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; 129 130 /// This is the current buffer index we're lexing from as managed by the 131 /// SourceMgr object. 132 unsigned CurBuffer; 133 134 AsmCond TheCondState; 135 std::vector<AsmCond> TheCondStack; 136 137 /// maps directive names to handler methods in parser 138 /// extensions. Extensions register themselves in this map by calling 139 /// addDirectiveHandler. 140 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 141 142 /// Stack of active macro instantiations. 143 std::vector<MacroInstantiation*> ActiveMacros; 144 145 /// List of bodies of anonymous macros. 146 std::deque<MCAsmMacro> MacroLikeBodies; 147 148 /// Boolean tracking whether macro substitution is enabled. 149 unsigned MacrosEnabledFlag : 1; 150 151 /// Keeps track of how many .macro's have been instantiated. 152 unsigned NumOfMacroInstantiations; 153 154 /// The values from the last parsed cpp hash file line comment if any. 155 struct CppHashInfoTy { 156 StringRef Filename; 157 int64_t LineNumber; 158 SMLoc Loc; 159 unsigned Buf; 160 CppHashInfoTy() : Filename(), LineNumber(0), Loc(), Buf(0) {} 161 }; 162 CppHashInfoTy CppHashInfo; 163 164 /// The filename from the first cpp hash file line comment, if any. 165 StringRef FirstCppHashFilename; 166 167 /// List of forward directional labels for diagnosis at the end. 168 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 169 170 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 171 unsigned AssemblerDialect = ~0U; 172 173 /// is Darwin compatibility enabled? 174 bool IsDarwin = false; 175 176 /// Are we parsing ms-style inline assembly? 177 bool ParsingMSInlineAsm = false; 178 179 /// Did we already inform the user about inconsistent MD5 usage? 180 bool ReportedInconsistentMD5 = false; 181 182 // Is alt macro mode enabled. 183 bool AltMacroMode = false; 184 185 public: 186 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 187 const MCAsmInfo &MAI, unsigned CB); 188 AsmParser(const AsmParser &) = delete; 189 AsmParser &operator=(const AsmParser &) = delete; 190 ~AsmParser() override; 191 192 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 193 194 void addDirectiveHandler(StringRef Directive, 195 ExtensionDirectiveHandler Handler) override { 196 ExtensionDirectiveMap[Directive] = Handler; 197 } 198 199 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 200 DirectiveKindMap[Directive.lower()] = DirectiveKindMap[Alias.lower()]; 201 } 202 203 /// @name MCAsmParser Interface 204 /// { 205 206 SourceMgr &getSourceManager() override { return SrcMgr; } 207 MCAsmLexer &getLexer() override { return Lexer; } 208 MCContext &getContext() override { return Ctx; } 209 MCStreamer &getStreamer() override { return Out; } 210 211 CodeViewContext &getCVContext() { return Ctx.getCVContext(); } 212 213 unsigned getAssemblerDialect() override { 214 if (AssemblerDialect == ~0U) 215 return MAI.getAssemblerDialect(); 216 else 217 return AssemblerDialect; 218 } 219 void setAssemblerDialect(unsigned i) override { 220 AssemblerDialect = i; 221 } 222 223 void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override; 224 bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override; 225 bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override; 226 227 const AsmToken &Lex() override; 228 229 void setParsingMSInlineAsm(bool V) override { 230 ParsingMSInlineAsm = V; 231 // When parsing MS inline asm, we must lex 0b1101 and 0ABCH as binary and 232 // hex integer literals. 233 Lexer.setLexMasmIntegers(V); 234 } 235 bool isParsingMSInlineAsm() override { return ParsingMSInlineAsm; } 236 237 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 238 unsigned &NumOutputs, unsigned &NumInputs, 239 SmallVectorImpl<std::pair<void *,bool>> &OpDecls, 240 SmallVectorImpl<std::string> &Constraints, 241 SmallVectorImpl<std::string> &Clobbers, 242 const MCInstrInfo *MII, const MCInstPrinter *IP, 243 MCAsmParserSemaCallback &SI) override; 244 245 bool parseExpression(const MCExpr *&Res); 246 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 247 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override; 248 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 249 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 250 SMLoc &EndLoc) override; 251 bool parseAbsoluteExpression(int64_t &Res) override; 252 253 /// Parse a floating point expression using the float \p Semantics 254 /// and set \p Res to the value. 255 bool parseRealValue(const fltSemantics &Semantics, APInt &Res); 256 257 /// Parse an identifier or string (as a quoted identifier) 258 /// and set \p Res to the identifier contents. 259 bool parseIdentifier(StringRef &Res) override; 260 void eatToEndOfStatement() override; 261 262 bool checkForValidSection() override; 263 264 /// } 265 266 private: 267 bool parseStatement(ParseStatementInfo &Info, 268 MCAsmParserSemaCallback *SI); 269 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 270 bool parseCppHashLineFilenameComment(SMLoc L, bool SaveLocInfo = true); 271 272 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 273 ArrayRef<MCAsmMacroParameter> Parameters); 274 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 275 ArrayRef<MCAsmMacroParameter> Parameters, 276 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 277 SMLoc L); 278 279 /// Are macros enabled in the parser? 280 bool areMacrosEnabled() {return MacrosEnabledFlag;} 281 282 /// Control a flag in the parser that enables or disables macros. 283 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 284 285 /// Are we inside a macro instantiation? 286 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 287 288 /// Handle entry to macro instantiation. 289 /// 290 /// \param M The macro. 291 /// \param NameLoc Instantiation location. 292 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 293 294 /// Handle exit from macro instantiation. 295 void handleMacroExit(); 296 297 /// Extract AsmTokens for a macro argument. 298 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 299 300 /// Parse all macro arguments for a given macro. 301 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 302 303 void printMacroInstantiations(); 304 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 305 SMRange Range = None) const { 306 ArrayRef<SMRange> Ranges(Range); 307 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 308 } 309 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 310 311 /// Should we emit DWARF describing this assembler source? (Returns false if 312 /// the source has .file directives, which means we don't want to generate 313 /// info describing the assembler source itself.) 314 bool enabledGenDwarfForAssembly(); 315 316 /// Enter the specified file. This returns true on failure. 317 bool enterIncludeFile(const std::string &Filename); 318 319 /// Process the specified file for the .incbin directive. 320 /// This returns true on failure. 321 bool processIncbinFile(const std::string &Filename, int64_t Skip = 0, 322 const MCExpr *Count = nullptr, SMLoc Loc = SMLoc()); 323 324 /// Reset the current lexer position to that given by \p Loc. The 325 /// current token is not set; clients should ensure Lex() is called 326 /// subsequently. 327 /// 328 /// \param InBuffer If not 0, should be the known buffer id that contains the 329 /// location. 330 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 331 332 /// Parse up to the end of statement and a return the contents from the 333 /// current token until the end of the statement; the current token on exit 334 /// will be either the EndOfStatement or EOF. 335 StringRef parseStringToEndOfStatement() override; 336 337 /// Parse until the end of a statement or a comma is encountered, 338 /// return the contents from the current token up to the end or comma. 339 StringRef parseStringToComma(); 340 341 bool parseAssignment(StringRef Name, bool allow_redef, 342 bool NoDeadStrip = false); 343 344 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 345 MCBinaryExpr::Opcode &Kind); 346 347 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 348 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 349 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 350 351 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 352 353 bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName); 354 bool parseCVFileId(int64_t &FileId, StringRef DirectiveName); 355 356 // Generic (target and platform independent) directive parsing. 357 enum DirectiveKind { 358 DK_NO_DIRECTIVE, // Placeholder 359 DK_SET, 360 DK_EQU, 361 DK_EQUIV, 362 DK_ASCII, 363 DK_ASCIZ, 364 DK_STRING, 365 DK_BYTE, 366 DK_SHORT, 367 DK_RELOC, 368 DK_VALUE, 369 DK_2BYTE, 370 DK_LONG, 371 DK_INT, 372 DK_4BYTE, 373 DK_QUAD, 374 DK_8BYTE, 375 DK_OCTA, 376 DK_DC, 377 DK_DC_A, 378 DK_DC_B, 379 DK_DC_D, 380 DK_DC_L, 381 DK_DC_S, 382 DK_DC_W, 383 DK_DC_X, 384 DK_DCB, 385 DK_DCB_B, 386 DK_DCB_D, 387 DK_DCB_L, 388 DK_DCB_S, 389 DK_DCB_W, 390 DK_DCB_X, 391 DK_DS, 392 DK_DS_B, 393 DK_DS_D, 394 DK_DS_L, 395 DK_DS_P, 396 DK_DS_S, 397 DK_DS_W, 398 DK_DS_X, 399 DK_SINGLE, 400 DK_FLOAT, 401 DK_DOUBLE, 402 DK_ALIGN, 403 DK_ALIGN32, 404 DK_BALIGN, 405 DK_BALIGNW, 406 DK_BALIGNL, 407 DK_P2ALIGN, 408 DK_P2ALIGNW, 409 DK_P2ALIGNL, 410 DK_ORG, 411 DK_FILL, 412 DK_ENDR, 413 DK_BUNDLE_ALIGN_MODE, 414 DK_BUNDLE_LOCK, 415 DK_BUNDLE_UNLOCK, 416 DK_ZERO, 417 DK_EXTERN, 418 DK_GLOBL, 419 DK_GLOBAL, 420 DK_LAZY_REFERENCE, 421 DK_NO_DEAD_STRIP, 422 DK_SYMBOL_RESOLVER, 423 DK_PRIVATE_EXTERN, 424 DK_REFERENCE, 425 DK_WEAK_DEFINITION, 426 DK_WEAK_REFERENCE, 427 DK_WEAK_DEF_CAN_BE_HIDDEN, 428 DK_COLD, 429 DK_COMM, 430 DK_COMMON, 431 DK_LCOMM, 432 DK_ABORT, 433 DK_INCLUDE, 434 DK_INCBIN, 435 DK_CODE16, 436 DK_CODE16GCC, 437 DK_REPT, 438 DK_IRP, 439 DK_IRPC, 440 DK_IF, 441 DK_IFEQ, 442 DK_IFGE, 443 DK_IFGT, 444 DK_IFLE, 445 DK_IFLT, 446 DK_IFNE, 447 DK_IFB, 448 DK_IFNB, 449 DK_IFC, 450 DK_IFEQS, 451 DK_IFNC, 452 DK_IFNES, 453 DK_IFDEF, 454 DK_IFNDEF, 455 DK_IFNOTDEF, 456 DK_ELSEIF, 457 DK_ELSE, 458 DK_ENDIF, 459 DK_SPACE, 460 DK_SKIP, 461 DK_FILE, 462 DK_LINE, 463 DK_LOC, 464 DK_STABS, 465 DK_CV_FILE, 466 DK_CV_FUNC_ID, 467 DK_CV_INLINE_SITE_ID, 468 DK_CV_LOC, 469 DK_CV_LINETABLE, 470 DK_CV_INLINE_LINETABLE, 471 DK_CV_DEF_RANGE, 472 DK_CV_STRINGTABLE, 473 DK_CV_STRING, 474 DK_CV_FILECHECKSUMS, 475 DK_CV_FILECHECKSUM_OFFSET, 476 DK_CV_FPO_DATA, 477 DK_CFI_SECTIONS, 478 DK_CFI_STARTPROC, 479 DK_CFI_ENDPROC, 480 DK_CFI_DEF_CFA, 481 DK_CFI_DEF_CFA_OFFSET, 482 DK_CFI_ADJUST_CFA_OFFSET, 483 DK_CFI_DEF_CFA_REGISTER, 484 DK_CFI_OFFSET, 485 DK_CFI_REL_OFFSET, 486 DK_CFI_PERSONALITY, 487 DK_CFI_LSDA, 488 DK_CFI_REMEMBER_STATE, 489 DK_CFI_RESTORE_STATE, 490 DK_CFI_SAME_VALUE, 491 DK_CFI_RESTORE, 492 DK_CFI_ESCAPE, 493 DK_CFI_RETURN_COLUMN, 494 DK_CFI_SIGNAL_FRAME, 495 DK_CFI_UNDEFINED, 496 DK_CFI_REGISTER, 497 DK_CFI_WINDOW_SAVE, 498 DK_CFI_B_KEY_FRAME, 499 DK_MACROS_ON, 500 DK_MACROS_OFF, 501 DK_ALTMACRO, 502 DK_NOALTMACRO, 503 DK_MACRO, 504 DK_EXITM, 505 DK_ENDM, 506 DK_ENDMACRO, 507 DK_PURGEM, 508 DK_SLEB128, 509 DK_ULEB128, 510 DK_ERR, 511 DK_ERROR, 512 DK_WARNING, 513 DK_PRINT, 514 DK_ADDRSIG, 515 DK_ADDRSIG_SYM, 516 DK_END 517 }; 518 519 /// Maps directive name --> DirectiveKind enum, for 520 /// directives parsed by this class. 521 StringMap<DirectiveKind> DirectiveKindMap; 522 523 // Codeview def_range type parsing. 524 enum CVDefRangeType { 525 CVDR_DEFRANGE = 0, // Placeholder 526 CVDR_DEFRANGE_REGISTER, 527 CVDR_DEFRANGE_FRAMEPOINTER_REL, 528 CVDR_DEFRANGE_SUBFIELD_REGISTER, 529 CVDR_DEFRANGE_REGISTER_REL 530 }; 531 532 /// Maps Codeview def_range types --> CVDefRangeType enum, for 533 /// Codeview def_range types parsed by this class. 534 StringMap<CVDefRangeType> CVDefRangeTypeMap; 535 536 // ".ascii", ".asciz", ".string" 537 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 538 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 539 bool parseDirectiveValue(StringRef IDVal, 540 unsigned Size); // ".byte", ".long", ... 541 bool parseDirectiveOctaValue(StringRef IDVal); // ".octa", ... 542 bool parseDirectiveRealValue(StringRef IDVal, 543 const fltSemantics &); // ".single", ... 544 bool parseDirectiveFill(); // ".fill" 545 bool parseDirectiveZero(); // ".zero" 546 // ".set", ".equ", ".equiv" 547 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 548 bool parseDirectiveOrg(); // ".org" 549 // ".align{,32}", ".p2align{,w,l}" 550 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 551 552 // ".file", ".line", ".loc", ".stabs" 553 bool parseDirectiveFile(SMLoc DirectiveLoc); 554 bool parseDirectiveLine(); 555 bool parseDirectiveLoc(); 556 bool parseDirectiveStabs(); 557 558 // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable", 559 // ".cv_inline_linetable", ".cv_def_range", ".cv_string" 560 bool parseDirectiveCVFile(); 561 bool parseDirectiveCVFuncId(); 562 bool parseDirectiveCVInlineSiteId(); 563 bool parseDirectiveCVLoc(); 564 bool parseDirectiveCVLinetable(); 565 bool parseDirectiveCVInlineLinetable(); 566 bool parseDirectiveCVDefRange(); 567 bool parseDirectiveCVString(); 568 bool parseDirectiveCVStringTable(); 569 bool parseDirectiveCVFileChecksums(); 570 bool parseDirectiveCVFileChecksumOffset(); 571 bool parseDirectiveCVFPOData(); 572 573 // .cfi directives 574 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 575 bool parseDirectiveCFIWindowSave(); 576 bool parseDirectiveCFISections(); 577 bool parseDirectiveCFIStartProc(); 578 bool parseDirectiveCFIEndProc(); 579 bool parseDirectiveCFIDefCfaOffset(); 580 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 581 bool parseDirectiveCFIAdjustCfaOffset(); 582 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 583 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 584 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 585 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 586 bool parseDirectiveCFIRememberState(); 587 bool parseDirectiveCFIRestoreState(); 588 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 589 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 590 bool parseDirectiveCFIEscape(); 591 bool parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc); 592 bool parseDirectiveCFISignalFrame(); 593 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 594 595 // macro directives 596 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 597 bool parseDirectiveExitMacro(StringRef Directive); 598 bool parseDirectiveEndMacro(StringRef Directive); 599 bool parseDirectiveMacro(SMLoc DirectiveLoc); 600 bool parseDirectiveMacrosOnOff(StringRef Directive); 601 // alternate macro mode directives 602 bool parseDirectiveAltmacro(StringRef Directive); 603 // ".bundle_align_mode" 604 bool parseDirectiveBundleAlignMode(); 605 // ".bundle_lock" 606 bool parseDirectiveBundleLock(); 607 // ".bundle_unlock" 608 bool parseDirectiveBundleUnlock(); 609 610 // ".space", ".skip" 611 bool parseDirectiveSpace(StringRef IDVal); 612 613 // ".dcb" 614 bool parseDirectiveDCB(StringRef IDVal, unsigned Size); 615 bool parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &); 616 // ".ds" 617 bool parseDirectiveDS(StringRef IDVal, unsigned Size); 618 619 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 620 bool parseDirectiveLEB128(bool Signed); 621 622 /// Parse a directive like ".globl" which 623 /// accepts a single symbol (which should be a label or an external). 624 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 625 626 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 627 628 bool parseDirectiveAbort(); // ".abort" 629 bool parseDirectiveInclude(); // ".include" 630 bool parseDirectiveIncbin(); // ".incbin" 631 632 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 633 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 634 // ".ifb" or ".ifnb", depending on ExpectBlank. 635 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 636 // ".ifc" or ".ifnc", depending on ExpectEqual. 637 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 638 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 639 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 640 // ".ifdef" or ".ifndef", depending on expect_defined 641 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 642 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 643 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 644 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 645 bool parseEscapedString(std::string &Data) override; 646 bool parseAngleBracketString(std::string &Data) override; 647 648 const MCExpr *applyModifierToExpr(const MCExpr *E, 649 MCSymbolRefExpr::VariantKind Variant); 650 651 // Macro-like directives 652 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 653 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 654 raw_svector_ostream &OS); 655 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 656 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 657 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 658 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 659 660 // "_emit" or "__emit" 661 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 662 size_t Len); 663 664 // "align" 665 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 666 667 // "end" 668 bool parseDirectiveEnd(SMLoc DirectiveLoc); 669 670 // ".err" or ".error" 671 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 672 673 // ".warning" 674 bool parseDirectiveWarning(SMLoc DirectiveLoc); 675 676 // .print <double-quotes-string> 677 bool parseDirectivePrint(SMLoc DirectiveLoc); 678 679 // Directives to support address-significance tables. 680 bool parseDirectiveAddrsig(); 681 bool parseDirectiveAddrsigSym(); 682 683 void initializeDirectiveKindMap(); 684 void initializeCVDefRangeTypeMap(); 685 }; 686 687 } // end anonymous namespace 688 689 namespace llvm { 690 691 extern MCAsmParserExtension *createDarwinAsmParser(); 692 extern MCAsmParserExtension *createELFAsmParser(); 693 extern MCAsmParserExtension *createCOFFAsmParser(); 694 extern MCAsmParserExtension *createWasmAsmParser(); 695 696 } // end namespace llvm 697 698 enum { DEFAULT_ADDRSPACE = 0 }; 699 700 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 701 const MCAsmInfo &MAI, unsigned CB = 0) 702 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 703 CurBuffer(CB ? CB : SM.getMainFileID()), MacrosEnabledFlag(true) { 704 HadError = false; 705 // Save the old handler. 706 SavedDiagHandler = SrcMgr.getDiagHandler(); 707 SavedDiagContext = SrcMgr.getDiagContext(); 708 // Set our own handler which calls the saved handler. 709 SrcMgr.setDiagHandler(DiagHandler, this); 710 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 711 712 // Initialize the platform / file format parser. 713 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 714 case MCObjectFileInfo::IsCOFF: 715 PlatformParser.reset(createCOFFAsmParser()); 716 break; 717 case MCObjectFileInfo::IsMachO: 718 PlatformParser.reset(createDarwinAsmParser()); 719 IsDarwin = true; 720 break; 721 case MCObjectFileInfo::IsELF: 722 PlatformParser.reset(createELFAsmParser()); 723 break; 724 case MCObjectFileInfo::IsWasm: 725 PlatformParser.reset(createWasmAsmParser()); 726 break; 727 case MCObjectFileInfo::IsXCOFF: 728 report_fatal_error( 729 "Need to implement createXCOFFAsmParser for XCOFF format."); 730 break; 731 } 732 733 PlatformParser->Initialize(*this); 734 initializeDirectiveKindMap(); 735 initializeCVDefRangeTypeMap(); 736 737 NumOfMacroInstantiations = 0; 738 } 739 740 AsmParser::~AsmParser() { 741 assert((HadError || ActiveMacros.empty()) && 742 "Unexpected active macro instantiation!"); 743 744 // Restore the saved diagnostics handler and context for use during 745 // finalization. 746 SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext); 747 } 748 749 void AsmParser::printMacroInstantiations() { 750 // Print the active macro instantiation stack. 751 for (std::vector<MacroInstantiation *>::const_reverse_iterator 752 it = ActiveMacros.rbegin(), 753 ie = ActiveMacros.rend(); 754 it != ie; ++it) 755 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 756 "while in macro instantiation"); 757 } 758 759 void AsmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) { 760 printPendingErrors(); 761 printMessage(L, SourceMgr::DK_Note, Msg, Range); 762 printMacroInstantiations(); 763 } 764 765 bool AsmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) { 766 if(getTargetParser().getTargetOptions().MCNoWarn) 767 return false; 768 if (getTargetParser().getTargetOptions().MCFatalWarnings) 769 return Error(L, Msg, Range); 770 printMessage(L, SourceMgr::DK_Warning, Msg, Range); 771 printMacroInstantiations(); 772 return false; 773 } 774 775 bool AsmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) { 776 HadError = true; 777 printMessage(L, SourceMgr::DK_Error, Msg, Range); 778 printMacroInstantiations(); 779 return true; 780 } 781 782 bool AsmParser::enterIncludeFile(const std::string &Filename) { 783 std::string IncludedFile; 784 unsigned NewBuf = 785 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 786 if (!NewBuf) 787 return true; 788 789 CurBuffer = NewBuf; 790 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 791 return false; 792 } 793 794 /// Process the specified .incbin file by searching for it in the include paths 795 /// then just emitting the byte contents of the file to the streamer. This 796 /// returns true on failure. 797 bool AsmParser::processIncbinFile(const std::string &Filename, int64_t Skip, 798 const MCExpr *Count, SMLoc Loc) { 799 std::string IncludedFile; 800 unsigned NewBuf = 801 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 802 if (!NewBuf) 803 return true; 804 805 // Pick up the bytes from the file and emit them. 806 StringRef Bytes = SrcMgr.getMemoryBuffer(NewBuf)->getBuffer(); 807 Bytes = Bytes.drop_front(Skip); 808 if (Count) { 809 int64_t Res; 810 if (!Count->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 811 return Error(Loc, "expected absolute expression"); 812 if (Res < 0) 813 return Warning(Loc, "negative count has no effect"); 814 Bytes = Bytes.take_front(Res); 815 } 816 getStreamer().emitBytes(Bytes); 817 return false; 818 } 819 820 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 821 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 822 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 823 Loc.getPointer()); 824 } 825 826 const AsmToken &AsmParser::Lex() { 827 if (Lexer.getTok().is(AsmToken::Error)) 828 Error(Lexer.getErrLoc(), Lexer.getErr()); 829 830 // if it's a end of statement with a comment in it 831 if (getTok().is(AsmToken::EndOfStatement)) { 832 // if this is a line comment output it. 833 if (!getTok().getString().empty() && getTok().getString().front() != '\n' && 834 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 835 Out.addExplicitComment(Twine(getTok().getString())); 836 } 837 838 const AsmToken *tok = &Lexer.Lex(); 839 840 // Parse comments here to be deferred until end of next statement. 841 while (tok->is(AsmToken::Comment)) { 842 if (MAI.preserveAsmComments()) 843 Out.addExplicitComment(Twine(tok->getString())); 844 tok = &Lexer.Lex(); 845 } 846 847 if (tok->is(AsmToken::Eof)) { 848 // If this is the end of an included file, pop the parent file off the 849 // include stack. 850 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 851 if (ParentIncludeLoc != SMLoc()) { 852 jumpToLoc(ParentIncludeLoc); 853 return Lex(); 854 } 855 } 856 857 return *tok; 858 } 859 860 bool AsmParser::enabledGenDwarfForAssembly() { 861 // Check whether the user specified -g. 862 if (!getContext().getGenDwarfForAssembly()) 863 return false; 864 // If we haven't encountered any .file directives (which would imply that 865 // the assembler source was produced with debug info already) then emit one 866 // describing the assembler source file itself. 867 if (getContext().getGenDwarfFileNumber() == 0) { 868 // Use the first #line directive for this, if any. It's preprocessed, so 869 // there is no checksum, and of course no source directive. 870 if (!FirstCppHashFilename.empty()) 871 getContext().setMCLineTableRootFile(/*CUID=*/0, 872 getContext().getCompilationDir(), 873 FirstCppHashFilename, 874 /*Cksum=*/None, /*Source=*/None); 875 const MCDwarfFile &RootFile = 876 getContext().getMCDwarfLineTable(/*CUID=*/0).getRootFile(); 877 getContext().setGenDwarfFileNumber(getStreamer().emitDwarfFileDirective( 878 /*CUID=*/0, getContext().getCompilationDir(), RootFile.Name, 879 RootFile.Checksum, RootFile.Source)); 880 } 881 return true; 882 } 883 884 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 885 // Create the initial section, if requested. 886 if (!NoInitialTextSection) 887 Out.InitSections(false); 888 889 // Prime the lexer. 890 Lex(); 891 892 HadError = false; 893 AsmCond StartingCondState = TheCondState; 894 SmallVector<AsmRewrite, 4> AsmStrRewrites; 895 896 // If we are generating dwarf for assembly source files save the initial text 897 // section. (Don't use enabledGenDwarfForAssembly() here, as we aren't 898 // emitting any actual debug info yet and haven't had a chance to parse any 899 // embedded .file directives.) 