1 //===- AsmParser.cpp - Parser for Assembly Files --------------------------===// 2 // 3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 4 // See https://llvm.org/LICENSE.txt for license information. 5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6 // 7 //===----------------------------------------------------------------------===// 8 // 9 // This class implements the parser for assembly files. 10 // 11 //===----------------------------------------------------------------------===// 12 13 #include "llvm/ADT/APFloat.h" 14 #include "llvm/ADT/APInt.h" 15 #include "llvm/ADT/ArrayRef.h" 16 #include "llvm/ADT/None.h" 17 #include "llvm/ADT/STLExtras.h" 18 #include "llvm/ADT/SmallString.h" 19 #include "llvm/ADT/SmallVector.h" 20 #include "llvm/ADT/StringExtras.h" 21 #include "llvm/ADT/StringMap.h" 22 #include "llvm/ADT/StringRef.h" 23 #include "llvm/ADT/Twine.h" 24 #include "llvm/BinaryFormat/Dwarf.h" 25 #include "llvm/MC/MCAsmInfo.h" 26 #include "llvm/MC/MCCodeView.h" 27 #include "llvm/MC/MCContext.h" 28 #include "llvm/MC/MCDirectives.h" 29 #include "llvm/MC/MCDwarf.h" 30 #include "llvm/MC/MCExpr.h" 31 #include "llvm/MC/MCInstPrinter.h" 32 #include "llvm/MC/MCInstrDesc.h" 33 #include "llvm/MC/MCInstrInfo.h" 34 #include "llvm/MC/MCObjectFileInfo.h" 35 #include "llvm/MC/MCParser/AsmCond.h" 36 #include "llvm/MC/MCParser/AsmLexer.h" 37 #include "llvm/MC/MCParser/MCAsmLexer.h" 38 #include "llvm/MC/MCParser/MCAsmParser.h" 39 #include "llvm/MC/MCParser/MCAsmParserExtension.h" 40 #include "llvm/MC/MCParser/MCAsmParserUtils.h" 41 #include "llvm/MC/MCParser/MCParsedAsmOperand.h" 42 #include "llvm/MC/MCParser/MCTargetAsmParser.h" 43 #include "llvm/MC/MCRegisterInfo.h" 44 #include "llvm/MC/MCSection.h" 45 #include "llvm/MC/MCStreamer.h" 46 #include "llvm/MC/MCSymbol.h" 47 #include "llvm/MC/MCTargetOptions.h" 48 #include "llvm/MC/MCValue.h" 49 #include "llvm/Support/Casting.h" 50 #include "llvm/Support/CommandLine.h" 51 #include "llvm/Support/ErrorHandling.h" 52 #include "llvm/Support/MD5.h" 53 #include "llvm/Support/MathExtras.h" 54 #include "llvm/Support/MemoryBuffer.h" 55 #include "llvm/Support/SMLoc.h" 56 #include "llvm/Support/SourceMgr.h" 57 #include "llvm/Support/raw_ostream.h" 58 #include <algorithm> 59 #include <cassert> 60 #include <cctype> 61 #include <climits> 62 #include <cstddef> 63 #include <cstdint> 64 #include <deque> 65 #include <memory> 66 #include <sstream> 67 #include <string> 68 #include <tuple> 69 #include <utility> 70 #include <vector> 71 72 using namespace llvm; 73 74 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() = default; 75 76 static cl::opt<unsigned> AsmMacroMaxNestingDepth( 77 "asm-macro-max-nesting-depth", cl::init(20), cl::Hidden, 78 cl::desc("The maximum nesting depth allowed for assembly macros.")); 79 80 namespace { 81 82 /// Helper types for tracking macro definitions. 83 typedef std::vector<AsmToken> MCAsmMacroArgument; 84 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 85 86 /// Helper class for storing information about an active macro 87 /// instantiation. 88 struct MacroInstantiation { 89 /// The location of the instantiation. 90 SMLoc InstantiationLoc; 91 92 /// The buffer where parsing should resume upon instantiation completion. 93 int ExitBuffer; 94 95 /// The location where parsing should resume upon instantiation completion. 96 SMLoc ExitLoc; 97 98 /// The depth of TheCondStack at the start of the instantiation. 99 size_t CondStackDepth; 100 101 public: 102 MacroInstantiation(SMLoc IL, int EB, SMLoc EL, size_t CondStackDepth); 103 }; 104 105 struct ParseStatementInfo { 106 /// The parsed operands from the last parsed statement. 107 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 108 109 /// The opcode from the last parsed instruction. 110 unsigned Opcode = ~0U; 111 112 /// Was there an error parsing the inline assembly? 113 bool ParseError = false; 114 115 SmallVectorImpl<AsmRewrite> *AsmRewrites = nullptr; 116 117 ParseStatementInfo() = delete; 118 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 119 : AsmRewrites(rewrites) {} 120 }; 121 122 /// The concrete assembly parser instance. 123 class AsmParser : public MCAsmParser { 124 private: 125 AsmLexer Lexer; 126 MCContext &Ctx; 127 MCStreamer &Out; 128 const MCAsmInfo &MAI; 129 SourceMgr &SrcMgr; 130 SourceMgr::DiagHandlerTy SavedDiagHandler; 131 void *SavedDiagContext; 132 std::unique_ptr<MCAsmParserExtension> PlatformParser; 133 134 /// This is the current buffer index we're lexing from as managed by the 135 /// SourceMgr object. 136 unsigned CurBuffer; 137 138 AsmCond TheCondState; 139 std::vector<AsmCond> TheCondStack; 140 141 /// maps directive names to handler methods in parser 142 /// extensions. Extensions register themselves in this map by calling 143 /// addDirectiveHandler. 144 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 145 146 /// Stack of active macro instantiations. 147 std::vector<MacroInstantiation*> ActiveMacros; 148 149 /// List of bodies of anonymous macros. 150 std::deque<MCAsmMacro> MacroLikeBodies; 151 152 /// Boolean tracking whether macro substitution is enabled. 153 unsigned MacrosEnabledFlag : 1; 154 155 /// Keeps track of how many .macro's have been instantiated. 156 unsigned NumOfMacroInstantiations; 157 158 /// The values from the last parsed cpp hash file line comment if any. 159 struct CppHashInfoTy { 160 StringRef Filename; 161 int64_t LineNumber; 162 SMLoc Loc; 163 unsigned Buf; 164 CppHashInfoTy() : Filename(), LineNumber(0), Loc(), Buf(0) {} 165 }; 166 CppHashInfoTy CppHashInfo; 167 168 /// List of forward directional labels for diagnosis at the end. 169 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 170 171 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 172 unsigned AssemblerDialect = ~0U; 173 174 /// is Darwin compatibility enabled? 175 bool IsDarwin = false; 176 177 /// Are we parsing ms-style inline assembly? 178 bool ParsingInlineAsm = false; 179 180 /// Did we already inform the user about inconsistent MD5 usage? 181 bool ReportedInconsistentMD5 = false; 182 183 // Is alt macro mode enabled. 184 bool AltMacroMode = false; 185 186 public: 187 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 188 const MCAsmInfo &MAI, unsigned CB); 189 AsmParser(const AsmParser &) = delete; 190 AsmParser &operator=(const AsmParser &) = delete; 191 ~AsmParser() override; 192 193 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 194 195 void addDirectiveHandler(StringRef Directive, 196 ExtensionDirectiveHandler Handler) override { 197 ExtensionDirectiveMap[Directive] = Handler; 198 } 199 200 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 201 DirectiveKindMap[Directive] = DirectiveKindMap[Alias]; 202 } 203 204 /// @name MCAsmParser Interface 205 /// { 206 207 SourceMgr &getSourceManager() override { return SrcMgr; } 208 MCAsmLexer &getLexer() override { return Lexer; } 209 MCContext &getContext() override { return Ctx; } 210 MCStreamer &getStreamer() override { return Out; } 211 212 CodeViewContext &getCVContext() { return Ctx.getCVContext(); } 213 214 unsigned getAssemblerDialect() override { 215 if (AssemblerDialect == ~0U) 216 return MAI.getAssemblerDialect(); 217 else 218 return AssemblerDialect; 219 } 220 void setAssemblerDialect(unsigned i) override { 221 AssemblerDialect = i; 222 } 223 224 void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override; 225 bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override; 226 bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override; 227 228 const AsmToken &Lex() override; 229 230 void setParsingInlineAsm(bool V) override { 231 ParsingInlineAsm = V; 232 // When parsing MS inline asm, we must lex 0b1101 and 0ABCH as binary and 233 // hex integer literals. 234 Lexer.setLexMasmIntegers(V); 235 } 236 bool isParsingInlineAsm() override { return ParsingInlineAsm; } 237 238 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 239 unsigned &NumOutputs, unsigned &NumInputs, 240 SmallVectorImpl<std::pair<void *,bool>> &OpDecls, 241 SmallVectorImpl<std::string> &Constraints, 242 SmallVectorImpl<std::string> &Clobbers, 243 const MCInstrInfo *MII, const MCInstPrinter *IP, 244 MCAsmParserSemaCallback &SI) override; 245 246 bool parseExpression(const MCExpr *&Res); 247 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 248 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override; 249 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 250 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 251 SMLoc &EndLoc) override; 252 bool parseAbsoluteExpression(int64_t &Res) override; 253 254 /// Parse a floating point expression using the float \p Semantics 255 /// and set \p Res to the value. 256 bool parseRealValue(const fltSemantics &Semantics, APInt &Res); 257 258 /// Parse an identifier or string (as a quoted identifier) 259 /// and set \p Res to the identifier contents. 260 bool parseIdentifier(StringRef &Res) override; 261 void eatToEndOfStatement() override; 262 263 bool checkForValidSection() override; 264 265 /// } 266 267 private: 268 bool parseStatement(ParseStatementInfo &Info, 269 MCAsmParserSemaCallback *SI); 270 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 271 bool parseCppHashLineFilenameComment(SMLoc L); 272 273 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 274 ArrayRef<MCAsmMacroParameter> Parameters); 275 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 276 ArrayRef<MCAsmMacroParameter> Parameters, 277 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 278 SMLoc L); 279 280 /// Are macros enabled in the parser? 281 bool areMacrosEnabled() {return MacrosEnabledFlag;} 282 283 /// Control a flag in the parser that enables or disables macros. 284 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 285 286 /// Are we inside a macro instantiation? 287 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 288 289 /// Handle entry to macro instantiation. 290 /// 291 /// \param M The macro. 292 /// \param NameLoc Instantiation location. 293 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 294 295 /// Handle exit from macro instantiation. 296 void handleMacroExit(); 297 298 /// Extract AsmTokens for a macro argument. 299 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 300 301 /// Parse all macro arguments for a given macro. 302 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 303 304 void printMacroInstantiations(); 305 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 306 SMRange Range = None) const { 307 ArrayRef<SMRange> Ranges(Range); 308 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 309 } 310 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 311 312 /// Should we emit DWARF describing this assembler source? (Returns false if 313 /// the source has .file directives, which means we don't want to generate 314 /// info describing the assembler source itself.) 315 bool enabledGenDwarfForAssembly(); 316 317 /// Enter the specified file. This returns true on failure. 318 bool enterIncludeFile(const std::string &Filename); 319 320 /// Process the specified file for the .incbin directive. 321 /// This returns true on failure. 322 bool processIncbinFile(const std::string &Filename, int64_t Skip = 0, 323 const MCExpr *Count = nullptr, SMLoc Loc = SMLoc()); 324 325 /// Reset the current lexer position to that given by \p Loc. The 326 /// current token is not set; clients should ensure Lex() is called 327 /// subsequently. 328 /// 329 /// \param InBuffer If not 0, should be the known buffer id that contains the 330 /// location. 331 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 332 333 /// Parse up to the end of statement and a return the contents from the 334 /// current token until the end of the statement; the current token on exit 335 /// will be either the EndOfStatement or EOF. 336 StringRef parseStringToEndOfStatement() override; 337 338 /// Parse until the end of a statement or a comma is encountered, 339 /// return the contents from the current token up to the end or comma. 340 StringRef parseStringToComma(); 341 342 bool parseAssignment(StringRef Name, bool allow_redef, 343 bool NoDeadStrip = false); 344 345 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 346 MCBinaryExpr::Opcode &Kind); 347 348 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 349 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 350 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 351 352 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 353 354 bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName); 355 bool parseCVFileId(int64_t &FileId, StringRef DirectiveName); 356 357 // Generic (target and platform independent) directive parsing. 358 enum DirectiveKind { 359 DK_NO_DIRECTIVE, // Placeholder 360 DK_SET, 361 DK_EQU, 362 DK_EQUIV, 363 DK_ASCII, 364 DK_ASCIZ, 365 DK_STRING, 366 DK_BYTE, 367 DK_SHORT, 368 DK_RELOC, 369 DK_VALUE, 370 DK_2BYTE, 371 DK_LONG, 372 DK_INT, 373 DK_4BYTE, 374 DK_QUAD, 375 DK_8BYTE, 376 DK_OCTA, 377 DK_DC, 378 DK_DC_A, 379 DK_DC_B, 380 DK_DC_D, 381 DK_DC_L, 382 DK_DC_S, 383 DK_DC_W, 384 DK_DC_X, 385 DK_DCB, 386 DK_DCB_B, 387 DK_DCB_D, 388 DK_DCB_L, 389 DK_DCB_S, 390 DK_DCB_W, 391 DK_DCB_X, 392 DK_DS, 393 DK_DS_B, 394 DK_DS_D, 395 DK_DS_L, 396 DK_DS_P, 397 DK_DS_S, 398 DK_DS_W, 399 DK_DS_X, 400 DK_SINGLE, 401 DK_FLOAT, 402 DK_DOUBLE, 403 DK_ALIGN, 404 DK_ALIGN32, 405 DK_BALIGN, 406 DK_BALIGNW, 407 DK_BALIGNL, 408 DK_P2ALIGN, 409 DK_P2ALIGNW, 410 DK_P2ALIGNL, 411 DK_ORG, 412 DK_FILL, 413 DK_ENDR, 414 DK_BUNDLE_ALIGN_MODE, 415 DK_BUNDLE_LOCK, 416 DK_BUNDLE_UNLOCK, 417 DK_ZERO, 418 DK_EXTERN, 419 DK_GLOBL, 420 DK_GLOBAL, 421 DK_LAZY_REFERENCE, 422 DK_NO_DEAD_STRIP, 423 DK_SYMBOL_RESOLVER, 424 DK_PRIVATE_EXTERN, 425 DK_REFERENCE, 426 DK_WEAK_DEFINITION, 427 DK_WEAK_REFERENCE, 428 DK_WEAK_DEF_CAN_BE_HIDDEN, 429 DK_COLD, 430 DK_COMM, 431 DK_COMMON, 432 DK_LCOMM, 433 DK_ABORT, 434 DK_INCLUDE, 435 DK_INCBIN, 436 DK_CODE16, 437 DK_CODE16GCC, 438 DK_REPT, 439 DK_IRP, 440 DK_IRPC, 441 DK_IF, 442 DK_IFEQ, 443 DK_IFGE, 444 DK_IFGT, 445 DK_IFLE, 446 DK_IFLT, 447 DK_IFNE, 448 DK_IFB, 449 DK_IFNB, 450 DK_IFC, 451 DK_IFEQS, 452 DK_IFNC, 453 DK_IFNES, 454 DK_IFDEF, 455 DK_IFNDEF, 456 DK_IFNOTDEF, 457 DK_ELSEIF, 458 DK_ELSE, 459 DK_ENDIF, 460 DK_SPACE, 461 DK_SKIP, 462 DK_FILE, 463 DK_LINE, 464 DK_LOC, 465 DK_STABS, 466 DK_CV_FILE, 467 DK_CV_FUNC_ID, 468 DK_CV_INLINE_SITE_ID, 469 DK_CV_LOC, 470 DK_CV_LINETABLE, 471 DK_CV_INLINE_LINETABLE, 472 DK_CV_DEF_RANGE, 473 DK_CV_STRINGTABLE, 474 DK_CV_STRING, 475 DK_CV_FILECHECKSUMS, 476 DK_CV_FILECHECKSUM_OFFSET, 477 DK_CV_FPO_DATA, 478 DK_CFI_SECTIONS, 479 DK_CFI_STARTPROC, 480 DK_CFI_ENDPROC, 481 DK_CFI_DEF_CFA, 482 DK_CFI_DEF_CFA_OFFSET, 483 DK_CFI_ADJUST_CFA_OFFSET, 484 DK_CFI_DEF_CFA_REGISTER, 485 DK_CFI_OFFSET, 486 DK_CFI_REL_OFFSET, 487 DK_CFI_PERSONALITY, 488 DK_CFI_LSDA, 489 DK_CFI_REMEMBER_STATE, 490 DK_CFI_RESTORE_STATE, 491 DK_CFI_SAME_VALUE, 492 DK_CFI_RESTORE, 493 DK_CFI_ESCAPE, 494 DK_CFI_RETURN_COLUMN, 495 DK_CFI_SIGNAL_FRAME, 496 DK_CFI_UNDEFINED, 497 DK_CFI_REGISTER, 498 DK_CFI_WINDOW_SAVE, 499 DK_CFI_B_KEY_FRAME, 500 DK_MACROS_ON, 501 DK_MACROS_OFF, 502 DK_ALTMACRO, 503 DK_NOALTMACRO, 504 DK_MACRO, 505 DK_EXITM, 506 DK_ENDM, 507 DK_ENDMACRO, 508 DK_PURGEM, 509 DK_SLEB128, 510 DK_ULEB128, 511 DK_ERR, 512 DK_ERROR, 513 DK_WARNING, 514 DK_PRINT, 515 DK_ADDRSIG, 516 DK_ADDRSIG_SYM, 517 DK_END 518 }; 519 520 /// Maps directive name --> DirectiveKind enum, for 521 /// directives parsed by this class. 522 StringMap<DirectiveKind> DirectiveKindMap; 523 524 // ".ascii", ".asciz", ".string" 525 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 526 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 527 bool parseDirectiveValue(StringRef IDVal, 528 unsigned Size); // ".byte", ".long", ... 529 bool parseDirectiveOctaValue(StringRef IDVal); // ".octa", ... 530 bool parseDirectiveRealValue(StringRef IDVal, 531 const fltSemantics &); // ".single", ... 532 bool parseDirectiveFill(); // ".fill" 533 bool parseDirectiveZero(); // ".zero" 534 // ".set", ".equ", ".equiv" 535 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 536 bool parseDirectiveOrg(); // ".org" 537 // ".align{,32}", ".p2align{,w,l}" 538 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 539 540 // ".file", ".line", ".loc", ".stabs" 541 bool parseDirectiveFile(SMLoc DirectiveLoc); 542 bool parseDirectiveLine(); 543 bool parseDirectiveLoc(); 544 bool parseDirectiveStabs(); 545 546 // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable", 547 // ".cv_inline_linetable", ".cv_def_range", ".cv_string" 548 bool parseDirectiveCVFile(); 549 bool parseDirectiveCVFuncId(); 550 bool parseDirectiveCVInlineSiteId(); 551 bool parseDirectiveCVLoc(); 552 bool parseDirectiveCVLinetable(); 553 bool parseDirectiveCVInlineLinetable(); 554 bool parseDirectiveCVDefRange(); 555 bool parseDirectiveCVString(); 556 bool parseDirectiveCVStringTable(); 557 bool parseDirectiveCVFileChecksums(); 558 bool parseDirectiveCVFileChecksumOffset(); 559 bool parseDirectiveCVFPOData(); 560 561 // .cfi directives 562 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 563 bool parseDirectiveCFIWindowSave(); 564 bool parseDirectiveCFISections(); 565 bool parseDirectiveCFIStartProc(); 566 bool parseDirectiveCFIEndProc(); 567 bool parseDirectiveCFIDefCfaOffset(); 568 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 569 bool parseDirectiveCFIAdjustCfaOffset(); 570 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 571 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 572 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 573 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 574 bool parseDirectiveCFIRememberState(); 575 bool parseDirectiveCFIRestoreState(); 576 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 577 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 578 bool parseDirectiveCFIEscape(); 579 bool parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc); 580 bool parseDirectiveCFISignalFrame(); 581 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 582 583 // macro directives 584 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 585 bool parseDirectiveExitMacro(StringRef Directive); 586 bool parseDirectiveEndMacro(StringRef Directive); 587 bool parseDirectiveMacro(SMLoc DirectiveLoc); 588 bool parseDirectiveMacrosOnOff(StringRef Directive); 589 // alternate macro mode directives 590 bool parseDirectiveAltmacro(StringRef Directive); 591 // ".bundle_align_mode" 592 bool parseDirectiveBundleAlignMode(); 593 // ".bundle_lock" 594 bool parseDirectiveBundleLock(); 595 // ".bundle_unlock" 596 bool parseDirectiveBundleUnlock(); 597 598 // ".space", ".skip" 599 bool parseDirectiveSpace(StringRef IDVal); 600 601 // ".dcb" 602 bool parseDirectiveDCB(StringRef IDVal, unsigned Size); 603 bool parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &); 604 // ".ds" 605 bool parseDirectiveDS(StringRef IDVal, unsigned Size); 606 607 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 608 bool parseDirectiveLEB128(bool Signed); 609 610 /// Parse a directive like ".globl" which 611 /// accepts a single symbol (which should be a label or an external). 612 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 613 614 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 615 616 bool parseDirectiveAbort(); // ".abort" 617 bool parseDirectiveInclude(); // ".include" 618 bool parseDirectiveIncbin(); // ".incbin" 619 620 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 621 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 622 // ".ifb" or ".ifnb", depending on ExpectBlank. 623 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 624 // ".ifc" or ".ifnc", depending on ExpectEqual. 625 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 626 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 627 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 628 // ".ifdef" or ".ifndef", depending on expect_defined 629 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 630 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 631 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 632 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 633 bool parseEscapedString(std::string &Data) override; 634 635 const MCExpr *applyModifierToExpr(const MCExpr *E, 636 MCSymbolRefExpr::VariantKind Variant); 637 638 // Macro-like directives 639 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 640 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 641 raw_svector_ostream &OS); 642 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 643 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 644 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 645 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 646 647 // "_emit" or "__emit" 648 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 649 size_t Len); 650 651 // "align" 652 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 653 654 // "end" 655 bool parseDirectiveEnd(SMLoc DirectiveLoc); 656 657 // ".err" or ".error" 658 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 659 660 // ".warning" 661 bool parseDirectiveWarning(SMLoc DirectiveLoc); 662 663 // .print <double-quotes-string> 664 bool parseDirectivePrint(SMLoc DirectiveLoc); 665 666 // Directives to support address-significance tables. 667 bool parseDirectiveAddrsig(); 668 bool parseDirectiveAddrsigSym(); 669 670 void initializeDirectiveKindMap(); 671 }; 672 673 } // end anonymous namespace 674 675 namespace llvm { 676 677 extern MCAsmParserExtension *createDarwinAsmParser(); 678 extern MCAsmParserExtension *createELFAsmParser(); 679 extern MCAsmParserExtension *createCOFFAsmParser(); 680 extern MCAsmParserExtension *createWasmAsmParser(); 681 682 } // end namespace llvm 683 684 enum { DEFAULT_ADDRSPACE = 0 }; 685 686 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 687 const MCAsmInfo &MAI, unsigned CB = 0) 688 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 689 CurBuffer(CB ? CB : SM.getMainFileID()), MacrosEnabledFlag(true) { 690 HadError = false; 691 // Save the old handler. 692 SavedDiagHandler = SrcMgr.getDiagHandler(); 693 SavedDiagContext = SrcMgr.getDiagContext(); 694 // Set our own handler which calls the saved handler. 695 SrcMgr.setDiagHandler(DiagHandler, this); 696 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 697 698 // Initialize the platform / file format parser. 699 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 700 case MCObjectFileInfo::IsCOFF: 701 PlatformParser.reset(createCOFFAsmParser()); 702 break; 703 case MCObjectFileInfo::IsMachO: 704 PlatformParser.reset(createDarwinAsmParser()); 705 IsDarwin = true; 706 break; 707 case MCObjectFileInfo::IsELF: 708 PlatformParser.reset(createELFAsmParser()); 709 break; 710 case MCObjectFileInfo::IsWasm: 711 PlatformParser.reset(createWasmAsmParser()); 712 break; 713 case MCObjectFileInfo::IsXCOFF: 714 // TODO: Need to implement createXCOFFAsmParser for XCOFF format. 715 break; 716 } 717 718 PlatformParser->Initialize(*this); 719 initializeDirectiveKindMap(); 720 721 NumOfMacroInstantiations = 0; 722 } 723 724 AsmParser::~AsmParser() { 725 assert((HadError || ActiveMacros.empty()) && 726 "Unexpected active macro instantiation!"); 727 728 // Restore the saved diagnostics handler and context for use during 729 // finalization. 730 SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext); 731 } 732 733 void AsmParser::printMacroInstantiations() { 734 // Print the active macro instantiation stack. 735 for (std::vector<MacroInstantiation *>::const_reverse_iterator 736 it = ActiveMacros.rbegin(), 737 ie = ActiveMacros.rend(); 738 it != ie; ++it) 739 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 740 "while in macro instantiation"); 741 } 742 743 void AsmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) { 744 printPendingErrors(); 745 printMessage(L, SourceMgr::DK_Note, Msg, Range); 746 printMacroInstantiations(); 747 } 748 749 bool AsmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) { 750 if(getTargetParser().getTargetOptions().MCNoWarn) 751 return false; 752 if (getTargetParser().getTargetOptions().MCFatalWarnings) 753 return Error(L, Msg, Range); 754 printMessage(L, SourceMgr::DK_Warning, Msg, Range); 755 printMacroInstantiations(); 756 return false; 757 } 758 759 bool AsmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) { 760 HadError = true; 761 printMessage(L, SourceMgr::DK_Error, Msg, Range); 762 printMacroInstantiations(); 763 return true; 764 } 765 766 bool AsmParser::enterIncludeFile(const std::string &Filename) { 767 std::string IncludedFile; 768 unsigned NewBuf = 769 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 770 if (!NewBuf) 771 return true; 772 773 CurBuffer = NewBuf; 774 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 775 return false; 776 } 777 778 /// Process the specified .incbin file by searching for it in the include paths 779 /// then just emitting the byte contents of the file to the streamer. This 780 /// returns true on failure. 