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