900 if (getContext().getGenDwarfForAssembly()) { 901 MCSection *Sec = getStreamer().getCurrentSectionOnly(); 902 if (!Sec->getBeginSymbol()) { 903 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 904 getStreamer().emitLabel(SectionStartSym); 905 Sec->setBeginSymbol(SectionStartSym); 906 } 907 bool InsertResult = getContext().addGenDwarfSection(Sec); 908 assert(InsertResult && ".text section should not have debug info yet"); 909 (void)InsertResult; 910 } 911 912 // While we have input, parse each statement. 913 while (Lexer.isNot(AsmToken::Eof)) { 914 ParseStatementInfo Info(&AsmStrRewrites); 915 bool Parsed = parseStatement(Info, nullptr); 916 917 // If we have a Lexer Error we are on an Error Token. Load in Lexer Error 918 // for printing ErrMsg via Lex() only if no (presumably better) parser error 919 // exists. 920 if (Parsed && !hasPendingError() && Lexer.getTok().is(AsmToken::Error)) { 921 Lex(); 922 } 923 924 // parseStatement returned true so may need to emit an error. 925 printPendingErrors(); 926 927 // Skipping to the next line if needed. 928 if (Parsed && !getLexer().isAtStartOfStatement()) 929 eatToEndOfStatement(); 930 } 931 932 getTargetParser().onEndOfFile(); 933 printPendingErrors(); 934 935 // All errors should have been emitted. 936 assert(!hasPendingError() && "unexpected error from parseStatement"); 937 938 getTargetParser().flushPendingInstructions(getStreamer()); 939 940 if (TheCondState.TheCond != StartingCondState.TheCond || 941 TheCondState.Ignore != StartingCondState.Ignore) 942 printError(getTok().getLoc(), "unmatched .ifs or .elses"); 943 // Check to see there are no empty DwarfFile slots. 944 const auto &LineTables = getContext().getMCDwarfLineTables(); 945 if (!LineTables.empty()) { 946 unsigned Index = 0; 947 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 948 if (File.Name.empty() && Index != 0) 949 printError(getTok().getLoc(), "unassigned file number: " + 950 Twine(Index) + 951 " for .file directives"); 952 ++Index; 953 } 954 } 955 956 // Check to see that all assembler local symbols were actually defined. 957 // Targets that don't do subsections via symbols may not want this, though, 958 // so conservatively exclude them. Only do this if we're finalizing, though, 959 // as otherwise we won't necessarilly have seen everything yet. 960 if (!NoFinalize) { 961 if (MAI.hasSubsectionsViaSymbols()) { 962 for (const auto &TableEntry : getContext().getSymbols()) { 963 MCSymbol *Sym = TableEntry.getValue(); 964 // Variable symbols may not be marked as defined, so check those 965 // explicitly. If we know it's a variable, we have a definition for 966 // the purposes of this check. 967 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 968 // FIXME: We would really like to refer back to where the symbol was 969 // first referenced for a source location. We need to add something 970 // to track that. Currently, we just point to the end of the file. 971 printError(getTok().getLoc(), "assembler local symbol '" + 972 Sym->getName() + "' not defined"); 973 } 974 } 975 976 // Temporary symbols like the ones for directional jumps don't go in the 977 // symbol table. They also need to be diagnosed in all (final) cases. 978 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 979 if (std::get<2>(LocSym)->isUndefined()) { 980 // Reset the state of any "# line file" directives we've seen to the 981 // context as it was at the diagnostic site. 982 CppHashInfo = std::get<1>(LocSym); 983 printError(std::get<0>(LocSym), "directional label undefined"); 984 } 985 } 986 } 987 988 // Finalize the output stream if there are no errors and if the client wants 989 // us to. 990 if (!HadError && !NoFinalize) 991 Out.Finish(); 992 993 return HadError || getContext().hadError(); 994 } 995 996 bool AsmParser::checkForValidSection() { 997 if (!ParsingMSInlineAsm && !getStreamer().getCurrentSectionOnly()) { 998 Out.InitSections(false); 999 return Error(getTok().getLoc(), 1000 "expected section directive before assembly directive"); 1001 } 1002 return false; 1003 } 1004 1005 /// Throw away the rest of the line for testing purposes. 1006 void AsmParser::eatToEndOfStatement() { 1007 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1008 Lexer.Lex(); 1009 1010 // Eat EOL. 1011 if (Lexer.is(AsmToken::EndOfStatement)) 1012 Lexer.Lex(); 1013 } 1014 1015 StringRef AsmParser::parseStringToEndOfStatement() { 1016 const char *Start = getTok().getLoc().getPointer(); 1017 1018 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1019 Lexer.Lex(); 1020 1021 const char *End = getTok().getLoc().getPointer(); 1022 return StringRef(Start, End - Start); 1023 } 1024 1025 StringRef AsmParser::parseStringToComma() { 1026 const char *Start = getTok().getLoc().getPointer(); 1027 1028 while (Lexer.isNot(AsmToken::EndOfStatement) && 1029 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 1030 Lexer.Lex(); 1031 1032 const char *End = getTok().getLoc().getPointer(); 1033 return StringRef(Start, End - Start); 1034 } 1035 1036 /// Parse a paren expression and return it. 1037 /// NOTE: This assumes the leading '(' has already been consumed. 1038 /// 1039 /// parenexpr ::= expr) 1040 /// 1041 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1042 if (parseExpression(Res)) 1043 return true; 1044 if (Lexer.isNot(AsmToken::RParen)) 1045 return TokError("expected ')' in parentheses expression"); 1046 EndLoc = Lexer.getTok().getEndLoc(); 1047 Lex(); 1048 return false; 1049 } 1050 1051 /// Parse a bracket expression and return it. 1052 /// NOTE: This assumes the leading '[' has already been consumed. 1053 /// 1054 /// bracketexpr ::= expr] 1055 /// 1056 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1057 if (parseExpression(Res)) 1058 return true; 1059 EndLoc = getTok().getEndLoc(); 1060 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 1061 return true; 1062 return false; 1063 } 1064 1065 /// Parse a primary expression and return it. 1066 /// primaryexpr ::= (parenexpr 1067 /// primaryexpr ::= symbol 1068 /// primaryexpr ::= number 1069 /// primaryexpr ::= '.' 1070 /// primaryexpr ::= ~,+,- primaryexpr 1071 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1072 SMLoc FirstTokenLoc = getLexer().getLoc(); 1073 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 1074 switch (FirstTokenKind) { 1075 default: 1076 return TokError("unknown token in expression"); 1077 // If we have an error assume that we've already handled it. 1078 case AsmToken::Error: 1079 return true; 1080 case AsmToken::Exclaim: 1081 Lex(); // Eat the operator. 1082 if (parsePrimaryExpr(Res, EndLoc)) 1083 return true; 1084 Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc); 1085 return false; 1086 case AsmToken::Dollar: 1087 case AsmToken::At: 1088 case AsmToken::String: 1089 case AsmToken::Identifier: { 1090 StringRef Identifier; 1091 if (parseIdentifier(Identifier)) { 1092 // We may have failed but $ may be a valid token. 1093 if (getTok().is(AsmToken::Dollar)) { 1094 if (Lexer.getMAI().getDollarIsPC()) { 1095 Lex(); 1096 // This is a '$' reference, which references the current PC. Emit a 1097 // temporary label to the streamer and refer to it. 1098 MCSymbol *Sym = Ctx.createTempSymbol(); 1099 Out.emitLabel(Sym); 1100 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 1101 getContext()); 1102 EndLoc = FirstTokenLoc; 1103 return false; 1104 } 1105 return Error(FirstTokenLoc, "invalid token in expression"); 1106 } 1107 } 1108 // Parse symbol variant 1109 std::pair<StringRef, StringRef> Split; 1110 if (!MAI.useParensForSymbolVariant()) { 1111 if (FirstTokenKind == AsmToken::String) { 1112 if (Lexer.is(AsmToken::At)) { 1113 Lex(); // eat @ 1114 SMLoc AtLoc = getLexer().getLoc(); 1115 StringRef VName; 1116 if (parseIdentifier(VName)) 1117 return Error(AtLoc, "expected symbol variant after '@'"); 1118 1119 Split = std::make_pair(Identifier, VName); 1120 } 1121 } else { 1122 Split = Identifier.split('@'); 1123 } 1124 } else if (Lexer.is(AsmToken::LParen)) { 1125 Lex(); // eat '('. 1126 StringRef VName; 1127 parseIdentifier(VName); 1128 // eat ')'. 1129 if (parseToken(AsmToken::RParen, 1130 "unexpected token in variant, expected ')'")) 1131 return true; 1132 Split = std::make_pair(Identifier, VName); 1133 } 1134 1135 EndLoc = SMLoc::getFromPointer(Identifier.end()); 1136 1137 // This is a symbol reference. 1138 StringRef SymbolName = Identifier; 1139 if (SymbolName.empty()) 1140 return Error(getLexer().getLoc(), "expected a symbol reference"); 1141 1142 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1143 1144 // Lookup the symbol variant if used. 1145 if (!Split.second.empty()) { 1146 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1147 if (Variant != MCSymbolRefExpr::VK_Invalid) { 1148 SymbolName = Split.first; 1149 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 1150 Variant = MCSymbolRefExpr::VK_None; 1151 } else { 1152 return Error(SMLoc::getFromPointer(Split.second.begin()), 1153 "invalid variant '" + Split.second + "'"); 1154 } 1155 } 1156 1157 MCSymbol *Sym = getContext().getInlineAsmLabel(SymbolName); 1158 if (!Sym) 1159 Sym = getContext().getOrCreateSymbol(SymbolName); 1160 1161 // If this is an absolute variable reference, substitute it now to preserve 1162 // semantics in the face of reassignment. 1163 if (Sym->isVariable()) { 1164 auto V = Sym->getVariableValue(/*SetUsed*/ false); 1165 bool DoInline = isa<MCConstantExpr>(V) && !Variant; 1166 if (auto TV = dyn_cast<MCTargetExpr>(V)) 1167 DoInline = TV->inlineAssignedExpr(); 1168 if (DoInline) { 1169 if (Variant) 1170 return Error(EndLoc, "unexpected modifier on variable reference"); 1171 Res = Sym->getVariableValue(/*SetUsed*/ false); 1172 return false; 1173 } 1174 } 1175 1176 // Otherwise create a symbol ref. 1177 Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc); 1178 return false; 1179 } 1180 case AsmToken::BigNum: 1181 return TokError("literal value out of range for directive"); 1182 case AsmToken::Integer: { 1183 SMLoc Loc = getTok().getLoc(); 1184 int64_t IntVal = getTok().getIntVal(); 1185 Res = MCConstantExpr::create(IntVal, getContext()); 1186 EndLoc = Lexer.getTok().getEndLoc(); 1187 Lex(); // Eat token. 1188 // Look for 'b' or 'f' following an Integer as a directional label 1189 if (Lexer.getKind() == AsmToken::Identifier) { 1190 StringRef IDVal = getTok().getString(); 1191 // Lookup the symbol variant if used. 1192 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 1193 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1194 if (Split.first.size() != IDVal.size()) { 1195 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1196 if (Variant == MCSymbolRefExpr::VK_Invalid) 1197 return TokError("invalid variant '" + Split.second + "'"); 1198 IDVal = Split.first; 1199 } 1200 if (IDVal == "f" || IDVal == "b") { 1201 MCSymbol *Sym = 1202 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 1203 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 1204 if (IDVal == "b" && Sym->isUndefined()) 1205 return Error(Loc, "directional label undefined"); 1206 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 1207 EndLoc = Lexer.getTok().getEndLoc(); 1208 Lex(); // Eat identifier. 1209 } 1210 } 1211 return false; 1212 } 1213 case AsmToken::Real: { 1214 APFloat RealVal(APFloat::IEEEdouble(), getTok().getString()); 1215 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1216 Res = MCConstantExpr::create(IntVal, getContext()); 1217 EndLoc = Lexer.getTok().getEndLoc(); 1218 Lex(); // Eat token. 1219 return false; 1220 } 1221 case AsmToken::Dot: { 1222 // This is a '.' reference, which references the current PC. Emit a 1223 // temporary label to the streamer and refer to it. 1224 MCSymbol *Sym = Ctx.createTempSymbol(); 1225 Out.emitLabel(Sym); 1226 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1227 EndLoc = Lexer.getTok().getEndLoc(); 1228 Lex(); // Eat identifier. 1229 return false; 1230 } 1231 case AsmToken::LParen: 1232 Lex(); // Eat the '('. 1233 return parseParenExpr(Res, EndLoc); 1234 case AsmToken::LBrac: 1235 if (!PlatformParser->HasBracketExpressions()) 1236 return TokError("brackets expression not supported on this target"); 1237 Lex(); // Eat the '['. 1238 return parseBracketExpr(Res, EndLoc); 1239 case AsmToken::Minus: 1240 Lex(); // Eat the operator. 1241 if (parsePrimaryExpr(Res, EndLoc)) 1242 return true; 1243 Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc); 1244 return false; 1245 case AsmToken::Plus: 1246 Lex(); // Eat the operator. 1247 if (parsePrimaryExpr(Res, EndLoc)) 1248 return true; 1249 Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc); 1250 return false; 1251 case AsmToken::Tilde: 1252 Lex(); // Eat the operator. 1253 if (parsePrimaryExpr(Res, EndLoc)) 1254 return true; 1255 Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc); 1256 return false; 1257 // MIPS unary expression operators. The lexer won't generate these tokens if 1258 // MCAsmInfo::HasMipsExpressions is false for the target. 1259 case AsmToken::PercentCall16: 1260 case AsmToken::PercentCall_Hi: 1261 case AsmToken::PercentCall_Lo: 1262 case AsmToken::PercentDtprel_Hi: 1263 case AsmToken::PercentDtprel_Lo: 1264 case AsmToken::PercentGot: 1265 case AsmToken::PercentGot_Disp: 1266 case AsmToken::PercentGot_Hi: 1267 case AsmToken::PercentGot_Lo: 1268 case AsmToken::PercentGot_Ofst: 1269 case AsmToken::PercentGot_Page: 1270 case AsmToken::PercentGottprel: 1271 case AsmToken::PercentGp_Rel: 1272 case AsmToken::PercentHi: 1273 case AsmToken::PercentHigher: 1274 case AsmToken::PercentHighest: 1275 case AsmToken::PercentLo: 1276 case AsmToken::PercentNeg: 1277 case AsmToken::PercentPcrel_Hi: 1278 case AsmToken::PercentPcrel_Lo: 1279 case AsmToken::PercentTlsgd: 1280 case AsmToken::PercentTlsldm: 1281 case AsmToken::PercentTprel_Hi: 1282 case AsmToken::PercentTprel_Lo: 1283 Lex(); // Eat the operator. 1284 if (Lexer.isNot(AsmToken::LParen)) 1285 return TokError("expected '(' after operator"); 1286 Lex(); // Eat the operator. 1287 if (parseExpression(Res, EndLoc)) 1288 return true; 1289 if (Lexer.isNot(AsmToken::RParen)) 1290 return TokError("expected ')'"); 1291 Lex(); // Eat the operator. 1292 Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx); 1293 return !Res; 1294 } 1295 } 1296 1297 bool AsmParser::parseExpression(const MCExpr *&Res) { 1298 SMLoc EndLoc; 1299 return parseExpression(Res, EndLoc); 1300 } 1301 1302 const MCExpr * 1303 AsmParser::applyModifierToExpr(const MCExpr *E, 1304 MCSymbolRefExpr::VariantKind Variant) { 1305 // Ask the target implementation about this expression first. 1306 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 1307 if (NewE) 1308 return NewE; 1309 // Recurse over the given expression, rebuilding it to apply the given variant 1310 // if there is exactly one symbol. 1311 switch (E->getKind()) { 1312 case MCExpr::Target: 1313 case MCExpr::Constant: 1314 return nullptr; 1315 1316 case MCExpr::SymbolRef: { 1317 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 1318 1319 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 1320 TokError("invalid variant on expression '" + getTok().getIdentifier() + 1321 "' (already modified)"); 1322 return E; 1323 } 1324 1325 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1326 } 1327 1328 case MCExpr::Unary: { 1329 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1330 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1331 if (!Sub) 1332 return nullptr; 1333 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1334 } 1335 1336 case MCExpr::Binary: { 1337 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1338 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1339 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1340 1341 if (!LHS && !RHS) 1342 return nullptr; 1343 1344 if (!LHS) 1345 LHS = BE->getLHS(); 1346 if (!RHS) 1347 RHS = BE->getRHS(); 1348 1349 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1350 } 1351 } 1352 1353 llvm_unreachable("Invalid expression kind!"); 1354 } 1355 1356 /// This function checks if the next token is <string> type or arithmetic. 1357 /// string that begin with character '<' must end with character '>'. 1358 /// otherwise it is arithmetics. 1359 /// If the function returns a 'true' value, 1360 /// the End argument will be filled with the last location pointed to the '>' 1361 /// character. 1362 1363 /// There is a gap between the AltMacro's documentation and the single quote 1364 /// implementation. GCC does not fully support this feature and so we will not 1365 /// support it. 1366 /// TODO: Adding single quote as a string. 1367 static bool isAngleBracketString(SMLoc &StrLoc, SMLoc &EndLoc) { 1368 assert((StrLoc.getPointer() != nullptr) && 1369 "Argument to the function cannot be a NULL value"); 1370 const char *CharPtr = StrLoc.getPointer(); 1371 while ((*CharPtr != '>') && (*CharPtr != '\n') && (*CharPtr != '\r') && 1372 (*CharPtr != '\0')) { 1373 if (*CharPtr == '!') 1374 CharPtr++; 1375 CharPtr++; 1376 } 1377 if (*CharPtr == '>') { 1378 EndLoc = StrLoc.getFromPointer(CharPtr + 1); 1379 return true; 1380 } 1381 return false; 1382 } 1383 1384 /// creating a string without the escape characters '!'. 1385 static std::string angleBracketString(StringRef AltMacroStr) { 1386 std::string Res; 1387 for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) { 1388 if (AltMacroStr[Pos] == '!') 1389 Pos++; 1390 Res += AltMacroStr[Pos]; 1391 } 1392 return Res; 1393 } 1394 1395 /// Parse an expression and return it. 1396 /// 1397 /// expr ::= expr &&,|| expr -> lowest. 1398 /// expr ::= expr |,^,&,! expr 1399 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1400 /// expr ::= expr <<,>> expr 1401 /// expr ::= expr +,- expr 1402 /// expr ::= expr *,/,% expr -> highest. 1403 /// expr ::= primaryexpr 1404 /// 1405 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1406 // Parse the expression. 1407 Res = nullptr; 1408 if (getTargetParser().parsePrimaryExpr(Res, EndLoc) || 1409 parseBinOpRHS(1, Res, EndLoc)) 1410 return true; 1411 1412 // As a special case, we support 'a op b @ modifier' by rewriting the 1413 // expression to include the modifier. This is inefficient, but in general we 1414 // expect users to use 'a@modifier op b'. 1415 if (Lexer.getKind() == AsmToken::At) { 1416 Lex(); 1417 1418 if (Lexer.isNot(AsmToken::Identifier)) 1419 return TokError("unexpected symbol modifier following '@'"); 1420 1421 MCSymbolRefExpr::VariantKind Variant = 1422 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1423 if (Variant == MCSymbolRefExpr::VK_Invalid) 1424 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1425 1426 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1427 if (!ModifiedRes) { 1428 return TokError("invalid modifier '" + getTok().getIdentifier() + 1429 "' (no symbols present)"); 1430 } 1431 1432 Res = ModifiedRes; 1433 Lex(); 1434 } 1435 1436 // Try to constant fold it up front, if possible. Do not exploit 1437 // assembler here. 1438 int64_t Value; 1439 if (Res->evaluateAsAbsolute(Value)) 1440 Res = MCConstantExpr::create(Value, getContext()); 1441 1442 return false; 1443 } 1444 1445 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1446 Res = nullptr; 1447 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1448 } 1449 1450 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1451 SMLoc &EndLoc) { 1452 if (parseParenExpr(Res, EndLoc)) 1453 return true; 1454 1455 for (; ParenDepth > 0; --ParenDepth) { 1456 if (parseBinOpRHS(1, Res, EndLoc)) 1457 return true; 1458 1459 // We don't Lex() the last RParen. 1460 // This is the same behavior as parseParenExpression(). 1461 if (ParenDepth - 1 > 0) { 1462 EndLoc = getTok().getEndLoc(); 1463 if (parseToken(AsmToken::RParen, 1464 "expected ')' in parentheses expression")) 1465 return true; 1466 } 1467 } 1468 return false; 1469 } 1470 1471 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1472 const MCExpr *Expr; 1473 1474 SMLoc StartLoc = Lexer.getLoc(); 1475 if (parseExpression(Expr)) 1476 return true; 1477 1478 if (!Expr->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 1479 return Error(StartLoc, "expected absolute expression"); 1480 1481 return false; 1482 } 1483 1484 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1485 MCBinaryExpr::Opcode &Kind, 1486 bool ShouldUseLogicalShr) { 1487 switch (K) { 1488 default: 1489 return 0; // not a binop. 1490 1491 // Lowest Precedence: &&, || 1492 case AsmToken::AmpAmp: 1493 Kind = MCBinaryExpr::LAnd; 1494 return 1; 1495 case AsmToken::PipePipe: 1496 Kind = MCBinaryExpr::LOr; 1497 return 1; 1498 1499 // Low Precedence: |, &, ^ 1500 // 1501 // FIXME: gas seems to support '!' as an infix operator? 1502 case AsmToken::Pipe: 1503 Kind = MCBinaryExpr::Or; 1504 return 2; 1505 case AsmToken::Caret: 1506 Kind = MCBinaryExpr::Xor; 1507 return 2; 1508 case AsmToken::Amp: 1509 Kind = MCBinaryExpr::And; 1510 return 2; 1511 1512 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1513 case AsmToken::EqualEqual: 1514 Kind = MCBinaryExpr::EQ; 1515 return 3; 1516 case AsmToken::ExclaimEqual: 1517 case AsmToken::LessGreater: 1518 Kind = MCBinaryExpr::NE; 1519 return 3; 1520 case AsmToken::Less: 1521 Kind = MCBinaryExpr::LT; 1522 return 3; 1523 case AsmToken::LessEqual: 1524 Kind = MCBinaryExpr::LTE; 1525 return 3; 1526 case AsmToken::Greater: 1527 Kind = MCBinaryExpr::GT; 1528 return 3; 1529 case AsmToken::GreaterEqual: 1530 Kind = MCBinaryExpr::GTE; 1531 return 3; 1532 1533 // Intermediate Precedence: <<, >> 1534 case AsmToken::LessLess: 1535 Kind = MCBinaryExpr::Shl; 1536 return 4; 1537 case AsmToken::GreaterGreater: 1538 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1539 return 4; 1540 1541 // High Intermediate Precedence: +, - 1542 case AsmToken::Plus: 1543 Kind = MCBinaryExpr::Add; 1544 return 5; 1545 case AsmToken::Minus: 1546 Kind = MCBinaryExpr::Sub; 1547 return 5; 1548 1549 // Highest Precedence: *, /, % 1550 case AsmToken::Star: 1551 Kind = MCBinaryExpr::Mul; 1552 return 6; 1553 case AsmToken::Slash: 1554 Kind = MCBinaryExpr::Div; 1555 return 6; 1556 case AsmToken::Percent: 1557 Kind = MCBinaryExpr::Mod; 1558 return 6; 1559 } 1560 } 1561 1562 static unsigned getGNUBinOpPrecedence(AsmToken::TokenKind K, 1563 MCBinaryExpr::Opcode &Kind, 1564 bool ShouldUseLogicalShr) { 1565 switch (K) { 1566 default: 1567 return 0; // not a binop. 1568 1569 // Lowest Precedence: &&, || 1570 case AsmToken::AmpAmp: 1571 Kind = MCBinaryExpr::LAnd; 1572 return 2; 1573 case AsmToken::PipePipe: 1574 Kind = MCBinaryExpr::LOr; 1575 return 1; 1576 1577 // Low Precedence: ==, !=, <>, <, <=, >, >= 1578 case AsmToken::EqualEqual: 1579 Kind = MCBinaryExpr::EQ; 1580 return 3; 1581 case AsmToken::ExclaimEqual: 1582 case AsmToken::LessGreater: 1583 Kind = MCBinaryExpr::NE; 1584 return 3; 1585 case AsmToken::Less: 1586 Kind = MCBinaryExpr::LT; 1587 return 3; 1588 case AsmToken::LessEqual: 1589 Kind = MCBinaryExpr::LTE; 1590 return 3; 1591 case AsmToken::Greater: 1592 Kind = MCBinaryExpr::GT; 1593 return 3; 1594 case AsmToken::GreaterEqual: 1595 Kind = MCBinaryExpr::GTE; 1596 return 3; 1597 1598 // Low Intermediate Precedence: +, - 1599 case AsmToken::Plus: 1600 Kind = MCBinaryExpr::Add; 1601 return 4; 1602 case AsmToken::Minus: 1603 Kind = MCBinaryExpr::Sub; 1604 return 4; 1605 1606 // High Intermediate Precedence: |, &, ^ 1607 // 1608 // FIXME: gas seems to support '!' as an infix operator? 1609 case AsmToken::Pipe: 1610 Kind = MCBinaryExpr::Or; 1611 return 5; 1612 case AsmToken::Caret: 1613 Kind = MCBinaryExpr::Xor; 1614 return 5; 1615 case AsmToken::Amp: 1616 Kind = MCBinaryExpr::And; 1617 return 5; 1618 1619 // Highest Precedence: *, /, %, <<, >> 1620 case AsmToken::Star: 1621 Kind = MCBinaryExpr::Mul; 1622 return 6; 1623 case AsmToken::Slash: 1624 Kind = MCBinaryExpr::Div; 1625 return 6; 1626 case AsmToken::Percent: 1627 Kind = MCBinaryExpr::Mod; 1628 return 6; 1629 case AsmToken::LessLess: 1630 Kind = MCBinaryExpr::Shl; 1631 return 6; 1632 case AsmToken::GreaterGreater: 1633 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1634 return 6; 1635 } 1636 } 1637 1638 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1639 MCBinaryExpr::Opcode &Kind) { 1640 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1641 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1642 : getGNUBinOpPrecedence(K, Kind, ShouldUseLogicalShr); 1643 } 1644 1645 /// Parse all binary operators with precedence >= 'Precedence'. 1646 /// Res contains the LHS of the expression on input. 1647 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1648 SMLoc &EndLoc) { 1649 SMLoc StartLoc = Lexer.getLoc(); 1650 while (true) { 1651 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1652 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1653 1654 // If the next token is lower precedence than we are allowed to eat, return 1655 // successfully with what we ate already. 1656 if (TokPrec < Precedence) 1657 return false; 1658 1659 Lex(); 1660 1661 // Eat the next primary expression. 