781 bool AsmParser::processIncbinFile(const std::string &Filename, int64_t Skip, 782 const MCExpr *Count, SMLoc Loc) { 783 std::string IncludedFile; 784 unsigned NewBuf = 785 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 786 if (!NewBuf) 787 return true; 788 789 // Pick up the bytes from the file and emit them. 790 StringRef Bytes = SrcMgr.getMemoryBuffer(NewBuf)->getBuffer(); 791 Bytes = Bytes.drop_front(Skip); 792 if (Count) { 793 int64_t Res; 794 if (!Count->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 795 return Error(Loc, "expected absolute expression"); 796 if (Res < 0) 797 return Warning(Loc, "negative count has no effect"); 798 Bytes = Bytes.take_front(Res); 799 } 800 getStreamer().EmitBytes(Bytes); 801 return false; 802 } 803 804 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 805 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 806 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 807 Loc.getPointer()); 808 } 809 810 const AsmToken &AsmParser::Lex() { 811 if (Lexer.getTok().is(AsmToken::Error)) 812 Error(Lexer.getErrLoc(), Lexer.getErr()); 813 814 // if it's a end of statement with a comment in it 815 if (getTok().is(AsmToken::EndOfStatement)) { 816 // if this is a line comment output it. 817 if (!getTok().getString().empty() && getTok().getString().front() != '\n' && 818 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 819 Out.addExplicitComment(Twine(getTok().getString())); 820 } 821 822 const AsmToken *tok = &Lexer.Lex(); 823 824 // Parse comments here to be deferred until end of next statement. 825 while (tok->is(AsmToken::Comment)) { 826 if (MAI.preserveAsmComments()) 827 Out.addExplicitComment(Twine(tok->getString())); 828 tok = &Lexer.Lex(); 829 } 830 831 if (tok->is(AsmToken::Eof)) { 832 // If this is the end of an included file, pop the parent file off the 833 // include stack. 834 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 835 if (ParentIncludeLoc != SMLoc()) { 836 jumpToLoc(ParentIncludeLoc); 837 return Lex(); 838 } 839 } 840 841 return *tok; 842 } 843 844 bool AsmParser::enabledGenDwarfForAssembly() { 845 // Check whether the user specified -g. 846 if (!getContext().getGenDwarfForAssembly()) 847 return false; 848 // If we haven't encountered any .file directives (which would imply that 849 // the assembler source was produced with debug info already) then emit one 850 // describing the assembler source file itself. 851 if (getContext().getGenDwarfFileNumber() == 0) { 852 const MCDwarfFile &RootFile = 853 getContext().getMCDwarfLineTable(/*CUID=*/0).getRootFile(); 854 getContext().setGenDwarfFileNumber(getStreamer().EmitDwarfFileDirective( 855 /*CUID=*/0, getContext().getCompilationDir(), RootFile.Name, 856 RootFile.Checksum, RootFile.Source)); 857 } 858 return true; 859 } 860 861 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 862 // Create the initial section, if requested. 863 if (!NoInitialTextSection) 864 Out.InitSections(false); 865 866 // Prime the lexer. 867 Lex(); 868 869 HadError = false; 870 AsmCond StartingCondState = TheCondState; 871 SmallVector<AsmRewrite, 4> AsmStrRewrites; 872 873 // If we are generating dwarf for assembly source files save the initial text 874 // section. (Don't use enabledGenDwarfForAssembly() here, as we aren't 875 // emitting any actual debug info yet and haven't had a chance to parse any 876 // embedded .file directives.) 877 if (getContext().getGenDwarfForAssembly()) { 878 MCSection *Sec = getStreamer().getCurrentSectionOnly(); 879 if (!Sec->getBeginSymbol()) { 880 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 881 getStreamer().EmitLabel(SectionStartSym); 882 Sec->setBeginSymbol(SectionStartSym); 883 } 884 bool InsertResult = getContext().addGenDwarfSection(Sec); 885 assert(InsertResult && ".text section should not have debug info yet"); 886 (void)InsertResult; 887 } 888 889 // While we have input, parse each statement. 890 while (Lexer.isNot(AsmToken::Eof)) { 891 ParseStatementInfo Info(&AsmStrRewrites); 892 if (!parseStatement(Info, nullptr)) 893 continue; 894 895 // If we have a Lexer Error we are on an Error Token. Load in Lexer Error 896 // for printing ErrMsg via Lex() only if no (presumably better) parser error 897 // exists. 898 if (!hasPendingError() && Lexer.getTok().is(AsmToken::Error)) { 899 Lex(); 900 } 901 902 // parseStatement returned true so may need to emit an error. 903 printPendingErrors(); 904 905 // Skipping to the next line if needed. 906 if (!getLexer().isAtStartOfStatement()) 907 eatToEndOfStatement(); 908 } 909 910 getTargetParser().onEndOfFile(); 911 printPendingErrors(); 912 913 // All errors should have been emitted. 914 assert(!hasPendingError() && "unexpected error from parseStatement"); 915 916 getTargetParser().flushPendingInstructions(getStreamer()); 917 918 if (TheCondState.TheCond != StartingCondState.TheCond || 919 TheCondState.Ignore != StartingCondState.Ignore) 920 printError(getTok().getLoc(), "unmatched .ifs or .elses"); 921 // Check to see there are no empty DwarfFile slots. 922 const auto &LineTables = getContext().getMCDwarfLineTables(); 923 if (!LineTables.empty()) { 924 unsigned Index = 0; 925 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 926 if (File.Name.empty() && Index != 0) 927 printError(getTok().getLoc(), "unassigned file number: " + 928 Twine(Index) + 929 " for .file directives"); 930 ++Index; 931 } 932 } 933 934 // Check to see that all assembler local symbols were actually defined. 935 // Targets that don't do subsections via symbols may not want this, though, 936 // so conservatively exclude them. Only do this if we're finalizing, though, 937 // as otherwise we won't necessarilly have seen everything yet. 938 if (!NoFinalize) { 939 if (MAI.hasSubsectionsViaSymbols()) { 940 for (const auto &TableEntry : getContext().getSymbols()) { 941 MCSymbol *Sym = TableEntry.getValue(); 942 // Variable symbols may not be marked as defined, so check those 943 // explicitly. If we know it's a variable, we have a definition for 944 // the purposes of this check. 945 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 946 // FIXME: We would really like to refer back to where the symbol was 947 // first referenced for a source location. We need to add something 948 // to track that. Currently, we just point to the end of the file. 949 printError(getTok().getLoc(), "assembler local symbol '" + 950 Sym->getName() + "' not defined"); 951 } 952 } 953 954 // Temporary symbols like the ones for directional jumps don't go in the 955 // symbol table. They also need to be diagnosed in all (final) cases. 956 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 957 if (std::get<2>(LocSym)->isUndefined()) { 958 // Reset the state of any "# line file" directives we've seen to the 959 // context as it was at the diagnostic site. 960 CppHashInfo = std::get<1>(LocSym); 961 printError(std::get<0>(LocSym), "directional label undefined"); 962 } 963 } 964 } 965 966 // Finalize the output stream if there are no errors and if the client wants 967 // us to. 968 if (!HadError && !NoFinalize) 969 Out.Finish(); 970 971 return HadError || getContext().hadError(); 972 } 973 974 bool AsmParser::checkForValidSection() { 975 if (!ParsingInlineAsm && !getStreamer().getCurrentSectionOnly()) { 976 Out.InitSections(false); 977 return Error(getTok().getLoc(), 978 "expected section directive before assembly directive"); 979 } 980 return false; 981 } 982 983 /// Throw away the rest of the line for testing purposes. 984 void AsmParser::eatToEndOfStatement() { 985 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 986 Lexer.Lex(); 987 988 // Eat EOL. 989 if (Lexer.is(AsmToken::EndOfStatement)) 990 Lexer.Lex(); 991 } 992 993 StringRef AsmParser::parseStringToEndOfStatement() { 994 const char *Start = getTok().getLoc().getPointer(); 995 996 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 997 Lexer.Lex(); 998 999 const char *End = getTok().getLoc().getPointer(); 1000 return StringRef(Start, End - Start); 1001 } 1002 1003 StringRef AsmParser::parseStringToComma() { 1004 const char *Start = getTok().getLoc().getPointer(); 1005 1006 while (Lexer.isNot(AsmToken::EndOfStatement) && 1007 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 1008 Lexer.Lex(); 1009 1010 const char *End = getTok().getLoc().getPointer(); 1011 return StringRef(Start, End - Start); 1012 } 1013 1014 /// Parse a paren expression and return it. 1015 /// NOTE: This assumes the leading '(' has already been consumed. 1016 /// 1017 /// parenexpr ::= expr) 1018 /// 1019 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1020 if (parseExpression(Res)) 1021 return true; 1022 if (Lexer.isNot(AsmToken::RParen)) 1023 return TokError("expected ')' in parentheses expression"); 1024 EndLoc = Lexer.getTok().getEndLoc(); 1025 Lex(); 1026 return false; 1027 } 1028 1029 /// Parse a bracket expression and return it. 1030 /// NOTE: This assumes the leading '[' has already been consumed. 1031 /// 1032 /// bracketexpr ::= expr] 1033 /// 1034 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1035 if (parseExpression(Res)) 1036 return true; 1037 EndLoc = getTok().getEndLoc(); 1038 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 1039 return true; 1040 return false; 1041 } 1042 1043 /// Parse a primary expression and return it. 1044 /// primaryexpr ::= (parenexpr 1045 /// primaryexpr ::= symbol 1046 /// primaryexpr ::= number 1047 /// primaryexpr ::= '.' 1048 /// primaryexpr ::= ~,+,- primaryexpr 1049 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1050 SMLoc FirstTokenLoc = getLexer().getLoc(); 1051 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 1052 switch (FirstTokenKind) { 1053 default: 1054 return TokError("unknown token in expression"); 1055 // If we have an error assume that we've already handled it. 1056 case AsmToken::Error: 1057 return true; 1058 case AsmToken::Exclaim: 1059 Lex(); // Eat the operator. 1060 if (parsePrimaryExpr(Res, EndLoc)) 1061 return true; 1062 Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc); 1063 return false; 1064 case AsmToken::Dollar: 1065 case AsmToken::At: 1066 case AsmToken::String: 1067 case AsmToken::Identifier: { 1068 StringRef Identifier; 1069 if (parseIdentifier(Identifier)) { 1070 // We may have failed but $ may be a valid token. 1071 if (getTok().is(AsmToken::Dollar)) { 1072 if (Lexer.getMAI().getDollarIsPC()) { 1073 Lex(); 1074 // This is a '$' reference, which references the current PC. Emit a 1075 // temporary label to the streamer and refer to it. 1076 MCSymbol *Sym = Ctx.createTempSymbol(); 1077 Out.EmitLabel(Sym); 1078 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 1079 getContext()); 1080 EndLoc = FirstTokenLoc; 1081 return false; 1082 } 1083 return Error(FirstTokenLoc, "invalid token in expression"); 1084 } 1085 } 1086 // Parse symbol variant 1087 std::pair<StringRef, StringRef> Split; 1088 if (!MAI.useParensForSymbolVariant()) { 1089 if (FirstTokenKind == AsmToken::String) { 1090 if (Lexer.is(AsmToken::At)) { 1091 Lex(); // eat @ 1092 SMLoc AtLoc = getLexer().getLoc(); 1093 StringRef VName; 1094 if (parseIdentifier(VName)) 1095 return Error(AtLoc, "expected symbol variant after '@'"); 1096 1097 Split = std::make_pair(Identifier, VName); 1098 } 1099 } else { 1100 Split = Identifier.split('@'); 1101 } 1102 } else if (Lexer.is(AsmToken::LParen)) { 1103 Lex(); // eat '('. 1104 StringRef VName; 1105 parseIdentifier(VName); 1106 // eat ')'. 1107 if (parseToken(AsmToken::RParen, 1108 "unexpected token in variant, expected ')'")) 1109 return true; 1110 Split = std::make_pair(Identifier, VName); 1111 } 1112 1113 EndLoc = SMLoc::getFromPointer(Identifier.end()); 1114 1115 // This is a symbol reference. 1116 StringRef SymbolName = Identifier; 1117 if (SymbolName.empty()) 1118 return Error(getLexer().getLoc(), "expected a symbol reference"); 1119 1120 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1121 1122 // Lookup the symbol variant if used. 1123 if (!Split.second.empty()) { 1124 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1125 if (Variant != MCSymbolRefExpr::VK_Invalid) { 1126 SymbolName = Split.first; 1127 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 1128 Variant = MCSymbolRefExpr::VK_None; 1129 } else { 1130 return Error(SMLoc::getFromPointer(Split.second.begin()), 1131 "invalid variant '" + Split.second + "'"); 1132 } 1133 } 1134 1135 MCSymbol *Sym = getContext().getOrCreateSymbol(SymbolName); 1136 1137 // If this is an absolute variable reference, substitute it now to preserve 1138 // semantics in the face of reassignment. 1139 if (Sym->isVariable()) { 1140 auto V = Sym->getVariableValue(/*SetUsed*/ false); 1141 bool DoInline = isa<MCConstantExpr>(V) && !Variant; 1142 if (auto TV = dyn_cast<MCTargetExpr>(V)) 1143 DoInline = TV->inlineAssignedExpr(); 1144 if (DoInline) { 1145 if (Variant) 1146 return Error(EndLoc, "unexpected modifier on variable reference"); 1147 Res = Sym->getVariableValue(/*SetUsed*/ false); 1148 return false; 1149 } 1150 } 1151 1152 // Otherwise create a symbol ref. 1153 Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc); 1154 return false; 1155 } 1156 case AsmToken::BigNum: 1157 return TokError("literal value out of range for directive"); 1158 case AsmToken::Integer: { 1159 SMLoc Loc = getTok().getLoc(); 1160 int64_t IntVal = getTok().getIntVal(); 1161 Res = MCConstantExpr::create(IntVal, getContext()); 1162 EndLoc = Lexer.getTok().getEndLoc(); 1163 Lex(); // Eat token. 1164 // Look for 'b' or 'f' following an Integer as a directional label 1165 if (Lexer.getKind() == AsmToken::Identifier) { 1166 StringRef IDVal = getTok().getString(); 1167 // Lookup the symbol variant if used. 1168 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 1169 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1170 if (Split.first.size() != IDVal.size()) { 1171 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1172 if (Variant == MCSymbolRefExpr::VK_Invalid) 1173 return TokError("invalid variant '" + Split.second + "'"); 1174 IDVal = Split.first; 1175 } 1176 if (IDVal == "f" || IDVal == "b") { 1177 MCSymbol *Sym = 1178 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 1179 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 1180 if (IDVal == "b" && Sym->isUndefined()) 1181 return Error(Loc, "directional label undefined"); 1182 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 1183 EndLoc = Lexer.getTok().getEndLoc(); 1184 Lex(); // Eat identifier. 1185 } 1186 } 1187 return false; 1188 } 1189 case AsmToken::Real: { 1190 APFloat RealVal(APFloat::IEEEdouble(), getTok().getString()); 1191 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1192 Res = MCConstantExpr::create(IntVal, getContext()); 1193 EndLoc = Lexer.getTok().getEndLoc(); 1194 Lex(); // Eat token. 1195 return false; 1196 } 1197 case AsmToken::Dot: { 1198 // This is a '.' reference, which references the current PC. Emit a 1199 // temporary label to the streamer and refer to it. 1200 MCSymbol *Sym = Ctx.createTempSymbol(); 1201 Out.EmitLabel(Sym); 1202 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1203 EndLoc = Lexer.getTok().getEndLoc(); 1204 Lex(); // Eat identifier. 1205 return false; 1206 } 1207 case AsmToken::LParen: 1208 Lex(); // Eat the '('. 1209 return parseParenExpr(Res, EndLoc); 1210 case AsmToken::LBrac: 1211 if (!PlatformParser->HasBracketExpressions()) 1212 return TokError("brackets expression not supported on this target"); 1213 Lex(); // Eat the '['. 1214 return parseBracketExpr(Res, EndLoc); 1215 case AsmToken::Minus: 1216 Lex(); // Eat the operator. 1217 if (parsePrimaryExpr(Res, EndLoc)) 1218 return true; 1219 Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc); 1220 return false; 1221 case AsmToken::Plus: 1222 Lex(); // Eat the operator. 1223 if (parsePrimaryExpr(Res, EndLoc)) 1224 return true; 1225 Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc); 1226 return false; 1227 case AsmToken::Tilde: 1228 Lex(); // Eat the operator. 1229 if (parsePrimaryExpr(Res, EndLoc)) 1230 return true; 1231 Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc); 1232 return false; 1233 // MIPS unary expression operators. The lexer won't generate these tokens if 1234 // MCAsmInfo::HasMipsExpressions is false for the target. 1235 case AsmToken::PercentCall16: 1236 case AsmToken::PercentCall_Hi: 1237 case AsmToken::PercentCall_Lo: 1238 case AsmToken::PercentDtprel_Hi: 1239 case AsmToken::PercentDtprel_Lo: 1240 case AsmToken::PercentGot: 1241 case AsmToken::PercentGot_Disp: 1242 case AsmToken::PercentGot_Hi: 1243 case AsmToken::PercentGot_Lo: 1244 case AsmToken::PercentGot_Ofst: 1245 case AsmToken::PercentGot_Page: 1246 case AsmToken::PercentGottprel: 1247 case AsmToken::PercentGp_Rel: 1248 case AsmToken::PercentHi: 1249 case AsmToken::PercentHigher: 1250 case AsmToken::PercentHighest: 1251 case AsmToken::PercentLo: 1252 case AsmToken::PercentNeg: 1253 case AsmToken::PercentPcrel_Hi: 1254 case AsmToken::PercentPcrel_Lo: 1255 case AsmToken::PercentTlsgd: 1256 case AsmToken::PercentTlsldm: 1257 case AsmToken::PercentTprel_Hi: 1258 case AsmToken::PercentTprel_Lo: 1259 Lex(); // Eat the operator. 1260 if (Lexer.isNot(AsmToken::LParen)) 1261 return TokError("expected '(' after operator"); 1262 Lex(); // Eat the operator. 1263 if (parseExpression(Res, EndLoc)) 1264 return true; 1265 if (Lexer.isNot(AsmToken::RParen)) 1266 return TokError("expected ')'"); 1267 Lex(); // Eat the operator. 1268 Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx); 1269 return !Res; 1270 } 1271 } 1272 1273 bool AsmParser::parseExpression(const MCExpr *&Res) { 1274 SMLoc EndLoc; 1275 return parseExpression(Res, EndLoc); 1276 } 1277 1278 const MCExpr * 1279 AsmParser::applyModifierToExpr(const MCExpr *E, 1280 MCSymbolRefExpr::VariantKind Variant) { 1281 // Ask the target implementation about this expression first. 1282 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 1283 if (NewE) 1284 return NewE; 1285 // Recurse over the given expression, rebuilding it to apply the given variant 1286 // if there is exactly one symbol. 1287 switch (E->getKind()) { 1288 case MCExpr::Target: 1289 case MCExpr::Constant: 1290 return nullptr; 1291 1292 case MCExpr::SymbolRef: { 1293 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 1294 1295 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 1296 TokError("invalid variant on expression '" + getTok().getIdentifier() + 1297 "' (already modified)"); 1298 return E; 1299 } 1300 1301 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1302 } 1303 1304 case MCExpr::Unary: { 1305 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1306 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1307 if (!Sub) 1308 return nullptr; 1309 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1310 } 1311 1312 case MCExpr::Binary: { 1313 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1314 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1315 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1316 1317 if (!LHS && !RHS) 1318 return nullptr; 1319 1320 if (!LHS) 1321 LHS = BE->getLHS(); 1322 if (!RHS) 1323 RHS = BE->getRHS(); 1324 1325 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1326 } 1327 } 1328 1329 llvm_unreachable("Invalid expression kind!"); 1330 } 1331 1332 /// This function checks if the next token is <string> type or arithmetic. 1333 /// string that begin with character '<' must end with character '>'. 1334 /// otherwise it is arithmetics. 1335 /// If the function returns a 'true' value, 1336 /// the End argument will be filled with the last location pointed to the '>' 1337 /// character. 1338 1339 /// There is a gap between the AltMacro's documentation and the single quote 1340 /// implementation. GCC does not fully support this feature and so we will not 1341 /// support it. 1342 /// TODO: Adding single quote as a string. 1343 static bool isAltmacroString(SMLoc &StrLoc, SMLoc &EndLoc) { 1344 assert((StrLoc.getPointer() != nullptr) && 1345 "Argument to the function cannot be a NULL value"); 1346 const char *CharPtr = StrLoc.getPointer(); 1347 while ((*CharPtr != '>') && (*CharPtr != '\n') && (*CharPtr != '\r') && 1348 (*CharPtr != '\0')) { 1349 if (*CharPtr == '!') 1350 CharPtr++; 1351 CharPtr++; 1352 } 1353 if (*CharPtr == '>') { 1354 EndLoc = StrLoc.getFromPointer(CharPtr + 1); 1355 return true; 1356 } 1357 return false; 1358 } 1359 1360 /// creating a string without the escape characters '!'. 1361 static std::string altMacroString(StringRef AltMacroStr) { 1362 std::string Res; 1363 for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) { 1364 if (AltMacroStr[Pos] == '!') 1365 Pos++; 1366 Res += AltMacroStr[Pos]; 1367 } 1368 return Res; 1369 } 1370 1371 /// Parse an expression and return it. 1372 /// 1373 /// expr ::= expr &&,|| expr -> lowest. 1374 /// expr ::= expr |,^,&,! expr 1375 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1376 /// expr ::= expr <<,>> expr 1377 /// expr ::= expr +,- expr 1378 /// expr ::= expr *,/,% expr -> highest. 1379 /// expr ::= primaryexpr 1380 /// 1381 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1382 // Parse the expression. 1383 Res = nullptr; 1384 if (getTargetParser().parsePrimaryExpr(Res, EndLoc) || 1385 parseBinOpRHS(1, Res, EndLoc)) 1386 return true; 1387 1388 // As a special case, we support 'a op b @ modifier' by rewriting the 1389 // expression to include the modifier. This is inefficient, but in general we 1390 // expect users to use 'a@modifier op b'. 1391 if (Lexer.getKind() == AsmToken::At) { 1392 Lex(); 1393 1394 if (Lexer.isNot(AsmToken::Identifier)) 1395 return TokError("unexpected symbol modifier following '@'"); 1396 1397 MCSymbolRefExpr::VariantKind Variant = 1398 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1399 if (Variant == MCSymbolRefExpr::VK_Invalid) 1400 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1401 1402 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1403 if (!ModifiedRes) { 1404 return TokError("invalid modifier '" + getTok().getIdentifier() + 1405 "' (no symbols present)"); 1406 } 1407 1408 Res = ModifiedRes; 1409 Lex(); 1410 } 1411 1412 // Try to constant fold it up front, if possible. Do not exploit 1413 // assembler here. 1414 int64_t Value; 1415 if (Res->evaluateAsAbsolute(Value)) 1416 Res = MCConstantExpr::create(Value, getContext()); 1417 1418 return false; 1419 } 1420 1421 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1422 Res = nullptr; 1423 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1424 } 1425 1426 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1427 SMLoc &EndLoc) { 1428 if (parseParenExpr(Res, EndLoc)) 1429 return true; 1430 1431 for (; ParenDepth > 0; --ParenDepth) { 1432 if (parseBinOpRHS(1, Res, EndLoc)) 1433 return true; 1434 1435 // We don't Lex() the last RParen. 1436 // This is the same behavior as parseParenExpression(). 1437 if (ParenDepth - 1 > 0) { 1438 EndLoc = getTok().getEndLoc(); 1439 if (parseToken(AsmToken::RParen, 1440 "expected ')' in parentheses expression")) 1441 return true; 1442 } 1443 } 1444 return false; 1445 } 1446 1447 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1448 const MCExpr *Expr; 1449 1450 SMLoc StartLoc = Lexer.getLoc(); 1451 if (parseExpression(Expr)) 1452 return true; 1453 1454 if (!Expr->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 1455 return Error(StartLoc, "expected absolute expression"); 1456 1457 return false; 1458 } 1459 1460 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1461 MCBinaryExpr::Opcode &Kind, 1462 bool ShouldUseLogicalShr) { 1463 switch (K) { 1464 default: 1465 return 0; // not a binop. 1466 1467 // Lowest Precedence: &&, || 1468 case AsmToken::AmpAmp: 1469 Kind = MCBinaryExpr::LAnd; 1470 return 1; 1471 case AsmToken::PipePipe: 1472 Kind = MCBinaryExpr::LOr; 1473 return 1; 1474 1475 // Low Precedence: |, &, ^ 1476 // 1477 // FIXME: gas seems to support '!' as an infix operator? 