• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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/STLExtras.h"
19 #include "llvm/ADT/SmallString.h"
20 #include "llvm/ADT/SmallVector.h"
21 #include "llvm/ADT/StringExtras.h"
22 #include "llvm/ADT/StringMap.h"
23 #include "llvm/ADT/StringRef.h"
24 #include "llvm/ADT/Twine.h"
25 #include "llvm/BinaryFormat/Dwarf.h"
26 #include "llvm/MC/MCAsmInfo.h"
27 #include "llvm/MC/MCCodeView.h"
28 #include "llvm/MC/MCContext.h"
29 #include "llvm/MC/MCDirectives.h"
30 #include "llvm/MC/MCDwarf.h"
31 #include "llvm/MC/MCExpr.h"
32 #include "llvm/MC/MCInstPrinter.h"
33 #include "llvm/MC/MCInstrDesc.h"
34 #include "llvm/MC/MCInstrInfo.h"
35 #include "llvm/MC/MCObjectFileInfo.h"
36 #include "llvm/MC/MCParser/AsmCond.h"
37 #include "llvm/MC/MCParser/AsmLexer.h"
38 #include "llvm/MC/MCParser/MCAsmLexer.h"
39 #include "llvm/MC/MCParser/MCAsmParser.h"
40 #include "llvm/MC/MCParser/MCAsmParserExtension.h"
41 #include "llvm/MC/MCParser/MCAsmParserUtils.h"
42 #include "llvm/MC/MCParser/MCParsedAsmOperand.h"
43 #include "llvm/MC/MCParser/MCTargetAsmParser.h"
44 #include "llvm/MC/MCRegisterInfo.h"
45 #include "llvm/MC/MCSection.h"
46 #include "llvm/MC/MCStreamer.h"
47 #include "llvm/MC/MCSymbol.h"
48 #include "llvm/MC/MCTargetOptions.h"
49 #include "llvm/MC/MCValue.h"
50 #include "llvm/Support/Casting.h"
51 #include "llvm/Support/CommandLine.h"
52 #include "llvm/Support/ErrorHandling.h"
53 #include "llvm/Support/MD5.h"
54 #include "llvm/Support/MathExtras.h"
55 #include "llvm/Support/MemoryBuffer.h"
56 #include "llvm/Support/SMLoc.h"
57 #include "llvm/Support/SourceMgr.h"
58 #include "llvm/Support/raw_ostream.h"
59 #include <algorithm>
60 #include <cassert>
61 #include <cctype>
62 #include <climits>
63 #include <cstddef>
64 #include <cstdint>
65 #include <deque>
66 #include <memory>
67 #include <sstream>
68 #include <string>
69 #include <tuple>
70 #include <utility>
71 #include <vector>
72 
73 using namespace llvm;
74 
75 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() = default;
76 
77 static cl::opt<unsigned> AsmMacroMaxNestingDepth(
78      "asm-macro-max-nesting-depth", cl::init(20), cl::Hidden,
79      cl::desc("The maximum nesting depth allowed for assembly macros."));
80 
81 namespace {
82 
83 /// Helper types for tracking macro definitions.
84 typedef std::vector<AsmToken> MCAsmMacroArgument;
85 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments;
86 
87 /// Helper class for storing information about an active macro
88 /// instantiation.
89 struct MacroInstantiation {
90   /// The location of the instantiation.
91   SMLoc InstantiationLoc;
92 
93   /// The buffer where parsing should resume upon instantiation completion.
94   int ExitBuffer;
95 
96   /// The location where parsing should resume upon instantiation completion.
97   SMLoc ExitLoc;
98 
99   /// The depth of TheCondStack at the start of the instantiation.
100   size_t CondStackDepth;
101 
102 public:
103   MacroInstantiation(SMLoc IL, int EB, SMLoc EL, size_t CondStackDepth);
104 };
105 
106 struct ParseStatementInfo {
107   /// The parsed operands from the last parsed statement.
108   SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands;
109 
110   /// The opcode from the last parsed instruction.
111   unsigned Opcode = ~0U;
112 
113   /// Was there an error parsing the inline assembly?
114   bool ParseError = false;
115 
116   SmallVectorImpl<AsmRewrite> *AsmRewrites = nullptr;
117 
118   ParseStatementInfo() = delete;
ParseStatementInfo__anon6f84d46e0111::ParseStatementInfo119   ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites)
120     : AsmRewrites(rewrites) {}
121 };
122 
123 /// The concrete assembly parser instance.
124 class AsmParser : public MCAsmParser {
125 private:
126   AsmLexer Lexer;
127   MCContext &Ctx;
128   MCStreamer &Out;
129   const MCAsmInfo &MAI;
130   SourceMgr &SrcMgr;
131   SourceMgr::DiagHandlerTy SavedDiagHandler;
132   void *SavedDiagContext;
133   std::unique_ptr<MCAsmParserExtension> PlatformParser;
134 
135   /// This is the current buffer index we're lexing from as managed by the
136   /// SourceMgr object.