1662 const MCExpr *RHS; 1663 if (getTargetParser().parsePrimaryExpr(RHS, EndLoc)) 1664 return true; 1665 1666 // If BinOp binds less tightly with RHS than the operator after RHS, let 1667 // the pending operator take RHS as its LHS. 1668 MCBinaryExpr::Opcode Dummy; 1669 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1670 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1671 return true; 1672 1673 // Merge LHS and RHS according to operator. 1674 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc); 1675 } 1676 } 1677 1678 /// ParseStatement: 1679 /// ::= EndOfStatement 1680 /// ::= Label* Directive ...Operands... EndOfStatement 1681 /// ::= Label* Identifier OperandList* EndOfStatement 1682 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1683 MCAsmParserSemaCallback *SI) { 1684 assert(!hasPendingError() && "parseStatement started with pending error"); 1685 // Eat initial spaces and comments 1686 while (Lexer.is(AsmToken::Space)) 1687 Lex(); 1688 if (Lexer.is(AsmToken::EndOfStatement)) { 1689 // if this is a line comment we can drop it safely 1690 if (getTok().getString().empty() || getTok().getString().front() == '\r' || 1691 getTok().getString().front() == '\n') 1692 Out.AddBlankLine(); 1693 Lex(); 1694 return false; 1695 } 1696 // Statements always start with an identifier. 1697 AsmToken ID = getTok(); 1698 SMLoc IDLoc = ID.getLoc(); 1699 StringRef IDVal; 1700 int64_t LocalLabelVal = -1; 1701 if (Lexer.is(AsmToken::HashDirective)) 1702 return parseCppHashLineFilenameComment(IDLoc, 1703 !isInsideMacroInstantiation()); 1704 1705 // Allow an integer followed by a ':' as a directional local label. 1706 if (Lexer.is(AsmToken::Integer)) { 1707 LocalLabelVal = getTok().getIntVal(); 1708 if (LocalLabelVal < 0) { 1709 if (!TheCondState.Ignore) { 1710 Lex(); // always eat a token 1711 return Error(IDLoc, "unexpected token at start of statement"); 1712 } 1713 IDVal = ""; 1714 } else { 1715 IDVal = getTok().getString(); 1716 Lex(); // Consume the integer token to be used as an identifier token. 1717 if (Lexer.getKind() != AsmToken::Colon) { 1718 if (!TheCondState.Ignore) { 1719 Lex(); // always eat a token 1720 return Error(IDLoc, "unexpected token at start of statement"); 1721 } 1722 } 1723 } 1724 } else if (Lexer.is(AsmToken::Dot)) { 1725 // Treat '.' as a valid identifier in this context. 1726 Lex(); 1727 IDVal = "."; 1728 } else if (Lexer.is(AsmToken::LCurly)) { 1729 // Treat '{' as a valid identifier in this context. 1730 Lex(); 1731 IDVal = "{"; 1732 1733 } else if (Lexer.is(AsmToken::RCurly)) { 1734 // Treat '}' as a valid identifier in this context. 1735 Lex(); 1736 IDVal = "}"; 1737 } else if (Lexer.is(AsmToken::Star) && 1738 getTargetParser().starIsStartOfStatement()) { 1739 // Accept '*' as a valid start of statement. 1740 Lex(); 1741 IDVal = "*"; 1742 } else if (parseIdentifier(IDVal)) { 1743 if (!TheCondState.Ignore) { 1744 Lex(); // always eat a token 1745 return Error(IDLoc, "unexpected token at start of statement"); 1746 } 1747 IDVal = ""; 1748 } 1749 1750 // Handle conditional assembly here before checking for skipping. We 1751 // have to do this so that .endif isn't skipped in a ".if 0" block for 1752 // example. 1753 StringMap<DirectiveKind>::const_iterator DirKindIt = 1754 DirectiveKindMap.find(IDVal.lower()); 1755 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1756 1757 ? DK_NO_DIRECTIVE 1758 : DirKindIt->getValue(); 1759 switch (DirKind) { 1760 default: 1761 break; 1762 case DK_IF: 1763 case DK_IFEQ: 1764 case DK_IFGE: 1765 case DK_IFGT: 1766 case DK_IFLE: 1767 case DK_IFLT: 1768 case DK_IFNE: 1769 return parseDirectiveIf(IDLoc, DirKind); 1770 case DK_IFB: 1771 return parseDirectiveIfb(IDLoc, true); 1772 case DK_IFNB: 1773 return parseDirectiveIfb(IDLoc, false); 1774 case DK_IFC: 1775 return parseDirectiveIfc(IDLoc, true); 1776 case DK_IFEQS: 1777 return parseDirectiveIfeqs(IDLoc, true); 1778 case DK_IFNC: 1779 return parseDirectiveIfc(IDLoc, false); 1780 case DK_IFNES: 1781 return parseDirectiveIfeqs(IDLoc, false); 1782 case DK_IFDEF: 1783 return parseDirectiveIfdef(IDLoc, true); 1784 case DK_IFNDEF: 1785 case DK_IFNOTDEF: 1786 return parseDirectiveIfdef(IDLoc, false); 1787 case DK_ELSEIF: 1788 return parseDirectiveElseIf(IDLoc); 1789 case DK_ELSE: 1790 return parseDirectiveElse(IDLoc); 1791 case DK_ENDIF: 1792 return parseDirectiveEndIf(IDLoc); 1793 } 1794 1795 // Ignore the statement if in the middle of inactive conditional 1796 // (e.g. ".if 0"). 1797 if (TheCondState.Ignore) { 1798 eatToEndOfStatement(); 1799 return false; 1800 } 1801 1802 // FIXME: Recurse on local labels? 1803 1804 // See what kind of statement we have. 1805 switch (Lexer.getKind()) { 1806 case AsmToken::Colon: { 1807 if (!getTargetParser().isLabel(ID)) 1808 break; 1809 if (checkForValidSection()) 1810 return true; 1811 1812 // identifier ':' -> Label. 1813 Lex(); 1814 1815 // Diagnose attempt to use '.' as a label. 1816 if (IDVal == ".") 1817 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1818 1819 // Diagnose attempt to use a variable as a label. 1820 // 1821 // FIXME: Diagnostics. Note the location of the definition as a label. 1822 // FIXME: This doesn't diagnose assignment to a symbol which has been 1823 // implicitly marked as external. 1824 MCSymbol *Sym; 1825 if (LocalLabelVal == -1) { 1826 if (ParsingMSInlineAsm && SI) { 1827 StringRef RewrittenLabel = 1828 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1829 assert(!RewrittenLabel.empty() && 1830 "We should have an internal name here."); 1831 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1832 RewrittenLabel); 1833 IDVal = RewrittenLabel; 1834 } 1835 Sym = getContext().getOrCreateSymbol(IDVal); 1836 } else 1837 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1838 // End of Labels should be treated as end of line for lexing 1839 // purposes but that information is not available to the Lexer who 1840 // does not understand Labels. This may cause us to see a Hash 1841 // here instead of a preprocessor line comment. 1842 if (getTok().is(AsmToken::Hash)) { 1843 StringRef CommentStr = parseStringToEndOfStatement(); 1844 Lexer.Lex(); 1845 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1846 } 1847 1848 // Consume any end of statement token, if present, to avoid spurious 1849 // AddBlankLine calls(). 1850 if (getTok().is(AsmToken::EndOfStatement)) { 1851 Lex(); 1852 } 1853 1854 getTargetParser().doBeforeLabelEmit(Sym); 1855 1856 // Emit the label. 1857 if (!getTargetParser().isParsingMSInlineAsm()) 1858 Out.emitLabel(Sym, IDLoc); 1859 1860 // If we are generating dwarf for assembly source files then gather the 1861 // info to make a dwarf label entry for this label if needed. 1862 if (enabledGenDwarfForAssembly()) 1863 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1864 IDLoc); 1865 1866 getTargetParser().onLabelParsed(Sym); 1867 1868 return false; 1869 } 1870 1871 case AsmToken::Equal: 1872 if (!getTargetParser().equalIsAsmAssignment()) 1873 break; 1874 // identifier '=' ... -> assignment statement 1875 Lex(); 1876 1877 return parseAssignment(IDVal, true); 1878 1879 default: // Normal instruction or directive. 1880 break; 1881 } 1882 1883 // If macros are enabled, check to see if this is a macro instantiation. 1884 if (areMacrosEnabled()) 1885 if (const MCAsmMacro *M = getContext().lookupMacro(IDVal)) { 1886 return handleMacroEntry(M, IDLoc); 1887 } 1888 1889 // Otherwise, we have a normal instruction or directive. 1890 1891 // Directives start with "." 1892 if (IDVal.startswith(".") && IDVal != ".") { 1893 // There are several entities interested in parsing directives: 1894 // 1895 // 1. The target-specific assembly parser. Some directives are target 1896 // specific or may potentially behave differently on certain targets. 1897 // 2. Asm parser extensions. For example, platform-specific parsers 1898 // (like the ELF parser) register themselves as extensions. 1899 // 3. The generic directive parser implemented by this class. These are 1900 // all the directives that behave in a target and platform independent 1901 // manner, or at least have a default behavior that's shared between 1902 // all targets and platforms. 1903 1904 getTargetParser().flushPendingInstructions(getStreamer()); 1905 1906 SMLoc StartTokLoc = getTok().getLoc(); 1907 bool TPDirectiveReturn = getTargetParser().ParseDirective(ID); 1908 1909 if (hasPendingError()) 1910 return true; 1911 // Currently the return value should be true if we are 1912 // uninterested but as this is at odds with the standard parsing 1913 // convention (return true = error) we have instances of a parsed 1914 // directive that fails returning true as an error. Catch these 1915 // cases as best as possible errors here. 1916 if (TPDirectiveReturn && StartTokLoc != getTok().getLoc()) 1917 return true; 1918 // Return if we did some parsing or believe we succeeded. 1919 if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc()) 1920 return false; 1921 1922 // Next, check the extension directive map to see if any extension has 1923 // registered itself to parse this directive. 1924 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1925 ExtensionDirectiveMap.lookup(IDVal); 1926 if (Handler.first) 1927 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1928 1929 // Finally, if no one else is interested in this directive, it must be 1930 // generic and familiar to this class. 1931 switch (DirKind) { 1932 default: 1933 break; 1934 case DK_SET: 1935 case DK_EQU: 1936 return parseDirectiveSet(IDVal, true); 1937 case DK_EQUIV: 1938 return parseDirectiveSet(IDVal, false); 1939 case DK_ASCII: 1940 return parseDirectiveAscii(IDVal, false); 1941 case DK_ASCIZ: 1942 case DK_STRING: 1943 return parseDirectiveAscii(IDVal, true); 1944 case DK_BYTE: 1945 case DK_DC_B: 1946 return parseDirectiveValue(IDVal, 1); 1947 case DK_DC: 1948 case DK_DC_W: 1949 case DK_SHORT: 1950 case DK_VALUE: 1951 case DK_2BYTE: 1952 return parseDirectiveValue(IDVal, 2); 1953 case DK_LONG: 1954 case DK_INT: 1955 case DK_4BYTE: 1956 case DK_DC_L: 1957 return parseDirectiveValue(IDVal, 4); 1958 case DK_QUAD: 1959 case DK_8BYTE: 1960 return parseDirectiveValue(IDVal, 8); 1961 case DK_DC_A: 1962 return parseDirectiveValue( 1963 IDVal, getContext().getAsmInfo()->getCodePointerSize()); 1964 case DK_OCTA: 1965 return parseDirectiveOctaValue(IDVal); 1966 case DK_SINGLE: 1967 case DK_FLOAT: 1968 case DK_DC_S: 1969 return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle()); 1970 case DK_DOUBLE: 1971 case DK_DC_D: 1972 return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble()); 1973 case DK_ALIGN: { 1974 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1975 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1976 } 1977 case DK_ALIGN32: { 1978 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1979 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1980 } 1981 case DK_BALIGN: 1982 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1983 case DK_BALIGNW: 1984 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1985 case DK_BALIGNL: 1986 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 1987 case DK_P2ALIGN: 1988 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 1989 case DK_P2ALIGNW: 1990 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 1991 case DK_P2ALIGNL: 1992 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 1993 case DK_ORG: 1994 return parseDirectiveOrg(); 1995 case DK_FILL: 1996 return parseDirectiveFill(); 1997 case DK_ZERO: 1998 return parseDirectiveZero(); 1999 case DK_EXTERN: 2000 eatToEndOfStatement(); // .extern is the default, ignore it. 2001 return false; 2002 case DK_GLOBL: 2003 case DK_GLOBAL: 2004 return parseDirectiveSymbolAttribute(MCSA_Global); 2005 case DK_LAZY_REFERENCE: 2006 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 2007 case DK_NO_DEAD_STRIP: 2008 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 2009 case DK_SYMBOL_RESOLVER: 2010 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 2011 case DK_PRIVATE_EXTERN: 2012 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 2013 case DK_REFERENCE: 2014 return parseDirectiveSymbolAttribute(MCSA_Reference); 2015 case DK_WEAK_DEFINITION: 2016 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 2017 case DK_WEAK_REFERENCE: 2018 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 2019 case DK_WEAK_DEF_CAN_BE_HIDDEN: 2020 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 2021 case DK_COLD: 2022 return parseDirectiveSymbolAttribute(MCSA_Cold); 2023 case DK_COMM: 2024 case DK_COMMON: 2025 return parseDirectiveComm(/*IsLocal=*/false); 2026 case DK_LCOMM: 2027 return parseDirectiveComm(/*IsLocal=*/true); 2028 case DK_ABORT: 2029 return parseDirectiveAbort(); 2030 case DK_INCLUDE: 2031 return parseDirectiveInclude(); 2032 case DK_INCBIN: 2033 return parseDirectiveIncbin(); 2034 case DK_CODE16: 2035 case DK_CODE16GCC: 2036 return TokError(Twine(IDVal) + 2037 " not currently supported for this target"); 2038 case DK_REPT: 2039 return parseDirectiveRept(IDLoc, IDVal); 2040 case DK_IRP: 2041 return parseDirectiveIrp(IDLoc); 2042 case DK_IRPC: 2043 return parseDirectiveIrpc(IDLoc); 2044 case DK_ENDR: 2045 return parseDirectiveEndr(IDLoc); 2046 case DK_BUNDLE_ALIGN_MODE: 2047 return parseDirectiveBundleAlignMode(); 2048 case DK_BUNDLE_LOCK: 2049 return parseDirectiveBundleLock(); 2050 case DK_BUNDLE_UNLOCK: 2051 return parseDirectiveBundleUnlock(); 2052 case DK_SLEB128: 2053 return parseDirectiveLEB128(true); 2054 case DK_ULEB128: 2055 return parseDirectiveLEB128(false); 2056 case DK_SPACE: 2057 case DK_SKIP: 2058 return parseDirectiveSpace(IDVal); 2059 case DK_FILE: 2060 return parseDirectiveFile(IDLoc); 2061 case DK_LINE: 2062 return parseDirectiveLine(); 2063 case DK_LOC: 2064 return parseDirectiveLoc(); 2065 case DK_STABS: 2066 return parseDirectiveStabs(); 2067 case DK_CV_FILE: 2068 return parseDirectiveCVFile(); 2069 case DK_CV_FUNC_ID: 2070 return parseDirectiveCVFuncId(); 2071 case DK_CV_INLINE_SITE_ID: 2072 return parseDirectiveCVInlineSiteId(); 2073 case DK_CV_LOC: 2074 return parseDirectiveCVLoc(); 2075 case DK_CV_LINETABLE: 2076 return parseDirectiveCVLinetable(); 2077 case DK_CV_INLINE_LINETABLE: 2078 return parseDirectiveCVInlineLinetable(); 2079 case DK_CV_DEF_RANGE: 2080 return parseDirectiveCVDefRange(); 2081 case DK_CV_STRING: 2082 return parseDirectiveCVString(); 2083 case DK_CV_STRINGTABLE: 2084 return parseDirectiveCVStringTable(); 2085 case DK_CV_FILECHECKSUMS: 2086 return parseDirectiveCVFileChecksums(); 2087 case DK_CV_FILECHECKSUM_OFFSET: 2088 return parseDirectiveCVFileChecksumOffset(); 2089 case DK_CV_FPO_DATA: 2090 return parseDirectiveCVFPOData(); 2091 case DK_CFI_SECTIONS: 2092 return parseDirectiveCFISections(); 2093 case DK_CFI_STARTPROC: 2094 return parseDirectiveCFIStartProc(); 2095 case DK_CFI_ENDPROC: 2096 return parseDirectiveCFIEndProc(); 2097 case DK_CFI_DEF_CFA: 2098 return parseDirectiveCFIDefCfa(IDLoc); 2099 case DK_CFI_DEF_CFA_OFFSET: 2100 return parseDirectiveCFIDefCfaOffset(); 2101 case DK_CFI_ADJUST_CFA_OFFSET: 2102 return parseDirectiveCFIAdjustCfaOffset(); 2103 case DK_CFI_DEF_CFA_REGISTER: 2104 return parseDirectiveCFIDefCfaRegister(IDLoc); 2105 case DK_CFI_OFFSET: 2106 return parseDirectiveCFIOffset(IDLoc); 2107 case DK_CFI_REL_OFFSET: 2108 return parseDirectiveCFIRelOffset(IDLoc); 2109 case DK_CFI_PERSONALITY: 2110 return parseDirectiveCFIPersonalityOrLsda(true); 2111 case DK_CFI_LSDA: 2112 return parseDirectiveCFIPersonalityOrLsda(false); 2113 case DK_CFI_REMEMBER_STATE: 2114 return parseDirectiveCFIRememberState(); 2115 case DK_CFI_RESTORE_STATE: 2116 return parseDirectiveCFIRestoreState(); 2117 case DK_CFI_SAME_VALUE: 2118 return parseDirectiveCFISameValue(IDLoc); 2119 case DK_CFI_RESTORE: 2120 return parseDirectiveCFIRestore(IDLoc); 2121 case DK_CFI_ESCAPE: 2122 return parseDirectiveCFIEscape(); 2123 case DK_CFI_RETURN_COLUMN: 2124 return parseDirectiveCFIReturnColumn(IDLoc); 2125 case DK_CFI_SIGNAL_FRAME: 2126 return parseDirectiveCFISignalFrame(); 2127 case DK_CFI_UNDEFINED: 2128 return parseDirectiveCFIUndefined(IDLoc); 2129 case DK_CFI_REGISTER: 2130 return parseDirectiveCFIRegister(IDLoc); 2131 case DK_CFI_WINDOW_SAVE: 2132 return parseDirectiveCFIWindowSave(); 2133 case DK_MACROS_ON: 2134 case DK_MACROS_OFF: 2135 return parseDirectiveMacrosOnOff(IDVal); 2136 case DK_MACRO: 2137 return parseDirectiveMacro(IDLoc); 2138 case DK_ALTMACRO: 2139 case DK_NOALTMACRO: 2140 return parseDirectiveAltmacro(IDVal); 2141 case DK_EXITM: 2142 return parseDirectiveExitMacro(IDVal); 2143 case DK_ENDM: 2144 case DK_ENDMACRO: 2145 return parseDirectiveEndMacro(IDVal); 2146 case DK_PURGEM: 2147 return parseDirectivePurgeMacro(IDLoc); 2148 case DK_END: 2149 return parseDirectiveEnd(IDLoc); 2150 case DK_ERR: 2151 return parseDirectiveError(IDLoc, false); 2152 case DK_ERROR: 2153 return parseDirectiveError(IDLoc, true); 2154 case DK_WARNING: 2155 return parseDirectiveWarning(IDLoc); 2156 case DK_RELOC: 2157 return parseDirectiveReloc(IDLoc); 2158 case DK_DCB: 2159 case DK_DCB_W: 2160 return parseDirectiveDCB(IDVal, 2); 2161 case DK_DCB_B: 2162 return parseDirectiveDCB(IDVal, 1); 2163 case DK_DCB_D: 2164 return parseDirectiveRealDCB(IDVal, APFloat::IEEEdouble()); 2165 case DK_DCB_L: 2166 return parseDirectiveDCB(IDVal, 4); 2167 case DK_DCB_S: 2168 return parseDirectiveRealDCB(IDVal, APFloat::IEEEsingle()); 2169 case DK_DC_X: 2170 case DK_DCB_X: 2171 return TokError(Twine(IDVal) + 2172 " not currently supported for this target"); 2173 case DK_DS: 2174 case DK_DS_W: 2175 return parseDirectiveDS(IDVal, 2); 2176 case DK_DS_B: 2177 return parseDirectiveDS(IDVal, 1); 2178 case DK_DS_D: 2179 return parseDirectiveDS(IDVal, 8); 2180 case DK_DS_L: 2181 case DK_DS_S: 2182 return parseDirectiveDS(IDVal, 4); 2183 case DK_DS_P: 2184 case DK_DS_X: 2185 return parseDirectiveDS(IDVal, 12); 2186 case DK_PRINT: 2187 return parseDirectivePrint(IDLoc); 2188 case DK_ADDRSIG: 2189 return parseDirectiveAddrsig(); 2190 case DK_ADDRSIG_SYM: 2191 return parseDirectiveAddrsigSym(); 2192 } 2193 2194 return Error(IDLoc, "unknown directive"); 2195 } 2196 2197 // __asm _emit or __asm __emit 2198 if (ParsingMSInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 2199 IDVal == "_EMIT" || IDVal == "__EMIT")) 2200 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 2201 2202 // __asm align 2203 if (ParsingMSInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 2204 return parseDirectiveMSAlign(IDLoc, Info); 2205 2206 if (ParsingMSInlineAsm && (IDVal == "even" || IDVal == "EVEN")) 2207 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 2208 if (checkForValidSection()) 2209 return true; 2210 2211 // Canonicalize the opcode to lower case. 2212 std::string OpcodeStr = IDVal.lower(); 2213 ParseInstructionInfo IInfo(Info.AsmRewrites); 2214 bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 2215 Info.ParsedOperands); 2216 Info.ParseError = ParseHadError; 2217 2218 // Dump the parsed representation, if requested. 2219 if (getShowParsedOperands()) { 2220 SmallString<256> Str; 2221 raw_svector_ostream OS(Str); 2222 OS << "parsed instruction: ["; 2223 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 2224 if (i != 0) 2225 OS << ", "; 2226 Info.ParsedOperands[i]->print(OS); 2227 } 2228 OS << "]"; 2229 2230 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 2231 } 2232 2233 // Fail even if ParseInstruction erroneously returns false. 2234 if (hasPendingError() || ParseHadError) 2235 return true; 2236 2237 // If we are generating dwarf for the current section then generate a .loc 2238 // directive for the instruction. 2239 if (!ParseHadError && enabledGenDwarfForAssembly() && 2240 getContext().getGenDwarfSectionSyms().count( 2241 getStreamer().getCurrentSectionOnly())) { 2242 unsigned Line; 2243 if (ActiveMacros.empty()) 2244 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 2245 else 2246 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 2247 ActiveMacros.front()->ExitBuffer); 2248 2249 // If we previously parsed a cpp hash file line comment then make sure the 2250 // current Dwarf File is for the CppHashFilename if not then emit the 2251 // Dwarf File table for it and adjust the line number for the .loc. 2252 if (!CppHashInfo.Filename.empty()) { 2253 unsigned FileNumber = getStreamer().emitDwarfFileDirective( 2254 0, StringRef(), CppHashInfo.Filename); 2255 getContext().setGenDwarfFileNumber(FileNumber); 2256 2257 unsigned CppHashLocLineNo = 2258 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 2259 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 2260 } 2261 2262 getStreamer().emitDwarfLocDirective( 2263 getContext().getGenDwarfFileNumber(), Line, 0, 2264 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 2265 StringRef()); 2266 } 2267 2268 // If parsing succeeded, match the instruction. 2269 if (!ParseHadError) { 2270 uint64_t ErrorInfo; 2271 if (getTargetParser().MatchAndEmitInstruction( 2272 IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo, 2273 getTargetParser().isParsingMSInlineAsm())) 2274 return true; 2275 } 2276 return false; 2277 } 2278 2279 // Parse and erase curly braces marking block start/end 2280 bool 2281 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 2282 // Identify curly brace marking block start/end 2283 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 2284 return false; 2285 2286 SMLoc StartLoc = Lexer.getLoc(); 2287 Lex(); // Eat the brace 2288 if (Lexer.is(AsmToken::EndOfStatement)) 2289 Lex(); // Eat EndOfStatement following the brace 2290 2291 // Erase the block start/end brace from the output asm string 2292 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 2293 StartLoc.getPointer()); 2294 return true; 2295 } 2296 2297 /// parseCppHashLineFilenameComment as this: 2298 /// ::= # number "filename" 2299 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L, bool SaveLocInfo) { 2300 Lex(); // Eat the hash token. 2301 // Lexer only ever emits HashDirective if it fully formed if it's 2302 // done the checking already so this is an internal error. 2303 assert(getTok().is(AsmToken::Integer) && 2304 "Lexing Cpp line comment: Expected Integer"); 2305 int64_t LineNumber = getTok().getIntVal(); 2306 Lex(); 2307 assert(getTok().is(AsmToken::String) && 2308 "Lexing Cpp line comment: Expected String"); 2309 StringRef Filename = getTok().getString(); 2310 Lex(); 2311 2312 if (!SaveLocInfo) 2313 return false; 2314 2315 // Get rid of the enclosing quotes. 2316 Filename = Filename.substr(1, Filename.size() - 2); 2317 2318 // Save the SMLoc, Filename and LineNumber for later use by diagnostics 2319 // and possibly DWARF file info. 2320 CppHashInfo.Loc = L; 2321 CppHashInfo.Filename = Filename; 2322 CppHashInfo.LineNumber = LineNumber; 2323 CppHashInfo.Buf = CurBuffer; 2324 if (FirstCppHashFilename.empty()) 2325 FirstCppHashFilename = Filename; 2326 return false; 2327 } 2328 2329 /// will use the last parsed cpp hash line filename comment 2330 /// for the Filename and LineNo if any in the diagnostic. 2331 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 2332 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 2333 raw_ostream &OS = errs(); 2334 2335 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 2336 SMLoc DiagLoc = Diag.getLoc(); 2337 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2338 unsigned CppHashBuf = 2339 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 2340 2341 // Like SourceMgr::printMessage() we need to print the include stack if any 2342 // before printing the message. 2343 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2344 if (!Parser->SavedDiagHandler && DiagCurBuffer && 2345 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 2346 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 2347 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 2348 } 2349 2350 // If we have not parsed a cpp hash line filename comment or the source 2351 // manager changed or buffer changed (like in a nested include) then just 2352 // print the normal diagnostic using its Filename and LineNo. 2353 if (!Parser->CppHashInfo.LineNumber || &DiagSrcMgr != &Parser->SrcMgr || 2354 DiagBuf != CppHashBuf) { 2355 if (Parser->SavedDiagHandler) 2356 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2357 else 2358 Diag.print(nullptr, OS); 2359 return; 2360 } 2361 2362 // Use the CppHashFilename and calculate a line number based on the 2363 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 2364 // for the diagnostic. 2365 const std::string &Filename = std::string(Parser->CppHashInfo.Filename); 2366 2367 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 2368 int CppHashLocLineNo = 2369 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 2370 int LineNo = 2371 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 2372 2373 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 2374 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 2375 Diag.getLineContents(), Diag.getRanges()); 2376 2377 if (Parser->SavedDiagHandler) 2378 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 2379 else 2380 NewDiag.print(nullptr, OS); 2381 } 2382 2383 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2384 // difference being that that function accepts '@' as part of identifiers and 2385 // we can't do that. AsmLexer.cpp should probably be changed to handle 2386 // '@' as a special case when needed. 