1478 case AsmToken::Pipe: 1479 Kind = MCBinaryExpr::Or; 1480 return 2; 1481 case AsmToken::Caret: 1482 Kind = MCBinaryExpr::Xor; 1483 return 2; 1484 case AsmToken::Amp: 1485 Kind = MCBinaryExpr::And; 1486 return 2; 1487 1488 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1489 case AsmToken::EqualEqual: 1490 Kind = MCBinaryExpr::EQ; 1491 return 3; 1492 case AsmToken::ExclaimEqual: 1493 case AsmToken::LessGreater: 1494 Kind = MCBinaryExpr::NE; 1495 return 3; 1496 case AsmToken::Less: 1497 Kind = MCBinaryExpr::LT; 1498 return 3; 1499 case AsmToken::LessEqual: 1500 Kind = MCBinaryExpr::LTE; 1501 return 3; 1502 case AsmToken::Greater: 1503 Kind = MCBinaryExpr::GT; 1504 return 3; 1505 case AsmToken::GreaterEqual: 1506 Kind = MCBinaryExpr::GTE; 1507 return 3; 1508 1509 // Intermediate Precedence: <<, >> 1510 case AsmToken::LessLess: 1511 Kind = MCBinaryExpr::Shl; 1512 return 4; 1513 case AsmToken::GreaterGreater: 1514 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1515 return 4; 1516 1517 // High Intermediate Precedence: +, - 1518 case AsmToken::Plus: 1519 Kind = MCBinaryExpr::Add; 1520 return 5; 1521 case AsmToken::Minus: 1522 Kind = MCBinaryExpr::Sub; 1523 return 5; 1524 1525 // Highest Precedence: *, /, % 1526 case AsmToken::Star: 1527 Kind = MCBinaryExpr::Mul; 1528 return 6; 1529 case AsmToken::Slash: 1530 Kind = MCBinaryExpr::Div; 1531 return 6; 1532 case AsmToken::Percent: 1533 Kind = MCBinaryExpr::Mod; 1534 return 6; 1535 } 1536 } 1537 1538 static unsigned getGNUBinOpPrecedence(AsmToken::TokenKind K, 1539 MCBinaryExpr::Opcode &Kind, 1540 bool ShouldUseLogicalShr) { 1541 switch (K) { 1542 default: 1543 return 0; // not a binop. 1544 1545 // Lowest Precedence: &&, || 1546 case AsmToken::AmpAmp: 1547 Kind = MCBinaryExpr::LAnd; 1548 return 2; 1549 case AsmToken::PipePipe: 1550 Kind = MCBinaryExpr::LOr; 1551 return 1; 1552 1553 // Low Precedence: ==, !=, <>, <, <=, >, >= 1554 case AsmToken::EqualEqual: 1555 Kind = MCBinaryExpr::EQ; 1556 return 3; 1557 case AsmToken::ExclaimEqual: 1558 case AsmToken::LessGreater: 1559 Kind = MCBinaryExpr::NE; 1560 return 3; 1561 case AsmToken::Less: 1562 Kind = MCBinaryExpr::LT; 1563 return 3; 1564 case AsmToken::LessEqual: 1565 Kind = MCBinaryExpr::LTE; 1566 return 3; 1567 case AsmToken::Greater: 1568 Kind = MCBinaryExpr::GT; 1569 return 3; 1570 case AsmToken::GreaterEqual: 1571 Kind = MCBinaryExpr::GTE; 1572 return 3; 1573 1574 // Low Intermediate Precedence: +, - 1575 case AsmToken::Plus: 1576 Kind = MCBinaryExpr::Add; 1577 return 4; 1578 case AsmToken::Minus: 1579 Kind = MCBinaryExpr::Sub; 1580 return 4; 1581 1582 // High Intermediate Precedence: |, &, ^ 1583 // 1584 // FIXME: gas seems to support '!' as an infix operator? 1585 case AsmToken::Pipe: 1586 Kind = MCBinaryExpr::Or; 1587 return 5; 1588 case AsmToken::Caret: 1589 Kind = MCBinaryExpr::Xor; 1590 return 5; 1591 case AsmToken::Amp: 1592 Kind = MCBinaryExpr::And; 1593 return 5; 1594 1595 // Highest Precedence: *, /, %, <<, >> 1596 case AsmToken::Star: 1597 Kind = MCBinaryExpr::Mul; 1598 return 6; 1599 case AsmToken::Slash: 1600 Kind = MCBinaryExpr::Div; 1601 return 6; 1602 case AsmToken::Percent: 1603 Kind = MCBinaryExpr::Mod; 1604 return 6; 1605 case AsmToken::LessLess: 1606 Kind = MCBinaryExpr::Shl; 1607 return 6; 1608 case AsmToken::GreaterGreater: 1609 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1610 return 6; 1611 } 1612 } 1613 1614 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1615 MCBinaryExpr::Opcode &Kind) { 1616 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1617 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1618 : getGNUBinOpPrecedence(K, Kind, ShouldUseLogicalShr); 1619 } 1620 1621 /// Parse all binary operators with precedence >= 'Precedence'. 1622 /// Res contains the LHS of the expression on input. 1623 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1624 SMLoc &EndLoc) { 1625 SMLoc StartLoc = Lexer.getLoc(); 1626 while (true) { 1627 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1628 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1629 1630 // If the next token is lower precedence than we are allowed to eat, return 1631 // successfully with what we ate already. 1632 if (TokPrec < Precedence) 1633 return false; 1634 1635 Lex(); 1636 1637 // Eat the next primary expression. 1638 const MCExpr *RHS; 1639 if (getTargetParser().parsePrimaryExpr(RHS, EndLoc)) 1640 return true; 1641 1642 // If BinOp binds less tightly with RHS than the operator after RHS, let 1643 // the pending operator take RHS as its LHS. 1644 MCBinaryExpr::Opcode Dummy; 1645 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1646 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1647 return true; 1648 1649 // Merge LHS and RHS according to operator. 1650 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc); 1651 } 1652 } 1653 1654 /// ParseStatement: 1655 /// ::= EndOfStatement 1656 /// ::= Label* Directive ...Operands... EndOfStatement 1657 /// ::= Label* Identifier OperandList* EndOfStatement 1658 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1659 MCAsmParserSemaCallback *SI) { 1660 assert(!hasPendingError() && "parseStatement started with pending error"); 1661 // Eat initial spaces and comments 1662 while (Lexer.is(AsmToken::Space)) 1663 Lex(); 1664 if (Lexer.is(AsmToken::EndOfStatement)) { 1665 // if this is a line comment we can drop it safely 1666 if (getTok().getString().empty() || getTok().getString().front() == '\r' || 1667 getTok().getString().front() == '\n') 1668 Out.AddBlankLine(); 1669 Lex(); 1670 return false; 1671 } 1672 // Statements always start with an identifier. 1673 AsmToken ID = getTok(); 1674 SMLoc IDLoc = ID.getLoc(); 1675 StringRef IDVal; 1676 int64_t LocalLabelVal = -1; 1677 if (Lexer.is(AsmToken::HashDirective)) 1678 return parseCppHashLineFilenameComment(IDLoc); 1679 // Allow an integer followed by a ':' as a directional local label. 1680 if (Lexer.is(AsmToken::Integer)) { 1681 LocalLabelVal = getTok().getIntVal(); 1682 if (LocalLabelVal < 0) { 1683 if (!TheCondState.Ignore) { 1684 Lex(); // always eat a token 1685 return Error(IDLoc, "unexpected token at start of statement"); 1686 } 1687 IDVal = ""; 1688 } else { 1689 IDVal = getTok().getString(); 1690 Lex(); // Consume the integer token to be used as an identifier token. 1691 if (Lexer.getKind() != AsmToken::Colon) { 1692 if (!TheCondState.Ignore) { 1693 Lex(); // always eat a token 1694 return Error(IDLoc, "unexpected token at start of statement"); 1695 } 1696 } 1697 } 1698 } else if (Lexer.is(AsmToken::Dot)) { 1699 // Treat '.' as a valid identifier in this context. 1700 Lex(); 1701 IDVal = "."; 1702 } else if (Lexer.is(AsmToken::LCurly)) { 1703 // Treat '{' as a valid identifier in this context. 1704 Lex(); 1705 IDVal = "{"; 1706 1707 } else if (Lexer.is(AsmToken::RCurly)) { 1708 // Treat '}' as a valid identifier in this context. 1709 Lex(); 1710 IDVal = "}"; 1711 } else if (Lexer.is(AsmToken::Star) && 1712 getTargetParser().starIsStartOfStatement()) { 1713 // Accept '*' as a valid start of statement. 1714 Lex(); 1715 IDVal = "*"; 1716 } else if (parseIdentifier(IDVal)) { 1717 if (!TheCondState.Ignore) { 1718 Lex(); // always eat a token 1719 return Error(IDLoc, "unexpected token at start of statement"); 1720 } 1721 IDVal = ""; 1722 } 1723 1724 // Handle conditional assembly here before checking for skipping. We 1725 // have to do this so that .endif isn't skipped in a ".if 0" block for 1726 // example. 1727 StringMap<DirectiveKind>::const_iterator DirKindIt = 1728 DirectiveKindMap.find(IDVal); 1729 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1730 ? DK_NO_DIRECTIVE 1731 : DirKindIt->getValue(); 1732 switch (DirKind) { 1733 default: 1734 break; 1735 case DK_IF: 1736 case DK_IFEQ: 1737 case DK_IFGE: 1738 case DK_IFGT: 1739 case DK_IFLE: 1740 case DK_IFLT: 1741 case DK_IFNE: 1742 return parseDirectiveIf(IDLoc, DirKind); 1743 case DK_IFB: 1744 return parseDirectiveIfb(IDLoc, true); 1745 case DK_IFNB: 1746 return parseDirectiveIfb(IDLoc, false); 1747 case DK_IFC: 1748 return parseDirectiveIfc(IDLoc, true); 1749 case DK_IFEQS: 1750 return parseDirectiveIfeqs(IDLoc, true); 1751 case DK_IFNC: 1752 return parseDirectiveIfc(IDLoc, false); 1753 case DK_IFNES: 1754 return parseDirectiveIfeqs(IDLoc, false); 1755 case DK_IFDEF: 1756 return parseDirectiveIfdef(IDLoc, true); 1757 case DK_IFNDEF: 1758 case DK_IFNOTDEF: 1759 return parseDirectiveIfdef(IDLoc, false); 1760 case DK_ELSEIF: 1761 return parseDirectiveElseIf(IDLoc); 1762 case DK_ELSE: 1763 return parseDirectiveElse(IDLoc); 1764 case DK_ENDIF: 1765 return parseDirectiveEndIf(IDLoc); 1766 } 1767 1768 // Ignore the statement if in the middle of inactive conditional 1769 // (e.g. ".if 0"). 1770 if (TheCondState.Ignore) { 1771 eatToEndOfStatement(); 1772 return false; 1773 } 1774 1775 // FIXME: Recurse on local labels? 1776 1777 // See what kind of statement we have. 1778 switch (Lexer.getKind()) { 1779 case AsmToken::Colon: { 1780 if (!getTargetParser().isLabel(ID)) 1781 break; 1782 if (checkForValidSection()) 1783 return true; 1784 1785 // identifier ':' -> Label. 1786 Lex(); 1787 1788 // Diagnose attempt to use '.' as a label. 1789 if (IDVal == ".") 1790 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1791 1792 // Diagnose attempt to use a variable as a label. 1793 // 1794 // FIXME: Diagnostics. Note the location of the definition as a label. 1795 // FIXME: This doesn't diagnose assignment to a symbol which has been 1796 // implicitly marked as external. 1797 MCSymbol *Sym; 1798 if (LocalLabelVal == -1) { 1799 if (ParsingInlineAsm && SI) { 1800 StringRef RewrittenLabel = 1801 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1802 assert(!RewrittenLabel.empty() && 1803 "We should have an internal name here."); 1804 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1805 RewrittenLabel); 1806 IDVal = RewrittenLabel; 1807 } 1808 Sym = getContext().getOrCreateSymbol(IDVal); 1809 } else 1810 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1811 // End of Labels should be treated as end of line for lexing 1812 // purposes but that information is not available to the Lexer who 1813 // does not understand Labels. This may cause us to see a Hash 1814 // here instead of a preprocessor line comment. 1815 if (getTok().is(AsmToken::Hash)) { 1816 StringRef CommentStr = parseStringToEndOfStatement(); 1817 Lexer.Lex(); 1818 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1819 } 1820 1821 // Consume any end of statement token, if present, to avoid spurious 1822 // AddBlankLine calls(). 1823 if (getTok().is(AsmToken::EndOfStatement)) { 1824 Lex(); 1825 } 1826 1827 getTargetParser().doBeforeLabelEmit(Sym); 1828 1829 // Emit the label. 1830 if (!getTargetParser().isParsingInlineAsm()) 1831 Out.EmitLabel(Sym, IDLoc); 1832 1833 // If we are generating dwarf for assembly source files then gather the 1834 // info to make a dwarf label entry for this label if needed. 1835 if (enabledGenDwarfForAssembly()) 1836 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1837 IDLoc); 1838 1839 getTargetParser().onLabelParsed(Sym); 1840 1841 return false; 1842 } 1843 1844 case AsmToken::Equal: 1845 if (!getTargetParser().equalIsAsmAssignment()) 1846 break; 1847 // identifier '=' ... -> assignment statement 1848 Lex(); 1849 1850 return parseAssignment(IDVal, true); 1851 1852 default: // Normal instruction or directive. 1853 break; 1854 } 1855 1856 // If macros are enabled, check to see if this is a macro instantiation. 1857 if (areMacrosEnabled()) 1858 if (const MCAsmMacro *M = getContext().lookupMacro(IDVal)) { 1859 return handleMacroEntry(M, IDLoc); 1860 } 1861 1862 // Otherwise, we have a normal instruction or directive. 1863 1864 // Directives start with "." 1865 if (IDVal.startswith(".") && IDVal != ".") { 1866 // There are several entities interested in parsing directives: 1867 // 1868 // 1. The target-specific assembly parser. Some directives are target 1869 // specific or may potentially behave differently on certain targets. 1870 // 2. Asm parser extensions. For example, platform-specific parsers 1871 // (like the ELF parser) register themselves as extensions. 1872 // 3. The generic directive parser implemented by this class. These are 1873 // all the directives that behave in a target and platform independent 1874 // manner, or at least have a default behavior that's shared between 1875 // all targets and platforms. 1876 1877 getTargetParser().flushPendingInstructions(getStreamer()); 1878 1879 SMLoc StartTokLoc = getTok().getLoc(); 1880 bool TPDirectiveReturn = getTargetParser().ParseDirective(ID); 1881 1882 if (hasPendingError()) 1883 return true; 1884 // Currently the return value should be true if we are 1885 // uninterested but as this is at odds with the standard parsing 1886 // convention (return true = error) we have instances of a parsed 1887 // directive that fails returning true as an error. Catch these 1888 // cases as best as possible errors here. 1889 if (TPDirectiveReturn && StartTokLoc != getTok().getLoc()) 1890 return true; 1891 // Return if we did some parsing or believe we succeeded. 1892 if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc()) 1893 return false; 1894 1895 // Next, check the extension directive map to see if any extension has 1896 // registered itself to parse this directive. 1897 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1898 ExtensionDirectiveMap.lookup(IDVal); 1899 if (Handler.first) 1900 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1901 1902 // Finally, if no one else is interested in this directive, it must be 1903 // generic and familiar to this class. 1904 switch (DirKind) { 1905 default: 1906 break; 1907 case DK_SET: 1908 case DK_EQU: 1909 return parseDirectiveSet(IDVal, true); 1910 case DK_EQUIV: 1911 return parseDirectiveSet(IDVal, false); 1912 case DK_ASCII: 1913 return parseDirectiveAscii(IDVal, false); 1914 case DK_ASCIZ: 1915 case DK_STRING: 1916 return parseDirectiveAscii(IDVal, true); 1917 case DK_BYTE: 1918 case DK_DC_B: 1919 return parseDirectiveValue(IDVal, 1); 1920 case DK_DC: 1921 case DK_DC_W: 1922 case DK_SHORT: 1923 case DK_VALUE: 1924 case DK_2BYTE: 1925 return parseDirectiveValue(IDVal, 2); 1926 case DK_LONG: 1927 case DK_INT: 1928 case DK_4BYTE: 1929 case DK_DC_L: 1930 return parseDirectiveValue(IDVal, 4); 1931 case DK_QUAD: 1932 case DK_8BYTE: 1933 return parseDirectiveValue(IDVal, 8); 1934 case DK_DC_A: 1935 return parseDirectiveValue( 1936 IDVal, getContext().getAsmInfo()->getCodePointerSize()); 1937 case DK_OCTA: 1938 return parseDirectiveOctaValue(IDVal); 1939 case DK_SINGLE: 1940 case DK_FLOAT: 1941 case DK_DC_S: 1942 return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle()); 1943 case DK_DOUBLE: 1944 case DK_DC_D: 1945 return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble()); 1946 case DK_ALIGN: { 1947 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1948 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1949 } 1950 case DK_ALIGN32: { 1951 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1952 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1953 } 1954 case DK_BALIGN: 1955 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1956 case DK_BALIGNW: 1957 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1958 case DK_BALIGNL: 1959 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 1960 case DK_P2ALIGN: 1961 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 1962 case DK_P2ALIGNW: 1963 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 1964 case DK_P2ALIGNL: 1965 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 1966 case DK_ORG: 1967 return parseDirectiveOrg(); 1968 case DK_FILL: 1969 return parseDirectiveFill(); 1970 case DK_ZERO: 1971 return parseDirectiveZero(); 1972 case DK_EXTERN: 1973 eatToEndOfStatement(); // .extern is the default, ignore it. 1974 return false; 1975 case DK_GLOBL: 1976 case DK_GLOBAL: 1977 return parseDirectiveSymbolAttribute(MCSA_Global); 1978 case DK_LAZY_REFERENCE: 1979 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 1980 case DK_NO_DEAD_STRIP: 1981 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 1982 case DK_SYMBOL_RESOLVER: 1983 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 1984 case DK_PRIVATE_EXTERN: 1985 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 1986 case DK_REFERENCE: 1987 return parseDirectiveSymbolAttribute(MCSA_Reference); 1988 case DK_WEAK_DEFINITION: 1989 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 1990 case DK_WEAK_REFERENCE: 1991 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 1992 case DK_WEAK_DEF_CAN_BE_HIDDEN: 1993 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 1994 case DK_COLD: 1995 return parseDirectiveSymbolAttribute(MCSA_Cold); 1996 case DK_COMM: 1997 case DK_COMMON: 1998 return parseDirectiveComm(/*IsLocal=*/false); 1999 case DK_LCOMM: 2000 return parseDirectiveComm(/*IsLocal=*/true); 2001 case DK_ABORT: 2002 return parseDirectiveAbort(); 2003 case DK_INCLUDE: 2004 return parseDirectiveInclude(); 2005 case DK_INCBIN: 2006 return parseDirectiveIncbin(); 2007 case DK_CODE16: 2008 case DK_CODE16GCC: 2009 return TokError(Twine(IDVal) + 2010 " not currently supported for this target"); 2011 case DK_REPT: 2012 return parseDirectiveRept(IDLoc, IDVal); 2013 case DK_IRP: 2014 return parseDirectiveIrp(IDLoc); 2015 case DK_IRPC: 2016 return parseDirectiveIrpc(IDLoc); 2017 case DK_ENDR: 2018 return parseDirectiveEndr(IDLoc); 2019 case DK_BUNDLE_ALIGN_MODE: 2020 return parseDirectiveBundleAlignMode(); 2021 case DK_BUNDLE_LOCK: 2022 return parseDirectiveBundleLock(); 2023 case DK_BUNDLE_UNLOCK: 2024 return parseDirectiveBundleUnlock(); 2025 case DK_SLEB128: 2026 return parseDirectiveLEB128(true); 2027 case DK_ULEB128: 2028 return parseDirectiveLEB128(false); 2029 case DK_SPACE: 2030 case DK_SKIP: 2031 return parseDirectiveSpace(IDVal); 2032 case DK_FILE: 2033 return parseDirectiveFile(IDLoc); 2034 case DK_LINE: 2035 return parseDirectiveLine(); 2036 case DK_LOC: 2037 return parseDirectiveLoc(); 2038 case DK_STABS: 2039 return parseDirectiveStabs(); 2040 case DK_CV_FILE: 2041 return parseDirectiveCVFile(); 2042 case DK_CV_FUNC_ID: 2043 return parseDirectiveCVFuncId(); 2044 case DK_CV_INLINE_SITE_ID: 2045 return parseDirectiveCVInlineSiteId(); 2046 case DK_CV_LOC: 2047 return parseDirectiveCVLoc(); 2048 case DK_CV_LINETABLE: 2049 return parseDirectiveCVLinetable(); 2050 case DK_CV_INLINE_LINETABLE: 2051 return parseDirectiveCVInlineLinetable(); 2052 case DK_CV_DEF_RANGE: 2053 return parseDirectiveCVDefRange(); 2054 case DK_CV_STRING: 2055 return parseDirectiveCVString(); 2056 case DK_CV_STRINGTABLE: 2057 return parseDirectiveCVStringTable(); 2058 case DK_CV_FILECHECKSUMS: 2059 return parseDirectiveCVFileChecksums(); 2060 case DK_CV_FILECHECKSUM_OFFSET: 2061 return parseDirectiveCVFileChecksumOffset(); 2062 case DK_CV_FPO_DATA: 2063 return parseDirectiveCVFPOData(); 2064 case DK_CFI_SECTIONS: 2065 return parseDirectiveCFISections(); 2066 case DK_CFI_STARTPROC: 2067 return parseDirectiveCFIStartProc(); 2068 case DK_CFI_ENDPROC: 2069 return parseDirectiveCFIEndProc(); 2070 case DK_CFI_DEF_CFA: 2071 return parseDirectiveCFIDefCfa(IDLoc); 2072 case DK_CFI_DEF_CFA_OFFSET: 2073 return parseDirectiveCFIDefCfaOffset(); 2074 case DK_CFI_ADJUST_CFA_OFFSET: 2075 return parseDirectiveCFIAdjustCfaOffset(); 2076 case DK_CFI_DEF_CFA_REGISTER: 2077 return parseDirectiveCFIDefCfaRegister(IDLoc); 2078 case DK_CFI_OFFSET: 2079 return parseDirectiveCFIOffset(IDLoc); 2080 case DK_CFI_REL_OFFSET: 2081 return parseDirectiveCFIRelOffset(IDLoc); 2082 case DK_CFI_PERSONALITY: 2083 return parseDirectiveCFIPersonalityOrLsda(true); 2084 case DK_CFI_LSDA: 2085 return parseDirectiveCFIPersonalityOrLsda(false); 2086 case DK_CFI_REMEMBER_STATE: 2087 return parseDirectiveCFIRememberState(); 2088 case DK_CFI_RESTORE_STATE: 2089 return parseDirectiveCFIRestoreState(); 2090 case DK_CFI_SAME_VALUE: 2091 return parseDirectiveCFISameValue(IDLoc); 2092 case DK_CFI_RESTORE: 2093 return parseDirectiveCFIRestore(IDLoc); 2094 case DK_CFI_ESCAPE: 2095 return parseDirectiveCFIEscape(); 2096 case DK_CFI_RETURN_COLUMN: 2097 return parseDirectiveCFIReturnColumn(IDLoc); 2098 case DK_CFI_SIGNAL_FRAME: 2099 return parseDirectiveCFISignalFrame(); 2100 case DK_CFI_UNDEFINED: 2101 return parseDirectiveCFIUndefined(IDLoc); 2102 case DK_CFI_REGISTER: 2103 return parseDirectiveCFIRegister(IDLoc); 2104 case DK_CFI_WINDOW_SAVE: 2105 return parseDirectiveCFIWindowSave(); 2106 case DK_MACROS_ON: 2107 case DK_MACROS_OFF: 2108 return parseDirectiveMacrosOnOff(IDVal); 2109 case DK_MACRO: 2110 return parseDirectiveMacro(IDLoc); 2111 case DK_ALTMACRO: 2112 case DK_NOALTMACRO: 2113 return parseDirectiveAltmacro(IDVal); 2114 case DK_EXITM: 2115 return parseDirectiveExitMacro(IDVal); 2116 case DK_ENDM: 2117 case DK_ENDMACRO: 2118 return parseDirectiveEndMacro(IDVal); 2119 case DK_PURGEM: 2120 return parseDirectivePurgeMacro(IDLoc); 2121 case DK_END: 2122 return parseDirectiveEnd(IDLoc); 2123 case DK_ERR: 2124 return parseDirectiveError(IDLoc, false); 2125 case DK_ERROR: 2126 return parseDirectiveError(IDLoc, true); 2127 case DK_WARNING: 2128 return parseDirectiveWarning(IDLoc); 2129 case DK_RELOC: 2130 return parseDirectiveReloc(IDLoc); 2131 case DK_DCB: 2132 case DK_DCB_W: 2133 return parseDirectiveDCB(IDVal, 2); 2134 case DK_DCB_B: 2135 return parseDirectiveDCB(IDVal, 1); 2136 case DK_DCB_D: 2137 return parseDirectiveRealDCB(IDVal, APFloat::IEEEdouble()); 2138 case DK_DCB_L: 2139 return parseDirectiveDCB(IDVal, 4); 2140 case DK_DCB_S: 2141 return parseDirectiveRealDCB(IDVal, APFloat::IEEEsingle()); 2142 case DK_DC_X: 2143 case DK_DCB_X: 2144 return TokError(Twine(IDVal) + 2145 " not currently supported for this target"); 2146 case DK_DS: 2147 case DK_DS_W: 2148 return parseDirectiveDS(IDVal, 2); 2149 case DK_DS_B: 2150 return parseDirectiveDS(IDVal, 1); 2151 case DK_DS_D: 2152 return parseDirectiveDS(IDVal, 8); 2153 case DK_DS_L: 2154 case DK_DS_S: 2155 return parseDirectiveDS(IDVal, 4); 2156 case DK_DS_P: 2157 case DK_DS_X: 2158 return parseDirectiveDS(IDVal, 12); 2159 case DK_PRINT: 2160 return parseDirectivePrint(IDLoc); 2161 case DK_ADDRSIG: 2162 return parseDirectiveAddrsig(); 2163 case DK_ADDRSIG_SYM: 2164 return parseDirectiveAddrsigSym(); 2165 } 2166 2167 return Error(IDLoc, "unknown directive"); 2168 } 2169 2170 // __asm _emit or __asm __emit 2171 if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 2172 IDVal == "_EMIT" || IDVal == "__EMIT")) 2173 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 2174 2175 // __asm align 2176 if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 2177 return parseDirectiveMSAlign(IDLoc, Info); 2178 2179 if (ParsingInlineAsm && (IDVal == "even" || IDVal == "EVEN")) 2180 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 2181 if (checkForValidSection()) 2182 return true; 2183 2184 // Canonicalize the opcode to lower case. 2185 std::string OpcodeStr = IDVal.lower(); 2186 ParseInstructionInfo IInfo(Info.AsmRewrites); 2187 bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 2188 Info.ParsedOperands); 2189 Info.ParseError = ParseHadError; 2190 2191 // Dump the parsed representation, if requested. 2192 if (getShowParsedOperands()) { 2193 SmallString<256> Str; 2194 raw_svector_ostream OS(Str); 2195 OS << "parsed instruction: ["; 2196 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 2197 if (i != 0) 2198 OS << ", "; 2199 Info.ParsedOperands[i]->print(OS); 2200 } 2201 OS << "]"; 2202 2203 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 2204 } 2205 2206 // Fail even if ParseInstruction erroneously returns false. 2207 if (hasPendingError() || ParseHadError) 2208 return true; 2209 2210 // If we are generating dwarf for the current section then generate a .loc 2211 // directive for the instruction. 2212 if (!ParseHadError && enabledGenDwarfForAssembly() && 2213 getContext().getGenDwarfSectionSyms().count( 2214 getStreamer().getCurrentSectionOnly())) { 2215 unsigned Line; 2216 if (ActiveMacros.empty()) 2217 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 2218 else 2219 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 2220 ActiveMacros.front()->ExitBuffer); 2221 2222 // If we previously parsed a cpp hash file line comment then make sure the 2223 // current Dwarf File is for the CppHashFilename if not then emit the 2224 // Dwarf File table for it and adjust the line number for the .loc. 2225 if (!CppHashInfo.Filename.empty()) { 2226 unsigned FileNumber = getStreamer().EmitDwarfFileDirective( 2227 0, StringRef(), CppHashInfo.Filename); 2228 getContext().setGenDwarfFileNumber(FileNumber); 2229 2230 unsigned CppHashLocLineNo = 2231 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 2232 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 2233 } 2234 2235 getStreamer().EmitDwarfLocDirective( 2236 getContext().getGenDwarfFileNumber(), Line, 0, 2237 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 2238 StringRef()); 2239 } 2240 2241 // If parsing succeeded, match the instruction. 