137   unsigned CurBuffer;
138 
139   AsmCond TheCondState;
140   std::vector<AsmCond> TheCondStack;
141 
142   /// maps directive names to handler methods in parser
143   /// extensions. Extensions register themselves in this map by calling
144   /// addDirectiveHandler.
145   StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap;
146 
147   /// Stack of active macro instantiations.
148   std::vector<MacroInstantiation*> ActiveMacros;
149 
150   /// List of bodies of anonymous macros.
151   std::deque<MCAsmMacro> MacroLikeBodies;
152 
153   /// Boolean tracking whether macro substitution is enabled.
154   unsigned MacrosEnabledFlag : 1;
155 
156   /// Keeps track of how many .macro's have been instantiated.
157   unsigned NumOfMacroInstantiations;
158 
159   /// The values from the last parsed cpp hash file line comment if any.
160   struct CppHashInfoTy {
161     StringRef Filename;
162     int64_t LineNumber = 0;
163     SMLoc Loc;
164     unsigned Buf = 0;
165   };
166   CppHashInfoTy CppHashInfo;
167 
168   /// List of forward directional labels for diagnosis at the end.
169   SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels;
170 
171   /// AssemblerDialect. ~OU means unset value and use value provided by MAI.
172   unsigned AssemblerDialect = ~0U;
173 
174   /// is Darwin compatibility enabled?
175   bool IsDarwin = false;
176 
177   /// Are we parsing ms-style inline assembly?
178   bool ParsingInlineAsm = false;
179 
180   /// Did we already inform the user about inconsistent MD5 usage?
181   bool ReportedInconsistentMD5 = false;
182 
183 public:
184   AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out,
185             const MCAsmInfo &MAI, unsigned CB);
186   AsmParser(const AsmParser &) = delete;
187   AsmParser &operator=(const AsmParser &) = delete;
188   ~AsmParser() override;
189 
190   bool Run(bool NoInitialTextSection, bool NoFinalize = false) override;
191 
addDirectiveHandler(StringRef Directive,ExtensionDirectiveHandler Handler)192   void addDirectiveHandler(StringRef Directive,
193                            ExtensionDirectiveHandler Handler) override {
194     ExtensionDirectiveMap[Directive] = Handler;
195   }
196 
addAliasForDirective(StringRef Directive,StringRef Alias)197   void addAliasForDirective(StringRef Directive, StringRef Alias) override {
198     DirectiveKindMap[Directive] = DirectiveKindMap[Alias];
199   }
200 
201   /// @name MCAsmParser Interface
202   /// {
203 
getSourceManager()204   SourceMgr &getSourceManager() override { return SrcMgr; }
getLexer()205   MCAsmLexer &getLexer() override { return Lexer; }
getContext()206   MCContext &getContext() override { return Ctx; }
getStreamer()207   MCStreamer &getStreamer() override { return Out; }
208 
getCVContext()209   CodeViewContext &getCVContext() { return Ctx.getCVContext(); }
210 
getAssemblerDialect()211   unsigned getAssemblerDialect() override {
212     if (AssemblerDialect == ~0U)
213       return MAI.getAssemblerDialect();
214     else
215       return AssemblerDialect;
216   }
setAssemblerDialect(unsigned i)217   void setAssemblerDialect(unsigned i) override {
218     AssemblerDialect = i;
219   }
220 
221   void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override;
222   bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override;
223   bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override;
224 
225   const AsmToken &Lex() override;
226 
setParsingInlineAsm(bool V)227   void setParsingInlineAsm(bool V) override {
228     ParsingInlineAsm = V;
229     Lexer.setParsingMSInlineAsm(V);
230   }
isParsingInlineAsm()231   bool isParsingInlineAsm() override { return ParsingInlineAsm; }
232 
233   bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString,
234                         unsigned &NumOutputs, unsigned &NumInputs,
235                         SmallVectorImpl<std::pair<void *,bool>> &OpDecls,
236                         SmallVectorImpl<std::string> &Constraints,
237                         SmallVectorImpl<std::string> &Clobbers,
238                         const MCInstrInfo *MII, const MCInstPrinter *IP,
239                         MCAsmParserSemaCallback &SI) override;
240 
241   bool parseExpression(const MCExpr *&Res);
242   bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override;
243   bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override;
244   bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override;
245   bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res,
246                              SMLoc &EndLoc) override;
247   bool parseAbsoluteExpression(int64_t &Res) override;
248 
249   /// Parse a floating point expression using the float \p Semantics
250   /// and set \p Res to the value.
251   bool parseRealValue(const fltSemantics &Semantics, APInt &Res);
252 
253   /// Parse an identifier or string (as a quoted identifier)
254   /// and set \p Res to the identifier contents.