2387 static bool isIdentifierChar(char c) { 2388 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2389 c == '.'; 2390 } 2391 2392 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2393 ArrayRef<MCAsmMacroParameter> Parameters, 2394 ArrayRef<MCAsmMacroArgument> A, 2395 bool EnableAtPseudoVariable, SMLoc L) { 2396 unsigned NParameters = Parameters.size(); 2397 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2398 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2399 return Error(L, "Wrong number of arguments"); 2400 2401 // A macro without parameters is handled differently on Darwin: 2402 // gas accepts no arguments and does no substitutions 2403 while (!Body.empty()) { 2404 // Scan for the next substitution. 2405 std::size_t End = Body.size(), Pos = 0; 2406 for (; Pos != End; ++Pos) { 2407 // Check for a substitution or escape. 2408 if (IsDarwin && !NParameters) { 2409 // This macro has no parameters, look for $0, $1, etc. 2410 if (Body[Pos] != '$' || Pos + 1 == End) 2411 continue; 2412 2413 char Next = Body[Pos + 1]; 2414 if (Next == '$' || Next == 'n' || 2415 isdigit(static_cast<unsigned char>(Next))) 2416 break; 2417 } else { 2418 // This macro has parameters, look for \foo, \bar, etc. 2419 if (Body[Pos] == '\\' && Pos + 1 != End) 2420 break; 2421 } 2422 } 2423 2424 // Add the prefix. 2425 OS << Body.slice(0, Pos); 2426 2427 // Check if we reached the end. 2428 if (Pos == End) 2429 break; 2430 2431 if (IsDarwin && !NParameters) { 2432 switch (Body[Pos + 1]) { 2433 // $$ => $ 2434 case '$': 2435 OS << '$'; 2436 break; 2437 2438 // $n => number of arguments 2439 case 'n': 2440 OS << A.size(); 2441 break; 2442 2443 // $[0-9] => argument 2444 default: { 2445 // Missing arguments are ignored. 2446 unsigned Index = Body[Pos + 1] - '0'; 2447 if (Index >= A.size()) 2448 break; 2449 2450 // Otherwise substitute with the token values, with spaces eliminated. 2451 for (const AsmToken &Token : A[Index]) 2452 OS << Token.getString(); 2453 break; 2454 } 2455 } 2456 Pos += 2; 2457 } else { 2458 unsigned I = Pos + 1; 2459 2460 // Check for the \@ pseudo-variable. 2461 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2462 ++I; 2463 else 2464 while (isIdentifierChar(Body[I]) && I + 1 != End) 2465 ++I; 2466 2467 const char *Begin = Body.data() + Pos + 1; 2468 StringRef Argument(Begin, I - (Pos + 1)); 2469 unsigned Index = 0; 2470 2471 if (Argument == "@") { 2472 OS << NumOfMacroInstantiations; 2473 Pos += 2; 2474 } else { 2475 for (; Index < NParameters; ++Index) 2476 if (Parameters[Index].Name == Argument) 2477 break; 2478 2479 if (Index == NParameters) { 2480 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2481 Pos += 3; 2482 else { 2483 OS << '\\' << Argument; 2484 Pos = I; 2485 } 2486 } else { 2487 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2488 for (const AsmToken &Token : A[Index]) 2489 // For altmacro mode, you can write '%expr'. 2490 // The prefix '%' evaluates the expression 'expr' 2491 // and uses the result as a string (e.g. replace %(1+2) with the 2492 // string "3"). 2493 // Here, we identify the integer token which is the result of the 2494 // absolute expression evaluation and replace it with its string 2495 // representation. 2496 if (AltMacroMode && Token.getString().front() == '%' && 2497 Token.is(AsmToken::Integer)) 2498 // Emit an integer value to the buffer. 2499 OS << Token.getIntVal(); 2500 // Only Token that was validated as a string and begins with '<' 2501 // is considered altMacroString!!! 2502 else if (AltMacroMode && Token.getString().front() == '<' && 2503 Token.is(AsmToken::String)) { 2504 OS << angleBracketString(Token.getStringContents()); 2505 } 2506 // We expect no quotes around the string's contents when 2507 // parsing for varargs. 2508 else if (Token.isNot(AsmToken::String) || VarargParameter) 2509 OS << Token.getString(); 2510 else 2511 OS << Token.getStringContents(); 2512 2513 Pos += 1 + Argument.size(); 2514 } 2515 } 2516 } 2517 // Update the scan point. 2518 Body = Body.substr(Pos); 2519 } 2520 2521 return false; 2522 } 2523 2524 static bool isOperator(AsmToken::TokenKind kind) { 2525 switch (kind) { 2526 default: 2527 return false; 2528 case AsmToken::Plus: 2529 case AsmToken::Minus: 2530 case AsmToken::Tilde: 2531 case AsmToken::Slash: 2532 case AsmToken::Star: 2533 case AsmToken::Dot: 2534 case AsmToken::Equal: 2535 case AsmToken::EqualEqual: 2536 case AsmToken::Pipe: 2537 case AsmToken::PipePipe: 2538 case AsmToken::Caret: 2539 case AsmToken::Amp: 2540 case AsmToken::AmpAmp: 2541 case AsmToken::Exclaim: 2542 case AsmToken::ExclaimEqual: 2543 case AsmToken::Less: 2544 case AsmToken::LessEqual: 2545 case AsmToken::LessLess: 2546 case AsmToken::LessGreater: 2547 case AsmToken::Greater: 2548 case AsmToken::GreaterEqual: 2549 case AsmToken::GreaterGreater: 2550 return true; 2551 } 2552 } 2553 2554 namespace { 2555 2556 class AsmLexerSkipSpaceRAII { 2557 public: 2558 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2559 Lexer.setSkipSpace(SkipSpace); 2560 } 2561 2562 ~AsmLexerSkipSpaceRAII() { 2563 Lexer.setSkipSpace(true); 2564 } 2565 2566 private: 2567 AsmLexer &Lexer; 2568 }; 2569 2570 } // end anonymous namespace 2571 2572 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2573 2574 if (Vararg) { 2575 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2576 StringRef Str = parseStringToEndOfStatement(); 2577 MA.emplace_back(AsmToken::String, Str); 2578 } 2579 return false; 2580 } 2581 2582 unsigned ParenLevel = 0; 2583 2584 // Darwin doesn't use spaces to delmit arguments. 2585 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2586 2587 bool SpaceEaten; 2588 2589 while (true) { 2590 SpaceEaten = false; 2591 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2592 return TokError("unexpected token in macro instantiation"); 2593 2594 if (ParenLevel == 0) { 2595 2596 if (Lexer.is(AsmToken::Comma)) 2597 break; 2598 2599 if (Lexer.is(AsmToken::Space)) { 2600 SpaceEaten = true; 2601 Lexer.Lex(); // Eat spaces 2602 } 2603 2604 // Spaces can delimit parameters, but could also be part an expression. 2605 // If the token after a space is an operator, add the token and the next 2606 // one into this argument 2607 if (!IsDarwin) { 2608 if (isOperator(Lexer.getKind())) { 2609 MA.push_back(getTok()); 2610 Lexer.Lex(); 2611 2612 // Whitespace after an operator can be ignored. 2613 if (Lexer.is(AsmToken::Space)) 2614 Lexer.Lex(); 2615 2616 continue; 2617 } 2618 } 2619 if (SpaceEaten) 2620 break; 2621 } 2622 2623 // handleMacroEntry relies on not advancing the lexer here 2624 // to be able to fill in the remaining default parameter values 2625 if (Lexer.is(AsmToken::EndOfStatement)) 2626 break; 2627 2628 // Adjust the current parentheses level. 2629 if (Lexer.is(AsmToken::LParen)) 2630 ++ParenLevel; 2631 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2632 --ParenLevel; 2633 2634 // Append the token to the current argument list. 2635 MA.push_back(getTok()); 2636 Lexer.Lex(); 2637 } 2638 2639 if (ParenLevel != 0) 2640 return TokError("unbalanced parentheses in macro argument"); 2641 return false; 2642 } 2643 2644 // Parse the macro instantiation arguments. 2645 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2646 MCAsmMacroArguments &A) { 2647 const unsigned NParameters = M ? M->Parameters.size() : 0; 2648 bool NamedParametersFound = false; 2649 SmallVector<SMLoc, 4> FALocs; 2650 2651 A.resize(NParameters); 2652 FALocs.resize(NParameters); 2653 2654 // Parse two kinds of macro invocations: 2655 // - macros defined without any parameters accept an arbitrary number of them 2656 // - macros defined with parameters accept at most that many of them 2657 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2658 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2659 ++Parameter) { 2660 SMLoc IDLoc = Lexer.getLoc(); 2661 MCAsmMacroParameter FA; 2662 2663 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2664 if (parseIdentifier(FA.Name)) 2665 return Error(IDLoc, "invalid argument identifier for formal argument"); 2666 2667 if (Lexer.isNot(AsmToken::Equal)) 2668 return TokError("expected '=' after formal parameter identifier"); 2669 2670 Lex(); 2671 2672 NamedParametersFound = true; 2673 } 2674 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2675 2676 if (NamedParametersFound && FA.Name.empty()) 2677 return Error(IDLoc, "cannot mix positional and keyword arguments"); 2678 2679 SMLoc StrLoc = Lexer.getLoc(); 2680 SMLoc EndLoc; 2681 if (AltMacroMode && Lexer.is(AsmToken::Percent)) { 2682 const MCExpr *AbsoluteExp; 2683 int64_t Value; 2684 /// Eat '%' 2685 Lex(); 2686 if (parseExpression(AbsoluteExp, EndLoc)) 2687 return false; 2688 if (!AbsoluteExp->evaluateAsAbsolute(Value, 2689 getStreamer().getAssemblerPtr())) 2690 return Error(StrLoc, "expected absolute expression"); 2691 const char *StrChar = StrLoc.getPointer(); 2692 const char *EndChar = EndLoc.getPointer(); 2693 AsmToken newToken(AsmToken::Integer, 2694 StringRef(StrChar, EndChar - StrChar), Value); 2695 FA.Value.push_back(newToken); 2696 } else if (AltMacroMode && Lexer.is(AsmToken::Less) && 2697 isAngleBracketString(StrLoc, EndLoc)) { 2698 const char *StrChar = StrLoc.getPointer(); 2699 const char *EndChar = EndLoc.getPointer(); 2700 jumpToLoc(EndLoc, CurBuffer); 2701 /// Eat from '<' to '>' 2702 Lex(); 2703 AsmToken newToken(AsmToken::String, 2704 StringRef(StrChar, EndChar - StrChar)); 2705 FA.Value.push_back(newToken); 2706 } else if(parseMacroArgument(FA.Value, Vararg)) 2707 return true; 2708 2709 unsigned PI = Parameter; 2710 if (!FA.Name.empty()) { 2711 unsigned FAI = 0; 2712 for (FAI = 0; FAI < NParameters; ++FAI) 2713 if (M->Parameters[FAI].Name == FA.Name) 2714 break; 2715 2716 if (FAI >= NParameters) { 2717 assert(M && "expected macro to be defined"); 2718 return Error(IDLoc, "parameter named '" + FA.Name + 2719 "' does not exist for macro '" + M->Name + "'"); 2720 } 2721 PI = FAI; 2722 } 2723 2724 if (!FA.Value.empty()) { 2725 if (A.size() <= PI) 2726 A.resize(PI + 1); 2727 A[PI] = FA.Value; 2728 2729 if (FALocs.size() <= PI) 2730 FALocs.resize(PI + 1); 2731 2732 FALocs[PI] = Lexer.getLoc(); 2733 } 2734 2735 // At the end of the statement, fill in remaining arguments that have 2736 // default values. If there aren't any, then the next argument is 2737 // required but missing 2738 if (Lexer.is(AsmToken::EndOfStatement)) { 2739 bool Failure = false; 2740 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2741 if (A[FAI].empty()) { 2742 if (M->Parameters[FAI].Required) { 2743 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2744 "missing value for required parameter " 2745 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2746 Failure = true; 2747 } 2748 2749 if (!M->Parameters[FAI].Value.empty()) 2750 A[FAI] = M->Parameters[FAI].Value; 2751 } 2752 } 2753 return Failure; 2754 } 2755 2756 if (Lexer.is(AsmToken::Comma)) 2757 Lex(); 2758 } 2759 2760 return TokError("too many positional arguments"); 2761 } 2762 2763 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2764 // Arbitrarily limit macro nesting depth (default matches 'as'). We can 2765 // eliminate this, although we should protect against infinite loops. 2766 unsigned MaxNestingDepth = AsmMacroMaxNestingDepth; 2767 if (ActiveMacros.size() == MaxNestingDepth) { 2768 std::ostringstream MaxNestingDepthError; 2769 MaxNestingDepthError << "macros cannot be nested more than " 2770 << MaxNestingDepth << " levels deep." 2771 << " Use -asm-macro-max-nesting-depth to increase " 2772 "this limit."; 2773 return TokError(MaxNestingDepthError.str()); 2774 } 2775 2776 MCAsmMacroArguments A; 2777 if (parseMacroArguments(M, A)) 2778 return true; 2779 2780 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2781 // to hold the macro body with substitutions. 2782 SmallString<256> Buf; 2783 StringRef Body = M->Body; 2784 raw_svector_ostream OS(Buf); 2785 2786 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2787 return true; 2788 2789 // We include the .endmacro in the buffer as our cue to exit the macro 2790 // instantiation. 2791 OS << ".endmacro\n"; 2792 2793 std::unique_ptr<MemoryBuffer> Instantiation = 2794 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2795 2796 // Create the macro instantiation object and add to the current macro 2797 // instantiation stack. 2798 MacroInstantiation *MI = new MacroInstantiation{ 2799 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 2800 ActiveMacros.push_back(MI); 2801 2802 ++NumOfMacroInstantiations; 2803 2804 // Jump to the macro instantiation and prime the lexer. 2805 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2806 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2807 Lex(); 2808 2809 return false; 2810 } 2811 2812 void AsmParser::handleMacroExit() { 2813 // Jump to the EndOfStatement we should return to, and consume it. 2814 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2815 Lex(); 2816 2817 // Pop the instantiation entry. 2818 delete ActiveMacros.back(); 2819 ActiveMacros.pop_back(); 2820 } 2821 2822 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2823 bool NoDeadStrip) { 2824 MCSymbol *Sym; 2825 const MCExpr *Value; 2826 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2827 Value)) 2828 return true; 2829 2830 if (!Sym) { 2831 // In the case where we parse an expression starting with a '.', we will 2832 // not generate an error, nor will we create a symbol. In this case we 2833 // should just return out. 2834 return false; 2835 } 2836 2837 // Do the assignment. 2838 Out.emitAssignment(Sym, Value); 2839 if (NoDeadStrip) 2840 Out.emitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2841 2842 return false; 2843 } 2844 2845 /// parseIdentifier: 2846 /// ::= identifier 2847 /// ::= string 2848 bool AsmParser::parseIdentifier(StringRef &Res) { 2849 // The assembler has relaxed rules for accepting identifiers, in particular we 2850 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2851 // separate tokens. At this level, we have already lexed so we cannot (currently) 2852 // handle this as a context dependent token, instead we detect adjacent tokens 2853 // and return the combined identifier. 2854 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2855 SMLoc PrefixLoc = getLexer().getLoc(); 2856 2857 // Consume the prefix character, and check for a following identifier. 2858 2859 AsmToken Buf[1]; 2860 Lexer.peekTokens(Buf, false); 2861 2862 if (Buf[0].isNot(AsmToken::Identifier) && Buf[0].isNot(AsmToken::Integer)) 2863 return true; 2864 2865 // We have a '$' or '@' followed by an identifier or integer token, make 2866 // sure they are adjacent. 2867 if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer()) 2868 return true; 2869 2870 // eat $ or @ 2871 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2872 // Construct the joined identifier and consume the token. 2873 Res = StringRef(PrefixLoc.getPointer(), getTok().getString().size() + 1); 2874 Lex(); // Parser Lex to maintain invariants. 2875 return false; 2876 } 2877 2878 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2879 return true; 2880 2881 Res = getTok().getIdentifier(); 2882 2883 Lex(); // Consume the identifier token. 2884 2885 return false; 2886 } 2887 2888 /// parseDirectiveSet: 2889 /// ::= .equ identifier ',' expression 2890 /// ::= .equiv identifier ',' expression 2891 /// ::= .set identifier ',' expression 2892 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2893 StringRef Name; 2894 if (check(parseIdentifier(Name), "expected identifier") || 2895 parseToken(AsmToken::Comma) || parseAssignment(Name, allow_redef, true)) 2896 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2897 return false; 2898 } 2899 2900 bool AsmParser::parseEscapedString(std::string &Data) { 2901 if (check(getTok().isNot(AsmToken::String), "expected string")) 2902 return true; 2903 2904 Data = ""; 2905 StringRef Str = getTok().getStringContents(); 2906 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2907 if (Str[i] != '\\') { 2908 Data += Str[i]; 2909 continue; 2910 } 2911 2912 // Recognize escaped characters. Note that this escape semantics currently 2913 // loosely follows Darwin 'as'. 2914 ++i; 2915 if (i == e) 2916 return TokError("unexpected backslash at end of string"); 2917 2918 // Recognize hex sequences similarly to GNU 'as'. 2919 if (Str[i] == 'x' || Str[i] == 'X') { 2920 size_t length = Str.size(); 2921 if (i + 1 >= length || !isHexDigit(Str[i + 1])) 2922 return TokError("invalid hexadecimal escape sequence"); 2923 2924 // Consume hex characters. GNU 'as' reads all hexadecimal characters and 2925 // then truncates to the lower 16 bits. Seems reasonable. 2926 unsigned Value = 0; 2927 while (i + 1 < length && isHexDigit(Str[i + 1])) 2928 Value = Value * 16 + hexDigitValue(Str[++i]); 2929 2930 Data += (unsigned char)(Value & 0xFF); 2931 continue; 2932 } 2933 2934 // Recognize octal sequences. 2935 if ((unsigned)(Str[i] - '0') <= 7) { 2936 // Consume up to three octal characters. 2937 unsigned Value = Str[i] - '0'; 2938 2939 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2940 ++i; 2941 Value = Value * 8 + (Str[i] - '0'); 2942 2943 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2944 ++i; 2945 Value = Value * 8 + (Str[i] - '0'); 2946 } 2947 } 2948 2949 if (Value > 255) 2950 return TokError("invalid octal escape sequence (out of range)"); 2951 2952 Data += (unsigned char)Value; 2953 continue; 2954 } 2955 2956 // Otherwise recognize individual escapes. 2957 switch (Str[i]) { 2958 default: 2959 // Just reject invalid escape sequences for now. 2960 return TokError("invalid escape sequence (unrecognized character)"); 2961 2962 case 'b': Data += '\b'; break; 2963 case 'f': Data += '\f'; break; 2964 case 'n': Data += '\n'; break; 2965 case 'r': Data += '\r'; break; 2966 case 't': Data += '\t'; break; 2967 case '"': Data += '"'; break; 2968 case '\\': Data += '\\'; break; 2969 } 2970 } 2971 2972 Lex(); 2973 return false; 2974 } 2975 2976 bool AsmParser::parseAngleBracketString(std::string &Data) { 2977 SMLoc EndLoc, StartLoc = getTok().getLoc(); 2978 if (isAngleBracketString(StartLoc, EndLoc)) { 2979 const char *StartChar = StartLoc.getPointer() + 1; 2980 const char *EndChar = EndLoc.getPointer() - 1; 2981 jumpToLoc(EndLoc, CurBuffer); 2982 /// Eat from '<' to '>' 2983 Lex(); 2984 2985 Data = angleBracketString(StringRef(StartChar, EndChar - StartChar)); 2986 return false; 2987 } 2988 return true; 2989 } 2990 2991 /// parseDirectiveAscii: 2992 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2993 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2994 auto parseOp = [&]() -> bool { 2995 std::string Data; 2996 if (checkForValidSection() || parseEscapedString(Data)) 2997 return true; 2998 getStreamer().emitBytes(Data); 2999 if (ZeroTerminated) 3000 getStreamer().emitBytes(StringRef("\0", 1)); 3001 return false; 3002 }; 3003 3004 if (parseMany(parseOp)) 3005 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3006 return false; 3007 } 3008 3009 /// parseDirectiveReloc 3010 /// ::= .reloc expression , identifier [ , expression ] 3011 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 3012 const MCExpr *Offset; 3013 const MCExpr *Expr = nullptr; 3014 int64_t OffsetValue; 3015 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 3016 3017 if (parseExpression(Offset)) 3018 return true; 3019 3020 if ((Offset->evaluateAsAbsolute(OffsetValue, 3021 getStreamer().getAssemblerPtr()) && 3022 check(OffsetValue < 0, OffsetLoc, "expression is negative")) || 3023 (check(Offset->getKind() != llvm::MCExpr::Constant && 3024 Offset->getKind() != llvm::MCExpr::SymbolRef, 3025 OffsetLoc, "expected non-negative number or a label")) || 3026 (parseToken(AsmToken::Comma, "expected comma") || 3027 check(getTok().isNot(AsmToken::Identifier), "expected relocation name"))) 3028 return true; 3029 3030 SMLoc NameLoc = Lexer.getTok().getLoc(); 3031 StringRef Name = Lexer.getTok().getIdentifier(); 3032 Lex(); 3033 3034 if (Lexer.is(AsmToken::Comma)) { 3035 Lex(); 3036 SMLoc ExprLoc = Lexer.getLoc(); 3037 if (parseExpression(Expr)) 3038 return true; 3039 3040 MCValue Value; 3041 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 3042 return Error(ExprLoc, "expression must be relocatable"); 3043 } 3044 3045 if (parseToken(AsmToken::EndOfStatement, 3046 "unexpected token in .reloc directive")) 3047 return true; 3048 3049 const MCTargetAsmParser &MCT = getTargetParser(); 3050 const MCSubtargetInfo &STI = MCT.getSTI(); 3051 if (getStreamer().emitRelocDirective(*Offset, Name, Expr, DirectiveLoc, STI)) 3052 return Error(NameLoc, "unknown relocation name"); 3053 3054 return false; 3055 } 3056 3057 /// parseDirectiveValue 3058 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 3059 bool AsmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) { 3060 auto parseOp = [&]() -> bool { 3061 const MCExpr *Value; 3062 SMLoc ExprLoc = getLexer().getLoc(); 3063 if (checkForValidSection() || parseExpression(Value)) 3064 return true; 3065 // Special case constant expressions to match code generator. 3066 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3067 assert(Size <= 8 && "Invalid size"); 3068 uint64_t IntValue = MCE->getValue(); 3069 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 3070 return Error(ExprLoc, "out of range literal value"); 3071 getStreamer().emitIntValue(IntValue, Size); 3072 } else 3073 getStreamer().emitValue(Value, Size, ExprLoc); 3074 return false; 3075 }; 3076 3077 if (parseMany(parseOp)) 3078 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3079 return false; 3080 } 3081 3082 static bool parseHexOcta(AsmParser &Asm, uint64_t &hi, uint64_t &lo) { 3083 if (Asm.getTok().isNot(AsmToken::Integer) && 3084 Asm.getTok().isNot(AsmToken::BigNum)) 3085 return Asm.TokError("unknown token in expression"); 3086 SMLoc ExprLoc = Asm.getTok().getLoc(); 3087 APInt IntValue = Asm.getTok().getAPIntVal(); 3088 Asm.Lex(); 3089 if (!IntValue.isIntN(128)) 3090 return Asm.Error(ExprLoc, "out of range literal value"); 3091 if (!IntValue.isIntN(64)) { 3092 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 3093 lo = IntValue.getLoBits(64).getZExtValue(); 3094 } else { 3095 hi = 0; 3096 lo = IntValue.getZExtValue(); 3097 } 3098 return false; 3099 } 3100 3101 /// ParseDirectiveOctaValue 3102 /// ::= .octa [ hexconstant (, hexconstant)* ] 3103 3104 bool AsmParser::parseDirectiveOctaValue(StringRef IDVal) { 3105 auto parseOp = [&]() -> bool { 3106 if (checkForValidSection()) 3107 return true; 3108 uint64_t hi, lo; 3109 if (parseHexOcta(*this, hi, lo)) 3110 return true; 3111 if (MAI.isLittleEndian()) { 3112 getStreamer().emitInt64(lo); 3113 getStreamer().emitInt64(hi); 3114 } else { 3115 getStreamer().emitInt64(hi); 3116 getStreamer().emitInt64(lo); 3117 } 3118 return false; 3119 }; 3120 3121 if (parseMany(parseOp)) 3122 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3123 return false; 3124 } 3125 3126 bool AsmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) { 3127 // We don't truly support arithmetic on floating point expressions, so we 3128 // have to manually parse unary prefixes. 3129 bool IsNeg = false; 3130 if (getLexer().is(AsmToken::Minus)) { 3131 Lexer.Lex(); 3132 IsNeg = true; 3133 } else if (getLexer().is(AsmToken::Plus)) 3134 Lexer.Lex(); 3135 3136 if (Lexer.is(AsmToken::Error)) 3137 return TokError(Lexer.getErr()); 3138 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 3139 Lexer.isNot(AsmToken::Identifier)) 3140 return TokError("unexpected token in directive"); 3141 3142 // Convert to an APFloat. 3143 APFloat Value(Semantics); 3144 StringRef IDVal = getTok().getString(); 3145 if (getLexer().is(AsmToken::Identifier)) { 3146 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 3147 Value = APFloat::getInf(Semantics); 3148 else if (!IDVal.compare_lower("nan")) 3149 Value = APFloat::getNaN(Semantics, false, ~0); 3150 else 3151 return TokError("invalid floating point literal"); 3152 } else if (errorToBool( 3153 Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) 3154 .takeError())) 3155 return TokError("invalid floating point literal"); 3156 if (IsNeg) 3157 Value.changeSign(); 3158 3159 // Consume the numeric token. 