2242 if (!ParseHadError) { 2243 uint64_t ErrorInfo; 2244 if (getTargetParser().MatchAndEmitInstruction( 2245 IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo, 2246 getTargetParser().isParsingInlineAsm())) 2247 return true; 2248 } 2249 return false; 2250 } 2251 2252 // Parse and erase curly braces marking block start/end 2253 bool 2254 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 2255 // Identify curly brace marking block start/end 2256 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 2257 return false; 2258 2259 SMLoc StartLoc = Lexer.getLoc(); 2260 Lex(); // Eat the brace 2261 if (Lexer.is(AsmToken::EndOfStatement)) 2262 Lex(); // Eat EndOfStatement following the brace 2263 2264 // Erase the block start/end brace from the output asm string 2265 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 2266 StartLoc.getPointer()); 2267 return true; 2268 } 2269 2270 /// parseCppHashLineFilenameComment as this: 2271 /// ::= # number "filename" 2272 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L) { 2273 Lex(); // Eat the hash token. 2274 // Lexer only ever emits HashDirective if it fully formed if it's 2275 // done the checking already so this is an internal error. 2276 assert(getTok().is(AsmToken::Integer) && 2277 "Lexing Cpp line comment: Expected Integer"); 2278 int64_t LineNumber = getTok().getIntVal(); 2279 Lex(); 2280 assert(getTok().is(AsmToken::String) && 2281 "Lexing Cpp line comment: Expected String"); 2282 StringRef Filename = getTok().getString(); 2283 Lex(); 2284 2285 // Get rid of the enclosing quotes. 2286 Filename = Filename.substr(1, Filename.size() - 2); 2287 2288 // Save the SMLoc, Filename and LineNumber for later use by diagnostics. 2289 CppHashInfo.Loc = L; 2290 CppHashInfo.Filename = Filename; 2291 CppHashInfo.LineNumber = LineNumber; 2292 CppHashInfo.Buf = CurBuffer; 2293 return false; 2294 } 2295 2296 /// will use the last parsed cpp hash line filename comment 2297 /// for the Filename and LineNo if any in the diagnostic. 2298 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 2299 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 2300 raw_ostream &OS = errs(); 2301 2302 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 2303 SMLoc DiagLoc = Diag.getLoc(); 2304 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2305 unsigned CppHashBuf = 2306 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 2307 2308 // Like SourceMgr::printMessage() we need to print the include stack if any 2309 // before printing the message. 2310 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2311 if (!Parser->SavedDiagHandler && DiagCurBuffer && 2312 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 2313 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 2314 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 2315 } 2316 2317 // If we have not parsed a cpp hash line filename comment or the source 2318 // manager changed or buffer changed (like in a nested include) then just 2319 // print the normal diagnostic using its Filename and LineNo. 2320 if (!Parser->CppHashInfo.LineNumber || &DiagSrcMgr != &Parser->SrcMgr || 2321 DiagBuf != CppHashBuf) { 2322 if (Parser->SavedDiagHandler) 2323 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2324 else 2325 Diag.print(nullptr, OS); 2326 return; 2327 } 2328 2329 // Use the CppHashFilename and calculate a line number based on the 2330 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 2331 // for the diagnostic. 2332 const std::string &Filename = Parser->CppHashInfo.Filename; 2333 2334 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 2335 int CppHashLocLineNo = 2336 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 2337 int LineNo = 2338 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 2339 2340 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 2341 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 2342 Diag.getLineContents(), Diag.getRanges()); 2343 2344 if (Parser->SavedDiagHandler) 2345 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 2346 else 2347 NewDiag.print(nullptr, OS); 2348 } 2349 2350 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2351 // difference being that that function accepts '@' as part of identifiers and 2352 // we can't do that. AsmLexer.cpp should probably be changed to handle 2353 // '@' as a special case when needed. 2354 static bool isIdentifierChar(char c) { 2355 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2356 c == '.'; 2357 } 2358 2359 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2360 ArrayRef<MCAsmMacroParameter> Parameters, 2361 ArrayRef<MCAsmMacroArgument> A, 2362 bool EnableAtPseudoVariable, SMLoc L) { 2363 unsigned NParameters = Parameters.size(); 2364 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2365 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2366 return Error(L, "Wrong number of arguments"); 2367 2368 // A macro without parameters is handled differently on Darwin: 2369 // gas accepts no arguments and does no substitutions 2370 while (!Body.empty()) { 2371 // Scan for the next substitution. 2372 std::size_t End = Body.size(), Pos = 0; 2373 for (; Pos != End; ++Pos) { 2374 // Check for a substitution or escape. 2375 if (IsDarwin && !NParameters) { 2376 // This macro has no parameters, look for $0, $1, etc. 2377 if (Body[Pos] != '$' || Pos + 1 == End) 2378 continue; 2379 2380 char Next = Body[Pos + 1]; 2381 if (Next == '$' || Next == 'n' || 2382 isdigit(static_cast<unsigned char>(Next))) 2383 break; 2384 } else { 2385 // This macro has parameters, look for \foo, \bar, etc. 2386 if (Body[Pos] == '\\' && Pos + 1 != End) 2387 break; 2388 } 2389 } 2390 2391 // Add the prefix. 2392 OS << Body.slice(0, Pos); 2393 2394 // Check if we reached the end. 2395 if (Pos == End) 2396 break; 2397 2398 if (IsDarwin && !NParameters) { 2399 switch (Body[Pos + 1]) { 2400 // $$ => $ 2401 case '$': 2402 OS << '$'; 2403 break; 2404 2405 // $n => number of arguments 2406 case 'n': 2407 OS << A.size(); 2408 break; 2409 2410 // $[0-9] => argument 2411 default: { 2412 // Missing arguments are ignored. 2413 unsigned Index = Body[Pos + 1] - '0'; 2414 if (Index >= A.size()) 2415 break; 2416 2417 // Otherwise substitute with the token values, with spaces eliminated. 2418 for (const AsmToken &Token : A[Index]) 2419 OS << Token.getString(); 2420 break; 2421 } 2422 } 2423 Pos += 2; 2424 } else { 2425 unsigned I = Pos + 1; 2426 2427 // Check for the \@ pseudo-variable. 2428 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2429 ++I; 2430 else 2431 while (isIdentifierChar(Body[I]) && I + 1 != End) 2432 ++I; 2433 2434 const char *Begin = Body.data() + Pos + 1; 2435 StringRef Argument(Begin, I - (Pos + 1)); 2436 unsigned Index = 0; 2437 2438 if (Argument == "@") { 2439 OS << NumOfMacroInstantiations; 2440 Pos += 2; 2441 } else { 2442 for (; Index < NParameters; ++Index) 2443 if (Parameters[Index].Name == Argument) 2444 break; 2445 2446 if (Index == NParameters) { 2447 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2448 Pos += 3; 2449 else { 2450 OS << '\\' << Argument; 2451 Pos = I; 2452 } 2453 } else { 2454 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2455 for (const AsmToken &Token : A[Index]) 2456 // For altmacro mode, you can write '%expr'. 2457 // The prefix '%' evaluates the expression 'expr' 2458 // and uses the result as a string (e.g. replace %(1+2) with the 2459 // string "3"). 2460 // Here, we identify the integer token which is the result of the 2461 // absolute expression evaluation and replace it with its string 2462 // representation. 2463 if (AltMacroMode && Token.getString().front() == '%' && 2464 Token.is(AsmToken::Integer)) 2465 // Emit an integer value to the buffer. 2466 OS << Token.getIntVal(); 2467 // Only Token that was validated as a string and begins with '<' 2468 // is considered altMacroString!!! 2469 else if (AltMacroMode && Token.getString().front() == '<' && 2470 Token.is(AsmToken::String)) { 2471 OS << altMacroString(Token.getStringContents()); 2472 } 2473 // We expect no quotes around the string's contents when 2474 // parsing for varargs. 2475 else if (Token.isNot(AsmToken::String) || VarargParameter) 2476 OS << Token.getString(); 2477 else 2478 OS << Token.getStringContents(); 2479 2480 Pos += 1 + Argument.size(); 2481 } 2482 } 2483 } 2484 // Update the scan point. 2485 Body = Body.substr(Pos); 2486 } 2487 2488 return false; 2489 } 2490 2491 MacroInstantiation::MacroInstantiation(SMLoc IL, int EB, SMLoc EL, 2492 size_t CondStackDepth) 2493 : InstantiationLoc(IL), ExitBuffer(EB), ExitLoc(EL), 2494 CondStackDepth(CondStackDepth) {} 2495 2496 static bool isOperator(AsmToken::TokenKind kind) { 2497 switch (kind) { 2498 default: 2499 return false; 2500 case AsmToken::Plus: 2501 case AsmToken::Minus: 2502 case AsmToken::Tilde: 2503 case AsmToken::Slash: 2504 case AsmToken::Star: 2505 case AsmToken::Dot: 2506 case AsmToken::Equal: 2507 case AsmToken::EqualEqual: 2508 case AsmToken::Pipe: 2509 case AsmToken::PipePipe: 2510 case AsmToken::Caret: 2511 case AsmToken::Amp: 2512 case AsmToken::AmpAmp: 2513 case AsmToken::Exclaim: 2514 case AsmToken::ExclaimEqual: 2515 case AsmToken::Less: 2516 case AsmToken::LessEqual: 2517 case AsmToken::LessLess: 2518 case AsmToken::LessGreater: 2519 case AsmToken::Greater: 2520 case AsmToken::GreaterEqual: 2521 case AsmToken::GreaterGreater: 2522 return true; 2523 } 2524 } 2525 2526 namespace { 2527 2528 class AsmLexerSkipSpaceRAII { 2529 public: 2530 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2531 Lexer.setSkipSpace(SkipSpace); 2532 } 2533 2534 ~AsmLexerSkipSpaceRAII() { 2535 Lexer.setSkipSpace(true); 2536 } 2537 2538 private: 2539 AsmLexer &Lexer; 2540 }; 2541 2542 } // end anonymous namespace 2543 2544 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2545 2546 if (Vararg) { 2547 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2548 StringRef Str = parseStringToEndOfStatement(); 2549 MA.emplace_back(AsmToken::String, Str); 2550 } 2551 return false; 2552 } 2553 2554 unsigned ParenLevel = 0; 2555 2556 // Darwin doesn't use spaces to delmit arguments. 2557 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2558 2559 bool SpaceEaten; 2560 2561 while (true) { 2562 SpaceEaten = false; 2563 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2564 return TokError("unexpected token in macro instantiation"); 2565 2566 if (ParenLevel == 0) { 2567 2568 if (Lexer.is(AsmToken::Comma)) 2569 break; 2570 2571 if (Lexer.is(AsmToken::Space)) { 2572 SpaceEaten = true; 2573 Lexer.Lex(); // Eat spaces 2574 } 2575 2576 // Spaces can delimit parameters, but could also be part an expression. 2577 // If the token after a space is an operator, add the token and the next 2578 // one into this argument 2579 if (!IsDarwin) { 2580 if (isOperator(Lexer.getKind())) { 2581 MA.push_back(getTok()); 2582 Lexer.Lex(); 2583 2584 // Whitespace after an operator can be ignored. 2585 if (Lexer.is(AsmToken::Space)) 2586 Lexer.Lex(); 2587 2588 continue; 2589 } 2590 } 2591 if (SpaceEaten) 2592 break; 2593 } 2594 2595 // handleMacroEntry relies on not advancing the lexer here 2596 // to be able to fill in the remaining default parameter values 2597 if (Lexer.is(AsmToken::EndOfStatement)) 2598 break; 2599 2600 // Adjust the current parentheses level. 2601 if (Lexer.is(AsmToken::LParen)) 2602 ++ParenLevel; 2603 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2604 --ParenLevel; 2605 2606 // Append the token to the current argument list. 2607 MA.push_back(getTok()); 2608 Lexer.Lex(); 2609 } 2610 2611 if (ParenLevel != 0) 2612 return TokError("unbalanced parentheses in macro argument"); 2613 return false; 2614 } 2615 2616 // Parse the macro instantiation arguments. 2617 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2618 MCAsmMacroArguments &A) { 2619 const unsigned NParameters = M ? M->Parameters.size() : 0; 2620 bool NamedParametersFound = false; 2621 SmallVector<SMLoc, 4> FALocs; 2622 2623 A.resize(NParameters); 2624 FALocs.resize(NParameters); 2625 2626 // Parse two kinds of macro invocations: 2627 // - macros defined without any parameters accept an arbitrary number of them 2628 // - macros defined with parameters accept at most that many of them 2629 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2630 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2631 ++Parameter) { 2632 SMLoc IDLoc = Lexer.getLoc(); 2633 MCAsmMacroParameter FA; 2634 2635 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2636 if (parseIdentifier(FA.Name)) 2637 return Error(IDLoc, "invalid argument identifier for formal argument"); 2638 2639 if (Lexer.isNot(AsmToken::Equal)) 2640 return TokError("expected '=' after formal parameter identifier"); 2641 2642 Lex(); 2643 2644 NamedParametersFound = true; 2645 } 2646 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2647 2648 if (NamedParametersFound && FA.Name.empty()) 2649 return Error(IDLoc, "cannot mix positional and keyword arguments"); 2650 2651 SMLoc StrLoc = Lexer.getLoc(); 2652 SMLoc EndLoc; 2653 if (AltMacroMode && Lexer.is(AsmToken::Percent)) { 2654 const MCExpr *AbsoluteExp; 2655 int64_t Value; 2656 /// Eat '%' 2657 Lex(); 2658 if (parseExpression(AbsoluteExp, EndLoc)) 2659 return false; 2660 if (!AbsoluteExp->evaluateAsAbsolute(Value, 2661 getStreamer().getAssemblerPtr())) 2662 return Error(StrLoc, "expected absolute expression"); 2663 const char *StrChar = StrLoc.getPointer(); 2664 const char *EndChar = EndLoc.getPointer(); 2665 AsmToken newToken(AsmToken::Integer, 2666 StringRef(StrChar, EndChar - StrChar), Value); 2667 FA.Value.push_back(newToken); 2668 } else if (AltMacroMode && Lexer.is(AsmToken::Less) && 2669 isAltmacroString(StrLoc, EndLoc)) { 2670 const char *StrChar = StrLoc.getPointer(); 2671 const char *EndChar = EndLoc.getPointer(); 2672 jumpToLoc(EndLoc, CurBuffer); 2673 /// Eat from '<' to '>' 2674 Lex(); 2675 AsmToken newToken(AsmToken::String, 2676 StringRef(StrChar, EndChar - StrChar)); 2677 FA.Value.push_back(newToken); 2678 } else if(parseMacroArgument(FA.Value, Vararg)) 2679 return true; 2680 2681 unsigned PI = Parameter; 2682 if (!FA.Name.empty()) { 2683 unsigned FAI = 0; 2684 for (FAI = 0; FAI < NParameters; ++FAI) 2685 if (M->Parameters[FAI].Name == FA.Name) 2686 break; 2687 2688 if (FAI >= NParameters) { 2689 assert(M && "expected macro to be defined"); 2690 return Error(IDLoc, "parameter named '" + FA.Name + 2691 "' does not exist for macro '" + M->Name + "'"); 2692 } 2693 PI = FAI; 2694 } 2695 2696 if (!FA.Value.empty()) { 2697 if (A.size() <= PI) 2698 A.resize(PI + 1); 2699 A[PI] = FA.Value; 2700 2701 if (FALocs.size() <= PI) 2702 FALocs.resize(PI + 1); 2703 2704 FALocs[PI] = Lexer.getLoc(); 2705 } 2706 2707 // At the end of the statement, fill in remaining arguments that have 2708 // default values. If there aren't any, then the next argument is 2709 // required but missing 2710 if (Lexer.is(AsmToken::EndOfStatement)) { 2711 bool Failure = false; 2712 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2713 if (A[FAI].empty()) { 2714 if (M->Parameters[FAI].Required) { 2715 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2716 "missing value for required parameter " 2717 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2718 Failure = true; 2719 } 2720 2721 if (!M->Parameters[FAI].Value.empty()) 2722 A[FAI] = M->Parameters[FAI].Value; 2723 } 2724 } 2725 return Failure; 2726 } 2727 2728 if (Lexer.is(AsmToken::Comma)) 2729 Lex(); 2730 } 2731 2732 return TokError("too many positional arguments"); 2733 } 2734 2735 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2736 // Arbitrarily limit macro nesting depth (default matches 'as'). We can 2737 // eliminate this, although we should protect against infinite loops. 2738 unsigned MaxNestingDepth = AsmMacroMaxNestingDepth; 2739 if (ActiveMacros.size() == MaxNestingDepth) { 2740 std::ostringstream MaxNestingDepthError; 2741 MaxNestingDepthError << "macros cannot be nested more than " 2742 << MaxNestingDepth << " levels deep." 2743 << " Use -asm-macro-max-nesting-depth to increase " 2744 "this limit."; 2745 return TokError(MaxNestingDepthError.str()); 2746 } 2747 2748 MCAsmMacroArguments A; 2749 if (parseMacroArguments(M, A)) 2750 return true; 2751 2752 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2753 // to hold the macro body with substitutions. 2754 SmallString<256> Buf; 2755 StringRef Body = M->Body; 2756 raw_svector_ostream OS(Buf); 2757 2758 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2759 return true; 2760 2761 // We include the .endmacro in the buffer as our cue to exit the macro 2762 // instantiation. 2763 OS << ".endmacro\n"; 2764 2765 std::unique_ptr<MemoryBuffer> Instantiation = 2766 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2767 2768 // Create the macro instantiation object and add to the current macro 2769 // instantiation stack. 2770 MacroInstantiation *MI = new MacroInstantiation( 2771 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 2772 ActiveMacros.push_back(MI); 2773 2774 ++NumOfMacroInstantiations; 2775 2776 // Jump to the macro instantiation and prime the lexer. 2777 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2778 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2779 Lex(); 2780 2781 return false; 2782 } 2783 2784 void AsmParser::handleMacroExit() { 2785 // Jump to the EndOfStatement we should return to, and consume it. 2786 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2787 Lex(); 2788 2789 // Pop the instantiation entry. 2790 delete ActiveMacros.back(); 2791 ActiveMacros.pop_back(); 2792 } 2793 2794 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2795 bool NoDeadStrip) { 2796 MCSymbol *Sym; 2797 const MCExpr *Value; 2798 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2799 Value)) 2800 return true; 2801 2802 if (!Sym) { 2803 // In the case where we parse an expression starting with a '.', we will 2804 // not generate an error, nor will we create a symbol. In this case we 2805 // should just return out. 2806 return false; 2807 } 2808 2809 // Do the assignment. 2810 Out.EmitAssignment(Sym, Value); 2811 if (NoDeadStrip) 2812 Out.EmitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2813 2814 return false; 2815 } 2816 2817 /// parseIdentifier: 2818 /// ::= identifier 2819 /// ::= string 2820 bool AsmParser::parseIdentifier(StringRef &Res) { 2821 // The assembler has relaxed rules for accepting identifiers, in particular we 2822 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2823 // separate tokens. At this level, we have already lexed so we cannot (currently) 2824 // handle this as a context dependent token, instead we detect adjacent tokens 2825 // and return the combined identifier. 2826 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2827 SMLoc PrefixLoc = getLexer().getLoc(); 2828 2829 // Consume the prefix character, and check for a following identifier. 2830 2831 AsmToken Buf[1]; 2832 Lexer.peekTokens(Buf, false); 2833 2834 if (Buf[0].isNot(AsmToken::Identifier)) 2835 return true; 2836 2837 // We have a '$' or '@' followed by an identifier, make sure they are adjacent. 2838 if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer()) 2839 return true; 2840 2841 // eat $ or @ 2842 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2843 // Construct the joined identifier and consume the token. 2844 Res = 2845 StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1); 2846 Lex(); // Parser Lex to maintain invariants. 2847 return false; 2848 } 2849 2850 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2851 return true; 2852 2853 Res = getTok().getIdentifier(); 2854 2855 Lex(); // Consume the identifier token. 2856 2857 return false; 2858 } 2859 2860 /// parseDirectiveSet: 2861 /// ::= .equ identifier ',' expression 2862 /// ::= .equiv identifier ',' expression 2863 /// ::= .set identifier ',' expression 2864 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2865 StringRef Name; 2866 if (check(parseIdentifier(Name), "expected identifier") || 2867 parseToken(AsmToken::Comma) || parseAssignment(Name, allow_redef, true)) 2868 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2869 return false; 2870 } 2871 2872 bool AsmParser::parseEscapedString(std::string &Data) { 2873 if (check(getTok().isNot(AsmToken::String), "expected string")) 2874 return true; 2875 2876 Data = ""; 2877 StringRef Str = getTok().getStringContents(); 2878 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2879 if (Str[i] != '\\') { 2880 Data += Str[i]; 2881 continue; 2882 } 2883 2884 // Recognize escaped characters. Note that this escape semantics currently 2885 // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes. 2886 ++i; 2887 if (i == e) 2888 return TokError("unexpected backslash at end of string"); 2889 2890 // Recognize octal sequences. 2891 if ((unsigned)(Str[i] - '0') <= 7) { 2892 // Consume up to three octal characters. 2893 unsigned Value = Str[i] - '0'; 2894 2895 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2896 ++i; 2897 Value = Value * 8 + (Str[i] - '0'); 2898 2899 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2900 ++i; 2901 Value = Value * 8 + (Str[i] - '0'); 2902 } 2903 } 2904 2905 if (Value > 255) 2906 return TokError("invalid octal escape sequence (out of range)"); 2907 2908 Data += (unsigned char)Value; 2909 continue; 2910 } 2911 2912 // Otherwise recognize individual escapes. 2913 switch (Str[i]) { 2914 default: 2915 // Just reject invalid escape sequences for now. 2916 return TokError("invalid escape sequence (unrecognized character)"); 2917 2918 case 'b': Data += '\b'; break; 2919 case 'f': Data += '\f'; break; 2920 case 'n': Data += '\n'; break; 2921 case 'r': Data += '\r'; break; 2922 case 't': Data += '\t'; break; 2923 case '"': Data += '"'; break; 2924 case '\\': Data += '\\'; break; 2925 } 2926 } 2927 2928 Lex(); 2929 return false; 2930 } 2931 2932 /// parseDirectiveAscii: 2933 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2934 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2935 auto parseOp = [&]() -> bool { 2936 std::string Data; 2937 if (checkForValidSection() || parseEscapedString(Data)) 2938 return true; 2939 getStreamer().EmitBytes(Data); 2940 if (ZeroTerminated) 2941 getStreamer().EmitBytes(StringRef("\0", 1)); 2942 return false; 2943 }; 2944 2945 if (parseMany(parseOp)) 2946 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2947 return false; 2948 } 2949 2950 /// parseDirectiveReloc 2951 /// ::= .reloc expression , identifier [ , expression ] 2952 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 2953 const MCExpr *Offset; 2954 const MCExpr *Expr = nullptr; 2955 int64_t OffsetValue; 2956 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 2957 2958 if (parseExpression(Offset)) 2959 return true; 2960 2961 if ((Offset->evaluateAsAbsolute(OffsetValue, 2962 getStreamer().getAssemblerPtr()) && 2963 check(OffsetValue < 0, OffsetLoc, "expression is negative")) || 2964 (check(Offset->getKind() != llvm::MCExpr::Constant && 2965 Offset->getKind() != llvm::MCExpr::SymbolRef, 2966 OffsetLoc, "expected non-negative number or a label")) || 2967 (parseToken(AsmToken::Comma, "expected comma") || 2968 check(getTok().isNot(AsmToken::Identifier), "expected relocation name"))) 2969 return true; 2970 2971 SMLoc NameLoc = Lexer.getTok().getLoc(); 2972 StringRef Name = Lexer.getTok().getIdentifier(); 2973 Lex(); 2974 2975 if (Lexer.is(AsmToken::Comma)) { 2976 Lex(); 2977 SMLoc ExprLoc = Lexer.getLoc(); 2978 if (parseExpression(Expr)) 2979 return true; 2980 2981 MCValue Value; 2982 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 2983 return Error(ExprLoc, "expression must be relocatable"); 2984 } 2985 2986 if (parseToken(AsmToken::EndOfStatement, 2987 "unexpected token in .reloc directive")) 2988 return true; 2989 2990 const MCTargetAsmParser &MCT = getTargetParser(); 2991 const MCSubtargetInfo &STI = MCT.getSTI(); 2992 if (getStreamer().EmitRelocDirective(*Offset, Name, Expr, DirectiveLoc, STI)) 2993 return Error(NameLoc, "unknown relocation name"); 2994 2995 return false; 2996 } 2997 2998 /// parseDirectiveValue 2999 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 3000 bool AsmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) { 3001 auto parseOp = [&]() -> bool { 3002 const MCExpr *Value; 3003 SMLoc ExprLoc = getLexer().getLoc(); 3004 if (checkForValidSection() || parseExpression(Value)) 3005 return true; 3006 // Special case constant expressions to match code generator. 