel expression"); 5961 if (AR.IntelExp.NeedBracs) 5962 OS << "["; 5963 if (AR.IntelExp.hasBaseReg()) 5964 OS << AR.IntelExp.BaseReg; 5965 if (AR.IntelExp.hasIndexReg()) 5966 OS << (AR.IntelExp.hasBaseReg() ? " + " : "") 5967 << AR.IntelExp.IndexReg; 5968 if (AR.IntelExp.Scale > 1) 5969 OS << " * $$" << AR.IntelExp.Scale; 5970 if (AR.IntelExp.hasOffset()) { 5971 if (AR.IntelExp.hasRegs()) 5972 OS << " + "; 5973 // Fuse this rewrite with a rewrite of the offset name, if present. 5974 StringRef OffsetName = AR.IntelExp.OffsetName; 5975 SMLoc OffsetLoc = SMLoc::getFromPointer(AR.IntelExp.OffsetName.data()); 5976 size_t OffsetLen = OffsetName.size(); 5977 auto rewrite_it = std::find_if( 5978 it, AsmStrRewrites.end(), [&](const AsmRewrite &FusingAR) { 5979 return FusingAR.Loc == OffsetLoc && FusingAR.Len == OffsetLen && 5980 (FusingAR.Kind == AOK_Input || 5981 FusingAR.Kind == AOK_CallInput); 5982 }); 5983 if (rewrite_it == AsmStrRewrites.end()) { 5984 OS << "offset " << OffsetName; 5985 } else if (rewrite_it->Kind == AOK_CallInput) { 5986 OS << "${" << InputIdx++ << ":P}"; 5987 rewrite_it->Done = true; 5988 } else { 5989 OS << '$' << InputIdx++; 5990 rewrite_it->Done = true; 5991 } 5992 } 5993 if (AR.IntelExp.Imm || AR.IntelExp.emitImm()) 5994 OS << (AR.IntelExp.emitImm() ? "$$" : " + $$") << AR.IntelExp.Imm; 5995 if (AR.IntelExp.NeedBracs) 5996 OS << "]"; 5997 break; 5998 case AOK_Label: 5999 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 6000 break; 6001 case AOK_Input: 6002 OS << '$' << InputIdx++; 6003 break; 6004 case AOK_CallInput: 6005 OS << "${" << InputIdx++ << ":P}"; 6006 break; 6007 case AOK_Output: 6008 OS << '$' << OutputIdx++; 6009 break; 6010 case AOK_SizeDirective: 6011 switch (AR.Val) { 6012 default: break; 6013 case 8: OS << "byte ptr "; break; 6014 case 16: OS << "word ptr "; break; 6015 case 32: OS << "dword ptr "; break; 6016 case 64: OS << "qword ptr "; break; 6017 case 80: OS << "xword ptr "; break; 6018 case 128: OS << "xmmword ptr "; break; 6019 case 256: OS << "ymmword ptr "; break; 6020 } 6021 break; 6022 case AOK_Emit: 6023 OS << ".byte"; 6024 break; 6025 case AOK_Align: { 6026 // MS alignment directives are measured in bytes. If the native assembler 6027 // measures alignment in bytes, we can pass it straight through. 6028 OS << ".align"; 6029 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 6030 break; 6031 6032 // Alignment is in log2 form, so print that instead and skip the original 6033 // immediate. 6034 unsigned Val = AR.Val; 6035 OS << ' ' << Val; 6036 assert(Val < 10 && "Expected alignment less then 2^10."); 6037 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 6038 break; 6039 } 6040 case AOK_EVEN: 6041 OS << ".even"; 6042 break; 6043 case AOK_EndOfStatement: 6044 OS << "\n\t"; 6045 break; 6046 } 6047 6048 // Skip the original expression. 6049 AsmStart = Loc + AR.Len + AdditionalSkip; 6050 } 6051 6052 // Emit the remainder of the asm string. 6053 if (AsmStart != AsmEnd) 6054 OS << StringRef(AsmStart, AsmEnd - AsmStart); 6055 6056 AsmString = OS.str(); 6057 return false; 6058 } 6059 6060 namespace llvm { 6061 namespace MCParserUtils { 6062 6063 /// Returns whether the given symbol is used anywhere in the given expression, 6064 /// or subexpressions. 6065 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 6066 switch (Value->getKind()) { 6067 case MCExpr::Binary: { 6068 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 6069 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 6070 isSymbolUsedInExpression(Sym, BE->getRHS()); 6071 } 6072 case MCExpr::Target: 6073 case MCExpr::Constant: 6074 return false; 6075 case MCExpr::SymbolRef: { 6076 const MCSymbol &S = 6077 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 6078 if (S.isVariable()) 6079 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 6080 return &S == Sym; 6081 } 6082 case MCExpr::Unary: 6083 return isSymbolUsedInExpression( 6084 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 6085 } 6086 6087 llvm_unreachable("Unknown expr kind!"); 6088 } 6089 6090 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 6091 MCAsmParser &Parser, MCSymbol *&Sym, 6092 const MCExpr *&Value) { 6093 6094 // FIXME: Use better location, we should use proper tokens. 6095 SMLoc EqualLoc = Parser.getTok().getLoc(); 6096 if (Parser.parseExpression(Value)) 6097 return Parser.TokError("missing expression"); 6098 6099 // Note: we don't count b as used in "a = b". This is to allow 6100 // a = b 6101 // b = c 6102 6103 if (Parser.parseToken(AsmToken::EndOfStatement)) 6104 return true; 6105 6106 // Validate that the LHS is allowed to be a variable (either it has not been 6107 // used as a symbol, or it is an absolute symbol). 6108 Sym = Parser.getContext().lookupSymbol(Name); 6109 if (Sym) { 6110 // Diagnose assignment to a label. 6111 // 6112 // FIXME: Diagnostics. Note the location of the definition as a label. 6113 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 6114 if (isSymbolUsedInExpression(Sym, Value)) 6115 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 6116 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 6117 !Sym->isVariable()) 6118 ; // Allow redefinitions of undefined symbols only used in directives. 6119 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 6120 ; // Allow redefinitions of variables that haven't yet been used. 6121 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 6122 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 6123 else if (!Sym->isVariable()) 6124 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 6125 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 6126 return Parser.Error(EqualLoc, 6127 "invalid reassignment of non-absolute variable '" + 6128 Name + "'"); 6129 } else if (Name == ".") { 6130 Parser.getStreamer().emitValueToOffset(Value, 0, EqualLoc); 6131 return false; 6132 } else 6133 Sym = Parser.getContext().getOrCreateSymbol(Name); 6134 6135 Sym->setRedefinable(allow_redef); 6136 6137 return false; 6138 } 6139 6140 } // end namespace MCParserUtils 6141 } // end namespace llvm 6142 6143 /// Create an MCAsmParser instance. 6144 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 6145 MCStreamer &Out, const MCAsmInfo &MAI, 6146 unsigned CB) { 6147 return new AsmParser(SM, C, Out, MAI, CB); 6148 } 6149