3160 Lex(); 3161 3162 Res = Value.bitcastToAPInt(); 3163 3164 return false; 3165 } 3166 3167 /// parseDirectiveRealValue 3168 /// ::= (.single | .double) [ expression (, expression)* ] 3169 bool AsmParser::parseDirectiveRealValue(StringRef IDVal, 3170 const fltSemantics &Semantics) { 3171 auto parseOp = [&]() -> bool { 3172 APInt AsInt; 3173 if (checkForValidSection() || parseRealValue(Semantics, AsInt)) 3174 return true; 3175 getStreamer().emitIntValue(AsInt.getLimitedValue(), 3176 AsInt.getBitWidth() / 8); 3177 return false; 3178 }; 3179 3180 if (parseMany(parseOp)) 3181 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3182 return false; 3183 } 3184 3185 /// parseDirectiveZero 3186 /// ::= .zero expression 3187 bool AsmParser::parseDirectiveZero() { 3188 SMLoc NumBytesLoc = Lexer.getLoc(); 3189 const MCExpr *NumBytes; 3190 if (checkForValidSection() || parseExpression(NumBytes)) 3191 return true; 3192 3193 int64_t Val = 0; 3194 if (getLexer().is(AsmToken::Comma)) { 3195 Lex(); 3196 if (parseAbsoluteExpression(Val)) 3197 return true; 3198 } 3199 3200 if (parseToken(AsmToken::EndOfStatement, 3201 "unexpected token in '.zero' directive")) 3202 return true; 3203 getStreamer().emitFill(*NumBytes, Val, NumBytesLoc); 3204 3205 return false; 3206 } 3207 3208 /// parseDirectiveFill 3209 /// ::= .fill expression [ , expression [ , expression ] ] 3210 bool AsmParser::parseDirectiveFill() { 3211 SMLoc NumValuesLoc = Lexer.getLoc(); 3212 const MCExpr *NumValues; 3213 if (checkForValidSection() || parseExpression(NumValues)) 3214 return true; 3215 3216 int64_t FillSize = 1; 3217 int64_t FillExpr = 0; 3218 3219 SMLoc SizeLoc, ExprLoc; 3220 3221 if (parseOptionalToken(AsmToken::Comma)) { 3222 SizeLoc = getTok().getLoc(); 3223 if (parseAbsoluteExpression(FillSize)) 3224 return true; 3225 if (parseOptionalToken(AsmToken::Comma)) { 3226 ExprLoc = getTok().getLoc(); 3227 if (parseAbsoluteExpression(FillExpr)) 3228 return true; 3229 } 3230 } 3231 if (parseToken(AsmToken::EndOfStatement, 3232 "unexpected token in '.fill' directive")) 3233 return true; 3234 3235 if (FillSize < 0) { 3236 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 3237 return false; 3238 } 3239 if (FillSize > 8) { 3240 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 3241 FillSize = 8; 3242 } 3243 3244 if (!isUInt<32>(FillExpr) && FillSize > 4) 3245 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 3246 3247 getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc); 3248 3249 return false; 3250 } 3251 3252 /// parseDirectiveOrg 3253 /// ::= .org expression [ , expression ] 3254 bool AsmParser::parseDirectiveOrg() { 3255 const MCExpr *Offset; 3256 SMLoc OffsetLoc = Lexer.getLoc(); 3257 if (checkForValidSection() || parseExpression(Offset)) 3258 return true; 3259 3260 // Parse optional fill expression. 3261 int64_t FillExpr = 0; 3262 if (parseOptionalToken(AsmToken::Comma)) 3263 if (parseAbsoluteExpression(FillExpr)) 3264 return addErrorSuffix(" in '.org' directive"); 3265 if (parseToken(AsmToken::EndOfStatement)) 3266 return addErrorSuffix(" in '.org' directive"); 3267 3268 getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc); 3269 return false; 3270 } 3271 3272 /// parseDirectiveAlign 3273 /// ::= {.align, ...} expression [ , expression [ , expression ]] 3274 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 3275 SMLoc AlignmentLoc = getLexer().getLoc(); 3276 int64_t Alignment; 3277 SMLoc MaxBytesLoc; 3278 bool HasFillExpr = false; 3279 int64_t FillExpr = 0; 3280 int64_t MaxBytesToFill = 0; 3281 3282 auto parseAlign = [&]() -> bool { 3283 if (parseAbsoluteExpression(Alignment)) 3284 return true; 3285 if (parseOptionalToken(AsmToken::Comma)) { 3286 // The fill expression can be omitted while specifying a maximum number of 3287 // alignment bytes, e.g: 3288 // .align 3,,4 3289 if (getTok().isNot(AsmToken::Comma)) { 3290 HasFillExpr = true; 3291 if (parseAbsoluteExpression(FillExpr)) 3292 return true; 3293 } 3294 if (parseOptionalToken(AsmToken::Comma)) 3295 if (parseTokenLoc(MaxBytesLoc) || 3296 parseAbsoluteExpression(MaxBytesToFill)) 3297 return true; 3298 } 3299 return parseToken(AsmToken::EndOfStatement); 3300 }; 3301 3302 if (checkForValidSection()) 3303 return addErrorSuffix(" in directive"); 3304 // Ignore empty '.p2align' directives for GNU-as compatibility 3305 if (IsPow2 && (ValueSize == 1) && getTok().is(AsmToken::EndOfStatement)) { 3306 Warning(AlignmentLoc, "p2align directive with no operand(s) is ignored"); 3307 return parseToken(AsmToken::EndOfStatement); 3308 } 3309 if (parseAlign()) 3310 return addErrorSuffix(" in directive"); 3311 3312 // Always emit an alignment here even if we thrown an error. 3313 bool ReturnVal = false; 3314 3315 // Compute alignment in bytes. 3316 if (IsPow2) { 3317 // FIXME: Diagnose overflow. 3318 if (Alignment >= 32) { 3319 ReturnVal |= Error(AlignmentLoc, "invalid alignment value"); 3320 Alignment = 31; 3321 } 3322 3323 Alignment = 1ULL << Alignment; 3324 } else { 3325 // Reject alignments that aren't either a power of two or zero, 3326 // for gas compatibility. Alignment of zero is silently rounded 3327 // up to one. 3328 if (Alignment == 0) 3329 Alignment = 1; 3330 if (!isPowerOf2_64(Alignment)) 3331 ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2"); 3332 } 3333 3334 // Diagnose non-sensical max bytes to align. 3335 if (MaxBytesLoc.isValid()) { 3336 if (MaxBytesToFill < 1) { 3337 ReturnVal |= Error(MaxBytesLoc, 3338 "alignment directive can never be satisfied in this " 3339 "many bytes, ignoring maximum bytes expression"); 3340 MaxBytesToFill = 0; 3341 } 3342 3343 if (MaxBytesToFill >= Alignment) { 3344 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 3345 "has no effect"); 3346 MaxBytesToFill = 0; 3347 } 3348 } 3349 3350 // Check whether we should use optimal code alignment for this .align 3351 // directive. 3352 const MCSection *Section = getStreamer().getCurrentSectionOnly(); 3353 assert(Section && "must have section to emit alignment"); 3354 bool UseCodeAlign = Section->UseCodeAlign(); 3355 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 3356 ValueSize == 1 && UseCodeAlign) { 3357 getStreamer().emitCodeAlignment(Alignment, MaxBytesToFill); 3358 } else { 3359 // FIXME: Target specific behavior about how the "extra" bytes are filled. 3360 getStreamer().emitValueToAlignment(Alignment, FillExpr, ValueSize, 3361 MaxBytesToFill); 3362 } 3363 3364 return ReturnVal; 3365 } 3366 3367 /// parseDirectiveFile 3368 /// ::= .file filename 3369 /// ::= .file number [directory] filename [md5 checksum] [source source-text] 3370 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 3371 // FIXME: I'm not sure what this is. 3372 int64_t FileNumber = -1; 3373 if (getLexer().is(AsmToken::Integer)) { 3374 FileNumber = getTok().getIntVal(); 3375 Lex(); 3376 3377 if (FileNumber < 0) 3378 return TokError("negative file number"); 3379 } 3380 3381 std::string Path; 3382 3383 // Usually the directory and filename together, otherwise just the directory. 3384 // Allow the strings to have escaped octal character sequence. 3385 if (check(getTok().isNot(AsmToken::String), 3386 "unexpected token in '.file' directive") || 3387 parseEscapedString(Path)) 3388 return true; 3389 3390 StringRef Directory; 3391 StringRef Filename; 3392 std::string FilenameData; 3393 if (getLexer().is(AsmToken::String)) { 3394 if (check(FileNumber == -1, 3395 "explicit path specified, but no file number") || 3396 parseEscapedString(FilenameData)) 3397 return true; 3398 Filename = FilenameData; 3399 Directory = Path; 3400 } else { 3401 Filename = Path; 3402 } 3403 3404 uint64_t MD5Hi, MD5Lo; 3405 bool HasMD5 = false; 3406 3407 Optional<StringRef> Source; 3408 bool HasSource = false; 3409 std::string SourceString; 3410 3411 while (!parseOptionalToken(AsmToken::EndOfStatement)) { 3412 StringRef Keyword; 3413 if (check(getTok().isNot(AsmToken::Identifier), 3414 "unexpected token in '.file' directive") || 3415 parseIdentifier(Keyword)) 3416 return true; 3417 if (Keyword == "md5") { 3418 HasMD5 = true; 3419 if (check(FileNumber == -1, 3420 "MD5 checksum specified, but no file number") || 3421 parseHexOcta(*this, MD5Hi, MD5Lo)) 3422 return true; 3423 } else if (Keyword == "source") { 3424 HasSource = true; 3425 if (check(FileNumber == -1, 3426 "source specified, but no file number") || 3427 check(getTok().isNot(AsmToken::String), 3428 "unexpected token in '.file' directive") || 3429 parseEscapedString(SourceString)) 3430 return true; 3431 } else { 3432 return TokError("unexpected token in '.file' directive"); 3433 } 3434 } 3435 3436 if (FileNumber == -1) { 3437 // Ignore the directive if there is no number and the target doesn't support 3438 // numberless .file directives. This allows some portability of assembler 3439 // between different object file formats. 3440 if (getContext().getAsmInfo()->hasSingleParameterDotFile()) 3441 getStreamer().emitFileDirective(Filename); 3442 } else { 3443 // In case there is a -g option as well as debug info from directive .file, 3444 // we turn off the -g option, directly use the existing debug info instead. 3445 // Throw away any implicit file table for the assembler source. 3446 if (Ctx.getGenDwarfForAssembly()) { 3447 Ctx.getMCDwarfLineTable(0).resetFileTable(); 3448 Ctx.setGenDwarfForAssembly(false); 3449 } 3450 3451 Optional<MD5::MD5Result> CKMem; 3452 if (HasMD5) { 3453 MD5::MD5Result Sum; 3454 for (unsigned i = 0; i != 8; ++i) { 3455 Sum.Bytes[i] = uint8_t(MD5Hi >> ((7 - i) * 8)); 3456 Sum.Bytes[i + 8] = uint8_t(MD5Lo >> ((7 - i) * 8)); 3457 } 3458 CKMem = Sum; 3459 } 3460 if (HasSource) { 3461 char *SourceBuf = static_cast<char *>(Ctx.allocate(SourceString.size())); 3462 memcpy(SourceBuf, SourceString.data(), SourceString.size()); 3463 Source = StringRef(SourceBuf, SourceString.size()); 3464 } 3465 if (FileNumber == 0) { 3466 if (Ctx.getDwarfVersion() < 5) 3467 return Warning(DirectiveLoc, "file 0 not supported prior to DWARF-5"); 3468 getStreamer().emitDwarfFile0Directive(Directory, Filename, CKMem, Source); 3469 } else { 3470 Expected<unsigned> FileNumOrErr = getStreamer().tryEmitDwarfFileDirective( 3471 FileNumber, Directory, Filename, CKMem, Source); 3472 if (!FileNumOrErr) 3473 return Error(DirectiveLoc, toString(FileNumOrErr.takeError())); 3474 } 3475 // Alert the user if there are some .file directives with MD5 and some not. 3476 // But only do that once. 3477 if (!ReportedInconsistentMD5 && !Ctx.isDwarfMD5UsageConsistent(0)) { 3478 ReportedInconsistentMD5 = true; 3479 return Warning(DirectiveLoc, "inconsistent use of MD5 checksums"); 3480 } 3481 } 3482 3483 return false; 3484 } 3485 3486 /// parseDirectiveLine 3487 /// ::= .line [number] 3488 bool AsmParser::parseDirectiveLine() { 3489 int64_t LineNumber; 3490 if (getLexer().is(AsmToken::Integer)) { 3491 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3492 return true; 3493 (void)LineNumber; 3494 // FIXME: Do something with the .line. 3495 } 3496 if (parseToken(AsmToken::EndOfStatement, 3497 "unexpected token in '.line' directive")) 3498 return true; 3499 3500 return false; 3501 } 3502 3503 /// parseDirectiveLoc 3504 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3505 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3506 /// The first number is a file number, must have been previously assigned with 3507 /// a .file directive, the second number is the line number and optionally the 3508 /// third number is a column position (zero if not specified). The remaining 3509 /// optional items are .loc sub-directives. 3510 bool AsmParser::parseDirectiveLoc() { 3511 int64_t FileNumber = 0, LineNumber = 0; 3512 SMLoc Loc = getTok().getLoc(); 3513 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3514 check(FileNumber < 1 && Ctx.getDwarfVersion() < 5, Loc, 3515 "file number less than one in '.loc' directive") || 3516 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3517 "unassigned file number in '.loc' directive")) 3518 return true; 3519 3520 // optional 3521 if (getLexer().is(AsmToken::Integer)) { 3522 LineNumber = getTok().getIntVal(); 3523 if (LineNumber < 0) 3524 return TokError("line number less than zero in '.loc' directive"); 3525 Lex(); 3526 } 3527 3528 int64_t ColumnPos = 0; 3529 if (getLexer().is(AsmToken::Integer)) { 3530 ColumnPos = getTok().getIntVal(); 3531 if (ColumnPos < 0) 3532 return TokError("column position less than zero in '.loc' directive"); 3533 Lex(); 3534 } 3535 3536 auto PrevFlags = getContext().getCurrentDwarfLoc().getFlags(); 3537 unsigned Flags = PrevFlags & DWARF2_FLAG_IS_STMT; 3538 unsigned Isa = 0; 3539 int64_t Discriminator = 0; 3540 3541 auto parseLocOp = [&]() -> bool { 3542 StringRef Name; 3543 SMLoc Loc = getTok().getLoc(); 3544 if (parseIdentifier(Name)) 3545 return TokError("unexpected token in '.loc' directive"); 3546 3547 if (Name == "basic_block") 3548 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3549 else if (Name == "prologue_end") 3550 Flags |= DWARF2_FLAG_PROLOGUE_END; 3551 else if (Name == "epilogue_begin") 3552 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3553 else if (Name == "is_stmt") { 3554 Loc = getTok().getLoc(); 3555 const MCExpr *Value; 3556 if (parseExpression(Value)) 3557 return true; 3558 // The expression must be the constant 0 or 1. 3559 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3560 int Value = MCE->getValue(); 3561 if (Value == 0) 3562 Flags &= ~DWARF2_FLAG_IS_STMT; 3563 else if (Value == 1) 3564 Flags |= DWARF2_FLAG_IS_STMT; 3565 else 3566 return Error(Loc, "is_stmt value not 0 or 1"); 3567 } else { 3568 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3569 } 3570 } else if (Name == "isa") { 3571 Loc = getTok().getLoc(); 3572 const MCExpr *Value; 3573 if (parseExpression(Value)) 3574 return true; 3575 // The expression must be a constant greater or equal to 0. 3576 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3577 int Value = MCE->getValue(); 3578 if (Value < 0) 3579 return Error(Loc, "isa number less than zero"); 3580 Isa = Value; 3581 } else { 3582 return Error(Loc, "isa number not a constant value"); 3583 } 3584 } else if (Name == "discriminator") { 3585 if (parseAbsoluteExpression(Discriminator)) 3586 return true; 3587 } else { 3588 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3589 } 3590 return false; 3591 }; 3592 3593 if (parseMany(parseLocOp, false /*hasComma*/)) 3594 return true; 3595 3596 getStreamer().emitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3597 Isa, Discriminator, StringRef()); 3598 3599 return false; 3600 } 3601 3602 /// parseDirectiveStabs 3603 /// ::= .stabs string, number, number, number 3604 bool AsmParser::parseDirectiveStabs() { 3605 return TokError("unsupported directive '.stabs'"); 3606 } 3607 3608 /// parseDirectiveCVFile 3609 /// ::= .cv_file number filename [checksum] [checksumkind] 3610 bool AsmParser::parseDirectiveCVFile() { 3611 SMLoc FileNumberLoc = getTok().getLoc(); 3612 int64_t FileNumber; 3613 std::string Filename; 3614 std::string Checksum; 3615 int64_t ChecksumKind = 0; 3616 3617 if (parseIntToken(FileNumber, 3618 "expected file number in '.cv_file' directive") || 3619 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3620 check(getTok().isNot(AsmToken::String), 3621 "unexpected token in '.cv_file' directive") || 3622 parseEscapedString(Filename)) 3623 return true; 3624 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3625 if (check(getTok().isNot(AsmToken::String), 3626 "unexpected token in '.cv_file' directive") || 3627 parseEscapedString(Checksum) || 3628 parseIntToken(ChecksumKind, 3629 "expected checksum kind in '.cv_file' directive") || 3630 parseToken(AsmToken::EndOfStatement, 3631 "unexpected token in '.cv_file' directive")) 3632 return true; 3633 } 3634 3635 Checksum = fromHex(Checksum); 3636 void *CKMem = Ctx.allocate(Checksum.size(), 1); 3637 memcpy(CKMem, Checksum.data(), Checksum.size()); 3638 ArrayRef<uint8_t> ChecksumAsBytes(reinterpret_cast<const uint8_t *>(CKMem), 3639 Checksum.size()); 3640 3641 if (!getStreamer().EmitCVFileDirective(FileNumber, Filename, ChecksumAsBytes, 3642 static_cast<uint8_t>(ChecksumKind))) 3643 return Error(FileNumberLoc, "file number already allocated"); 3644 3645 return false; 3646 } 3647 3648 bool AsmParser::parseCVFunctionId(int64_t &FunctionId, 3649 StringRef DirectiveName) { 3650 SMLoc Loc; 3651 return parseTokenLoc(Loc) || 3652 parseIntToken(FunctionId, "expected function id in '" + DirectiveName + 3653 "' directive") || 3654 check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc, 3655 "expected function id within range [0, UINT_MAX)"); 3656 } 3657 3658 bool AsmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) { 3659 SMLoc Loc; 3660 return parseTokenLoc(Loc) || 3661 parseIntToken(FileNumber, "expected integer in '" + DirectiveName + 3662 "' directive") || 3663 check(FileNumber < 1, Loc, "file number less than one in '" + 3664 DirectiveName + "' directive") || 3665 check(!getCVContext().isValidFileNumber(FileNumber), Loc, 3666 "unassigned file number in '" + DirectiveName + "' directive"); 3667 } 3668 3669 /// parseDirectiveCVFuncId 3670 /// ::= .cv_func_id FunctionId 3671 /// 3672 /// Introduces a function ID that can be used with .cv_loc. 3673 bool AsmParser::parseDirectiveCVFuncId() { 3674 SMLoc FunctionIdLoc = getTok().getLoc(); 3675 int64_t FunctionId; 3676 3677 if (parseCVFunctionId(FunctionId, ".cv_func_id") || 3678 parseToken(AsmToken::EndOfStatement, 3679 "unexpected token in '.cv_func_id' directive")) 3680 return true; 3681 3682 if (!getStreamer().EmitCVFuncIdDirective(FunctionId)) 3683 return Error(FunctionIdLoc, "function id already allocated"); 3684 3685 return false; 3686 } 3687 3688 /// parseDirectiveCVInlineSiteId 3689 /// ::= .cv_inline_site_id FunctionId 3690 /// "within" IAFunc 3691 /// "inlined_at" IAFile IALine [IACol] 3692 /// 3693 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined 3694 /// at" source location information for use in the line table of the caller, 3695 /// whether the caller is a real function or another inlined call site. 3696 bool AsmParser::parseDirectiveCVInlineSiteId() { 3697 SMLoc FunctionIdLoc = getTok().getLoc(); 3698 int64_t FunctionId; 3699 int64_t IAFunc; 3700 int64_t IAFile; 3701 int64_t IALine; 3702 int64_t IACol = 0; 3703 3704 // FunctionId 3705 if (parseCVFunctionId(FunctionId, ".cv_inline_site_id")) 3706 return true; 3707 3708 // "within" 3709 if (check((getLexer().isNot(AsmToken::Identifier) || 3710 getTok().getIdentifier() != "within"), 3711 "expected 'within' identifier in '.cv_inline_site_id' directive")) 3712 return true; 3713 Lex(); 3714 3715 // IAFunc 3716 if (parseCVFunctionId(IAFunc, ".cv_inline_site_id")) 3717 return true; 3718 3719 // "inlined_at" 3720 if (check((getLexer().isNot(AsmToken::Identifier) || 3721 getTok().getIdentifier() != "inlined_at"), 3722 "expected 'inlined_at' identifier in '.cv_inline_site_id' " 3723 "directive") ) 3724 return true; 3725 Lex(); 3726 3727 // IAFile IALine 3728 if (parseCVFileId(IAFile, ".cv_inline_site_id") || 3729 parseIntToken(IALine, "expected line number after 'inlined_at'")) 3730 return true; 3731 3732 // [IACol] 3733 if (getLexer().is(AsmToken::Integer)) { 3734 IACol = getTok().getIntVal(); 3735 Lex(); 3736 } 3737 3738 if (parseToken(AsmToken::EndOfStatement, 3739 "unexpected token in '.cv_inline_site_id' directive")) 3740 return true; 3741 3742 if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile, 3743 IALine, IACol, FunctionIdLoc)) 3744 return Error(FunctionIdLoc, "function id already allocated"); 3745 3746 return false; 3747 } 3748 3749 /// parseDirectiveCVLoc 3750 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3751 /// [is_stmt VALUE] 3752 /// The first number is a file number, must have been previously assigned with 3753 /// a .file directive, the second number is the line number and optionally the 3754 /// third number is a column position (zero if not specified). The remaining 3755 /// optional items are .loc sub-directives. 3756 bool AsmParser::parseDirectiveCVLoc() { 3757 SMLoc DirectiveLoc = getTok().getLoc(); 3758 int64_t FunctionId, FileNumber; 3759 if (parseCVFunctionId(FunctionId, ".cv_loc") || 3760 parseCVFileId(FileNumber, ".cv_loc")) 3761 return true; 3762 3763 int64_t LineNumber = 0; 3764 if (getLexer().is(AsmToken::Integer)) { 3765 LineNumber = getTok().getIntVal(); 3766 if (LineNumber < 0) 3767 return TokError("line number less than zero in '.cv_loc' directive"); 3768 Lex(); 3769 } 3770 3771 int64_t ColumnPos = 0; 3772 if (getLexer().is(AsmToken::Integer)) { 3773 ColumnPos = getTok().getIntVal(); 3774 if (ColumnPos < 0) 3775 return TokError("column position less than zero in '.cv_loc' directive"); 3776 Lex(); 3777 } 3778 3779 bool PrologueEnd = false; 3780 uint64_t IsStmt = 0; 3781 3782 auto parseOp = [&]() -> bool { 3783 StringRef Name; 3784 SMLoc Loc = getTok().getLoc(); 3785 if (parseIdentifier(Name)) 3786 return TokError("unexpected token in '.cv_loc' directive"); 3787 if (Name == "prologue_end") 3788 PrologueEnd = true; 3789 else if (Name == "is_stmt") { 3790 Loc = getTok().getLoc(); 3791 const MCExpr *Value; 3792 if (parseExpression(Value)) 3793 return true; 3794 // The expression must be the constant 0 or 1. 3795 IsStmt = ~0ULL; 3796 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3797 IsStmt = MCE->getValue(); 3798 3799 if (IsStmt > 1) 3800 return Error(Loc, "is_stmt value not 0 or 1"); 3801 } else { 3802 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3803 } 3804 return false; 3805 }; 3806 3807 if (parseMany(parseOp, false /*hasComma*/)) 3808 return true; 3809 3810 getStreamer().emitCVLocDirective(FunctionId, FileNumber, LineNumber, 3811 ColumnPos, PrologueEnd, IsStmt, StringRef(), 3812 DirectiveLoc); 3813 return false; 3814 } 3815 3816 /// parseDirectiveCVLinetable 3817 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3818 bool AsmParser::parseDirectiveCVLinetable() { 3819 int64_t FunctionId; 3820 StringRef FnStartName, FnEndName; 3821 SMLoc Loc = getTok().getLoc(); 3822 if (parseCVFunctionId(FunctionId, ".cv_linetable") || 3823 parseToken(AsmToken::Comma, 3824 "unexpected token in '.cv_linetable' directive") || 3825 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3826 "expected identifier in directive") || 3827 parseToken(AsmToken::Comma, 3828 "unexpected token in '.cv_linetable' directive") || 3829 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3830 "expected identifier in directive")) 3831 return true; 3832 3833 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3834 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3835 3836 getStreamer().emitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3837 return false; 3838 } 3839 3840 /// parseDirectiveCVInlineLinetable 3841 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3842 bool AsmParser::parseDirectiveCVInlineLinetable() { 3843 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3844 StringRef FnStartName, FnEndName; 3845 SMLoc Loc = getTok().getLoc(); 3846 if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") || 3847 parseTokenLoc(Loc) || 3848 parseIntToken( 3849 SourceFileId, 3850 "expected SourceField in '.cv_inline_linetable' directive") || 3851 check(SourceFileId <= 0, Loc, 3852 "File id less than zero in '.cv_inline_linetable' directive") || 3853 parseTokenLoc(Loc) || 3854 parseIntToken( 3855 SourceLineNum, 3856 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3857 check(SourceLineNum < 0, Loc, 3858 "Line number less than zero in '.cv_inline_linetable' directive") || 3859 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3860 "expected identifier in directive") || 3861 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3862 "expected identifier in directive")) 3863 return true; 3864 3865 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3866 return true; 3867 3868 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3869 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3870 getStreamer().emitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3871 SourceLineNum, FnStartSym, 3872 FnEndSym); 3873 return false; 3874 } 3875 3876 void AsmParser::initializeCVDefRangeTypeMap() { 3877 CVDefRangeTypeMap["reg"] = CVDR_DEFRANGE_REGISTER; 3878 CVDefRangeTypeMap["frame_ptr_rel"] = CVDR_DEFRANGE_FRAMEPOINTER_REL; 3879 CVDefRangeTypeMap["subfield_reg"] = CVDR_DEFRANGE_SUBFIELD_REGISTER; 3880 CVDefRangeTypeMap["reg_rel"] = CVDR_DEFRANGE_REGISTER_REL; 3881 } 3882 3883 /// parseDirectiveCVDefRange 3884 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3885 bool AsmParser::parseDirectiveCVDefRange() { 3886 SMLoc Loc; 3887 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3888 while (getLexer().is(AsmToken::Identifier)) { 3889 Loc = getLexer().getLoc(); 3890 StringRef GapStartName; 3891 if (parseIdentifier(GapStartName)) 3892 return Error(Loc, "expected identifier in directive"); 3893 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3894 3895 Loc = getLexer().getLoc(); 3896 StringRef GapEndName; 3897 if (parseIdentifier(GapEndName)) 3898 return Error(Loc, "expected identifier in directive"); 3899 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3900 3901 Ranges.push_back({GapStartSym, GapEndSym}); 3902 } 3903 3904 StringRef CVDefRangeTypeStr; 3905 if (parseToken( 3906 AsmToken::Comma, 3907 "expected comma before def_range type in .cv_def_range directive") || 3908 parseIdentifier(CVDefRangeTypeStr)) 3909 return Error(Loc, "expected def_range type in directive"); 3910 3911 StringMap<CVDefRangeType>::const_iterator CVTypeIt = 3912 CVDefRangeTypeMap.find(CVDefRangeTypeStr); 3913 CVDefRangeType CVDRType = (CVTypeIt == CVDefRangeTypeMap.end()) 3914 ? CVDR_DEFRANGE 3915 : CVTypeIt->getValue(); 3916 switch (CVDRType) { 3917 case CVDR_DEFRANGE_REGISTER: { 3918 int64_t DRRegister; 3919 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3920 ".cv_def_range directive") || 3921 parseAbsoluteExpression(DRRegister)) 3922 return Error(Loc, "expected register number"); 3923 3924 codeview::DefRangeRegisterHeader DRHdr; 3925 DRHdr.Register = DRRegister; 3926 DRHdr.MayHaveNoName = 0; 3927 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3928 break; 3929 } 3930 case CVDR_DEFRANGE_FRAMEPOINTER_REL: { 3931 int64_t DROffset; 3932 if (parseToken(AsmToken::Comma, 3933 "expected comma before offset in .cv_def_range directive") || 3934 parseAbsoluteExpression(DROffset)) 3935 return Error(Loc, "expected offset value"); 3936 3937 codeview::DefRangeFramePointerRelHeader DRHdr; 3938 DRHdr.Offset = DROffset; 3939 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3940 break; 3941 } 3942 case CVDR_DEFRANGE_SUBFIELD_REGISTER: { 3943 int64_t DRRegister; 3944 int64_t DROffsetInParent; 3945 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3946 ".cv_def_range directive") || 3947 parseAbsoluteExpression(DRRegister)) 3948 return Error(Loc, "expected register number"); 3949 if (parseToken(AsmToken::Comma, 3950 "expected comma before offset in .cv_def_range directive") || 3951 parseAbsoluteExpression(DROffsetInParent)) 3952 return Error(Loc, "expected offset value"); 3953 3954 codeview::DefRangeSubfieldRegisterHeader DRHdr; 3955 DRHdr.Register = DRRegister; 3956 DRHdr.MayHaveNoName = 0; 3957 DRHdr.OffsetInParent = DROffsetInParent; 3958 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3959 break; 3960 } 3961 case CVDR_DEFRANGE_REGISTER_REL: { 3962 int64_t DRRegister; 3963 int64_t DRFlags; 3964 int64_t DRBasePointerOffset; 3965 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3966 ".cv_def_range directive") || 3967 parseAbsoluteExpression(DRRegister)) 3968 return Error(Loc, "expected register value"); 3969 if (parseToken( 3970 AsmToken::Comma, 3971 "expected comma before flag value in .cv_def_range directive") || 3972 parseAbsoluteExpression(DRFlags)) 3973 return Error(Loc, "expected flag value"); 3974 if (parseToken(AsmToken::Comma, "expected comma before base pointer offset " 3975 "in .cv_def_range directive") || 3976 parseAbsoluteExpression(DRBasePointerOffset)) 3977 return Error(Loc, "expected base pointer offset value"); 3978 3979 codeview::DefRangeRegisterRelHeader DRHdr; 3980 DRHdr.Register = DRRegister; 3981 DRHdr.Flags = DRFlags; 3982 DRHdr.BasePointerOffset = DRBasePointerOffset; 3983 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3984 break; 3985 } 3986 default: 3987 return Error(Loc, "unexpected def_range type in .cv_def_range directive"); 3988 } 3989 return true; 3990 } 3991 3992 /// parseDirectiveCVString 3993 /// ::= .cv_stringtable "string" 3994 bool AsmParser::parseDirectiveCVString() { 3995 std::string Data; 3996 if (checkForValidSection() || parseEscapedString(Data)) 3997 return addErrorSuffix(" in '.cv_string' directive"); 3998 3999 // Put the string in the table and emit the offset. 