3007 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3008 assert(Size <= 8 && "Invalid size"); 3009 uint64_t IntValue = MCE->getValue(); 3010 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 3011 return Error(ExprLoc, "out of range literal value"); 3012 getStreamer().EmitIntValue(IntValue, Size); 3013 } else 3014 getStreamer().EmitValue(Value, Size, ExprLoc); 3015 return false; 3016 }; 3017 3018 if (parseMany(parseOp)) 3019 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3020 return false; 3021 } 3022 3023 static bool parseHexOcta(AsmParser &Asm, uint64_t &hi, uint64_t &lo) { 3024 if (Asm.getTok().isNot(AsmToken::Integer) && 3025 Asm.getTok().isNot(AsmToken::BigNum)) 3026 return Asm.TokError("unknown token in expression"); 3027 SMLoc ExprLoc = Asm.getTok().getLoc(); 3028 APInt IntValue = Asm.getTok().getAPIntVal(); 3029 Asm.Lex(); 3030 if (!IntValue.isIntN(128)) 3031 return Asm.Error(ExprLoc, "out of range literal value"); 3032 if (!IntValue.isIntN(64)) { 3033 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 3034 lo = IntValue.getLoBits(64).getZExtValue(); 3035 } else { 3036 hi = 0; 3037 lo = IntValue.getZExtValue(); 3038 } 3039 return false; 3040 } 3041 3042 /// ParseDirectiveOctaValue 3043 /// ::= .octa [ hexconstant (, hexconstant)* ] 3044 3045 bool AsmParser::parseDirectiveOctaValue(StringRef IDVal) { 3046 auto parseOp = [&]() -> bool { 3047 if (checkForValidSection()) 3048 return true; 3049 uint64_t hi, lo; 3050 if (parseHexOcta(*this, hi, lo)) 3051 return true; 3052 if (MAI.isLittleEndian()) { 3053 getStreamer().EmitIntValue(lo, 8); 3054 getStreamer().EmitIntValue(hi, 8); 3055 } else { 3056 getStreamer().EmitIntValue(hi, 8); 3057 getStreamer().EmitIntValue(lo, 8); 3058 } 3059 return false; 3060 }; 3061 3062 if (parseMany(parseOp)) 3063 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3064 return false; 3065 } 3066 3067 bool AsmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) { 3068 // We don't truly support arithmetic on floating point expressions, so we 3069 // have to manually parse unary prefixes. 3070 bool IsNeg = false; 3071 if (getLexer().is(AsmToken::Minus)) { 3072 Lexer.Lex(); 3073 IsNeg = true; 3074 } else if (getLexer().is(AsmToken::Plus)) 3075 Lexer.Lex(); 3076 3077 if (Lexer.is(AsmToken::Error)) 3078 return TokError(Lexer.getErr()); 3079 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 3080 Lexer.isNot(AsmToken::Identifier)) 3081 return TokError("unexpected token in directive"); 3082 3083 // Convert to an APFloat. 3084 APFloat Value(Semantics); 3085 StringRef IDVal = getTok().getString(); 3086 if (getLexer().is(AsmToken::Identifier)) { 3087 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 3088 Value = APFloat::getInf(Semantics); 3089 else if (!IDVal.compare_lower("nan")) 3090 Value = APFloat::getNaN(Semantics, false, ~0); 3091 else 3092 return TokError("invalid floating point literal"); 3093 } else if (Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) == 3094 APFloat::opInvalidOp) 3095 return TokError("invalid floating point literal"); 3096 if (IsNeg) 3097 Value.changeSign(); 3098 3099 // Consume the numeric token. 3100 Lex(); 3101 3102 Res = Value.bitcastToAPInt(); 3103 3104 return false; 3105 } 3106 3107 /// parseDirectiveRealValue 3108 /// ::= (.single | .double) [ expression (, expression)* ] 3109 bool AsmParser::parseDirectiveRealValue(StringRef IDVal, 3110 const fltSemantics &Semantics) { 3111 auto parseOp = [&]() -> bool { 3112 APInt AsInt; 3113 if (checkForValidSection() || parseRealValue(Semantics, AsInt)) 3114 return true; 3115 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 3116 AsInt.getBitWidth() / 8); 3117 return false; 3118 }; 3119 3120 if (parseMany(parseOp)) 3121 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3122 return false; 3123 } 3124 3125 /// parseDirectiveZero 3126 /// ::= .zero expression 3127 bool AsmParser::parseDirectiveZero() { 3128 SMLoc NumBytesLoc = Lexer.getLoc(); 3129 const MCExpr *NumBytes; 3130 if (checkForValidSection() || parseExpression(NumBytes)) 3131 return true; 3132 3133 int64_t Val = 0; 3134 if (getLexer().is(AsmToken::Comma)) { 3135 Lex(); 3136 if (parseAbsoluteExpression(Val)) 3137 return true; 3138 } 3139 3140 if (parseToken(AsmToken::EndOfStatement, 3141 "unexpected token in '.zero' directive")) 3142 return true; 3143 getStreamer().emitFill(*NumBytes, Val, NumBytesLoc); 3144 3145 return false; 3146 } 3147 3148 /// parseDirectiveFill 3149 /// ::= .fill expression [ , expression [ , expression ] ] 3150 bool AsmParser::parseDirectiveFill() { 3151 SMLoc NumValuesLoc = Lexer.getLoc(); 3152 const MCExpr *NumValues; 3153 if (checkForValidSection() || parseExpression(NumValues)) 3154 return true; 3155 3156 int64_t FillSize = 1; 3157 int64_t FillExpr = 0; 3158 3159 SMLoc SizeLoc, ExprLoc; 3160 3161 if (parseOptionalToken(AsmToken::Comma)) { 3162 SizeLoc = getTok().getLoc(); 3163 if (parseAbsoluteExpression(FillSize)) 3164 return true; 3165 if (parseOptionalToken(AsmToken::Comma)) { 3166 ExprLoc = getTok().getLoc(); 3167 if (parseAbsoluteExpression(FillExpr)) 3168 return true; 3169 } 3170 } 3171 if (parseToken(AsmToken::EndOfStatement, 3172 "unexpected token in '.fill' directive")) 3173 return true; 3174 3175 if (FillSize < 0) { 3176 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 3177 return false; 3178 } 3179 if (FillSize > 8) { 3180 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 3181 FillSize = 8; 3182 } 3183 3184 if (!isUInt<32>(FillExpr) && FillSize > 4) 3185 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 3186 3187 getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc); 3188 3189 return false; 3190 } 3191 3192 /// parseDirectiveOrg 3193 /// ::= .org expression [ , expression ] 3194 bool AsmParser::parseDirectiveOrg() { 3195 const MCExpr *Offset; 3196 SMLoc OffsetLoc = Lexer.getLoc(); 3197 if (checkForValidSection() || parseExpression(Offset)) 3198 return true; 3199 3200 // Parse optional fill expression. 3201 int64_t FillExpr = 0; 3202 if (parseOptionalToken(AsmToken::Comma)) 3203 if (parseAbsoluteExpression(FillExpr)) 3204 return addErrorSuffix(" in '.org' directive"); 3205 if (parseToken(AsmToken::EndOfStatement)) 3206 return addErrorSuffix(" in '.org' directive"); 3207 3208 getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc); 3209 return false; 3210 } 3211 3212 /// parseDirectiveAlign 3213 /// ::= {.align, ...} expression [ , expression [ , expression ]] 3214 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 3215 SMLoc AlignmentLoc = getLexer().getLoc(); 3216 int64_t Alignment; 3217 SMLoc MaxBytesLoc; 3218 bool HasFillExpr = false; 3219 int64_t FillExpr = 0; 3220 int64_t MaxBytesToFill = 0; 3221 3222 auto parseAlign = [&]() -> bool { 3223 if (parseAbsoluteExpression(Alignment)) 3224 return true; 3225 if (parseOptionalToken(AsmToken::Comma)) { 3226 // The fill expression can be omitted while specifying a maximum number of 3227 // alignment bytes, e.g: 3228 // .align 3,,4 3229 if (getTok().isNot(AsmToken::Comma)) { 3230 HasFillExpr = true; 3231 if (parseAbsoluteExpression(FillExpr)) 3232 return true; 3233 } 3234 if (parseOptionalToken(AsmToken::Comma)) 3235 if (parseTokenLoc(MaxBytesLoc) || 3236 parseAbsoluteExpression(MaxBytesToFill)) 3237 return true; 3238 } 3239 return parseToken(AsmToken::EndOfStatement); 3240 }; 3241 3242 if (checkForValidSection()) 3243 return addErrorSuffix(" in directive"); 3244 // Ignore empty '.p2align' directives for GNU-as compatibility 3245 if (IsPow2 && (ValueSize == 1) && getTok().is(AsmToken::EndOfStatement)) { 3246 Warning(AlignmentLoc, "p2align directive with no operand(s) is ignored"); 3247 return parseToken(AsmToken::EndOfStatement); 3248 } 3249 if (parseAlign()) 3250 return addErrorSuffix(" in directive"); 3251 3252 // Always emit an alignment here even if we thrown an error. 3253 bool ReturnVal = false; 3254 3255 // Compute alignment in bytes. 3256 if (IsPow2) { 3257 // FIXME: Diagnose overflow. 3258 if (Alignment >= 32) { 3259 ReturnVal |= Error(AlignmentLoc, "invalid alignment value"); 3260 Alignment = 31; 3261 } 3262 3263 Alignment = 1ULL << Alignment; 3264 } else { 3265 // Reject alignments that aren't either a power of two or zero, 3266 // for gas compatibility. Alignment of zero is silently rounded 3267 // up to one. 3268 if (Alignment == 0) 3269 Alignment = 1; 3270 if (!isPowerOf2_64(Alignment)) 3271 ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2"); 3272 } 3273 3274 // Diagnose non-sensical max bytes to align. 3275 if (MaxBytesLoc.isValid()) { 3276 if (MaxBytesToFill < 1) { 3277 ReturnVal |= Error(MaxBytesLoc, 3278 "alignment directive can never be satisfied in this " 3279 "many bytes, ignoring maximum bytes expression"); 3280 MaxBytesToFill = 0; 3281 } 3282 3283 if (MaxBytesToFill >= Alignment) { 3284 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 3285 "has no effect"); 3286 MaxBytesToFill = 0; 3287 } 3288 } 3289 3290 // Check whether we should use optimal code alignment for this .align 3291 // directive. 3292 const MCSection *Section = getStreamer().getCurrentSectionOnly(); 3293 assert(Section && "must have section to emit alignment"); 3294 bool UseCodeAlign = Section->UseCodeAlign(); 3295 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 3296 ValueSize == 1 && UseCodeAlign) { 3297 getStreamer().EmitCodeAlignment(Alignment, MaxBytesToFill); 3298 } else { 3299 // FIXME: Target specific behavior about how the "extra" bytes are filled. 3300 getStreamer().EmitValueToAlignment(Alignment, FillExpr, ValueSize, 3301 MaxBytesToFill); 3302 } 3303 3304 return ReturnVal; 3305 } 3306 3307 /// parseDirectiveFile 3308 /// ::= .file filename 3309 /// ::= .file number [directory] filename [md5 checksum] [source source-text] 3310 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 3311 // FIXME: I'm not sure what this is. 3312 int64_t FileNumber = -1; 3313 if (getLexer().is(AsmToken::Integer)) { 3314 FileNumber = getTok().getIntVal(); 3315 Lex(); 3316 3317 if (FileNumber < 0) 3318 return TokError("negative file number"); 3319 } 3320 3321 std::string Path; 3322 3323 // Usually the directory and filename together, otherwise just the directory. 3324 // Allow the strings to have escaped octal character sequence. 3325 if (check(getTok().isNot(AsmToken::String), 3326 "unexpected token in '.file' directive") || 3327 parseEscapedString(Path)) 3328 return true; 3329 3330 StringRef Directory; 3331 StringRef Filename; 3332 std::string FilenameData; 3333 if (getLexer().is(AsmToken::String)) { 3334 if (check(FileNumber == -1, 3335 "explicit path specified, but no file number") || 3336 parseEscapedString(FilenameData)) 3337 return true; 3338 Filename = FilenameData; 3339 Directory = Path; 3340 } else { 3341 Filename = Path; 3342 } 3343 3344 uint64_t MD5Hi, MD5Lo; 3345 bool HasMD5 = false; 3346 3347 Optional<StringRef> Source; 3348 bool HasSource = false; 3349 std::string SourceString; 3350 3351 while (!parseOptionalToken(AsmToken::EndOfStatement)) { 3352 StringRef Keyword; 3353 if (check(getTok().isNot(AsmToken::Identifier), 3354 "unexpected token in '.file' directive") || 3355 parseIdentifier(Keyword)) 3356 return true; 3357 if (Keyword == "md5") { 3358 HasMD5 = true; 3359 if (check(FileNumber == -1, 3360 "MD5 checksum specified, but no file number") || 3361 parseHexOcta(*this, MD5Hi, MD5Lo)) 3362 return true; 3363 } else if (Keyword == "source") { 3364 HasSource = true; 3365 if (check(FileNumber == -1, 3366 "source specified, but no file number") || 3367 check(getTok().isNot(AsmToken::String), 3368 "unexpected token in '.file' directive") || 3369 parseEscapedString(SourceString)) 3370 return true; 3371 } else { 3372 return TokError("unexpected token in '.file' directive"); 3373 } 3374 } 3375 3376 if (FileNumber == -1) { 3377 // Ignore the directive if there is no number and the target doesn't support 3378 // numberless .file directives. This allows some portability of assembler 3379 // between different object file formats. 3380 if (getContext().getAsmInfo()->hasSingleParameterDotFile()) 3381 getStreamer().EmitFileDirective(Filename); 3382 } else { 3383 // In case there is a -g option as well as debug info from directive .file, 3384 // we turn off the -g option, directly use the existing debug info instead. 3385 // Also reset any implicit ".file 0" for the assembler source. 3386 if (Ctx.getGenDwarfForAssembly()) { 3387 Ctx.getMCDwarfLineTable(0).resetRootFile(); 3388 Ctx.setGenDwarfForAssembly(false); 3389 } 3390 3391 Optional<MD5::MD5Result> CKMem; 3392 if (HasMD5) { 3393 MD5::MD5Result Sum; 3394 for (unsigned i = 0; i != 8; ++i) { 3395 Sum.Bytes[i] = uint8_t(MD5Hi >> ((7 - i) * 8)); 3396 Sum.Bytes[i + 8] = uint8_t(MD5Lo >> ((7 - i) * 8)); 3397 } 3398 CKMem = Sum; 3399 } 3400 if (HasSource) { 3401 char *SourceBuf = static_cast<char *>(Ctx.allocate(SourceString.size())); 3402 memcpy(SourceBuf, SourceString.data(), SourceString.size()); 3403 Source = StringRef(SourceBuf, SourceString.size()); 3404 } 3405 if (FileNumber == 0) { 3406 if (Ctx.getDwarfVersion() < 5) 3407 return Warning(DirectiveLoc, "file 0 not supported prior to DWARF-5"); 3408 getStreamer().emitDwarfFile0Directive(Directory, Filename, CKMem, Source); 3409 } else { 3410 Expected<unsigned> FileNumOrErr = getStreamer().tryEmitDwarfFileDirective( 3411 FileNumber, Directory, Filename, CKMem, Source); 3412 if (!FileNumOrErr) 3413 return Error(DirectiveLoc, toString(FileNumOrErr.takeError())); 3414 FileNumber = FileNumOrErr.get(); 3415 } 3416 // Alert the user if there are some .file directives with MD5 and some not. 3417 // But only do that once. 3418 if (!ReportedInconsistentMD5 && !Ctx.isDwarfMD5UsageConsistent(0)) { 3419 ReportedInconsistentMD5 = true; 3420 return Warning(DirectiveLoc, "inconsistent use of MD5 checksums"); 3421 } 3422 } 3423 3424 return false; 3425 } 3426 3427 /// parseDirectiveLine 3428 /// ::= .line [number] 3429 bool AsmParser::parseDirectiveLine() { 3430 int64_t LineNumber; 3431 if (getLexer().is(AsmToken::Integer)) { 3432 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3433 return true; 3434 (void)LineNumber; 3435 // FIXME: Do something with the .line. 3436 } 3437 if (parseToken(AsmToken::EndOfStatement, 3438 "unexpected token in '.line' directive")) 3439 return true; 3440 3441 return false; 3442 } 3443 3444 /// parseDirectiveLoc 3445 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3446 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3447 /// The first number is a file number, must have been previously assigned with 3448 /// a .file directive, the second number is the line number and optionally the 3449 /// third number is a column position (zero if not specified). The remaining 3450 /// optional items are .loc sub-directives. 3451 bool AsmParser::parseDirectiveLoc() { 3452 int64_t FileNumber = 0, LineNumber = 0; 3453 SMLoc Loc = getTok().getLoc(); 3454 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3455 check(FileNumber < 1 && Ctx.getDwarfVersion() < 5, Loc, 3456 "file number less than one in '.loc' directive") || 3457 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3458 "unassigned file number in '.loc' directive")) 3459 return true; 3460 3461 // optional 3462 if (getLexer().is(AsmToken::Integer)) { 3463 LineNumber = getTok().getIntVal(); 3464 if (LineNumber < 0) 3465 return TokError("line number less than zero in '.loc' directive"); 3466 Lex(); 3467 } 3468 3469 int64_t ColumnPos = 0; 3470 if (getLexer().is(AsmToken::Integer)) { 3471 ColumnPos = getTok().getIntVal(); 3472 if (ColumnPos < 0) 3473 return TokError("column position less than zero in '.loc' directive"); 3474 Lex(); 3475 } 3476 3477 unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0; 3478 unsigned Isa = 0; 3479 int64_t Discriminator = 0; 3480 3481 auto parseLocOp = [&]() -> bool { 3482 StringRef Name; 3483 SMLoc Loc = getTok().getLoc(); 3484 if (parseIdentifier(Name)) 3485 return TokError("unexpected token in '.loc' directive"); 3486 3487 if (Name == "basic_block") 3488 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3489 else if (Name == "prologue_end") 3490 Flags |= DWARF2_FLAG_PROLOGUE_END; 3491 else if (Name == "epilogue_begin") 3492 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3493 else if (Name == "is_stmt") { 3494 Loc = getTok().getLoc(); 3495 const MCExpr *Value; 3496 if (parseExpression(Value)) 3497 return true; 3498 // The expression must be the constant 0 or 1. 3499 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3500 int Value = MCE->getValue(); 3501 if (Value == 0) 3502 Flags &= ~DWARF2_FLAG_IS_STMT; 3503 else if (Value == 1) 3504 Flags |= DWARF2_FLAG_IS_STMT; 3505 else 3506 return Error(Loc, "is_stmt value not 0 or 1"); 3507 } else { 3508 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3509 } 3510 } else if (Name == "isa") { 3511 Loc = getTok().getLoc(); 3512 const MCExpr *Value; 3513 if (parseExpression(Value)) 3514 return true; 3515 // The expression must be a constant greater or equal to 0. 3516 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3517 int Value = MCE->getValue(); 3518 if (Value < 0) 3519 return Error(Loc, "isa number less than zero"); 3520 Isa = Value; 3521 } else { 3522 return Error(Loc, "isa number not a constant value"); 3523 } 3524 } else if (Name == "discriminator") { 3525 if (parseAbsoluteExpression(Discriminator)) 3526 return true; 3527 } else { 3528 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3529 } 3530 return false; 3531 }; 3532 3533 if (parseMany(parseLocOp, false /*hasComma*/)) 3534 return true; 3535 3536 getStreamer().EmitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3537 Isa, Discriminator, StringRef()); 3538 3539 return false; 3540 } 3541 3542 /// parseDirectiveStabs 3543 /// ::= .stabs string, number, number, number 3544 bool AsmParser::parseDirectiveStabs() { 3545 return TokError("unsupported directive '.stabs'"); 3546 } 3547 3548 /// parseDirectiveCVFile 3549 /// ::= .cv_file number filename [checksum] [checksumkind] 3550 bool AsmParser::parseDirectiveCVFile() { 3551 SMLoc FileNumberLoc = getTok().getLoc(); 3552 int64_t FileNumber; 3553 std::string Filename; 3554 std::string Checksum; 3555 int64_t ChecksumKind = 0; 3556 3557 if (parseIntToken(FileNumber, 3558 "expected file number in '.cv_file' directive") || 3559 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3560 check(getTok().isNot(AsmToken::String), 3561 "unexpected token in '.cv_file' directive") || 3562 parseEscapedString(Filename)) 3563 return true; 3564 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3565 if (check(getTok().isNot(AsmToken::String), 3566 "unexpected token in '.cv_file' directive") || 3567 parseEscapedString(Checksum) || 3568 parseIntToken(ChecksumKind, 3569 "expected checksum kind in '.cv_file' directive") || 3570 parseToken(AsmToken::EndOfStatement, 3571 "unexpected token in '.cv_file' directive")) 3572 return true; 3573 } 3574 3575 Checksum = fromHex(Checksum); 3576 void *CKMem = Ctx.allocate(Checksum.size(), 1); 3577 memcpy(CKMem, Checksum.data(), Checksum.size()); 3578 ArrayRef<uint8_t> ChecksumAsBytes(reinterpret_cast<const uint8_t *>(CKMem), 3579 Checksum.size()); 3580 3581 if (!getStreamer().EmitCVFileDirective(FileNumber, Filename, ChecksumAsBytes, 3582 static_cast<uint8_t>(ChecksumKind))) 3583 return Error(FileNumberLoc, "file number already allocated"); 3584 3585 return false; 3586 } 3587 3588 bool AsmParser::parseCVFunctionId(int64_t &FunctionId, 3589 StringRef DirectiveName) { 3590 SMLoc Loc; 3591 return parseTokenLoc(Loc) || 3592 parseIntToken(FunctionId, "expected function id in '" + DirectiveName + 3593 "' directive") || 3594 check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc, 3595 "expected function id within range [0, UINT_MAX)"); 3596 } 3597 3598 bool AsmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) { 3599 SMLoc Loc; 3600 return parseTokenLoc(Loc) || 3601 parseIntToken(FileNumber, "expected integer in '" + DirectiveName + 3602 "' directive") || 3603 check(FileNumber < 1, Loc, "file number less than one in '" + 3604 DirectiveName + "' directive") || 3605 check(!getCVContext().isValidFileNumber(FileNumber), Loc, 3606 "unassigned file number in '" + DirectiveName + "' directive"); 3607 } 3608 3609 /// parseDirectiveCVFuncId 3610 /// ::= .cv_func_id FunctionId 3611 /// 3612 /// Introduces a function ID that can be used with .cv_loc. 3613 bool AsmParser::parseDirectiveCVFuncId() { 3614 SMLoc FunctionIdLoc = getTok().getLoc(); 3615 int64_t FunctionId; 3616 3617 if (parseCVFunctionId(FunctionId, ".cv_func_id") || 3618 parseToken(AsmToken::EndOfStatement, 3619 "unexpected token in '.cv_func_id' directive")) 3620 return true; 3621 3622 if (!getStreamer().EmitCVFuncIdDirective(FunctionId)) 3623 return Error(FunctionIdLoc, "function id already allocated"); 3624 3625 return false; 3626 } 3627 3628 /// parseDirectiveCVInlineSiteId 3629 /// ::= .cv_inline_site_id FunctionId 3630 /// "within" IAFunc 3631 /// "inlined_at" IAFile IALine [IACol] 3632 /// 3633 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined 3634 /// at" source location information for use in the line table of the caller, 3635 /// whether the caller is a real function or another inlined call site. 3636 bool AsmParser::parseDirectiveCVInlineSiteId() { 3637 SMLoc FunctionIdLoc = getTok().getLoc(); 3638 int64_t FunctionId; 3639 int64_t IAFunc; 3640 int64_t IAFile; 3641 int64_t IALine; 3642 int64_t IACol = 0; 3643 3644 // FunctionId 3645 if (parseCVFunctionId(FunctionId, ".cv_inline_site_id")) 3646 return true; 3647 3648 // "within" 3649 if (check((getLexer().isNot(AsmToken::Identifier) || 3650 getTok().getIdentifier() != "within"), 3651 "expected 'within' identifier in '.cv_inline_site_id' directive")) 3652 return true; 3653 Lex(); 3654 3655 // IAFunc 3656 if (parseCVFunctionId(IAFunc, ".cv_inline_site_id")) 3657 return true; 3658 3659 // "inlined_at" 3660 if (check((getLexer().isNot(AsmToken::Identifier) || 3661 getTok().getIdentifier() != "inlined_at"), 3662 "expected 'inlined_at' identifier in '.cv_inline_site_id' " 3663 "directive") ) 3664 return true; 3665 Lex(); 3666 3667 // IAFile IALine 3668 if (parseCVFileId(IAFile, ".cv_inline_site_id") || 3669 parseIntToken(IALine, "expected line number after 'inlined_at'")) 3670 return true; 3671 3672 // [IACol] 3673 if (getLexer().is(AsmToken::Integer)) { 3674 IACol = getTok().getIntVal(); 3675 Lex(); 3676 } 3677 3678 if (parseToken(AsmToken::EndOfStatement, 3679 "unexpected token in '.cv_inline_site_id' directive")) 3680 return true; 3681 3682 if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile, 3683 IALine, IACol, FunctionIdLoc)) 3684 return Error(FunctionIdLoc, "function id already allocated"); 3685 3686 return false; 3687 } 3688 3689 /// parseDirectiveCVLoc 3690 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3691 /// [is_stmt VALUE] 3692 /// The first number is a file number, must have been previously assigned with 3693 /// a .file directive, the second number is the line number and optionally the 3694 /// third number is a column position (zero if not specified). The remaining 3695 /// optional items are .loc sub-directives. 3696 bool AsmParser::parseDirectiveCVLoc() { 3697 SMLoc DirectiveLoc = getTok().getLoc(); 3698 int64_t FunctionId, FileNumber; 3699 if (parseCVFunctionId(FunctionId, ".cv_loc") || 3700 parseCVFileId(FileNumber, ".cv_loc")) 3701 return true; 3702 3703 int64_t LineNumber = 0; 3704 if (getLexer().is(AsmToken::Integer)) { 3705 LineNumber = getTok().getIntVal(); 3706 if (LineNumber < 0) 3707 return TokError("line number less than zero in '.cv_loc' directive"); 3708 Lex(); 3709 } 3710 3711 int64_t ColumnPos = 0; 3712 if (getLexer().is(AsmToken::Integer)) { 3713 ColumnPos = getTok().getIntVal(); 3714 if (ColumnPos < 0) 3715 return TokError("column position less than zero in '.cv_loc' directive"); 3716 Lex(); 3717 } 3718 3719 bool PrologueEnd = false; 3720 uint64_t IsStmt = 0; 3721 3722 auto parseOp = [&]() -> bool { 3723 StringRef Name; 3724 SMLoc Loc = getTok().getLoc(); 3725 if (parseIdentifier(Name)) 3726 return TokError("unexpected token in '.cv_loc' directive"); 3727 if (Name == "prologue_end") 3728 PrologueEnd = true; 3729 else if (Name == "is_stmt") { 3730 Loc = getTok().getLoc(); 3731 const MCExpr *Value; 3732 if (parseExpression(Value)) 3733 return true; 3734 // The expression must be the constant 0 or 1. 3735 IsStmt = ~0ULL; 3736 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3737 IsStmt = MCE->getValue(); 3738 3739 if (IsStmt > 1) 3740 return Error(Loc, "is_stmt value not 0 or 1"); 3741 } else { 3742 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3743 } 3744 return false; 3745 }; 3746 3747 if (parseMany(parseOp, false /*hasComma*/)) 3748 return true; 3749 3750 getStreamer().EmitCVLocDirective(FunctionId, FileNumber, LineNumber, 3751 ColumnPos, PrologueEnd, IsStmt, StringRef(), 3752 DirectiveLoc); 3753 return false; 3754 } 3755 3756 /// parseDirectiveCVLinetable 3757 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3758 bool AsmParser::parseDirectiveCVLinetable() { 3759 int64_t FunctionId; 3760 StringRef FnStartName, FnEndName; 3761 SMLoc Loc = getTok().getLoc(); 3762 if (parseCVFunctionId(FunctionId, ".cv_linetable") || 3763 parseToken(AsmToken::Comma, 3764 "unexpected token in '.cv_linetable' directive") || 3765 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3766 "expected identifier in directive") || 3767 parseToken(AsmToken::Comma, 3768 "unexpected token in '.cv_linetable' directive") || 3769 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3770 "expected identifier in directive")) 3771 return true; 3772 3773 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3774 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3775 3776 getStreamer().EmitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3777 return false; 3778 } 3779 3780 /// parseDirectiveCVInlineLinetable 3781 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3782 bool AsmParser::parseDirectiveCVInlineLinetable() { 3783 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3784 StringRef FnStartName, FnEndName; 3785 SMLoc Loc = getTok().getLoc(); 3786 if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") || 3787 parseTokenLoc(Loc) || 3788 parseIntToken( 3789 SourceFileId, 3790 "expected SourceField in '.cv_inline_linetable' directive") || 3791 check(SourceFileId <= 0, Loc, 3792 "File id less than zero in '.cv_inline_linetable' directive") || 3793 parseTokenLoc(Loc) || 3794 parseIntToken( 3795 SourceLineNum, 3796 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3797 check(SourceLineNum < 0, Loc, 3798 "Line number less than zero in '.cv_inline_linetable' directive") || 3799 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3800 "expected identifier in directive") || 3801 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3802 "expected identifier in directive")) 3803 return true; 3804 3805 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3806 return true; 3807 3808 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3809 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3810 getStreamer().EmitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3811 SourceLineNum, FnStartSym, 3812 FnEndSym); 3813 return false; 3814 } 3815 3816 /// parseDirectiveCVDefRange 3817 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3818 bool AsmParser::parseDirectiveCVDefRange() { 3819 SMLoc Loc; 3820 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3821 while (getLexer().is(AsmToken::Identifier)) { 3822 Loc = getLexer().getLoc(); 3823 StringRef GapStartName; 3824 if (parseIdentifier(GapStartName)) 3825 return Error(Loc, "expected identifier in directive"); 3826 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3827 3828 Loc = getLexer().getLoc(); 3829 StringRef GapEndName; 3830 if (parseIdentifier(GapEndName)) 3831 return Error(Loc, "expected identifier in directive"); 3832 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3833 3834 Ranges.push_back({GapStartSym, GapEndSym}); 3835 } 3836 3837 std::string FixedSizePortion; 3838 if (parseToken(AsmToken::Comma, "unexpected token in directive") || 3839 parseEscapedString(FixedSizePortion)) 3840 return true; 3841 3842 getStreamer().EmitCVDefRangeDirective(Ranges, FixedSizePortion); 3843 return false; 3844 } 3845 3846 /// parseDirectiveCVString 3847 /// ::= .cv_stringtable "string" 3848 bool AsmParser::parseDirectiveCVString() { 3849 std::string Data; 3850 if (checkForValidSection() || parseEscapedString(Data)) 3851 return addErrorSuffix(" in '.cv_string' directive"); 3852 3853 // Put the string in the table and emit the offset. 3854 std::pair<StringRef, unsigned> Insertion = 3855 getCVContext().addToStringTable(Data); 3856 getStreamer().EmitIntValue(Insertion.second, 4); 3857 return false; 3858 } 3859 3860 /// parseDirectiveCVStringTable 3861 /// ::= .cv_stringtable 3862 bool AsmParser::parseDirectiveCVStringTable() { 3863 getStreamer().EmitCVStringTableDirective(); 3864 return false; 3865 } 3866 3867 /// parseDirectiveCVFileChecksums 3868 /// ::= .cv_filechecksums 3869 bool AsmParser::parseDirectiveCVFileChecksums() { 3870 getStreamer().EmitCVFileChecksumsDirective(); 3871 return false; 3872 } 3873 3874 /// parseDirectiveCVFileChecksumOffset 3875 /// ::= .cv_filechecksumoffset fileno 3876 bool AsmParser::parseDirectiveCVFileChecksumOffset() { 3877 int64_t FileNo; 3878 if (parseIntToken(FileNo, "expected identifier in directive")) 3879 return true; 3880 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3881 return true; 3882 getStreamer().EmitCVFileChecksumOffsetDirective(FileNo); 3883 return false; 3884 } 3885 3886 /// parseDirectiveCVFPOData 3887 /// ::= .cv_fpo_data procsym 3888 bool AsmParser::parseDirectiveCVFPOData() { 3889 SMLoc DirLoc = getLexer().getLoc(); 3890 StringRef ProcName; 3891 if (parseIdentifier(ProcName)) 3892 return TokError("expected symbol name"); 3893 if (parseEOL("unexpected tokens")) 3894 return addErrorSuffix(" in '.cv_fpo_data' directive"); 3895 MCSymbol *ProcSym = getContext().getOrCreateSymbol(ProcName); 3896 getStreamer().EmitCVFPOData(ProcSym, DirLoc); 3897 return false; 3898 } 3899 3900 /// parseDirectiveCFISections 3901 /// ::= .cfi_sections section [, section] 3902 bool AsmParser::parseDirectiveCFISections() { 3903 StringRef Name; 3904 bool EH = false; 3905 bool Debug = false; 3906 3907 if (parseIdentifier(Name)) 3908 return TokError("Expected an identifier"); 3909 3910 if (Name == ".eh_frame") 3911 EH = true; 3912 else if (Name == ".debug_frame") 3913 Debug = true; 3914 3915 if (getLexer().is(AsmToken::Comma)) { 3916 Lex(); 3917 3918 if (parseIdentifier(Name)) 3919 return TokError("Expected an identifier"); 3920 3921 if (Name == ".eh_frame") 3922 EH = true; 3923 else if (Name == ".debug_frame") 3924 Debug = true; 3925 } 3926 3927 getStreamer().EmitCFISections(EH, Debug); 3928 return false; 3929 } 3930 3931 /// parseDirectiveCFIStartProc 3932 /// ::= .cfi_startproc [simple] 3933 bool AsmParser::parseDirectiveCFIStartProc() { 3934 StringRef Simple; 3935 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3936 if (check(parseIdentifier(Simple) || Simple != "simple", 3937 "unexpected token") || 3938 parseToken(AsmToken::EndOfStatement)) 3939 return addErrorSuffix(" in '.cfi_startproc' directive"); 3940 } 3941 3942 // TODO(kristina): Deal with a corner case of incorrect diagnostic context 3943 // being produced if this directive is emitted as part of preprocessor macro 3944 // expansion which can *ONLY* happen if Clang's cc1as is the API consumer. 3945 // Tools like llvm-mc on the other hand are not affected by it, and report 3946 // correct context information. 3947 getStreamer().EmitCFIStartProc(!Simple.empty(), Lexer.getLoc()); 3948 return false; 3949 } 3950 3951 /// parseDirectiveCFIEndProc 3952 /// ::= .cfi_endproc 3953 bool AsmParser::parseDirectiveCFIEndProc() { 3954 getStreamer().EmitCFIEndProc(); 3955 return false; 3956 } 3957 3958 /// parse register name or number. 3959 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 3960 SMLoc DirectiveLoc) { 3961 unsigned RegNo; 3962 3963 if (getLexer().isNot(AsmToken::Integer)) { 3964 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 3965 return true; 3966 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 3967 } else 3968 return parseAbsoluteExpression(Register); 3969 3970 return false; 3971 } 3972 3973 /// parseDirectiveCFIDefCfa 3974 /// ::= .cfi_def_cfa register, offset 3975 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 3976 int64_t Register = 0, Offset = 0; 3977 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3978 parseToken(AsmToken::Comma, "unexpected token in directive") || 3979 parseAbsoluteExpression(Offset)) 3980 return true; 3981 3982 getStreamer().EmitCFIDefCfa(Register, Offset); 3983 return false; 3984 } 3985 3986 /// parseDirectiveCFIDefCfaOffset 3987 /// ::= .cfi_def_cfa_offset offset 3988 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 3989 int64_t Offset = 0; 3990 if (parseAbsoluteExpression(Offset)) 3991 return true; 3992 3993 getStreamer().EmitCFIDefCfaOffset(Offset); 3994 return false; 3995 } 3996 3997 /// parseDirectiveCFIRegister 3998 /// ::= .cfi_register register, register 3999 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 4000 int64_t Register1 = 0, Register2 = 0; 4001 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || 4002 parseToken(AsmToken::Comma, "unexpected token in directive") || 4003 parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 4004 return true; 4005 4006 getStreamer().EmitCFIRegister(Register1, Register2); 4007 return false; 4008 } 4009 4010 /// parseDirectiveCFIWindowSave 4011 /// ::= .cfi_window_save 4012 bool AsmParser::parseDirectiveCFIWindowSave() { 4013 getStreamer().EmitCFIWindowSave(); 4014 return false; 4015 } 4016 4017 /// parseDirectiveCFIAdjustCfaOffset 4018 /// ::= .cfi_adjust_cfa_offset adjustment 4019 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 4020 int64_t Adjustment = 0; 4021 if (parseAbsoluteExpression(Adjustment)) 4022 return true; 4023 4024 getStreamer().EmitCFIAdjustCfaOffset(Adjustment); 4025 return false; 4026 } 4027 4028 /// parseDirectiveCFIDefCfaRegister 4029 /// ::= .cfi_def_cfa_register register 4030 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 4031 int64_t Register = 0; 4032 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4033 return true; 4034 4035 getStreamer().EmitCFIDefCfaRegister(Register); 4036 return false; 4037 } 4038 4039 /// parseDirectiveCFIOffset 4040 /// ::= .cfi_offset register, offset 4041 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 4042 int64_t Register = 0; 4043 int64_t Offset = 0; 4044 4045 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4046 parseToken(AsmToken::Comma, "unexpected token in directive") || 4047 parseAbsoluteExpression(Offset)) 4048 return true; 4049 4050 getStreamer().EmitCFIOffset(Register, Offset); 4051 return false; 4052 } 4053 4054 /// parseDirectiveCFIRelOffset 4055 /// ::= .cfi_rel_offset register, offset 4056 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 4057 int64_t Register = 0, Offset = 0; 4058 4059 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4060 parseToken(AsmToken::Comma, "unexpected token in directive") || 4061 parseAbsoluteExpression(Offset)) 4062 return true; 4063 4064 getStreamer().EmitCFIRelOffset(Register, Offset); 4065 return false; 4066 } 4067 4068 static bool isValidEncoding(int64_t Encoding) { 4069 if (Encoding & ~0xff) 4070 return false; 4071 4072 if (Encoding == dwarf::DW_EH_PE_omit) 4073 return true; 4074 4075 const unsigned Format = Encoding & 0xf; 4076 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 4077 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 4078 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 4079 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 4080 return false; 4081 4082 const unsigned Application = Encoding & 0x70; 4083 if (Application != dwarf::DW_EH_PE_absptr && 4084 Application != dwarf::DW_EH_PE_pcrel) 4085 return false; 4086 4087 return true; 4088 } 4089 4090 /// parseDirectiveCFIPersonalityOrLsda 4091 /// IsPersonality true for cfi_personality, false for cfi_lsda 4092 /// ::= .cfi_personality encoding, [symbol_name] 4093 /// ::= .cfi_lsda encoding, [symbol_name] 4094 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 4095 int64_t Encoding = 0; 4096 if (parseAbsoluteExpression(Encoding)) 4097 return true; 4098 if (Encoding == dwarf::DW_EH_PE_omit) 4099 return false; 4100 4101 StringRef Name; 4102 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 4103 parseToken(AsmToken::Comma, "unexpected token in directive") || 4104 check(parseIdentifier(Name), "expected identifier in directive")) 4105 return true; 4106 4107 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4108 4109 if (IsPersonality) 4110 getStreamer().EmitCFIPersonality(Sym, Encoding); 4111 else 4112 getStreamer().EmitCFILsda(Sym, Encoding); 4113 return false; 4114 } 4115 4116 /// parseDirectiveCFIRememberState 4117 /// ::= .cfi_remember_state 4118 bool AsmParser::parseDirectiveCFIRememberState() { 4119 getStreamer().EmitCFIRememberState(); 4120 return false; 4121 } 4122 4123 /// parseDirectiveCFIRestoreState 4124 /// ::= .cfi_remember_state 4125 bool AsmParser::parseDirectiveCFIRestoreState() { 4126 getStreamer().EmitCFIRestoreState(); 4127 return false; 4128 } 4129 4130 /// parseDirectiveCFISameValue 4131 /// ::= .cfi_same_value register 4132 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 4133 int64_t Register = 0; 4134 4135 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4136 return true; 4137 4138 getStreamer().EmitCFISameValue(Register); 4139 return false; 4140 } 4141 4142 /// parseDirectiveCFIRestore 4143 /// ::= .cfi_restore register 4144 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 4145 int64_t Register = 0; 4146 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4147 return true; 4148 4149 getStreamer().EmitCFIRestore(Register); 4150 return false; 4151 } 4152 4153 /// parseDirectiveCFIEscape 4154 /// ::= .cfi_escape expression[,...] 4155 bool AsmParser::parseDirectiveCFIEscape() { 4156 std::string Values; 4157 int64_t CurrValue; 4158 if (parseAbsoluteExpression(CurrValue)) 4159 return true; 4160 4161 Values.push_back((uint8_t)CurrValue); 4162 4163 while (getLexer().is(AsmToken::Comma)) { 4164 Lex(); 4165 4166 if (parseAbsoluteExpression(CurrValue)) 4167 return true; 4168 4169 Values.push_back((uint8_t)CurrValue); 4170 } 4171 4172 getStreamer().EmitCFIEscape(Values); 4173 return false; 4174 } 4175 4176 /// parseDirectiveCFIReturnColumn 4177 /// ::= .cfi_return_column register 4178 bool AsmParser::parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc) { 4179 int64_t Register = 0; 4180 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4181 return true; 4182 getStreamer().EmitCFIReturnColumn(Register); 4183 return false; 4184 } 4185 4186 /// parseDirectiveCFISignalFrame 4187 /// ::= .cfi_signal_frame 4188 bool AsmParser::parseDirectiveCFISignalFrame() { 4189 if (parseToken(AsmToken::EndOfStatement, 4190 "unexpected token in '.cfi_signal_frame'")) 4191 return true; 4192 4193 getStreamer().EmitCFISignalFrame(); 4194 return false; 4195 } 4196 4197 /// parseDirectiveCFIUndefined 4198 /// ::= .cfi_undefined register 4199 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 4200 int64_t Register = 0; 4201 4202 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4203 return true; 4204 4205 getStreamer().EmitCFIUndefined(Register); 4206 return false; 4207 } 4208 4209 /// parseDirectiveAltmacro 4210 /// ::= .altmacro 4211 /// ::= .noaltmacro 4212 bool AsmParser::parseDirectiveAltmacro(StringRef Directive) { 4213 if (getLexer().isNot(AsmToken::EndOfStatement)) 4214 return TokError("unexpected token in '" + Directive + "' directive"); 4215 AltMacroMode = (Directive == ".altmacro"); 4216 return false; 4217 } 4218 4219 /// parseDirectiveMacrosOnOff 4220 /// ::= .macros_on 4221 /// ::= .macros_off 4222 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 4223 if (parseToken(AsmToken::EndOfStatement, 4224 "unexpected token in '" + Directive + "' directive")) 4225 return true; 4226 4227 setMacrosEnabled(Directive == ".macros_on"); 4228 return false; 4229 } 4230 4231 /// parseDirectiveMacro 4232 /// ::= .macro name[,] [parameters] 4233 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 4234 StringRef Name; 4235 if (parseIdentifier(Name)) 4236 return TokError("expected identifier in '.macro' directive"); 4237 4238 if (getLexer().is(AsmToken::Comma)) 4239 Lex(); 4240 4241 MCAsmMacroParameters Parameters; 4242 while (getLexer().isNot(AsmToken::EndOfStatement)) { 4243 4244 if (!Parameters.empty() && Parameters.back().Vararg) 4245 return Error(Lexer.getLoc(), 4246 "Vararg parameter '" + Parameters.back().Name + 4247 "' should be last one in the list of parameters."); 4248 4249 MCAsmMacroParameter Parameter; 4250 if (parseIdentifier(Parameter.Name)) 4251 return TokError("expected identifier in '.macro' directive"); 4252 4253 // Emit an error if two (or more) named parameters share the same name 4254 for (const MCAsmMacroParameter& CurrParam : Parameters) 4255 if (CurrParam.Name.equals(Parameter.Name)) 4256 return TokError("macro '" + Name + "' has multiple parameters" 4257 " named '" + Parameter.Name + "'"); 4258 4259 if (Lexer.is(AsmToken::Colon)) { 4260 Lex(); // consume ':' 4261 4262 SMLoc QualLoc; 4263 StringRef Qualifier; 4264 4265 QualLoc = Lexer.getLoc(); 4266 if (parseIdentifier(Qualifier)) 4267 return Error(QualLoc, "missing parameter qualifier for " 4268 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4269 4270 if (Qualifier == "req") 4271 Parameter.Required = true; 4272 else if (Qualifier == "vararg") 4273 Parameter.Vararg = true; 4274 else 4275 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 4276 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 4277 } 4278 4279 if (getLexer().is(AsmToken::Equal)) { 4280 Lex(); 4281 4282 SMLoc ParamLoc; 4283 4284 ParamLoc = Lexer.getLoc(); 4285 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 4286 return true; 4287 4288 if (Parameter.Required) 4289 Warning(ParamLoc, "pointless default value for required parameter " 4290 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4291 } 4292 4293 Parameters.push_back(std::move(Parameter)); 4294 4295 if (getLexer().is(AsmToken::Comma)) 4296 Lex(); 4297 } 4298 4299 // Eat just the end of statement. 4300 Lexer.Lex(); 4301 4302 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 4303 AsmToken EndToken, StartToken = getTok(); 4304 unsigned MacroDepth = 0; 4305 // Lex the macro definition. 4306 while (true) { 4307 // Ignore Lexing errors in macros. 4308 while (Lexer.is(AsmToken::Error)) { 4309 Lexer.Lex(); 4310 } 4311 4312 // Check whether we have reached the end of the file. 4313 if (getLexer().is(AsmToken::Eof)) 4314 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 4315 4316 // Otherwise, check whether we have reach the .endmacro. 4317 if (getLexer().is(AsmToken::Identifier)) { 4318 if (getTok().getIdentifier() == ".endm" || 4319 getTok().getIdentifier() == ".endmacro") { 4320 if (MacroDepth == 0) { // Outermost macro. 4321 EndToken = getTok(); 4322 Lexer.Lex(); 4323 if (getLexer().isNot(AsmToken::EndOfStatement)) 4324 return TokError("unexpected token in '" + EndToken.getIdentifier() + 4325 "' directive"); 4326 break; 4327 } else { 4328 // Otherwise we just found the end of an inner macro. 4329 --MacroDepth; 4330 } 4331 } else if (getTok().getIdentifier() == ".macro") { 4332 // We allow nested macros. Those aren't instantiated until the outermost 4333 // macro is expanded so just ignore them for now. 4334 ++MacroDepth; 4335 } 4336 } 4337 4338 // Otherwise, scan til the end of the statement. 4339 eatToEndOfStatement(); 4340 } 4341 4342 if (getContext().lookupMacro(Name)) { 4343 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 4344 } 4345 4346 const char *BodyStart = StartToken.getLoc().getPointer(); 4347 const char *BodyEnd = EndToken.getLoc().getPointer(); 4348 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4349 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 4350 MCAsmMacro Macro(Name, Body, std::move(Parameters)); 4351 DEBUG_WITH_TYPE("asm-macros", dbgs() << "Defining new macro:\n"; 4352 Macro.dump()); 4353 getContext().defineMacro(Name, std::move(Macro)); 4354 return false; 4355 } 4356 4357 /// checkForBadMacro 4358 /// 4359 /// With the support added for named parameters there may be code out there that 4360 /// is transitioning from positional parameters. In versions of gas that did 4361 /// not support named parameters they would be ignored on the macro definition. 4362 /// But to support both styles of parameters this is not possible so if a macro 4363 /// definition has named parameters but does not use them and has what appears 4364 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 4365 /// warning that the positional parameter found in body which have no effect. 4366 /// Hoping the developer will either remove the named parameters from the macro 4367 /// definition so the positional parameters get used if that was what was 4368 /// intended or change the macro to use the named parameters. It is possible 4369 /// this warning will trigger when the none of the named parameters are used 4370 /// and the strings like $1 are infact to simply to be passed trough unchanged. 4371 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 4372 StringRef Body, 4373 ArrayRef<MCAsmMacroParameter> Parameters) { 4374 // If this macro is not defined with named parameters the warning we are 4375 // checking for here doesn't apply. 4376 unsigned NParameters = Parameters.size(); 4377 if (NParameters == 0) 4378 return; 4379 4380 bool NamedParametersFound = false; 4381 bool PositionalParametersFound = false; 4382 4383 // Look at the body of the macro for use of both the named parameters and what 4384 // are likely to be positional parameters. This is what expandMacro() is 4385 // doing when it finds the parameters in the body. 4386 while (!Body.empty()) { 4387 // Scan for the next possible parameter. 4388 std::size_t End = Body.size(), Pos = 0; 4389 for (; Pos != End; ++Pos) { 4390 // Check for a substitution or escape. 4391 // This macro is defined with parameters, look for \foo, \bar, etc. 4392 if (Body[Pos] == '\\' && Pos + 1 != End) 4393 break; 4394 4395 // This macro should have parameters, but look for $0, $1, ..., $n too. 4396 if (Body[Pos] != '$' || Pos + 1 == End) 4397 continue; 4398 char Next = Body[Pos + 1]; 4399 if (Next == '$' || Next == 'n' || 4400 isdigit(static_cast<unsigned char>(Next))) 4401 break; 4402 } 4403 4404 // Check if we reached the end. 4405 if (Pos == End) 4406 break; 4407 4408 if (Body[Pos] == '$') { 4409 switch (Body[Pos + 1]) { 4410 // $$ => $ 4411 case '$': 4412 break; 4413 4414 // $n => number of arguments 4415 case 'n': 4416 PositionalParametersFound = true; 4417 break; 4418 4419 // $[0-9] => argument 4420 default: { 4421 PositionalParametersFound = true; 4422 break; 4423 } 4424 } 4425 Pos += 2; 4426 } else { 4427 unsigned I = Pos + 1; 4428 while (isIdentifierChar(Body[I]) && I + 1 != End) 4429 ++I; 4430 4431 const char *Begin = Body.data() + Pos + 1; 4432 StringRef Argument(Begin, I - (Pos + 1)); 4433 unsigned Index = 0; 4434 for (; Index < NParameters; ++Index) 4435 if (Parameters[Index].Name == Argument) 4436 break; 4437 4438 if (Index == NParameters) { 4439 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 4440 Pos += 3; 4441 else { 4442 Pos = I; 4443 } 4444 } else { 4445 NamedParametersFound = true; 4446 Pos += 1 + Argument.size(); 4447 } 4448 } 4449 // Update the scan point. 4450 Body = Body.substr(Pos); 4451 } 4452 4453 if (!NamedParametersFound && PositionalParametersFound) 4454 Warning(DirectiveLoc, "macro defined with named parameters which are not " 4455 "used in macro body, possible positional parameter " 4456 "found in body which will have no effect"); 4457 } 4458 4459 /// parseDirectiveExitMacro 4460 /// ::= .exitm 4461 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 4462 if (parseToken(AsmToken::EndOfStatement, 4463 "unexpected token in '" + Directive + "' directive")) 4464 return true; 4465 4466 if (!isInsideMacroInstantiation()) 4467 return TokError("unexpected '" + Directive + "' in file, " 4468 "no current macro definition"); 4469 4470 // Exit all conditionals that are active in the current macro. 4471 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 4472 TheCondState = TheCondStack.back(); 4473 TheCondStack.pop_back(); 4474 } 4475 4476 handleMacroExit(); 4477 return false; 4478 } 4479 4480 /// parseDirectiveEndMacro 4481 /// ::= .endm 4482 /// ::= .endmacro 4483 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 4484 if (getLexer().isNot(AsmToken::EndOfStatement)) 4485 return TokError("unexpected token in '" + Directive + "' directive"); 4486 4487 // If we are inside a macro instantiation, terminate the current 4488 // instantiation. 4489 if (isInsideMacroInstantiation()) { 4490 handleMacroExit(); 4491 return false; 4492 } 4493 4494 // Otherwise, this .endmacro is a stray entry in the file; well formed 4495 // .endmacro directives are handled during the macro definition parsing. 