4000 std::pair<StringRef, unsigned> Insertion = 4001 getCVContext().addToStringTable(Data); 4002 getStreamer().emitInt32(Insertion.second); 4003 return false; 4004 } 4005 4006 /// parseDirectiveCVStringTable 4007 /// ::= .cv_stringtable 4008 bool AsmParser::parseDirectiveCVStringTable() { 4009 getStreamer().emitCVStringTableDirective(); 4010 return false; 4011 } 4012 4013 /// parseDirectiveCVFileChecksums 4014 /// ::= .cv_filechecksums 4015 bool AsmParser::parseDirectiveCVFileChecksums() { 4016 getStreamer().emitCVFileChecksumsDirective(); 4017 return false; 4018 } 4019 4020 /// parseDirectiveCVFileChecksumOffset 4021 /// ::= .cv_filechecksumoffset fileno 4022 bool AsmParser::parseDirectiveCVFileChecksumOffset() { 4023 int64_t FileNo; 4024 if (parseIntToken(FileNo, "expected identifier in directive")) 4025 return true; 4026 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 4027 return true; 4028 getStreamer().emitCVFileChecksumOffsetDirective(FileNo); 4029 return false; 4030 } 4031 4032 /// parseDirectiveCVFPOData 4033 /// ::= .cv_fpo_data procsym 4034 bool AsmParser::parseDirectiveCVFPOData() { 4035 SMLoc DirLoc = getLexer().getLoc(); 4036 StringRef ProcName; 4037 if (parseIdentifier(ProcName)) 4038 return TokError("expected symbol name"); 4039 if (parseEOL("unexpected tokens")) 4040 return addErrorSuffix(" in '.cv_fpo_data' directive"); 4041 MCSymbol *ProcSym = getContext().getOrCreateSymbol(ProcName); 4042 getStreamer().EmitCVFPOData(ProcSym, DirLoc); 4043 return false; 4044 } 4045 4046 /// parseDirectiveCFISections 4047 /// ::= .cfi_sections section [, section] 4048 bool AsmParser::parseDirectiveCFISections() { 4049 StringRef Name; 4050 bool EH = false; 4051 bool Debug = false; 4052 4053 if (parseIdentifier(Name)) 4054 return TokError("Expected an identifier"); 4055 4056 if (Name == ".eh_frame") 4057 EH = true; 4058 else if (Name == ".debug_frame") 4059 Debug = true; 4060 4061 if (getLexer().is(AsmToken::Comma)) { 4062 Lex(); 4063 4064 if (parseIdentifier(Name)) 4065 return TokError("Expected an identifier"); 4066 4067 if (Name == ".eh_frame") 4068 EH = true; 4069 else if (Name == ".debug_frame") 4070 Debug = true; 4071 } 4072 4073 getStreamer().emitCFISections(EH, Debug); 4074 return false; 4075 } 4076 4077 /// parseDirectiveCFIStartProc 4078 /// ::= .cfi_startproc [simple] 4079 bool AsmParser::parseDirectiveCFIStartProc() { 4080 StringRef Simple; 4081 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4082 if (check(parseIdentifier(Simple) || Simple != "simple", 4083 "unexpected token") || 4084 parseToken(AsmToken::EndOfStatement)) 4085 return addErrorSuffix(" in '.cfi_startproc' directive"); 4086 } 4087 4088 // TODO(kristina): Deal with a corner case of incorrect diagnostic context 4089 // being produced if this directive is emitted as part of preprocessor macro 4090 // expansion which can *ONLY* happen if Clang's cc1as is the API consumer. 4091 // Tools like llvm-mc on the other hand are not affected by it, and report 4092 // correct context information. 4093 getStreamer().emitCFIStartProc(!Simple.empty(), Lexer.getLoc()); 4094 return false; 4095 } 4096 4097 /// parseDirectiveCFIEndProc 4098 /// ::= .cfi_endproc 4099 bool AsmParser::parseDirectiveCFIEndProc() { 4100 getStreamer().emitCFIEndProc(); 4101 return false; 4102 } 4103 4104 /// parse register name or number. 4105 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 4106 SMLoc DirectiveLoc) { 4107 unsigned RegNo; 4108 4109 if (getLexer().isNot(AsmToken::Integer)) { 4110 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 4111 return true; 4112 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 4113 } else 4114 return parseAbsoluteExpression(Register); 4115 4116 return false; 4117 } 4118 4119 /// parseDirectiveCFIDefCfa 4120 /// ::= .cfi_def_cfa register, offset 4121 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 4122 int64_t Register = 0, Offset = 0; 4123 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4124 parseToken(AsmToken::Comma, "unexpected token in directive") || 4125 parseAbsoluteExpression(Offset)) 4126 return true; 4127 4128 getStreamer().emitCFIDefCfa(Register, Offset); 4129 return false; 4130 } 4131 4132 /// parseDirectiveCFIDefCfaOffset 4133 /// ::= .cfi_def_cfa_offset offset 4134 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 4135 int64_t Offset = 0; 4136 if (parseAbsoluteExpression(Offset)) 4137 return true; 4138 4139 getStreamer().emitCFIDefCfaOffset(Offset); 4140 return false; 4141 } 4142 4143 /// parseDirectiveCFIRegister 4144 /// ::= .cfi_register register, register 4145 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 4146 int64_t Register1 = 0, Register2 = 0; 4147 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || 4148 parseToken(AsmToken::Comma, "unexpected token in directive") || 4149 parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 4150 return true; 4151 4152 getStreamer().emitCFIRegister(Register1, Register2); 4153 return false; 4154 } 4155 4156 /// parseDirectiveCFIWindowSave 4157 /// ::= .cfi_window_save 4158 bool AsmParser::parseDirectiveCFIWindowSave() { 4159 getStreamer().emitCFIWindowSave(); 4160 return false; 4161 } 4162 4163 /// parseDirectiveCFIAdjustCfaOffset 4164 /// ::= .cfi_adjust_cfa_offset adjustment 4165 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 4166 int64_t Adjustment = 0; 4167 if (parseAbsoluteExpression(Adjustment)) 4168 return true; 4169 4170 getStreamer().emitCFIAdjustCfaOffset(Adjustment); 4171 return false; 4172 } 4173 4174 /// parseDirectiveCFIDefCfaRegister 4175 /// ::= .cfi_def_cfa_register register 4176 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 4177 int64_t Register = 0; 4178 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4179 return true; 4180 4181 getStreamer().emitCFIDefCfaRegister(Register); 4182 return false; 4183 } 4184 4185 /// parseDirectiveCFIOffset 4186 /// ::= .cfi_offset register, offset 4187 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 4188 int64_t Register = 0; 4189 int64_t Offset = 0; 4190 4191 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4192 parseToken(AsmToken::Comma, "unexpected token in directive") || 4193 parseAbsoluteExpression(Offset)) 4194 return true; 4195 4196 getStreamer().emitCFIOffset(Register, Offset); 4197 return false; 4198 } 4199 4200 /// parseDirectiveCFIRelOffset 4201 /// ::= .cfi_rel_offset register, offset 4202 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 4203 int64_t Register = 0, Offset = 0; 4204 4205 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4206 parseToken(AsmToken::Comma, "unexpected token in directive") || 4207 parseAbsoluteExpression(Offset)) 4208 return true; 4209 4210 getStreamer().emitCFIRelOffset(Register, Offset); 4211 return false; 4212 } 4213 4214 static bool isValidEncoding(int64_t Encoding) { 4215 if (Encoding & ~0xff) 4216 return false; 4217 4218 if (Encoding == dwarf::DW_EH_PE_omit) 4219 return true; 4220 4221 const unsigned Format = Encoding & 0xf; 4222 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 4223 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 4224 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 4225 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 4226 return false; 4227 4228 const unsigned Application = Encoding & 0x70; 4229 if (Application != dwarf::DW_EH_PE_absptr && 4230 Application != dwarf::DW_EH_PE_pcrel) 4231 return false; 4232 4233 return true; 4234 } 4235 4236 /// parseDirectiveCFIPersonalityOrLsda 4237 /// IsPersonality true for cfi_personality, false for cfi_lsda 4238 /// ::= .cfi_personality encoding, [symbol_name] 4239 /// ::= .cfi_lsda encoding, [symbol_name] 4240 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 4241 int64_t Encoding = 0; 4242 if (parseAbsoluteExpression(Encoding)) 4243 return true; 4244 if (Encoding == dwarf::DW_EH_PE_omit) 4245 return false; 4246 4247 StringRef Name; 4248 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 4249 parseToken(AsmToken::Comma, "unexpected token in directive") || 4250 check(parseIdentifier(Name), "expected identifier in directive")) 4251 return true; 4252 4253 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4254 4255 if (IsPersonality) 4256 getStreamer().emitCFIPersonality(Sym, Encoding); 4257 else 4258 getStreamer().emitCFILsda(Sym, Encoding); 4259 return false; 4260 } 4261 4262 /// parseDirectiveCFIRememberState 4263 /// ::= .cfi_remember_state 4264 bool AsmParser::parseDirectiveCFIRememberState() { 4265 getStreamer().emitCFIRememberState(); 4266 return false; 4267 } 4268 4269 /// parseDirectiveCFIRestoreState 4270 /// ::= .cfi_remember_state 4271 bool AsmParser::parseDirectiveCFIRestoreState() { 4272 getStreamer().emitCFIRestoreState(); 4273 return false; 4274 } 4275 4276 /// parseDirectiveCFISameValue 4277 /// ::= .cfi_same_value register 4278 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 4279 int64_t Register = 0; 4280 4281 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4282 return true; 4283 4284 getStreamer().emitCFISameValue(Register); 4285 return false; 4286 } 4287 4288 /// parseDirectiveCFIRestore 4289 /// ::= .cfi_restore register 4290 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 4291 int64_t Register = 0; 4292 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4293 return true; 4294 4295 getStreamer().emitCFIRestore(Register); 4296 return false; 4297 } 4298 4299 /// parseDirectiveCFIEscape 4300 /// ::= .cfi_escape expression[,...] 4301 bool AsmParser::parseDirectiveCFIEscape() { 4302 std::string Values; 4303 int64_t CurrValue; 4304 if (parseAbsoluteExpression(CurrValue)) 4305 return true; 4306 4307 Values.push_back((uint8_t)CurrValue); 4308 4309 while (getLexer().is(AsmToken::Comma)) { 4310 Lex(); 4311 4312 if (parseAbsoluteExpression(CurrValue)) 4313 return true; 4314 4315 Values.push_back((uint8_t)CurrValue); 4316 } 4317 4318 getStreamer().emitCFIEscape(Values); 4319 return false; 4320 } 4321 4322 /// parseDirectiveCFIReturnColumn 4323 /// ::= .cfi_return_column register 4324 bool AsmParser::parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc) { 4325 int64_t Register = 0; 4326 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4327 return true; 4328 getStreamer().emitCFIReturnColumn(Register); 4329 return false; 4330 } 4331 4332 /// parseDirectiveCFISignalFrame 4333 /// ::= .cfi_signal_frame 4334 bool AsmParser::parseDirectiveCFISignalFrame() { 4335 if (parseToken(AsmToken::EndOfStatement, 4336 "unexpected token in '.cfi_signal_frame'")) 4337 return true; 4338 4339 getStreamer().emitCFISignalFrame(); 4340 return false; 4341 } 4342 4343 /// parseDirectiveCFIUndefined 4344 /// ::= .cfi_undefined register 4345 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 4346 int64_t Register = 0; 4347 4348 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4349 return true; 4350 4351 getStreamer().emitCFIUndefined(Register); 4352 return false; 4353 } 4354 4355 /// parseDirectiveAltmacro 4356 /// ::= .altmacro 4357 /// ::= .noaltmacro 4358 bool AsmParser::parseDirectiveAltmacro(StringRef Directive) { 4359 if (getLexer().isNot(AsmToken::EndOfStatement)) 4360 return TokError("unexpected token in '" + Directive + "' directive"); 4361 AltMacroMode = (Directive == ".altmacro"); 4362 return false; 4363 } 4364 4365 /// parseDirectiveMacrosOnOff 4366 /// ::= .macros_on 4367 /// ::= .macros_off 4368 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 4369 if (parseToken(AsmToken::EndOfStatement, 4370 "unexpected token in '" + Directive + "' directive")) 4371 return true; 4372 4373 setMacrosEnabled(Directive == ".macros_on"); 4374 return false; 4375 } 4376 4377 /// parseDirectiveMacro 4378 /// ::= .macro name[,] [parameters] 4379 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 4380 StringRef Name; 4381 if (parseIdentifier(Name)) 4382 return TokError("expected identifier in '.macro' directive"); 4383 4384 if (getLexer().is(AsmToken::Comma)) 4385 Lex(); 4386 4387 MCAsmMacroParameters Parameters; 4388 while (getLexer().isNot(AsmToken::EndOfStatement)) { 4389 4390 if (!Parameters.empty() && Parameters.back().Vararg) 4391 return Error(Lexer.getLoc(), "vararg parameter '" + 4392 Parameters.back().Name + 4393 "' should be the last parameter"); 4394 4395 MCAsmMacroParameter Parameter; 4396 if (parseIdentifier(Parameter.Name)) 4397 return TokError("expected identifier in '.macro' directive"); 4398 4399 // Emit an error if two (or more) named parameters share the same name 4400 for (const MCAsmMacroParameter& CurrParam : Parameters) 4401 if (CurrParam.Name.equals(Parameter.Name)) 4402 return TokError("macro '" + Name + "' has multiple parameters" 4403 " named '" + Parameter.Name + "'"); 4404 4405 if (Lexer.is(AsmToken::Colon)) { 4406 Lex(); // consume ':' 4407 4408 SMLoc QualLoc; 4409 StringRef Qualifier; 4410 4411 QualLoc = Lexer.getLoc(); 4412 if (parseIdentifier(Qualifier)) 4413 return Error(QualLoc, "missing parameter qualifier for " 4414 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4415 4416 if (Qualifier == "req") 4417 Parameter.Required = true; 4418 else if (Qualifier == "vararg") 4419 Parameter.Vararg = true; 4420 else 4421 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 4422 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 4423 } 4424 4425 if (getLexer().is(AsmToken::Equal)) { 4426 Lex(); 4427 4428 SMLoc ParamLoc; 4429 4430 ParamLoc = Lexer.getLoc(); 4431 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 4432 return true; 4433 4434 if (Parameter.Required) 4435 Warning(ParamLoc, "pointless default value for required parameter " 4436 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4437 } 4438 4439 Parameters.push_back(std::move(Parameter)); 4440 4441 if (getLexer().is(AsmToken::Comma)) 4442 Lex(); 4443 } 4444 4445 // Eat just the end of statement. 4446 Lexer.Lex(); 4447 4448 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 4449 AsmToken EndToken, StartToken = getTok(); 4450 unsigned MacroDepth = 0; 4451 // Lex the macro definition. 4452 while (true) { 4453 // Ignore Lexing errors in macros. 4454 while (Lexer.is(AsmToken::Error)) { 4455 Lexer.Lex(); 4456 } 4457 4458 // Check whether we have reached the end of the file. 4459 if (getLexer().is(AsmToken::Eof)) 4460 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 4461 4462 // Otherwise, check whether we have reach the .endmacro or the start of a 4463 // preprocessor line marker. 4464 if (getLexer().is(AsmToken::Identifier)) { 4465 if (getTok().getIdentifier() == ".endm" || 4466 getTok().getIdentifier() == ".endmacro") { 4467 if (MacroDepth == 0) { // Outermost macro. 4468 EndToken = getTok(); 4469 Lexer.Lex(); 4470 if (getLexer().isNot(AsmToken::EndOfStatement)) 4471 return TokError("unexpected token in '" + EndToken.getIdentifier() + 4472 "' directive"); 4473 break; 4474 } else { 4475 // Otherwise we just found the end of an inner macro. 4476 --MacroDepth; 4477 } 4478 } else if (getTok().getIdentifier() == ".macro") { 4479 // We allow nested macros. Those aren't instantiated until the outermost 4480 // macro is expanded so just ignore them for now. 4481 ++MacroDepth; 4482 } 4483 } else if (Lexer.is(AsmToken::HashDirective)) { 4484 (void)parseCppHashLineFilenameComment(getLexer().getLoc()); 4485 } 4486 4487 // Otherwise, scan til the end of the statement. 4488 eatToEndOfStatement(); 4489 } 4490 4491 if (getContext().lookupMacro(Name)) { 4492 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 4493 } 4494 4495 const char *BodyStart = StartToken.getLoc().getPointer(); 4496 const char *BodyEnd = EndToken.getLoc().getPointer(); 4497 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4498 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 4499 MCAsmMacro Macro(Name, Body, std::move(Parameters)); 4500 DEBUG_WITH_TYPE("asm-macros", dbgs() << "Defining new macro:\n"; 4501 Macro.dump()); 4502 getContext().defineMacro(Name, std::move(Macro)); 4503 return false; 4504 } 4505 4506 /// checkForBadMacro 4507 /// 4508 /// With the support added for named parameters there may be code out there that 4509 /// is transitioning from positional parameters. In versions of gas that did 4510 /// not support named parameters they would be ignored on the macro definition. 4511 /// But to support both styles of parameters this is not possible so if a macro 4512 /// definition has named parameters but does not use them and has what appears 4513 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 4514 /// warning that the positional parameter found in body which have no effect. 4515 /// Hoping the developer will either remove the named parameters from the macro 4516 /// definition so the positional parameters get used if that was what was 4517 /// intended or change the macro to use the named parameters. It is possible 4518 /// this warning will trigger when the none of the named parameters are used 4519 /// and the strings like $1 are infact to simply to be passed trough unchanged. 4520 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 4521 StringRef Body, 4522 ArrayRef<MCAsmMacroParameter> Parameters) { 4523 // If this macro is not defined with named parameters the warning we are 4524 // checking for here doesn't apply. 4525 unsigned NParameters = Parameters.size(); 4526 if (NParameters == 0) 4527 return; 4528 4529 bool NamedParametersFound = false; 4530 bool PositionalParametersFound = false; 4531 4532 // Look at the body of the macro for use of both the named parameters and what 4533 // are likely to be positional parameters. This is what expandMacro() is 4534 // doing when it finds the parameters in the body. 4535 while (!Body.empty()) { 4536 // Scan for the next possible parameter. 4537 std::size_t End = Body.size(), Pos = 0; 4538 for (; Pos != End; ++Pos) { 4539 // Check for a substitution or escape. 4540 // This macro is defined with parameters, look for \foo, \bar, etc. 4541 if (Body[Pos] == '\\' && Pos + 1 != End) 4542 break; 4543 4544 // This macro should have parameters, but look for $0, $1, ..., $n too. 4545 if (Body[Pos] != '$' || Pos + 1 == End) 4546 continue; 4547 char Next = Body[Pos + 1]; 4548 if (Next == '$' || Next == 'n' || 4549 isdigit(static_cast<unsigned char>(Next))) 4550 break; 4551 } 4552 4553 // Check if we reached the end. 4554 if (Pos == End) 4555 break; 4556 4557 if (Body[Pos] == '$') { 4558 switch (Body[Pos + 1]) { 4559 // $$ => $ 4560 case '$': 4561 break; 4562 4563 // $n => number of arguments 4564 case 'n': 4565 PositionalParametersFound = true; 4566 break; 4567 4568 // $[0-9] => argument 4569 default: { 4570 PositionalParametersFound = true; 4571 break; 4572 } 4573 } 4574 Pos += 2; 4575 } else { 4576 unsigned I = Pos + 1; 4577 while (isIdentifierChar(Body[I]) && I + 1 != End) 4578 ++I; 4579 4580 const char *Begin = Body.data() + Pos + 1; 4581 StringRef Argument(Begin, I - (Pos + 1)); 4582 unsigned Index = 0; 4583 for (; Index < NParameters; ++Index) 4584 if (Parameters[Index].Name == Argument) 4585 break; 4586 4587 if (Index == NParameters) { 4588 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 4589 Pos += 3; 4590 else { 4591 Pos = I; 4592 } 4593 } else { 4594 NamedParametersFound = true; 4595 Pos += 1 + Argument.size(); 4596 } 4597 } 4598 // Update the scan point. 4599 Body = Body.substr(Pos); 4600 } 4601 4602 if (!NamedParametersFound && PositionalParametersFound) 4603 Warning(DirectiveLoc, "macro defined with named parameters which are not " 4604 "used in macro body, possible positional parameter " 4605 "found in body which will have no effect"); 4606 } 4607 4608 /// parseDirectiveExitMacro 4609 /// ::= .exitm 4610 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 4611 if (parseToken(AsmToken::EndOfStatement, 4612 "unexpected token in '" + Directive + "' directive")) 4613 return true; 4614 4615 if (!isInsideMacroInstantiation()) 4616 return TokError("unexpected '" + Directive + "' in file, " 4617 "no current macro definition"); 4618 4619 // Exit all conditionals that are active in the current macro. 4620 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 4621 TheCondState = TheCondStack.back(); 4622 TheCondStack.pop_back(); 4623 } 4624 4625 handleMacroExit(); 4626 return false; 4627 } 4628 4629 /// parseDirectiveEndMacro 4630 /// ::= .endm 4631 /// ::= .endmacro 4632 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 4633 if (getLexer().isNot(AsmToken::EndOfStatement)) 4634 return TokError("unexpected token in '" + Directive + "' directive"); 4635 4636 // If we are inside a macro instantiation, terminate the current 4637 // instantiation. 4638 if (isInsideMacroInstantiation()) { 4639 handleMacroExit(); 4640 return false; 4641 } 4642 4643 // Otherwise, this .endmacro is a stray entry in the file; well formed 4644 // .endmacro directives are handled during the macro definition parsing. 4645 return TokError("unexpected '" + Directive + "' in file, " 4646 "no current macro definition"); 4647 } 4648 4649 /// parseDirectivePurgeMacro 4650 /// ::= .purgem 4651 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 4652 StringRef Name; 4653 SMLoc Loc; 4654 if (parseTokenLoc(Loc) || 4655 check(parseIdentifier(Name), Loc, 4656 "expected identifier in '.purgem' directive") || 4657 parseToken(AsmToken::EndOfStatement, 4658 "unexpected token in '.purgem' directive")) 4659 return true; 4660 4661 if (!getContext().lookupMacro(Name)) 4662 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 4663 4664 getContext().undefineMacro(Name); 4665 DEBUG_WITH_TYPE("asm-macros", dbgs() 4666 << "Un-defining macro: " << Name << "\n"); 4667 return false; 4668 } 4669 4670 /// parseDirectiveBundleAlignMode 4671 /// ::= {.bundle_align_mode} expression 4672 bool AsmParser::parseDirectiveBundleAlignMode() { 4673 // Expect a single argument: an expression that evaluates to a constant 4674 // in the inclusive range 0-30. 