4496 return TokError("unexpected '" + Directive + "' in file, " 4497 "no current macro definition"); 4498 } 4499 4500 /// parseDirectivePurgeMacro 4501 /// ::= .purgem 4502 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 4503 StringRef Name; 4504 SMLoc Loc; 4505 if (parseTokenLoc(Loc) || 4506 check(parseIdentifier(Name), Loc, 4507 "expected identifier in '.purgem' directive") || 4508 parseToken(AsmToken::EndOfStatement, 4509 "unexpected token in '.purgem' directive")) 4510 return true; 4511 4512 if (!getContext().lookupMacro(Name)) 4513 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 4514 4515 getContext().undefineMacro(Name); 4516 DEBUG_WITH_TYPE("asm-macros", dbgs() 4517 << "Un-defining macro: " << Name << "\n"); 4518 return false; 4519 } 4520 4521 /// parseDirectiveBundleAlignMode 4522 /// ::= {.bundle_align_mode} expression 4523 bool AsmParser::parseDirectiveBundleAlignMode() { 4524 // Expect a single argument: an expression that evaluates to a constant 4525 // in the inclusive range 0-30. 4526 SMLoc ExprLoc = getLexer().getLoc(); 4527 int64_t AlignSizePow2; 4528 if (checkForValidSection() || parseAbsoluteExpression(AlignSizePow2) || 4529 parseToken(AsmToken::EndOfStatement, "unexpected token after expression " 4530 "in '.bundle_align_mode' " 4531 "directive") || 4532 check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc, 4533 "invalid bundle alignment size (expected between 0 and 30)")) 4534 return true; 4535 4536 // Because of AlignSizePow2's verified range we can safely truncate it to 4537 // unsigned. 4538 getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 4539 return false; 4540 } 4541 4542 /// parseDirectiveBundleLock 4543 /// ::= {.bundle_lock} [align_to_end] 4544 bool AsmParser::parseDirectiveBundleLock() { 4545 if (checkForValidSection()) 4546 return true; 4547 bool AlignToEnd = false; 4548 4549 StringRef Option; 4550 SMLoc Loc = getTok().getLoc(); 4551 const char *kInvalidOptionError = 4552 "invalid option for '.bundle_lock' directive"; 4553 4554 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4555 if (check(parseIdentifier(Option), Loc, kInvalidOptionError) || 4556 check(Option != "align_to_end", Loc, kInvalidOptionError) || 4557 parseToken(AsmToken::EndOfStatement, 4558 "unexpected token after '.bundle_lock' directive option")) 4559 return true; 4560 AlignToEnd = true; 4561 } 4562 4563 getStreamer().EmitBundleLock(AlignToEnd); 4564 return false; 4565 } 4566 4567 /// parseDirectiveBundleLock 4568 /// ::= {.bundle_lock} 4569 bool AsmParser::parseDirectiveBundleUnlock() { 4570 if (checkForValidSection() || 4571 parseToken(AsmToken::EndOfStatement, 4572 "unexpected token in '.bundle_unlock' directive")) 4573 return true; 4574 4575 getStreamer().EmitBundleUnlock(); 4576 return false; 4577 } 4578 4579 /// parseDirectiveSpace 4580 /// ::= (.skip | .space) expression [ , expression ] 4581 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4582 SMLoc NumBytesLoc = Lexer.getLoc(); 4583 const MCExpr *NumBytes; 4584 if (checkForValidSection() || parseExpression(NumBytes)) 4585 return true; 4586 4587 int64_t FillExpr = 0; 4588 if (parseOptionalToken(AsmToken::Comma)) 4589 if (parseAbsoluteExpression(FillExpr)) 4590 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4591 if (parseToken(AsmToken::EndOfStatement)) 4592 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4593 4594 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4595 getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc); 4596 4597 return false; 4598 } 4599 4600 /// parseDirectiveDCB 4601 /// ::= .dcb.{b, l, w} expression, expression 4602 bool AsmParser::parseDirectiveDCB(StringRef IDVal, unsigned Size) { 4603 SMLoc NumValuesLoc = Lexer.getLoc(); 4604 int64_t NumValues; 4605 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4606 return true; 4607 4608 if (NumValues < 0) { 4609 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4610 return false; 4611 } 4612 4613 if (parseToken(AsmToken::Comma, 4614 "unexpected token in '" + Twine(IDVal) + "' directive")) 4615 return true; 4616 4617 const MCExpr *Value; 4618 SMLoc ExprLoc = getLexer().getLoc(); 4619 if (parseExpression(Value)) 4620 return true; 4621 4622 // Special case constant expressions to match code generator. 4623 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 4624 assert(Size <= 8 && "Invalid size"); 4625 uint64_t IntValue = MCE->getValue(); 4626 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 4627 return Error(ExprLoc, "literal value out of range for directive"); 4628 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4629 getStreamer().EmitIntValue(IntValue, Size); 4630 } else { 4631 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4632 getStreamer().EmitValue(Value, Size, ExprLoc); 4633 } 4634 4635 if (parseToken(AsmToken::EndOfStatement, 4636 "unexpected token in '" + Twine(IDVal) + "' directive")) 4637 return true; 4638 4639 return false; 4640 } 4641 4642 /// parseDirectiveRealDCB 4643 /// ::= .dcb.{d, s} expression, expression 4644 bool AsmParser::parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &Semantics) { 4645 SMLoc NumValuesLoc = Lexer.getLoc(); 4646 int64_t NumValues; 4647 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4648 return true; 4649 4650 if (NumValues < 0) { 4651 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4652 return false; 4653 } 4654 4655 if (parseToken(AsmToken::Comma, 4656 "unexpected token in '" + Twine(IDVal) + "' directive")) 4657 return true; 4658 4659 APInt AsInt; 4660 if (parseRealValue(Semantics, AsInt)) 4661 return true; 4662 4663 if (parseToken(AsmToken::EndOfStatement, 4664 "unexpected token in '" + Twine(IDVal) + "' directive")) 4665 return true; 4666 4667 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4668 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 4669 AsInt.getBitWidth() / 8); 4670 4671 return false; 4672 } 4673 4674 /// parseDirectiveDS 4675 /// ::= .ds.{b, d, l, p, s, w, x} expression 4676 bool AsmParser::parseDirectiveDS(StringRef IDVal, unsigned Size) { 4677 SMLoc NumValuesLoc = Lexer.getLoc(); 4678 int64_t NumValues; 4679 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4680 return true; 4681 4682 if (NumValues < 0) { 4683 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4684 return false; 4685 } 4686 4687 if (parseToken(AsmToken::EndOfStatement, 4688 "unexpected token in '" + Twine(IDVal) + "' directive")) 4689 return true; 4690 4691 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4692 getStreamer().emitFill(Size, 0); 4693 4694 return false; 4695 } 4696 4697 /// parseDirectiveLEB128 4698 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4699 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4700 if (checkForValidSection()) 4701 return true; 4702 4703 auto parseOp = [&]() -> bool { 4704 const MCExpr *Value; 4705 if (parseExpression(Value)) 4706 return true; 4707 if (Signed) 4708 getStreamer().EmitSLEB128Value(Value); 4709 else 4710 getStreamer().EmitULEB128Value(Value); 4711 return false; 4712 }; 4713 4714 if (parseMany(parseOp)) 4715 return addErrorSuffix(" in directive"); 4716 4717 return false; 4718 } 4719 4720 /// parseDirectiveSymbolAttribute 4721 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4722 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4723 auto parseOp = [&]() -> bool { 4724 StringRef Name; 4725 SMLoc Loc = getTok().getLoc(); 4726 if (parseIdentifier(Name)) 4727 return Error(Loc, "expected identifier"); 4728 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4729 4730 // Assembler local symbols don't make any sense here. Complain loudly. 4731 if (Sym->isTemporary()) 4732 return Error(Loc, "non-local symbol required"); 4733 4734 if (!getStreamer().EmitSymbolAttribute(Sym, Attr)) 4735 return Error(Loc, "unable to emit symbol attribute"); 4736 return false; 4737 }; 4738 4739 if (parseMany(parseOp)) 4740 return addErrorSuffix(" in directive"); 4741 return false; 4742 } 4743 4744 /// parseDirectiveComm 4745 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4746 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4747 if (checkForValidSection()) 4748 return true; 4749 4750 SMLoc IDLoc = getLexer().getLoc(); 4751 StringRef Name; 4752 if (parseIdentifier(Name)) 4753 return TokError("expected identifier in directive"); 4754 4755 // Handle the identifier as the key symbol. 4756 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4757 4758 if (getLexer().isNot(AsmToken::Comma)) 4759 return TokError("unexpected token in directive"); 4760 Lex(); 4761 4762 int64_t Size; 4763 SMLoc SizeLoc = getLexer().getLoc(); 4764 if (parseAbsoluteExpression(Size)) 4765 return true; 4766 4767 int64_t Pow2Alignment = 0; 4768 SMLoc Pow2AlignmentLoc; 4769 if (getLexer().is(AsmToken::Comma)) { 4770 Lex(); 4771 Pow2AlignmentLoc = getLexer().getLoc(); 4772 if (parseAbsoluteExpression(Pow2Alignment)) 4773 return true; 4774 4775 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4776 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4777 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4778 4779 // If this target takes alignments in bytes (not log) validate and convert. 4780 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4781 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4782 if (!isPowerOf2_64(Pow2Alignment)) 4783 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4784 Pow2Alignment = Log2_64(Pow2Alignment); 4785 } 4786 } 4787 4788 if (parseToken(AsmToken::EndOfStatement, 4789 "unexpected token in '.comm' or '.lcomm' directive")) 4790 return true; 4791 4792 // NOTE: a size of zero for a .comm should create a undefined symbol 4793 // but a size of .lcomm creates a bss symbol of size zero. 4794 if (Size < 0) 4795 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4796 "be less than zero"); 4797 4798 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4799 // may internally end up wanting an alignment in bytes. 4800 // FIXME: Diagnose overflow. 4801 if (Pow2Alignment < 0) 4802 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4803 "alignment, can't be less than zero"); 4804 4805 Sym->redefineIfPossible(); 4806 if (!Sym->isUndefined()) 4807 return Error(IDLoc, "invalid symbol redefinition"); 4808 4809 // Create the Symbol as a common or local common with Size and Pow2Alignment 4810 if (IsLocal) { 4811 getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4812 return false; 4813 } 4814 4815 getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4816 return false; 4817 } 4818 4819 /// parseDirectiveAbort 4820 /// ::= .abort [... message ...] 4821 bool AsmParser::parseDirectiveAbort() { 4822 // FIXME: Use loc from directive. 4823 SMLoc Loc = getLexer().getLoc(); 4824 4825 StringRef Str = parseStringToEndOfStatement(); 4826 if (parseToken(AsmToken::EndOfStatement, 4827 "unexpected token in '.abort' directive")) 4828 return true; 4829 4830 if (Str.empty()) 4831 return Error(Loc, ".abort detected. Assembly stopping."); 4832 else 4833 return Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 4834 // FIXME: Actually abort assembly here. 4835 4836 return false; 4837 } 4838 4839 /// parseDirectiveInclude 4840 /// ::= .include "filename" 4841 bool AsmParser::parseDirectiveInclude() { 4842 // Allow the strings to have escaped octal character sequence. 4843 std::string Filename; 4844 SMLoc IncludeLoc = getTok().getLoc(); 4845 4846 if (check(getTok().isNot(AsmToken::String), 4847 "expected string in '.include' directive") || 4848 parseEscapedString(Filename) || 4849 check(getTok().isNot(AsmToken::EndOfStatement), 4850 "unexpected token in '.include' directive") || 4851 // Attempt to switch the lexer to the included file before consuming the 4852 // end of statement to avoid losing it when we switch. 4853 check(enterIncludeFile(Filename), IncludeLoc, 4854 "Could not find include file '" + Filename + "'")) 4855 return true; 4856 4857 return false; 4858 } 4859 4860 /// parseDirectiveIncbin 4861 /// ::= .incbin "filename" [ , skip [ , count ] ] 4862 bool AsmParser::parseDirectiveIncbin() { 4863 // Allow the strings to have escaped octal character sequence. 4864 std::string Filename; 4865 SMLoc IncbinLoc = getTok().getLoc(); 4866 if (check(getTok().isNot(AsmToken::String), 4867 "expected string in '.incbin' directive") || 4868 parseEscapedString(Filename)) 4869 return true; 4870 4871 int64_t Skip = 0; 4872 const MCExpr *Count = nullptr; 4873 SMLoc SkipLoc, CountLoc; 4874 if (parseOptionalToken(AsmToken::Comma)) { 4875 // The skip expression can be omitted while specifying the count, e.g: 4876 // .incbin "filename",,4 4877 if (getTok().isNot(AsmToken::Comma)) { 4878 if (parseTokenLoc(SkipLoc) || parseAbsoluteExpression(Skip)) 4879 return true; 4880 } 4881 if (parseOptionalToken(AsmToken::Comma)) { 4882 CountLoc = getTok().getLoc(); 4883 if (parseExpression(Count)) 4884 return true; 4885 } 4886 } 4887 4888 if (parseToken(AsmToken::EndOfStatement, 4889 "unexpected token in '.incbin' directive")) 4890 return true; 4891 4892 if (check(Skip < 0, SkipLoc, "skip is negative")) 4893 return true; 4894 4895 // Attempt to process the included file. 4896 if (processIncbinFile(Filename, Skip, Count, CountLoc)) 4897 return Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 4898 return false; 4899 } 4900 4901 /// parseDirectiveIf 4902 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 4903 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 4904 TheCondStack.push_back(TheCondState); 4905 TheCondState.TheCond = AsmCond::IfCond; 4906 if (TheCondState.Ignore) { 4907 eatToEndOfStatement(); 4908 } else { 4909 int64_t ExprValue; 4910 if (parseAbsoluteExpression(ExprValue) || 4911 parseToken(AsmToken::EndOfStatement, 4912 "unexpected token in '.if' directive")) 4913 return true; 4914 4915 switch (DirKind) { 4916 default: 4917 llvm_unreachable("unsupported directive"); 4918 case DK_IF: 4919 case DK_IFNE: 4920 break; 4921 case DK_IFEQ: 4922 ExprValue = ExprValue == 0; 4923 break; 4924 case DK_IFGE: 4925 ExprValue = ExprValue >= 0; 4926 break; 4927 case DK_IFGT: 4928 ExprValue = ExprValue > 0; 4929 break; 4930 case DK_IFLE: 4931 ExprValue = ExprValue <= 0; 4932 break; 4933 case DK_IFLT: 4934 ExprValue = ExprValue < 0; 4935 break; 4936 } 4937 4938 TheCondState.CondMet = ExprValue; 4939 TheCondState.Ignore = !TheCondState.CondMet; 4940 } 4941 4942 return false; 4943 } 4944 4945 /// parseDirectiveIfb 4946 /// ::= .ifb string 4947 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 4948 TheCondStack.push_back(TheCondState); 4949 TheCondState.TheCond = AsmCond::IfCond; 4950 4951 if (TheCondState.Ignore) { 4952 eatToEndOfStatement(); 4953 } else { 4954 StringRef Str = parseStringToEndOfStatement(); 4955 4956 if (parseToken(AsmToken::EndOfStatement, 4957 "unexpected token in '.ifb' directive")) 4958 return true; 4959 4960 TheCondState.CondMet = ExpectBlank == Str.empty(); 4961 TheCondState.Ignore = !TheCondState.CondMet; 4962 } 4963 4964 return false; 4965 } 4966 4967 /// parseDirectiveIfc 4968 /// ::= .ifc string1, string2 4969 /// ::= .ifnc string1, string2 4970 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 4971 TheCondStack.push_back(TheCondState); 4972 TheCondState.TheCond = AsmCond::IfCond; 4973 4974 if (TheCondState.Ignore) { 4975 eatToEndOfStatement(); 4976 } else { 4977 StringRef Str1 = parseStringToComma(); 4978 4979 if (parseToken(AsmToken::Comma, "unexpected token in '.ifc' directive")) 4980 return true; 4981 4982 StringRef Str2 = parseStringToEndOfStatement(); 4983 4984 if (parseToken(AsmToken::EndOfStatement, 4985 "unexpected token in '.ifc' directive")) 4986 return true; 4987 4988 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 4989 TheCondState.Ignore = !TheCondState.CondMet; 4990 } 4991 4992 return false; 4993 } 4994 4995 /// parseDirectiveIfeqs 4996 /// ::= .ifeqs string1, string2 4997 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 4998 if (Lexer.isNot(AsmToken::String)) { 4999 if (ExpectEqual) 5000 return TokError("expected string parameter for '.ifeqs' directive"); 5001 return TokError("expected string parameter for '.ifnes' directive"); 5002 } 5003 5004 StringRef String1 = getTok().getStringContents(); 5005 Lex(); 5006 5007 if (Lexer.isNot(AsmToken::Comma)) { 5008 if (ExpectEqual) 5009 return TokError( 5010 "expected comma after first string for '.ifeqs' directive"); 5011 return TokError("expected comma after first string for '.ifnes' directive"); 5012 } 5013 5014 Lex(); 5015 5016 if (Lexer.isNot(AsmToken::String)) { 5017 if (ExpectEqual) 5018 return TokError("expected string parameter for '.ifeqs' directive"); 5019 return TokError("expected string parameter for '.ifnes' directive"); 5020 } 5021 5022 StringRef String2 = getTok().getStringContents(); 5023 Lex(); 5024 5025 TheCondStack.push_back(TheCondState); 5026 TheCondState.TheCond = AsmCond::IfCond; 5027 TheCondState.CondMet = ExpectEqual == (String1 == String2); 5028 TheCondState.Ignore = !TheCondState.CondMet; 5029 5030 return false; 5031 } 5032 5033 /// parseDirectiveIfdef 5034 /// ::= .ifdef symbol 5035 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 5036 StringRef Name; 5037 TheCondStack.push_back(TheCondState); 5038 TheCondState.TheCond = AsmCond::IfCond; 5039 5040 if (TheCondState.Ignore) { 5041 eatToEndOfStatement(); 5042 } else { 5043 if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") || 5044 parseToken(AsmToken::EndOfStatement, "unexpected token in '.ifdef'")) 5045 return true; 5046 5047 MCSymbol *Sym = getContext().lookupSymbol(Name); 5048 5049 if (expect_defined) 5050 TheCondState.CondMet = (Sym && !Sym->isUndefined(false)); 5051 else 5052 TheCondState.CondMet = (!Sym || Sym->isUndefined(false)); 5053 TheCondState.Ignore = !TheCondState.CondMet; 5054 } 5055 5056 return false; 5057 } 5058 5059 /// parseDirectiveElseIf 5060 /// ::= .elseif expression 5061 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 5062 if (TheCondState.TheCond != AsmCond::IfCond && 5063 TheCondState.TheCond != AsmCond::ElseIfCond) 5064 return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an" 5065 " .if or an .elseif"); 5066 TheCondState.TheCond = AsmCond::ElseIfCond; 5067 5068 bool LastIgnoreState = false; 5069 if (!TheCondStack.empty()) 5070 LastIgnoreState = TheCondStack.back().Ignore; 5071 if (LastIgnoreState || TheCondState.CondMet) { 5072 TheCondState.Ignore = true; 5073 eatToEndOfStatement(); 5074 } else { 5075 int64_t ExprValue; 5076 if (parseAbsoluteExpression(ExprValue)) 5077 return true; 5078 5079 if (parseToken(AsmToken::EndOfStatement, 5080 "unexpected token in '.elseif' directive")) 5081 return true; 5082 5083 TheCondState.CondMet = ExprValue; 5084 TheCondState.Ignore = !TheCondState.CondMet; 5085 } 5086 5087 return false; 5088 } 5089 5090 /// parseDirectiveElse 5091 /// ::= .else 5092 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 5093 if (parseToken(AsmToken::EndOfStatement, 5094 "unexpected token in '.else' directive")) 5095 return true; 5096 5097 if (TheCondState.TheCond != AsmCond::IfCond && 5098 TheCondState.TheCond != AsmCond::ElseIfCond) 5099 return Error(DirectiveLoc, "Encountered a .else that doesn't follow " 5100 " an .if or an .elseif"); 5101 TheCondState.TheCond = AsmCond::ElseCond; 5102 bool LastIgnoreState = false; 5103 if (!TheCondStack.empty()) 5104 LastIgnoreState = TheCondStack.back().Ignore; 5105 if (LastIgnoreState || TheCondState.CondMet) 5106 TheCondState.Ignore = true; 5107 else 5108 TheCondState.Ignore = false; 5109 5110 return false; 5111 } 5112 5113 /// parseDirectiveEnd 5114 /// ::= .end 5115 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 5116 if (parseToken(AsmToken::EndOfStatement, 5117 "unexpected token in '.end' directive")) 5118 return true; 5119 5120 while (Lexer.isNot(AsmToken::Eof)) 5121 Lexer.Lex(); 5122 5123 return false; 5124 } 5125 5126 /// parseDirectiveError 5127 /// ::= .err 5128 /// ::= .error [string] 5129 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 5130 if (!TheCondStack.empty()) { 5131 if (TheCondStack.back().Ignore) { 5132 eatToEndOfStatement(); 5133 return false; 5134 } 5135 } 5136 5137 if (!WithMessage) 5138 return Error(L, ".err encountered"); 5139 5140 StringRef Message = ".error directive invoked in source file"; 5141 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5142 if (Lexer.isNot(AsmToken::String)) 5143 return TokError(".error argument must be a string"); 5144 5145 Message = getTok().getStringContents(); 5146 Lex(); 5147 } 5148 5149 return Error(L, Message); 5150 } 5151 5152 /// parseDirectiveWarning 5153 /// ::= .warning [string] 5154 bool AsmParser::parseDirectiveWarning(SMLoc L) { 5155 if (!TheCondStack.empty()) { 5156 if (TheCondStack.back().Ignore) { 5157 eatToEndOfStatement(); 5158 return false; 5159 } 5160 } 5161 5162 StringRef Message = ".warning directive invoked in source file"; 5163 5164 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 5165 if (Lexer.isNot(AsmToken::String)) 5166 return TokError(".warning argument must be a string"); 5167 5168 Message = getTok().getStringContents(); 5169 Lex(); 5170 if (parseToken(AsmToken::EndOfStatement, 5171 "expected end of statement in '.warning' directive")) 5172 return true; 5173 } 5174 5175 return Warning(L, Message); 5176 } 5177 5178 /// parseDirectiveEndIf 5179 /// ::= .endif 5180 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 5181 if (parseToken(AsmToken::EndOfStatement, 5182 "unexpected token in '.endif' directive")) 5183 return true; 5184 5185 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 5186 return Error(DirectiveLoc, "Encountered a .endif that doesn't follow " 5187 "an .if or .else"); 5188 if (!TheCondStack.empty()) { 5189 TheCondState = TheCondStack.back(); 5190 TheCondStack.pop_back(); 5191 } 5192 5193 return false; 5194 } 5195 5196 void AsmParser::initializeDirectiveKindMap() { 5197 DirectiveKindMap[".set"] = DK_SET; 5198 DirectiveKindMap[".equ"] = DK_EQU; 5199 DirectiveKindMap[".equiv"] = DK_EQUIV; 5200 DirectiveKindMap[".ascii"] = DK_ASCII; 5201 DirectiveKindMap[".asciz"] = DK_ASCIZ; 5202 DirectiveKindMap[".string"] = DK_STRING; 5203 DirectiveKindMap[".byte"] = DK_BYTE; 5204 DirectiveKindMap[".short"] = DK_SHORT; 5205 DirectiveKindMap[".value"] = DK_VALUE; 5206 DirectiveKindMap[".2byte"] = DK_2BYTE; 5207 DirectiveKindMap[".long"] = DK_LONG; 5208 DirectiveKindMap[".int"] = DK_INT; 5209 DirectiveKindMap[".4byte"] = DK_4BYTE; 5210 DirectiveKindMap[".quad"] = DK_QUAD; 5211 DirectiveKindMap[".8byte"] = DK_8BYTE; 5212 DirectiveKindMap[".octa"] = DK_OCTA; 5213 DirectiveKindMap[".single"] = DK_SINGLE; 5214 DirectiveKindMap[".float"] = DK_FLOAT; 5215 DirectiveKindMap[".double"] = DK_DOUBLE; 5216 DirectiveKindMap[".align"] = DK_ALIGN; 5217 DirectiveKindMap[".align32"] = DK_ALIGN32; 5218 DirectiveKindMap[".balign"] = DK_BALIGN; 5219 DirectiveKindMap[".balignw"] = DK_BALIGNW; 5220 DirectiveKindMap[".balignl"] = DK_BALIGNL; 5221 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 5222 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 5223 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 5224 DirectiveKindMap[".org"] = DK_ORG; 5225 DirectiveKindMap[".fill"] = DK_FILL; 5226 DirectiveKindMap[".zero"] = DK_ZERO; 5227 DirectiveKindMap[".extern"] = DK_EXTERN; 5228 DirectiveKindMap[".globl"] = DK_GLOBL; 5229 DirectiveKindMap[".global"] = DK_GLOBAL; 5230 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 5231 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 5232 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 5233 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 5234 DirectiveKindMap[".reference"] = DK_REFERENCE; 5235 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 5236 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 5237 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 5238 DirectiveKindMap[".cold"] = DK_COLD; 5239 DirectiveKindMap[".comm"] = DK_COMM; 5240 DirectiveKindMap[".common"] = DK_COMMON; 5241 DirectiveKindMap[".lcomm"] = DK_LCOMM; 5242 DirectiveKindMap[".abort"] = DK_ABORT; 5243 DirectiveKindMap[".include"] = DK_INCLUDE; 5244 DirectiveKindMap[".incbin"] = DK_INCBIN; 5245 DirectiveKindMap[".code16"] = DK_CODE16; 5246 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 5247 DirectiveKindMap[".rept"] = DK_REPT; 5248 DirectiveKindMap[".rep"] = DK_REPT; 5249 DirectiveKindMap[".irp"] = DK_IRP; 5250 DirectiveKindMap[".irpc"] = DK_IRPC; 5251 DirectiveKindMap[".endr"] = DK_ENDR; 5252 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 5253 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 5254 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 5255 DirectiveKindMap[".if"] = DK_IF; 5256 DirectiveKindMap[".ifeq"] = DK_IFEQ; 5257 DirectiveKindMap[".ifge"] = DK_IFGE; 5258 DirectiveKindMap[".ifgt"] = DK_IFGT; 5259 DirectiveKindMap[".ifle"] = DK_IFLE; 5260 DirectiveKindMap[".iflt"] = DK_IFLT; 5261 DirectiveKindMap[".ifne"] = DK_IFNE; 5262 DirectiveKindMap[".ifb"] = DK_IFB; 5263 DirectiveKindMap[".ifnb"] = DK_IFNB; 5264 DirectiveKindMap[".ifc"] = DK_IFC; 5265 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 5266 DirectiveKindMap[".ifnc"] = DK_IFNC; 5267 DirectiveKindMap[".ifnes"] = DK_IFNES; 5268 DirectiveKindMap[".ifdef"] = DK_IFDEF; 5269 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 5270 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 5271 DirectiveKindMap[".elseif"] = DK_ELSEIF; 5272 DirectiveKindMap[".else"] = DK_ELSE; 5273 DirectiveKindMap[".end"] = DK_END; 5274 DirectiveKindMap[".endif"] = DK_ENDIF; 5275 DirectiveKindMap[".skip"] = DK_SKIP; 5276 DirectiveKindMap[".space"] = DK_SPACE; 5277 DirectiveKindMap[".file"] = DK_FILE; 5278 DirectiveKindMap[".line"] = DK_LINE; 5279 DirectiveKindMap[".loc"] = DK_LOC; 5280 DirectiveKindMap[".stabs"] = DK_STABS; 5281 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 5282 DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID; 5283 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 5284 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 5285 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 5286 DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID; 5287 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 5288 DirectiveKindMap[".cv_string"] = DK_CV_STRING; 5289 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 5290 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 5291 DirectiveKindMap[".cv_filechecksumoffset"] = DK_CV_FILECHECKSUM_OFFSET; 5292 DirectiveKindMap[".cv_fpo_data"] = DK_CV_FPO_DATA; 5293 DirectiveKindMap[".sleb128"] = DK_SLEB128; 5294 DirectiveKindMap[".uleb128"] = DK_ULEB128; 5295 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 5296 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 5297 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 5298 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 5299 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 5300 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 5301 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 5302 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 5303 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 5304 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 5305 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 5306 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 5307 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 5308 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 5309 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 5310 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 5311 DirectiveKindMap[".cfi_return_column"] = DK_CFI_RETURN_COLUMN; 5312 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 5313 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 5314 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 5315 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 5316 DirectiveKindMap[".cfi_b_key_frame"] = DK_CFI_B_KEY_FRAME; 5317 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 5318 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 5319 DirectiveKindMap[".macro"] = DK_MACRO; 5320 DirectiveKindMap[".exitm"] = DK_EXITM; 5321 DirectiveKindMap[".endm"] = DK_ENDM; 5322 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 5323 DirectiveKindMap[".purgem"] = DK_PURGEM; 5324 DirectiveKindMap[".err"] = DK_ERR; 5325 DirectiveKindMap[".error"] = DK_ERROR; 5326 DirectiveKindMap[".warning"] = DK_WARNING; 5327 DirectiveKindMap[".altmacro"] = DK_ALTMACRO; 5328 DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO; 5329 DirectiveKindMap[".reloc"] = DK_RELOC; 5330 DirectiveKindMap[".dc"] = DK_DC; 5331 DirectiveKindMap[".dc.a"] = DK_DC_A; 5332 DirectiveKindMap[".dc.b"] = DK_DC_B; 5333 DirectiveKindMap[".dc.d"] = DK_DC_D; 5334 DirectiveKindMap[".dc.l"] = DK_DC_L; 5335 DirectiveKindMap[".dc.s"] = DK_DC_S; 5336 DirectiveKindMap[".dc.w"] = DK_DC_W; 5337 DirectiveKindMap[".dc.x"] = DK_DC_X; 5338 DirectiveKindMap[".dcb"] = DK_DCB; 5339 DirectiveKindMap[".dcb.b"] = DK_DCB_B; 5340 DirectiveKindMap[".dcb.d"] = DK_DCB_D; 5341 DirectiveKindMap[".dcb.l"] = DK_DCB_L; 5342 DirectiveKindMap[".dcb.s"] = DK_DCB_S; 5343 DirectiveKindMap[".dcb.w"] = DK_DCB_W; 5344 DirectiveKindMap[".dcb.x"] = DK_DCB_X; 5345 DirectiveKindMap[".ds"] = DK_DS; 5346 DirectiveKindMap[".ds.b"] = DK_DS_B; 5347 DirectiveKindMap[".ds.d"] = DK_DS_D; 5348 DirectiveKindMap[".ds.l"] = DK_DS_L; 5349 DirectiveKindMap[".ds.p"] = DK_DS_P; 5350 DirectiveKindMap[".ds.s"] = DK_DS_S; 5351 DirectiveKindMap[".ds.w"] = DK_DS_W; 5352 DirectiveKindMap[".ds.x"] = DK_DS_X; 5353 DirectiveKindMap[".print"] = DK_PRINT; 5354 DirectiveKindMap[".addrsig"] = DK_ADDRSIG; 5355 DirectiveKindMap[".addrsig_sym"] = DK_ADDRSIG_SYM; 5356 } 5357 5358 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 5359 AsmToken EndToken, StartToken = getTok(); 5360 5361 unsigned NestLevel = 0; 5362 while (true) { 5363 // Check whether we have reached the end of the file. 5364 if (getLexer().is(AsmToken::Eof)) { 5365 printError(DirectiveLoc, "no matching '.endr' in definition"); 5366 return nullptr; 5367 } 5368 5369 if (Lexer.is(AsmToken::Identifier) && 5370 (getTok().getIdentifier() == ".rep" || 5371 getTok().getIdentifier() == ".rept" || 5372 getTok().getIdentifier() == ".irp" || 5373 getTok().getIdentifier() == ".irpc")) { 5374 ++NestLevel; 5375 } 5376 5377 // Otherwise, check whether we have reached the .endr. 5378 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 5379 if (NestLevel == 0) { 5380 EndToken = getTok(); 5381 Lex(); 5382 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5383 printError(getTok().getLoc(), 5384 "unexpected token in '.endr' directive"); 5385 return nullptr; 5386 } 5387 break; 5388 } 5389 --NestLevel; 5390 } 5391 5392 // Otherwise, scan till the end of the statement. 5393 eatToEndOfStatement(); 5394 } 5395 5396 const char *BodyStart = StartToken.getLoc().getPointer(); 5397 const char *BodyEnd = EndToken.getLoc().getPointer(); 5398 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 5399 5400 // We Are Anonymous. 5401 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 5402 return &MacroLikeBodies.back(); 5403 } 5404 5405 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 5406 raw_svector_ostream &OS) { 5407 OS << ".endr\n"; 5408 5409 std::unique_ptr<MemoryBuffer> Instantiation = 5410 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 5411 5412 // Create the macro instantiation object and add to the current macro 5413 // instantiation stack. 5414 MacroInstantiation *MI = new MacroInstantiation( 5415 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 5416 ActiveMacros.push_back(MI); 5417 5418 // Jump to the macro instantiation and prime the lexer. 5419 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 5420 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 5421 Lex(); 5422 } 5423 5424 /// parseDirectiveRept 5425 /// ::= .rep | .rept count 5426 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 5427 const MCExpr *CountExpr; 5428 SMLoc CountLoc = getTok().getLoc(); 5429 if (parseExpression(CountExpr)) 5430 return true; 5431 5432 int64_t Count; 5433 if (!CountExpr->evaluateAsAbsolute(Count, getStreamer().getAssemblerPtr())) { 5434 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 5435 } 5436 5437 if (check(Count < 0, CountLoc, "Count is negative") || 5438 parseToken(AsmToken::EndOfStatement, 5439 "unexpected token in '" + Dir + "' directive")) 5440 return true; 5441 5442 // Lex the rept definition. 5443 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5444 if (!M) 5445 return true; 5446 5447 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5448 // to hold the macro body with substitutions. 5449 SmallString<256> Buf; 5450 raw_svector_ostream OS(Buf); 5451 while (Count--) { 5452 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 5453 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 5454 return true; 5455 } 5456 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5457 5458 return false; 5459 } 5460 5461 /// parseDirectiveIrp 5462 /// ::= .irp symbol,values 5463 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 5464 MCAsmMacroParameter Parameter; 5465 MCAsmMacroArguments A; 5466 if (check(parseIdentifier(Parameter.Name), 5467 "expected identifier in '.irp' directive") || 5468 parseToken(AsmToken::Comma, "expected comma in '.irp' directive") || 5469 parseMacroArguments(nullptr, A) || 5470 parseToken(AsmToken::EndOfStatement, "expected End of Statement")) 5471 return true; 5472 5473 // Lex the irp definition. 5474 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5475 if (!M) 5476 return true; 5477 5478 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5479 // to hold the macro body with substitutions. 5480 SmallString<256> Buf; 5481 raw_svector_ostream OS(Buf); 5482 5483 for (const MCAsmMacroArgument &Arg : A) { 5484 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 5485 // This is undocumented, but GAS seems to support it. 5486 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5487 return true; 5488 } 5489 5490 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5491 5492 return false; 5493 } 5494 5495 /// parseDirectiveIrpc 5496 /// ::= .irpc symbol,values 5497 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 5498 MCAsmMacroParameter Parameter; 5499 MCAsmMacroArguments A; 5500 5501 if (check(parseIdentifier(Parameter.Name), 5502 "expected identifier in '.irpc' directive") || 5503 parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") || 5504 parseMacroArguments(nullptr, A)) 5505 return true; 5506 5507 if (A.size() != 1 || A.front().size() != 1) 5508 return TokError("unexpected token in '.irpc' directive"); 5509 5510 // Eat the end of statement. 5511 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5512 return true; 5513 5514 // Lex the irpc definition. 5515 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5516 if (!M) 5517 return true; 5518 5519 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5520 // to hold the macro body with substitutions. 5521 SmallString<256> Buf; 5522 raw_svector_ostream OS(Buf); 5523 5524 StringRef Values = A.front().front().getString(); 5525 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 5526 MCAsmMacroArgument Arg; 5527 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 5528 5529 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 5530 // This is undocumented, but GAS seems to support it. 5531 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5532 return true; 5533 } 5534 5535 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5536 5537 return false; 5538 } 5539 5540 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 5541 if (ActiveMacros.empty()) 5542 return TokError("unmatched '.endr' directive"); 5543 5544 // The only .repl that should get here are the ones created by 5545 // instantiateMacroLikeBody. 5546 assert(getLexer().is(AsmToken::EndOfStatement)); 5547 5548 handleMacroExit(); 5549 return false; 5550 } 5551 5552 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 5553 size_t Len) { 5554 const MCExpr *Value; 5555 SMLoc ExprLoc = getLexer().getLoc(); 5556 if (parseExpression(Value)) 5557 return true; 5558 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5559 if (!MCE) 5560 return Error(ExprLoc, "unexpected expression in _emit"); 5561 uint64_t IntValue = MCE->getValue(); 5562 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 5563 return Error(ExprLoc, "literal value out of range for directive"); 5564 5565 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 5566 return false; 5567 } 5568 5569 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 5570 const MCExpr *Value; 5571 SMLoc ExprLoc = getLexer().getLoc(); 5572 if (parseExpression(Value)) 5573 return true; 5574 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5575 if (!MCE) 5576 return Error(ExprLoc, "unexpected expression in align"); 5577 uint64_t IntValue = MCE->getValue(); 5578 if (!isPowerOf2_64(IntValue)) 5579 return Error(ExprLoc, "literal value not a power of two greater then zero"); 5580 5581 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 5582 return false; 5583 } 5584 5585 bool AsmParser::parseDirectivePrint(SMLoc DirectiveLoc) { 5586 const AsmToken StrTok = getTok(); 5587 Lex(); 5588 if (StrTok.isNot(AsmToken::String) || StrTok.getString().front() != '"') 5589 return Error(DirectiveLoc, "expected double quoted string after .print"); 5590 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5591 return true; 5592 llvm::outs() << StrTok.getStringContents() << '\n'; 5593 return false; 5594 } 5595 5596 bool AsmParser::parseDirectiveAddrsig() { 5597 getStreamer().EmitAddrsig(); 5598 return false; 5599 } 5600 5601 bool AsmParser::parseDirectiveAddrsigSym() { 5602 StringRef Name; 5603 if (check(parseIdentifier(Name), 5604 "expected identifier in '.addrsig_sym' directive")) 5605 return true; 5606 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 5607 getStreamer().EmitAddrsigSym(Sym); 5608 return false; 5609 } 5610 5611 // We are comparing pointers, but the pointers are relative to a single string. 5612 // Thus, this should always be deterministic. 5613 static int rewritesSort(const AsmRewrite *AsmRewriteA, 5614 const AsmRewrite *AsmRewriteB) { 5615 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 5616 return -1; 5617 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 5618 return 1; 5619 5620 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 5621 // rewrite to the same location. Make sure the SizeDirective rewrite is 5622 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 5623 // ensures the sort algorithm is stable. 5624 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 5625 AsmRewritePrecedence[AsmRewriteB->Kind]) 5626 return -1; 5627 5628 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 5629 AsmRewritePrecedence[AsmRewriteB->Kind]) 5630 return 1; 5631 llvm_unreachable("Unstable rewrite sort."); 5632 } 5633 5634 bool AsmParser::parseMSInlineAsm( 5635 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 5636 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls, 5637 SmallVectorImpl<std::string> &Constraints, 5638 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 5639 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 5640 SmallVector<void *, 4> InputDecls; 5641 SmallVector<void *, 4> OutputDecls; 5642 SmallVector<bool, 4> InputDeclsAddressOf; 5643 SmallVector<bool, 4> OutputDeclsAddressOf; 5644 SmallVector<std::string, 4> InputConstraints; 5645 SmallVector<std::string, 4> OutputConstraints; 5646 SmallVector<unsigned, 4> ClobberRegs; 5647 5648 SmallVector<AsmRewrite, 4> AsmStrRewrites; 5649 5650 // Prime the lexer. 5651 Lex(); 5652 5653 // While we have input, parse each statement. 5654 unsigned InputIdx = 0; 5655 unsigned OutputIdx = 0; 5656 while (getLexer().isNot(AsmToken::Eof)) { 5657 // Parse curly braces marking block start/end 5658 if (parseCurlyBlockScope(AsmStrRewrites)) 5659 continue; 5660 5661 ParseStatementInfo Info(&AsmStrRewrites); 5662 bool StatementErr = parseStatement(Info, &SI); 5663 5664 if (StatementErr || Info.ParseError) { 5665 // Emit pending errors if any exist. 5666 printPendingErrors(); 5667 return true; 5668 } 5669 5670 // No pending error should exist here. 5671 assert(!hasPendingError() && "unexpected error from parseStatement"); 5672 5673 if (Info.Opcode == ~0U) 5674 continue; 5675 5676 const MCInstrDesc &Desc = MII->get(Info.Opcode); 5677 5678 // Build the list of clobbers, outputs and inputs. 5679 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 5680 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 5681 5682 // Immediate. 5683 if (Operand.isImm()) 5684 continue; 5685 5686 // Register operand. 5687 if (Operand.isReg() && !Operand.needAddressOf() && 5688 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5689 unsigned NumDefs = Desc.getNumDefs(); 5690 // Clobber. 5691 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5692 ClobberRegs.push_back(Operand.getReg()); 5693 continue; 5694 } 5695 5696 // Expr/Input or Output. 5697 StringRef SymName = Operand.getSymName(); 5698 if (SymName.empty()) 5699 continue; 5700 5701 void *OpDecl = Operand.getOpDecl(); 5702 if (!OpDecl) 5703 continue; 5704 5705 bool isOutput = (i == 1) && Desc.mayStore(); 5706 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5707 if (isOutput) { 5708 ++InputIdx; 5709 OutputDecls.push_back(OpDecl); 5710 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5711 OutputConstraints.push_back(("=" + Operand.getConstraint()).str()); 5712 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5713 } else { 5714 InputDecls.push_back(OpDecl); 5715 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5716 InputConstraints.push_back(Operand.getConstraint().str()); 5717 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5718 } 5719 } 5720 5721 // Consider implicit defs to be clobbers. Think of cpuid and push. 5722 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5723 Desc.getNumImplicitDefs()); 5724 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 5725 } 5726 5727 // Set the number of Outputs and Inputs. 5728 NumOutputs = OutputDecls.size(); 5729 NumInputs = InputDecls.size(); 5730 5731 // Set the unique clobbers. 5732 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5733 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5734 ClobberRegs.end()); 5735 Clobbers.assign(ClobberRegs.size(), std::string()); 5736 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5737 raw_string_ostream OS(Clobbers[I]); 5738 IP->printRegName(OS, ClobberRegs[I]); 5739 } 5740 5741 // Merge the various outputs and inputs. Output are expected first. 5742 if (NumOutputs || NumInputs) { 5743 unsigned NumExprs = NumOutputs + NumInputs; 5744 OpDecls.resize(NumExprs); 5745 Constraints.resize(NumExprs); 5746 for (unsigned i = 0; i < NumOutputs; ++i) { 5747 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5748 Constraints[i] = OutputConstraints[i]; 5749 } 5750 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5751 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5752 Constraints[j] = InputConstraints[i]; 5753 } 5754 } 5755 5756 // Build the IR assembly string. 5757 std::string AsmStringIR; 5758 raw_string_ostream OS(AsmStringIR); 5759 StringRef ASMString = 5760 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5761 const char *AsmStart = ASMString.begin(); 5762 const char *AsmEnd = ASMString.end(); 5763 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 5764 for (const AsmRewrite &AR : AsmStrRewrites) { 5765 AsmRewriteKind Kind = AR.Kind; 5766 5767 const char *Loc = AR.Loc.getPointer(); 5768 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 5769 5770 // Emit everything up to the immediate/expression. 5771 if (unsigned Len = Loc - AsmStart) 5772 OS << StringRef(AsmStart, Len); 5773 5774 // Skip the original expression. 5775 if (Kind == AOK_Skip) { 5776 AsmStart = Loc + AR.Len; 5777 continue; 5778 } 5779 5780 unsigned AdditionalSkip = 0; 5781 // Rewrite expressions in $N notation. 5782 switch (Kind) { 5783 default: 5784 break; 5785 case AOK_IntelExpr: 5786 assert(AR.IntelExp.isValid() && "cannot write invalid intel expression"); 5787 if (AR.IntelExp.NeedBracs) 5788 OS << "["; 5789 if (AR.IntelExp.hasBaseReg()) 5790 OS << AR.IntelExp.BaseReg; 5791 if (AR.IntelExp.hasIndexReg()) 5792 OS << (AR.IntelExp.hasBaseReg() ? " + " : "") 5793 << AR.IntelExp.IndexReg; 5794 if (AR.IntelExp.Scale > 1) 5795 OS << " * $$" << AR.IntelExp.Scale; 5796 if (AR.IntelExp.Imm || !AR.IntelExp.hasRegs()) 5797 OS << (AR.IntelExp.hasRegs() ? " + $$" : "$$") << AR.IntelExp.Imm; 5798 if (AR.IntelExp.NeedBracs) 5799 OS << "]"; 5800 break; 5801 case AOK_Label: 5802 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 5803 break; 5804 case AOK_Input: 5805 OS << '$' << InputIdx++; 5806 break; 5807 case AOK_Output: 5808 OS << '$' << OutputIdx++; 5809 break; 5810 case AOK_SizeDirective: 5811 switch (AR.Val) { 5812 default: break; 5813 case 8: OS << "byte ptr "; break; 5814 case 16: OS << "word ptr "; break; 5815 case 32: OS << "dword ptr "; break; 5816 case 64: OS << "qword ptr "; break; 5817 case 80: OS << "xword ptr "; break; 5818 case 128: OS << "xmmword ptr "; break; 5819 case 256: OS << "ymmword ptr "; break; 5820 } 5821 break; 5822 case AOK_Emit: 5823 OS << ".byte"; 5824 break; 5825 case AOK_Align: { 5826 // MS alignment directives are measured in bytes. If the native assembler 5827 // measures alignment in bytes, we can pass it straight through. 5828 OS << ".align"; 5829 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 5830 break; 5831 5832 // Alignment is in log2 form, so print that instead and skip the original 5833 // immediate. 5834 unsigned Val = AR.Val; 5835 OS << ' ' << Val; 5836 assert(Val < 10 && "Expected alignment less then 2^10."); 5837 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 5838 break; 5839 } 5840 case AOK_EVEN: 5841 OS << ".even"; 5842 break; 5843 case AOK_EndOfStatement: 5844 OS << "\n\t"; 5845 break; 5846 } 5847 5848 // Skip the original expression. 5849 AsmStart = Loc + AR.Len + AdditionalSkip; 5850 } 5851 5852 // Emit the remainder of the asm string. 5853 if (AsmStart != AsmEnd) 5854 OS << StringRef(AsmStart, AsmEnd - AsmStart); 5855 5856 AsmString = OS.str(); 5857 return false; 5858 } 5859 5860 namespace llvm { 5861 namespace MCParserUtils { 5862 5863 /// Returns whether the given symbol is used anywhere in the given expression, 5864 /// or subexpressions. 5865 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 5866 switch (Value->getKind()) { 5867 case MCExpr::Binary: { 5868 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 5869 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 5870 isSymbolUsedInExpression(Sym, BE->getRHS()); 5871 } 5872 case MCExpr::Target: 5873 case MCExpr::Constant: 5874 return false; 5875 case MCExpr::SymbolRef: { 5876 const MCSymbol &S = 5877 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 5878 if (S.isVariable()) 5879 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 5880 return &S == Sym; 5881 } 5882 case MCExpr::Unary: 5883 return isSymbolUsedInExpression( 5884 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 5885 } 5886 5887 llvm_unreachable("Unknown expr kind!"); 5888 } 5889 5890 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 5891 MCAsmParser &Parser, MCSymbol *&Sym, 5892 const MCExpr *&Value) { 5893 5894 // FIXME: Use better location, we should use proper tokens. 5895 SMLoc EqualLoc = Parser.getTok().getLoc(); 5896 if (Parser.parseExpression(Value)) 5897 return Parser.TokError("missing expression"); 5898 5899 // Note: we don't count b as used in "a = b". This is to allow 5900 // a = b 5901 // b = c 5902 5903 if (Parser.parseToken(AsmToken::EndOfStatement)) 5904 return true; 5905 5906 // Validate that the LHS is allowed to be a variable (either it has not been 5907 // used as a symbol, or it is an absolute symbol). 5908 Sym = Parser.getContext().lookupSymbol(Name); 5909 if (Sym) { 5910 // Diagnose assignment to a label. 5911 // 5912 // FIXME: Diagnostics. Note the location of the definition as a label. 5913 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 5914 if (isSymbolUsedInExpression(Sym, Value)) 5915 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 5916 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 5917 !Sym->isVariable()) 5918 ; // Allow redefinitions of undefined symbols only used in directives. 5919 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 5920 ; // Allow redefinitions of variables that haven't yet been used. 5921 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 5922 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 5923 else if (!Sym->isVariable()) 5924 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 5925 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 5926 return Parser.Error(EqualLoc, 5927 "invalid reassignment of non-absolute variable '" + 5928 Name + "'"); 5929 } else if (Name == ".") { 5930 Parser.getStreamer().emitValueToOffset(Value, 0, EqualLoc); 5931 return false; 5932 } else 5933 Sym = Parser.getContext().getOrCreateSymbol(Name); 5934 5935 Sym->setRedefinable(allow_redef); 5936 5937 return false; 5938 } 5939 5940 } // end namespace MCParserUtils 5941 } // end namespace llvm 5942 5943 /// Create an MCAsmParser instance. 5944 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 5945 MCStreamer &Out, const MCAsmInfo &MAI, 5946 unsigned CB) { 5947 return new AsmParser(SM, C, Out, MAI, CB); 5948 } 5949