4675 SMLoc ExprLoc = getLexer().getLoc(); 4676 int64_t AlignSizePow2; 4677 if (checkForValidSection() || parseAbsoluteExpression(AlignSizePow2) || 4678 parseToken(AsmToken::EndOfStatement, "unexpected token after expression " 4679 "in '.bundle_align_mode' " 4680 "directive") || 4681 check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc, 4682 "invalid bundle alignment size (expected between 0 and 30)")) 4683 return true; 4684 4685 // Because of AlignSizePow2's verified range we can safely truncate it to 4686 // unsigned. 4687 getStreamer().emitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 4688 return false; 4689 } 4690 4691 /// parseDirectiveBundleLock 4692 /// ::= {.bundle_lock} [align_to_end] 4693 bool AsmParser::parseDirectiveBundleLock() { 4694 if (checkForValidSection()) 4695 return true; 4696 bool AlignToEnd = false; 4697 4698 StringRef Option; 4699 SMLoc Loc = getTok().getLoc(); 4700 const char *kInvalidOptionError = 4701 "invalid option for '.bundle_lock' directive"; 4702 4703 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4704 if (check(parseIdentifier(Option), Loc, kInvalidOptionError) || 4705 check(Option != "align_to_end", Loc, kInvalidOptionError) || 4706 parseToken(AsmToken::EndOfStatement, 4707 "unexpected token after '.bundle_lock' directive option")) 4708 return true; 4709 AlignToEnd = true; 4710 } 4711 4712 getStreamer().emitBundleLock(AlignToEnd); 4713 return false; 4714 } 4715 4716 /// parseDirectiveBundleLock 4717 /// ::= {.bundle_lock} 4718 bool AsmParser::parseDirectiveBundleUnlock() { 4719 if (checkForValidSection() || 4720 parseToken(AsmToken::EndOfStatement, 4721 "unexpected token in '.bundle_unlock' directive")) 4722 return true; 4723 4724 getStreamer().emitBundleUnlock(); 4725 return false; 4726 } 4727 4728 /// parseDirectiveSpace 4729 /// ::= (.skip | .space) expression [ , expression ] 4730 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4731 SMLoc NumBytesLoc = Lexer.getLoc(); 4732 const MCExpr *NumBytes; 4733 if (checkForValidSection() || parseExpression(NumBytes)) 4734 return true; 4735 4736 int64_t FillExpr = 0; 4737 if (parseOptionalToken(AsmToken::Comma)) 4738 if (parseAbsoluteExpression(FillExpr)) 4739 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4740 if (parseToken(AsmToken::EndOfStatement)) 4741 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4742 4743 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4744 getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc); 4745 4746 return false; 4747 } 4748 4749 /// parseDirectiveDCB 4750 /// ::= .dcb.{b, l, w} expression, expression 4751 bool AsmParser::parseDirectiveDCB(StringRef IDVal, unsigned Size) { 4752 SMLoc NumValuesLoc = Lexer.getLoc(); 4753 int64_t NumValues; 4754 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4755 return true; 4756 4757 if (NumValues < 0) { 4758 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4759 return false; 4760 } 4761 4762 if (parseToken(AsmToken::Comma, 4763 "unexpected token in '" + Twine(IDVal) + "' directive")) 4764 return true; 4765 4766 const MCExpr *Value; 4767 SMLoc ExprLoc = getLexer().getLoc(); 4768 if (parseExpression(Value)) 4769 return true; 4770 4771 // Special case constant expressions to match code generator. 4772 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 4773 assert(Size <= 8 && "Invalid size"); 4774 uint64_t IntValue = MCE->getValue(); 4775 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 4776 return Error(ExprLoc, "literal value out of range for directive"); 4777 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4778 getStreamer().emitIntValue(IntValue, Size); 4779 } else { 4780 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4781 getStreamer().emitValue(Value, Size, ExprLoc); 4782 } 4783 4784 if (parseToken(AsmToken::EndOfStatement, 4785 "unexpected token in '" + Twine(IDVal) + "' directive")) 4786 return true; 4787 4788 return false; 4789 } 4790 4791 /// parseDirectiveRealDCB 4792 /// ::= .dcb.{d, s} expression, expression 4793 bool AsmParser::parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &Semantics) { 4794 SMLoc NumValuesLoc = Lexer.getLoc(); 4795 int64_t NumValues; 4796 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4797 return true; 4798 4799 if (NumValues < 0) { 4800 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4801 return false; 4802 } 4803 4804 if (parseToken(AsmToken::Comma, 4805 "unexpected token in '" + Twine(IDVal) + "' directive")) 4806 return true; 4807 4808 APInt AsInt; 4809 if (parseRealValue(Semantics, AsInt)) 4810 return true; 4811 4812 if (parseToken(AsmToken::EndOfStatement, 4813 "unexpected token in '" + Twine(IDVal) + "' directive")) 4814 return true; 4815 4816 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4817 getStreamer().emitIntValue(AsInt.getLimitedValue(), 4818 AsInt.getBitWidth() / 8); 4819 4820 return false; 4821 } 4822 4823 /// parseDirectiveDS 4824 /// ::= .ds.{b, d, l, p, s, w, x} expression 4825 bool AsmParser::parseDirectiveDS(StringRef IDVal, unsigned Size) { 4826 SMLoc NumValuesLoc = Lexer.getLoc(); 4827 int64_t NumValues; 4828 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4829 return true; 4830 4831 if (NumValues < 0) { 4832 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4833 return false; 4834 } 4835 4836 if (parseToken(AsmToken::EndOfStatement, 4837 "unexpected token in '" + Twine(IDVal) + "' directive")) 4838 return true; 4839 4840 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4841 getStreamer().emitFill(Size, 0); 4842 4843 return false; 4844 } 4845 4846 /// parseDirectiveLEB128 4847 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4848 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4849 if (checkForValidSection()) 4850 return true; 4851 4852 auto parseOp = [&]() -> bool { 4853 const MCExpr *Value; 4854 if (parseExpression(Value)) 4855 return true; 4856 if (Signed) 4857 getStreamer().emitSLEB128Value(Value); 4858 else 4859 getStreamer().emitULEB128Value(Value); 4860 return false; 4861 }; 4862 4863 if (parseMany(parseOp)) 4864 return addErrorSuffix(" in directive"); 4865 4866 return false; 4867 } 4868 4869 /// parseDirectiveSymbolAttribute 4870 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4871 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4872 auto parseOp = [&]() -> bool { 4873 StringRef Name; 4874 SMLoc Loc = getTok().getLoc(); 4875 if (parseIdentifier(Name)) 4876 return Error(Loc, "expected identifier"); 4877 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4878 4879 // Assembler local symbols don't make any sense here. Complain loudly. 4880 if (Sym->isTemporary()) 4881 return Error(Loc, "non-local symbol required"); 4882 4883 if (!getStreamer().emitSymbolAttribute(Sym, Attr)) 4884 return Error(Loc, "unable to emit symbol attribute"); 4885 return false; 4886 }; 4887 4888 if (parseMany(parseOp)) 4889 return addErrorSuffix(" in directive"); 4890 return false; 4891 } 4892 4893 /// parseDirectiveComm 4894 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4895 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4896 if (checkForValidSection()) 4897 return true; 4898 4899 SMLoc IDLoc = getLexer().getLoc(); 4900 StringRef Name; 4901 if (parseIdentifier(Name)) 4902 return TokError("expected identifier in directive"); 4903 4904 // Handle the identifier as the key symbol. 4905 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4906 4907 if (getLexer().isNot(AsmToken::Comma)) 4908 return TokError("unexpected token in directive"); 4909 Lex(); 4910 4911 int64_t Size; 4912 SMLoc SizeLoc = getLexer().getLoc(); 4913 if (parseAbsoluteExpression(Size)) 4914 return true; 4915 4916 int64_t Pow2Alignment = 0; 4917 SMLoc Pow2AlignmentLoc; 4918 if (getLexer().is(AsmToken::Comma)) { 4919 Lex(); 4920 Pow2AlignmentLoc = getLexer().getLoc(); 4921 if (parseAbsoluteExpression(Pow2Alignment)) 4922 return true; 4923 4924 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4925 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4926 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4927 4928 // If this target takes alignments in bytes (not log) validate and convert. 4929 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4930 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4931 if (!isPowerOf2_64(Pow2Alignment)) 4932 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4933 Pow2Alignment = Log2_64(Pow2Alignment); 4934 } 4935 } 4936 4937 if (parseToken(AsmToken::EndOfStatement, 4938 "unexpected token in '.comm' or '.lcomm' directive")) 4939 return true; 4940 4941 // NOTE: a size of zero for a .comm should create a undefined symbol 4942 // but a size of .lcomm creates a bss symbol of size zero. 4943 if (Size < 0) 4944 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4945 "be less than zero"); 4946 4947 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4948 // may internally end up wanting an alignment in bytes. 4949 // FIXME: Diagnose overflow. 4950 if (Pow2Alignment < 0) 4951 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4952 "alignment, can't be less than zero"); 4953 4954 Sym->redefineIfPossible(); 4955 if (!Sym->isUndefined()) 4956 return Error(IDLoc, "invalid symbol redefinition"); 4957 4958 // Create the Symbol as a common or local common with Size and Pow2Alignment 4959 if (IsLocal) { 4960 getStreamer().emitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4961 return false; 4962 } 4963 4964 getStreamer().emitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4965 return false; 4966 } 4967 4968 /// parseDirectiveAbort 4969 /// ::= .abort [... message ...] 4970 bool AsmParser::parseDirectiveAbort() { 4971 // FIXME: Use loc from directive. 4972 SMLoc Loc = getLexer().getLoc(); 4973 4974 StringRef Str = parseStringToEndOfStatement(); 4975 if (parseToken(AsmToken::EndOfStatement, 4976 "unexpected token in '.abort' directive")) 4977 return true; 4978 4979 if (Str.empty()) 4980 return Error(Loc, ".abort detected. Assembly stopping."); 4981 else 4982 return Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 4983 // FIXME: Actually abort assembly here. 4984 4985 return false; 4986 } 4987 4988 /// parseDirectiveInclude 4989 /// ::= .include "filename" 4990 bool AsmParser::parseDirectiveInclude() { 4991 // Allow the strings to have escaped octal character sequence. 4992 std::string Filename; 4993 SMLoc IncludeLoc = getTok().getLoc(); 4994 4995 if (check(getTok().isNot(AsmToken::String), 4996 "expected string in '.include' directive") || 4997 parseEscapedString(Filename) || 4998 check(getTok().isNot(AsmToken::EndOfStatement), 4999 "unexpected token in '.include' directive") || 5000 // Attempt to switch the lexer to the included file before consuming the 5001 // end of statement to avoid losing it when we switch. 5002 check(enterIncludeFile(Filename), IncludeLoc, 5003 "Could not find include file '" + Filename + "'")) 5004 return true; 5005 5006 return false; 5007 } 5008 5009 /// parseDirectiveIncbin 5010 /// ::= .incbin "filename" [ , skip [ , count ] ] 5011 bool AsmParser::parseDirectiveIncbin() { 5012 // Allow the strings to have escaped octal character sequence. 5013 std::string Filename; 5014 SMLoc IncbinLoc = getTok().getLoc(); 5015 if (check(getTok().isNot(AsmToken::String), 5016 "expected string in '.incbin' directive") || 5017 parseEscapedString(Filename)) 5018 return true; 5019 5020 int64_t Skip = 0; 5021 const MCExpr *Count = nullptr; 5022 SMLoc SkipLoc, CountLoc; 5023 if (parseOptionalToken(AsmToken::Comma)) { 5024 // The skip expression can be omitted while specifying the count, e.g: 5025 // .incbin "filename",,4 5026 if (getTok().isNot(AsmToken::Comma)) { 5027 if (parseTokenLoc(SkipLoc) || parseAbsoluteExpression(Skip)) 5028 return true; 5029 } 5030 if (parseOptionalToken(AsmToken::Comma)) { 5031 CountLoc = getTok().getLoc(); 5032 if (parseExpression(Count)) 5033 return true; 5034 } 5035 } 5036 5037 if (parseToken(AsmToken::EndOfStatement, 5038 "unexpected token in '.incbin' directive")) 5039 return true; 5040 5041 if (check(Skip < 0, SkipLoc, "skip is negative")) 5042 return true; 5043 5044 // Attempt to process the included file. 5045 if (processIncbinFile(Filename, Skip, Count, CountLoc)) 5046 return Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 5047 return false; 5048 } 5049 5050 /// parseDirectiveIf 5051 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 5052 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 5053 TheCondStack.push_back(TheCondState); 5054 TheCondState.TheCond = AsmCond::IfCond; 5055 if (TheCondState.Ignore) { 5056 eatToEndOfStatement(); 5057 } else { 5058 int64_t ExprValue; 5059 if (parseAbsoluteExpression(ExprValue) || 5060 parseToken(AsmToken::EndOfStatement, 5061 "unexpected token in '.if' directive")) 5062 return true; 5063 5064 switch (DirKind) { 5065 default: 5066 llvm_unreachable("unsupported directive"); 5067 case DK_IF: 5068 case DK_IFNE: 5069 break; 5070 case DK_IFEQ: 5071 ExprValue = ExprValue == 0; 5072 break; 5073 case DK_IFGE: 5074 ExprValue = ExprValue >= 0; 5075 break; 5076 case DK_IFGT: 5077 ExprValue = ExprValue > 0; 5078 break; 5079 case DK_IFLE: 5080 ExprValue = ExprValue <= 0; 5081 break; 5082 case DK_IFLT: 5083 ExprValue = ExprValue < 0; 5084 break; 5085 } 5086 5087 TheCondState.CondMet = ExprValue; 5088 TheCondState.Ignore = !TheCondState.CondMet; 5089 } 5090 5091 return false; 5092 } 5093 5094 /// parseDirectiveIfb 5095 /// ::= .ifb string 5096 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 5097 TheCondStack.push_back(TheCondState); 5098 TheCondState.TheCond = AsmCond::IfCond; 5099 5100 if (TheCondState.Ignore) { 5101 eatToEndOfStatement(); 5102 } else { 5103 StringRef Str = parseStringToEndOfStatement(); 5104 5105 if (parseToken(AsmToken::EndOfStatement, 5106 "unexpected token in '.ifb' directive")) 5107 return true; 5108 5109 TheCondState.CondMet = ExpectBlank == Str.empty(); 5110 TheCondState.Ignore = !TheCondState.CondMet; 5111 } 5112 5113 return false; 5114 } 5115 5116 /// parseDirectiveIfc 5117 /// ::= .ifc string1, string2 5118 /// ::= .ifnc string1, string2 5119 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 5120 TheCondStack.push_back(TheCondState); 5121 TheCondState.TheCond = AsmCond::IfCond; 5122 5123 if (TheCondState.Ignore) { 5124 eatToEndOfStatement(); 5125 } else { 5126 StringRef Str1 = parseStringToComma(); 5127 5128 if (parseToken(AsmToken::Comma, "unexpected token in '.ifc' directive")) 5129 return true; 5130 5131 StringRef Str2 = parseStringToEndOfStatement(); 5132 5133 if (parseToken(AsmToken::EndOfStatement, 5134 "unexpected token in '.ifc' directive")) 5135 return true; 5136 5137 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 5138 TheCondState.Ignore = !TheCondState.CondMet; 5139 } 5140 5141 return false; 5142 } 5143 5144 /// parseDirectiveIfeqs 5145 /// ::= .ifeqs string1, string2 5146 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 5147 if (Lexer.isNot(AsmToken::String)) { 5148 if (ExpectEqual) 5149 return TokError("expected string parameter for '.ifeqs' directive"); 5150 return TokError("expected string parameter for '.ifnes' directive"); 5151 } 5152 5153 StringRef String1 = getTok().getStringContents(); 5154 Lex(); 5155 5156 if (Lexer.isNot(AsmToken::Comma)) { 5157 if (ExpectEqual) 5158 return TokError( 5159 "expected comma after first string for '.ifeqs' directive"); 5160 return TokError("expected comma after first string for '.ifnes' directive"); 5161 } 5162 5163 Lex(); 5164 5165 if (Lexer.isNot(AsmToken::String)) { 5166 if (ExpectEqual) 5167 return TokError("expected string parameter for '.ifeqs' directive"); 5168 return TokError("expected string parameter for '.ifnes' directive"); 5169 } 5170 5171 StringRef String2 = getTok().getStringContents(); 5172 Lex(); 5173 5174 TheCondStack.push_back(TheCondState); 5175 TheCondState.TheCond = AsmCond::IfCond; 5176 TheCondState.CondMet = ExpectEqual == (String1 == String2); 5177 TheCondState.Ignore = !TheCondState.CondMet; 5178 5179 return false; 5180 } 5181 5182 /// parseDirectiveIfdef 5183 /// ::= .ifdef symbol 5184 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 5185 StringRef Name; 5186 TheCondStack.push_back(TheCondState); 5187 TheCondState.TheCond = AsmCond::IfCond; 5188 5189 if (TheCondState.Ignore) { 5190 eatToEndOfStatement(); 5191 } else { 5192 if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") || 5193 parseToken(AsmToken::EndOfStatement, "unexpected token in '.ifdef'")) 5194 return true; 5195 5196 MCSymbol *Sym = getContext().lookupSymbol(Name); 5197 5198 if (expect_defined) 5199 TheCondState.CondMet = (Sym && !Sym->isUndefined(false)); 5200 else 5201 TheCondState.CondMet = (!Sym || Sym->isUndefined(false)); 5202 TheCondState.Ignore = !TheCondState.CondMet; 5203 } 5204 5205 return false; 5206 } 5207 5208 /// parseDirectiveElseIf 5209 /// ::= .elseif expression 5210 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 5211 if (TheCondState.TheCond != AsmCond::IfCond && 5212 TheCondState.TheCond != AsmCond::ElseIfCond) 5213 return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an" 5214 " .if or an .elseif"); 5215 TheCondState.TheCond = AsmCond::ElseIfCond; 5216 5217 bool LastIgnoreState = false; 5218 if (!TheCondStack.empty()) 5219 LastIgnoreState = TheCondStack.back().Ignore; 5220 if (LastIgnoreState || TheCondState.CondMet) { 5221 TheCondState.Ignore = true; 5222 eatToEndOfStatement(); 5223 } else { 5224 int64_t ExprValue; 5225 if (parseAbsoluteExpression(ExprValue)) 5226 return true; 5227 5228 if (parseToken(AsmToken::EndOfStatement, 5229 "unexpected token in '.elseif' directive")) 5230 return true; 5231 5232 TheCondState.CondMet = ExprValue; 5233 TheCondState.Ignore = !TheCondState.CondMet; 5234 } 5235 5236 return false; 5237 } 5238 5239 /// parseDirectiveElse 5240 /// ::= .else 5241 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 5242 if (parseToken(AsmToken::EndOfStatement, 5243 "unexpected token in '.else' directive")) 5244 return true; 5245 5246 if (TheCondState.TheCond != AsmCond::IfCond && 5247 TheCondState.TheCond != AsmCond::ElseIfCond) 5248 return Error(DirectiveLoc, "Encountered a .else that doesn't follow " 5249 " an .if or an .elseif"); 5250 TheCondState.TheCond = AsmCond::ElseCond; 5251 bool LastIgnoreState = false; 5252 if (!TheCondStack.empty()) 5253 LastIgnoreState = TheCondStack.back().Ignore; 5254 if (LastIgnoreState || TheCondState.CondMet) 5255 TheCondState.Ignore = true; 5256 else 5257 TheCondState.Ignore = false; 5258 5259 return false; 5260 } 5261 5262 /// parseDirectiveEnd 5263 /// ::= .end 5264 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 5265 if (parseToken(AsmToken::EndOfStatement, 5266 "unexpected token in '.end' directive")) 5267 return true; 5268 5269 while (Lexer.isNot(AsmToken::Eof)) 5270 Lexer.Lex(); 5271 5272 return false; 5273 } 5274 5275 /// parseDirectiveError 5276 /// ::= .err 5277 /// ::= .error [string] 5278 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 5279 if (!TheCondStack.empty()) { 5280 if (TheCondStack.back().Ignore) { 5281 eatToEndOfStatement(); 5282 return false; 5283 } 5284 } 5285 5286 if (!WithMessage) 5287 return Error(L, ".err encountered"); 5288 5289 StringRef Message = ".error directive invoked in source file"; 5290 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5291 if (Lexer.isNot(AsmToken::String)) 5292 return TokError(".error argument must be a string"); 5293 5294 Message = getTok().getStringContents(); 5295 Lex(); 5296 } 5297 5298 return Error(L, Message); 5299 } 5300 5301 /// parseDirectiveWarning 5302 /// ::= .warning [string] 5303 bool AsmParser::parseDirectiveWarning(SMLoc L) { 5304 if (!TheCondStack.empty()) { 5305 if (TheCondStack.back().Ignore) { 5306 eatToEndOfStatement(); 5307 return false; 5308 } 5309 } 5310 5311 StringRef Message = ".warning directive invoked in source file"; 5312 5313 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 5314 if (Lexer.isNot(AsmToken::String)) 5315 return TokError(".warning argument must be a string"); 5316 5317 Message = getTok().getStringContents(); 5318 Lex(); 5319 if (parseToken(AsmToken::EndOfStatement, 5320 "expected end of statement in '.warning' directive")) 5321 return true; 5322 } 5323 5324 return Warning(L, Message); 5325 } 5326 5327 /// parseDirectiveEndIf 5328 /// ::= .endif 5329 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 5330 if (parseToken(AsmToken::EndOfStatement, 5331 "unexpected token in '.endif' directive")) 5332 return true; 5333 5334 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 5335 return Error(DirectiveLoc, "Encountered a .endif that doesn't follow " 5336 "an .if or .else"); 5337 if (!TheCondStack.empty()) { 5338 TheCondState = TheCondStack.back(); 5339 TheCondStack.pop_back(); 5340 } 5341 5342 return false; 5343 } 5344 5345 void AsmParser::initializeDirectiveKindMap() { 5346 /* Lookup will be done with the directive 5347 * converted to lower case, so all these 5348 * keys should be lower case. 5349 * (target specific directives are handled 5350 * elsewhere) 5351 */ 5352 DirectiveKindMap[".set"] = DK_SET; 5353 DirectiveKindMap[".equ"] = DK_EQU; 5354 DirectiveKindMap[".equiv"] = DK_EQUIV; 5355 DirectiveKindMap[".ascii"] = DK_ASCII; 5356 DirectiveKindMap[".asciz"] = DK_ASCIZ; 5357 DirectiveKindMap[".string"] = DK_STRING; 5358 DirectiveKindMap[".byte"] = DK_BYTE; 5359 DirectiveKindMap[".short"] = DK_SHORT; 5360 DirectiveKindMap[".value"] = DK_VALUE; 5361 DirectiveKindMap[".2byte"] = DK_2BYTE; 5362 DirectiveKindMap[".long"] = DK_LONG; 5363 DirectiveKindMap[".int"] = DK_INT; 5364 DirectiveKindMap[".4byte"] = DK_4BYTE; 5365 DirectiveKindMap[".quad"] = DK_QUAD; 5366 DirectiveKindMap[".8byte"] = DK_8BYTE; 5367 DirectiveKindMap[".octa"] = DK_OCTA; 5368 DirectiveKindMap[".single"] = DK_SINGLE; 5369 DirectiveKindMap[".float"] = DK_FLOAT; 5370 DirectiveKindMap[".double"] = DK_DOUBLE; 5371 DirectiveKindMap[".align"] = DK_ALIGN; 5372 DirectiveKindMap[".align32"] = DK_ALIGN32; 5373 DirectiveKindMap[".balign"] = DK_BALIGN; 5374 DirectiveKindMap[".balignw"] = DK_BALIGNW; 5375 DirectiveKindMap[".balignl"] = DK_BALIGNL; 5376 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 5377 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 5378 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 5379 DirectiveKindMap[".org"] = DK_ORG; 5380 DirectiveKindMap[".fill"] = DK_FILL; 5381 DirectiveKindMap[".zero"] = DK_ZERO; 5382 DirectiveKindMap[".extern"] = DK_EXTERN; 5383 DirectiveKindMap[".globl"] = DK_GLOBL; 5384 DirectiveKindMap[".global"] = DK_GLOBAL; 5385 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 5386 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 5387 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 5388 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 5389 DirectiveKindMap[".reference"] = DK_REFERENCE; 5390 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 5391 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 5392 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 5393 DirectiveKindMap[".cold"] = DK_COLD; 5394 DirectiveKindMap[".comm"] = DK_COMM; 5395 DirectiveKindMap[".common"] = DK_COMMON; 5396 DirectiveKindMap[".lcomm"] = DK_LCOMM; 5397 DirectiveKindMap[".abort"] = DK_ABORT; 5398 DirectiveKindMap[".include"] = DK_INCLUDE; 5399 DirectiveKindMap[".incbin"] = DK_INCBIN; 5400 DirectiveKindMap[".code16"] = DK_CODE16; 5401 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 5402 DirectiveKindMap[".rept"] = DK_REPT; 5403 DirectiveKindMap[".rep"] = DK_REPT; 5404 DirectiveKindMap[".irp"] = DK_IRP; 5405 DirectiveKindMap[".irpc"] = DK_IRPC; 5406 DirectiveKindMap[".endr"] = DK_ENDR; 5407 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 5408 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 5409 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 5410 DirectiveKindMap[".if"] = DK_IF; 5411 DirectiveKindMap[".ifeq"] = DK_IFEQ; 5412 DirectiveKindMap[".ifge"] = DK_IFGE; 5413 DirectiveKindMap[".ifgt"] = DK_IFGT; 5414 DirectiveKindMap[".ifle"] = DK_IFLE; 5415 DirectiveKindMap[".iflt"] = DK_IFLT; 5416 DirectiveKindMap[".ifne"] = DK_IFNE; 5417 DirectiveKindMap[".ifb"] = DK_IFB; 5418 DirectiveKindMap[".ifnb"] = DK_IFNB; 5419 DirectiveKindMap[".ifc"] = DK_IFC; 5420 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 5421 DirectiveKindMap[".ifnc"] = DK_IFNC; 5422 DirectiveKindMap[".ifnes"] = DK_IFNES; 5423 DirectiveKindMap[".ifdef"] = DK_IFDEF; 5424 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 5425 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 5426 DirectiveKindMap[".elseif"] = DK_ELSEIF; 5427 DirectiveKindMap[".else"] = DK_ELSE; 5428 DirectiveKindMap[".end"] = DK_END; 5429 DirectiveKindMap[".endif"] = DK_ENDIF; 5430 DirectiveKindMap[".skip"] = DK_SKIP; 5431 DirectiveKindMap[".space"] = DK_SPACE; 5432 DirectiveKindMap[".file"] = DK_FILE; 5433 DirectiveKindMap[".line"] = DK_LINE; 5434 DirectiveKindMap[".loc"] = DK_LOC; 5435 DirectiveKindMap[".stabs"] = DK_STABS; 5436 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 5437 DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID; 5438 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 5439 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 5440 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 5441 DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID; 5442 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 5443 DirectiveKindMap[".cv_string"] = DK_CV_STRING; 5444 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 5445 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 5446 DirectiveKindMap[".cv_filechecksumoffset"] = DK_CV_FILECHECKSUM_OFFSET; 5447 DirectiveKindMap[".cv_fpo_data"] = DK_CV_FPO_DATA; 5448 DirectiveKindMap[".sleb128"] = DK_SLEB128; 5449 DirectiveKindMap[".uleb128"] = DK_ULEB128; 5450 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 5451 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 5452 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 5453 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 5454 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 5455 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 5456 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 5457 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 5458 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 5459 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 5460 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 5461 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 5462 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 5463 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 5464 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 5465 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 5466 DirectiveKindMap[".cfi_return_column"] = DK_CFI_RETURN_COLUMN; 5467 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 5468 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 5469 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 5470 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 5471 DirectiveKindMap[".cfi_b_key_frame"] = DK_CFI_B_KEY_FRAME; 5472 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 5473 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 5474 DirectiveKindMap[".macro"] = DK_MACRO; 5475 DirectiveKindMap[".exitm"] = DK_EXITM; 5476 DirectiveKindMap[".endm"] = DK_ENDM; 5477 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 5478 DirectiveKindMap[".purgem"] = DK_PURGEM; 5479 DirectiveKindMap[".err"] = DK_ERR; 5480 DirectiveKindMap[".error"] = DK_ERROR; 5481 DirectiveKindMap[".warning"] = DK_WARNING; 5482 DirectiveKindMap[".altmacro"] = DK_ALTMACRO; 5483 DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO; 5484 DirectiveKindMap[".reloc"] = DK_RELOC; 5485 DirectiveKindMap[".dc"] = DK_DC; 5486 DirectiveKindMap[".dc.a"] = DK_DC_A; 5487 DirectiveKindMap[".dc.b"] = DK_DC_B; 5488 DirectiveKindMap[".dc.d"] = DK_DC_D; 5489 DirectiveKindMap[".dc.l"] = DK_DC_L; 5490 DirectiveKindMap[".dc.s"] = DK_DC_S; 5491 DirectiveKindMap[".dc.w"] = DK_DC_W; 5492 DirectiveKindMap[".dc.x"] = DK_DC_X; 5493 DirectiveKindMap[".dcb"] = DK_DCB; 5494 DirectiveKindMap[".dcb.b"] = DK_DCB_B; 5495 DirectiveKindMap[".dcb.d"] = DK_DCB_D; 5496 DirectiveKindMap[".dcb.l"] = DK_DCB_L; 5497 DirectiveKindMap[".dcb.s"] = DK_DCB_S; 5498 DirectiveKindMap[".dcb.w"] = DK_DCB_W; 5499 DirectiveKindMap[".dcb.x"] = DK_DCB_X; 5500 DirectiveKindMap[".ds"] = DK_DS; 5501 DirectiveKindMap[".ds.b"] = DK_DS_B; 5502 DirectiveKindMap[".ds.d"] = DK_DS_D; 5503 DirectiveKindMap[".ds.l"] = DK_DS_L; 5504 DirectiveKindMap[".ds.p"] = DK_DS_P; 5505 DirectiveKindMap[".ds.s"] = DK_DS_S; 5506 DirectiveKindMap[".ds.w"] = DK_DS_W; 5507 DirectiveKindMap[".ds.x"] = DK_DS_X; 5508 DirectiveKindMap[".print"] = DK_PRINT; 5509 DirectiveKindMap[".addrsig"] = DK_ADDRSIG; 5510 DirectiveKindMap[".addrsig_sym"] = DK_ADDRSIG_SYM; 5511 } 5512 5513 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 5514 AsmToken EndToken, StartToken = getTok(); 5515 5516 unsigned NestLevel = 0; 5517 while (true) { 5518 // Check whether we have reached the end of the file. 5519 if (getLexer().is(AsmToken::Eof)) { 5520 printError(DirectiveLoc, "no matching '.endr' in definition"); 5521 return nullptr; 5522 } 5523 5524 if (Lexer.is(AsmToken::Identifier) && 5525 (getTok().getIdentifier() == ".rep" || 5526 getTok().getIdentifier() == ".rept" || 5527 getTok().getIdentifier() == ".irp" || 5528 getTok().getIdentifier() == ".irpc")) { 5529 ++NestLevel; 5530 } 5531 5532 // Otherwise, check whether we have reached the .endr. 5533 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 5534 if (NestLevel == 0) { 5535 EndToken = getTok(); 5536 Lex(); 5537 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5538 printError(getTok().getLoc(), 5539 "unexpected token in '.endr' directive"); 5540 return nullptr; 5541 } 5542 break; 5543 } 5544 --NestLevel; 5545 } 5546 5547 // Otherwise, scan till the end of the statement. 5548 eatToEndOfStatement(); 5549 } 5550 5551 const char *BodyStart = StartToken.getLoc().getPointer(); 5552 const char *BodyEnd = EndToken.getLoc().getPointer(); 5553 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 5554 5555 // We Are Anonymous. 5556 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 5557 return &MacroLikeBodies.back(); 5558 } 5559 5560 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 5561 raw_svector_ostream &OS) { 5562 OS << ".endr\n"; 5563 5564 std::unique_ptr<MemoryBuffer> Instantiation = 5565 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 5566 5567 // Create the macro instantiation object and add to the current macro 5568 // instantiation stack. 5569 MacroInstantiation *MI = new MacroInstantiation{ 5570 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 5571 ActiveMacros.push_back(MI); 5572 5573 // Jump to the macro instantiation and prime the lexer. 5574 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 5575 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 5576 Lex(); 5577 } 5578 5579 /// parseDirectiveRept 5580 /// ::= .rep | .rept count 5581 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 5582 const MCExpr *CountExpr; 5583 SMLoc CountLoc = getTok().getLoc(); 5584 if (parseExpression(CountExpr)) 5585 return true; 5586 5587 int64_t Count; 5588 if (!CountExpr->evaluateAsAbsolute(Count, getStreamer().getAssemblerPtr())) { 5589 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 5590 } 5591 5592 if (check(Count < 0, CountLoc, "Count is negative") || 5593 parseToken(AsmToken::EndOfStatement, 5594 "unexpected token in '" + Dir + "' directive")) 5595 return true; 5596 5597 // Lex the rept definition. 5598 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5599 if (!M) 5600 return true; 5601 5602 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5603 // to hold the macro body with substitutions. 5604 SmallString<256> Buf; 5605 raw_svector_ostream OS(Buf); 5606 while (Count--) { 5607 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 5608 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 5609 return true; 5610 } 5611 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5612 5613 return false; 5614 } 5615 5616 /// parseDirectiveIrp 5617 /// ::= .irp symbol,values 5618 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 5619 MCAsmMacroParameter Parameter; 5620 MCAsmMacroArguments A; 5621 if (check(parseIdentifier(Parameter.Name), 5622 "expected identifier in '.irp' directive") || 5623 parseToken(AsmToken::Comma, "expected comma in '.irp' directive") || 5624 parseMacroArguments(nullptr, A) || 5625 parseToken(AsmToken::EndOfStatement, "expected End of Statement")) 5626 return true; 5627 5628 // Lex the irp definition. 5629 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5630 if (!M) 5631 return true; 5632 5633 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5634 // to hold the macro body with substitutions. 5635 SmallString<256> Buf; 5636 raw_svector_ostream OS(Buf); 5637 5638 for (const MCAsmMacroArgument &Arg : A) { 5639 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 5640 // This is undocumented, but GAS seems to support it. 5641 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5642 return true; 5643 } 5644 5645 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5646 5647 return false; 5648 } 5649 5650 /// parseDirectiveIrpc 5651 /// ::= .irpc symbol,values 5652 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 5653 MCAsmMacroParameter Parameter; 5654 MCAsmMacroArguments A; 5655 5656 if (check(parseIdentifier(Parameter.Name), 5657 "expected identifier in '.irpc' directive") || 5658 parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") || 5659 parseMacroArguments(nullptr, A)) 5660 return true; 5661 5662 if (A.size() != 1 || A.front().size() != 1) 5663 return TokError("unexpected token in '.irpc' directive"); 5664 5665 // Eat the end of statement. 5666 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5667 return true; 5668 5669 // Lex the irpc definition. 5670 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5671 if (!M) 5672 return true; 5673 5674 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5675 // to hold the macro body with substitutions. 5676 SmallString<256> Buf; 5677 raw_svector_ostream OS(Buf); 5678 5679 StringRef Values = A.front().front().getString(); 5680 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 5681 MCAsmMacroArgument Arg; 5682 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 5683 5684 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 5685 // This is undocumented, but GAS seems to support it. 5686 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5687 return true; 5688 } 5689 5690 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5691 5692 return false; 5693 } 5694 5695 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 5696 if (ActiveMacros.empty()) 5697 return TokError("unmatched '.endr' directive"); 5698 5699 // The only .repl that should get here are the ones created by 5700 // instantiateMacroLikeBody. 5701 assert(getLexer().is(AsmToken::EndOfStatement)); 5702 5703 handleMacroExit(); 5704 return false; 5705 } 5706 5707 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 5708 size_t Len) { 5709 const MCExpr *Value; 5710 SMLoc ExprLoc = getLexer().getLoc(); 5711 if (parseExpression(Value)) 5712 return true; 5713 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5714 if (!MCE) 5715 return Error(ExprLoc, "unexpected expression in _emit"); 5716 uint64_t IntValue = MCE->getValue(); 5717 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 5718 return Error(ExprLoc, "literal value out of range for directive"); 5719 5720 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 5721 return false; 5722 } 5723 5724 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 5725 const MCExpr *Value; 5726 SMLoc ExprLoc = getLexer().getLoc(); 5727 if (parseExpression(Value)) 5728 return true; 5729 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5730 if (!MCE) 5731 return Error(ExprLoc, "unexpected expression in align"); 5732 uint64_t IntValue = MCE->getValue(); 5733 if (!isPowerOf2_64(IntValue)) 5734 return Error(ExprLoc, "literal value not a power of two greater then zero"); 5735 5736 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 5737 return false; 5738 } 5739 5740 bool AsmParser::parseDirectivePrint(SMLoc DirectiveLoc) { 5741 const AsmToken StrTok = getTok(); 5742 Lex(); 5743 if (StrTok.isNot(AsmToken::String) || StrTok.getString().front() != '"') 5744 return Error(DirectiveLoc, "expected double quoted string after .print"); 5745 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5746 return true; 5747 llvm::outs() << StrTok.getStringContents() << '\n'; 5748 return false; 5749 } 5750 5751 bool AsmParser::parseDirectiveAddrsig() { 5752 getStreamer().emitAddrsig(); 5753 return false; 5754 } 5755 5756 bool AsmParser::parseDirectiveAddrsigSym() { 5757 StringRef Name; 5758 if (check(parseIdentifier(Name), 5759 "expected identifier in '.addrsig_sym' directive")) 5760 return true; 5761 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 5762 getStreamer().emitAddrsigSym(Sym); 5763 return false; 5764 } 5765 5766 // We are comparing pointers, but the pointers are relative to a single string. 5767 // Thus, this should always be deterministic. 5768 static int rewritesSort(const AsmRewrite *AsmRewriteA, 5769 const AsmRewrite *AsmRewriteB) { 5770 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 5771 return -1; 5772 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 5773 return 1; 5774 5775 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 5776 // rewrite to the same location. Make sure the SizeDirective rewrite is 5777 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 5778 // ensures the sort algorithm is stable. 5779 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 5780 AsmRewritePrecedence[AsmRewriteB->Kind]) 5781 return -1; 5782 5783 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 5784 AsmRewritePrecedence[AsmRewriteB->Kind]) 5785 return 1; 5786 llvm_unreachable("Unstable rewrite sort."); 5787 } 5788 5789 bool AsmParser::parseMSInlineAsm( 5790 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 5791 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls, 5792 SmallVectorImpl<std::string> &Constraints, 5793 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 5794 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 5795 SmallVector<void *, 4> InputDecls; 5796 SmallVector<void *, 4> OutputDecls; 5797 SmallVector<bool, 4> InputDeclsAddressOf; 5798 SmallVector<bool, 4> OutputDeclsAddressOf; 5799 SmallVector<std::string, 4> InputConstraints; 5800 SmallVector<std::string, 4> OutputConstraints; 5801 SmallVector<unsigned, 4> ClobberRegs; 5802 5803 SmallVector<AsmRewrite, 4> AsmStrRewrites; 5804 5805 // Prime the lexer. 5806 Lex(); 5807 5808 // While we have input, parse each statement. 5809 unsigned InputIdx = 0; 5810 unsigned OutputIdx = 0; 5811 while (getLexer().isNot(AsmToken::Eof)) { 5812 // Parse curly braces marking block start/end 5813 if (parseCurlyBlockScope(AsmStrRewrites)) 5814 continue; 5815 5816 ParseStatementInfo Info(&AsmStrRewrites); 5817 bool StatementErr = parseStatement(Info, &SI); 5818 5819 if (StatementErr || Info.ParseError) { 5820 // Emit pending errors if any exist. 5821 printPendingErrors(); 5822 return true; 5823 } 5824 5825 // No pending error should exist here. 5826 assert(!hasPendingError() && "unexpected error from parseStatement"); 5827 5828 if (Info.Opcode == ~0U) 5829 continue; 5830 5831 const MCInstrDesc &Desc = MII->get(Info.Opcode); 5832 5833 // Build the list of clobbers, outputs and inputs. 5834 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 5835 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 5836 5837 // Register operand. 5838 if (Operand.isReg() && !Operand.needAddressOf() && 5839 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5840 unsigned NumDefs = Desc.getNumDefs(); 5841 // Clobber. 5842 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5843 ClobberRegs.push_back(Operand.getReg()); 5844 continue; 5845 } 5846 5847 // Expr/Input or Output. 5848 StringRef SymName = Operand.getSymName(); 5849 if (SymName.empty()) 5850 continue; 5851 5852 void *OpDecl = Operand.getOpDecl(); 5853 if (!OpDecl) 5854 continue; 5855 5856 StringRef Constraint = Operand.getConstraint(); 5857 if (Operand.isImm()) { 5858 // Offset as immediate 5859 if (Operand.isOffsetOfLocal()) 5860 Constraint = "r"; 5861 else 5862 Constraint = "i"; 5863 } 5864 5865 bool isOutput = (i == 1) && Desc.mayStore(); 5866 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5867 if (isOutput) { 5868 ++InputIdx; 5869 OutputDecls.push_back(OpDecl); 5870 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5871 OutputConstraints.push_back(("=" + Constraint).str()); 5872 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5873 } else { 5874 InputDecls.push_back(OpDecl); 5875 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5876 InputConstraints.push_back(Constraint.str()); 5877 if (Desc.OpInfo[i - 1].isBranchTarget()) 5878 AsmStrRewrites.emplace_back(AOK_CallInput, Start, SymName.size()); 5879 else 5880 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5881 } 5882 } 5883 5884 // Consider implicit defs to be clobbers. Think of cpuid and push. 5885 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5886 Desc.getNumImplicitDefs()); 5887 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 5888 } 5889 5890 // Set the number of Outputs and Inputs. 5891 NumOutputs = OutputDecls.size(); 5892 NumInputs = InputDecls.size(); 5893 5894 // Set the unique clobbers. 5895 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5896 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5897 ClobberRegs.end()); 5898 Clobbers.assign(ClobberRegs.size(), std::string()); 5899 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5900 raw_string_ostream OS(Clobbers[I]); 5901 IP->printRegName(OS, ClobberRegs[I]); 5902 } 5903 5904 // Merge the various outputs and inputs. Output are expected first. 5905 if (NumOutputs || NumInputs) { 5906 unsigned NumExprs = NumOutputs + NumInputs; 5907 OpDecls.resize(NumExprs); 5908 Constraints.resize(NumExprs); 5909 for (unsigned i = 0; i < NumOutputs; ++i) { 5910 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5911 Constraints[i] = OutputConstraints[i]; 5912 } 5913 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5914 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5915 Constraints[j] = InputConstraints[i]; 5916 } 5917 } 5918 5919 // Build the IR assembly string. 5920 std::string AsmStringIR; 5921 raw_string_ostream OS(AsmStringIR); 5922 StringRef ASMString = 5923 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5924 const char *AsmStart = ASMString.begin(); 5925 const char *AsmEnd = ASMString.end(); 5926 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 5927 for (auto it = AsmStrRewrites.begin(); it != AsmStrRewrites.end(); ++it) { 5928 const AsmRewrite &AR = *it; 5929 // Check if this has already been covered by another rewrite... 5930 if (AR.Done) 5931 continue; 5932 AsmRewriteKind Kind = AR.Kind; 5933 5934 const char *Loc = AR.Loc.getPointer(); 5935 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 5936 5937 // Emit everything up to the immediate/expression. 5938 if (unsigned Len = Loc - AsmStart) 5939 OS << StringRef(AsmStart, Len); 5940 5941 // Skip the original expression. 5942 if (Kind == AOK_Skip) { 5943 AsmStart = Loc + AR.Len; 5944 continue; 5945 } 5946 5947 unsigned AdditionalSkip = 0; 5948 // Rewrite expressions in $N notation. 5949 switch (Kind) { 5950 default: 5951 break; 5952 case AOK_IntelExpr: 5953 assert(AR.IntelExp.isValid() && "cannot write invalid intel expression"); 5954 if (AR.IntelExp.NeedBracs) 5955 OS << "["; 5956 if (AR.IntelExp.hasBaseReg()) 5957 OS << AR.IntelExp.BaseReg; 5958 if (AR.IntelExp.hasIndexReg()) 5959 OS << (AR.IntelExp.hasBaseReg() ? " + " : "") 5960 << AR.IntelExp.IndexReg; 5961 if (AR.IntelExp.Scale > 1) 5962 OS << " * $$" << AR.IntelExp.Scale; 5963 if (AR.IntelExp.hasOffset()) { 5964 if (AR.IntelExp.hasRegs()) 5965 OS << " + "; 5966 // Fuse this rewrite with a rewrite of the offset name, if present. 5967 StringRef OffsetName = AR.IntelExp.OffsetName; 5968 SMLoc OffsetLoc = SMLoc::getFromPointer(AR.IntelExp.OffsetName.data()); 5969 size_t OffsetLen = OffsetName.size(); 5970 auto rewrite_it = std::find_if( 5971 it, AsmStrRewrites.end(), [&](const AsmRewrite &FusingAR) { 5972 return FusingAR.Loc == OffsetLoc && FusingAR.Len == OffsetLen && 5973 (FusingAR.Kind == AOK_Input || 5974 FusingAR.Kind == AOK_CallInput); 5975 }); 5976 if (rewrite_it == AsmStrRewrites.end()) { 5977 OS << "offset " << OffsetName; 5978 } else if (rewrite_it->Kind == AOK_CallInput) { 5979 OS << "${" << InputIdx++ << ":P}"; 5980 rewrite_it->Done = true; 5981 } else { 5982 OS << '$' << InputIdx++; 5983 rewrite_it->Done = true; 5984 } 5985 } 5986 if (AR.IntelExp.Imm || AR.IntelExp.emitImm()) 5987 OS << (AR.IntelExp.emitImm() ? "$$" : " + $$") << AR.IntelExp.Imm; 5988 if (AR.IntelExp.NeedBracs) 5989 OS << "]"; 5990 break; 5991 case AOK_Label: 5992 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 5993 break; 5994 case AOK_Input: 5995 OS << '$' << InputIdx++; 5996 break; 5997 case AOK_CallInput: 5998 OS << "${" << InputIdx++ << ":P}"; 5999 break; 6000 case AOK_Output: 6001 OS << '$' << OutputIdx++; 6002 break; 6003 case AOK_SizeDirective: 6004 switch (AR.Val) { 6005 default: break; 6006 case 8: OS << "byte ptr "; break; 6007 case 16: OS << "word ptr "; break; 6008 case 32: OS << "dword ptr "; break; 6009 case 64: OS << "qword ptr "; break; 6010 case 80: OS << "xword ptr "; break; 6011 case 128: OS << "xmmword ptr "; break; 6012 case 256: OS << "ymmword ptr "; break; 6013 } 6014 break; 6015 case AOK_Emit: 6016 OS << ".byte"; 6017 break; 6018 case AOK_Align: { 6019 // MS alignment directives are measured in bytes. If the native assembler 6020 // measures alignment in bytes, we can pass it straight through. 6021 OS << ".align"; 6022 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 6023 break; 6024 6025 // Alignment is in log2 form, so print that instead and skip the original 6026 // immediate. 6027 unsigned Val = AR.Val; 6028 OS << ' ' << Val; 6029 assert(Val < 10 && "Expected alignment less then 2^10."); 6030 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 6031 break; 6032 } 6033 case AOK_EVEN: 6034 OS << ".even"; 6035 break; 6036 case AOK_EndOfStatement: 6037 OS << "\n\t"; 6038 break; 6039 } 6040 6041 // Skip the original expression. 6042 AsmStart = Loc + AR.Len + AdditionalSkip; 6043 } 6044 6045 // Emit the remainder of the asm string. 6046 if (AsmStart != AsmEnd) 6047 OS << StringRef(AsmStart, AsmEnd - AsmStart); 6048 6049 AsmString = OS.str(); 6050 return false; 6051 } 6052 6053 namespace llvm { 6054 namespace MCParserUtils { 6055 6056 /// Returns whether the given symbol is used anywhere in the given expression, 6057 /// or subexpressions. 6058 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 6059 switch (Value->getKind()) { 6060 case MCExpr::Binary: { 6061 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 6062 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 6063 isSymbolUsedInExpression(Sym, BE->getRHS()); 6064 } 6065 case MCExpr::Target: 6066 case MCExpr::Constant: 6067 return false; 6068 case MCExpr::SymbolRef: { 6069 const MCSymbol &S = 6070 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 6071 if (S.isVariable()) 6072 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 6073 return &S == Sym; 6074 } 6075 case MCExpr::Unary: 6076 return isSymbolUsedInExpression( 6077 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 6078 } 6079 6080 llvm_unreachable("Unknown expr kind!"); 6081 } 6082 6083 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 6084 MCAsmParser &Parser, MCSymbol *&Sym, 6085 const MCExpr *&Value) { 6086 6087 // FIXME: Use better location, we should use proper tokens. 6088 SMLoc EqualLoc = Parser.getTok().getLoc(); 6089 if (Parser.parseExpression(Value)) 6090 return Parser.TokError("missing expression"); 6091 6092 // Note: we don't count b as used in "a = b". This is to allow 6093 // a = b 6094 // b = c 6095 6096 if (Parser.parseToken(AsmToken::EndOfStatement)) 6097 return true; 6098 6099 // Validate that the LHS is allowed to be a variable (either it has not been 6100 // used as a symbol, or it is an absolute symbol). 6101 Sym = Parser.getContext().lookupSymbol(Name); 6102 if (Sym) { 6103 // Diagnose assignment to a label. 6104 // 6105 // FIXME: Diagnostics. Note the location of the definition as a label. 6106 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 6107 if (isSymbolUsedInExpression(Sym, Value)) 6108 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 6109 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 6110 !Sym->isVariable()) 6111 ; // Allow redefinitions of undefined symbols only used in directives. 6112 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 6113 ; // Allow redefinitions of variables that haven't yet been used. 6114 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 6115 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 6116 else if (!Sym->isVariable()) 6117 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 6118 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 6119 return Parser.Error(EqualLoc, 6120 "invalid reassignment of non-absolute variable '" + 6121 Name + "'"); 6122 } else if (Name == ".") { 6123 Parser.getStreamer().emitValueToOffset(Value, 0, EqualLoc); 6124 return false; 6125 } else 6126 Sym = Parser.getContext().getOrCreateSymbol(Name); 6127 6128 Sym->setRedefinable(allow_redef); 6129 6130 return false; 6131 } 6132 6133 } // end namespace MCParserUtils 6134 } // end namespace llvm 6135 6136 /// Create an MCAsmParser instance. 6137 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 6138 MCStreamer &Out, const MCAsmInfo &MAI, 6139 unsigned CB) { 6140 return new AsmParser(SM, C, Out, MAI, CB); 6141 } 6142