255   bool parseIdentifier(StringRef &Res) override;
256   void eatToEndOfStatement() override;
257 
258   bool checkForValidSection() override;
259 
260   /// }
261 
262 private:
263   bool isAltmacroString(SMLoc &StrLoc, SMLoc &EndLoc);
264   void altMacroString(StringRef AltMacroStr, std::string &Res);
265   bool parseStatement(ParseStatementInfo &Info,
266                       MCAsmParserSemaCallback *SI);
267   bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites);
268   bool parseCppHashLineFilenameComment(SMLoc L);
269 
270   void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body,
271                         ArrayRef<MCAsmMacroParameter> Parameters);
272   bool expandMacro(raw_svector_ostream &OS, StringRef Body,
273                    ArrayRef<MCAsmMacroParameter> Parameters,
274                    ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable,
275                    SMLoc L);
276 
277   /// Are macros enabled in the parser?
areMacrosEnabled()278   bool areMacrosEnabled() {return MacrosEnabledFlag;}
279 
280   /// Control a flag in the parser that enables or disables macros.
setMacrosEnabled(bool Flag)281   void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;}
282 
283   /// Are we inside a macro instantiation?
isInsideMacroInstantiation()284   bool isInsideMacroInstantiation() {return !ActiveMacros.empty();}
285 
286   /// Handle entry to macro instantiation.
287   ///
288   /// \param M The macro.
289   /// \param NameLoc Instantiation location.
290   bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc);
291 
292   /// Handle exit from macro instantiation.
293   void handleMacroExit();
294 
295   /// Extract AsmTokens for a macro argument.
296   bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg);
297 
298   /// Parse all macro arguments for a given macro.
299   bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A);
300 
301   void printMacroInstantiations();
printMessage(SMLoc Loc,SourceMgr::DiagKind Kind,const Twine & Msg,SMRange Range=None) const302   void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg,
303                     SMRange Range = None) const {
304     ArrayRef<SMRange> Ranges(Range);
305     SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges);
306   }
307   static void DiagHandler(const SMDiagnostic &Diag, void *Context);
308 
309   /// Should we emit DWARF describing this assembler source?  (Returns false if
310   /// the source has .file directives, which means we don't want to generate
311   /// info describing the assembler source itself.)
312   bool enabledGenDwarfForAssembly();
313 
314   /// Enter the specified file. This returns true on failure.
315   bool enterIncludeFile(const std::string &Filename);
316 
317   /// Process the specified file for the .incbin directive.
318   /// This returns true on failure.
319   bool processIncbinFile(const std::string &Filename, int64_t Skip = 0,
320                          const MCExpr *Count = nullptr, SMLoc Loc = SMLoc());
321 
322   /// Reset the current lexer position to that given by \p Loc. The
323   /// current token is not set; clients should ensure Lex() is called
324   /// subsequently.
325   ///
326   /// \param InBuffer If not 0, should be the known buffer id that contains the
327   /// location.
328   void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0);
329 
330   /// Parse up to the end of statement and a return the contents from the
331   /// current token until the end of the statement; the current token on exit
332   /// will be either the EndOfStatement or EOF.
333   StringRef parseStringToEndOfStatement() override;
334 
335   /// Parse until the end of a statement or a comma is encountered,
336   /// return the contents from the current token up to the end or comma.
337   StringRef parseStringToComma();
338 
339   bool parseAssignment(StringRef Name, bool allow_redef,
340                        bool NoDeadStrip = false);
341 
342   unsigned getBinOpPrecedence(AsmToken::TokenKind K,
343                               MCBinaryExpr::Opcode &Kind);
344 
345   bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc);
346   bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc);
347   bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc);
348 
349   bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc);
350 
351   bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName);
352   bool parseCVFileId(int64_t &FileId, StringRef DirectiveName);
353 
354   // Generic (target and platform independent) directive parsing.
355   enum DirectiveKind {
356     DK_NO_DIRECTIVE, // Placeholder
357     DK_SET,
358     DK_EQU,
359     DK_EQUIV,
360     DK_ASCII,
361     DK_ASCIZ,
362     DK_STRING,
363     DK_BYTE,
364     DK_SHORT,
365     DK_RELOC,
366     DK_VALUE,
367     DK_2BYTE,
368     DK_LONG,
369     DK_INT,
370     DK_4BYTE,
371     DK_QUAD,
372     DK_8BYTE,
373     DK_OCTA,
374     DK_DC,
375     DK_DC_A,
376     DK_DC_B,
377     DK_DC_D,
378     DK_DC_L,
379     DK_DC_S,
380     DK_DC_W,
381     DK_DC_X,
382     DK_DCB,
383     DK_DCB_B,
384     DK_DCB_D,
385     DK_DCB_L,
386     DK_DCB_S,
387     DK_DCB_W,
388     DK_DCB_X,
389     DK_DS,
390     DK_DS_B,
391     DK_DS_D,
392     DK_DS_L,
393     DK_DS_P,
394     DK_DS_S,
395     DK_DS_W,
396     DK_DS_X,
397     DK_SINGLE,
398     DK_FLOAT,
399     DK_DOUBLE,
400     DK_ALIGN,
401     DK_ALIGN32,
402     DK_BALIGN,
403     DK_BALIGNW,
404     DK_BALIGNL,
405     DK_P2ALIGN,
406     DK_P2ALIGNW,
407     DK_P2ALIGNL,
408     DK_ORG,
409     DK_FILL,
410     DK_ENDR,
411     DK_BUNDLE_ALIGN_MODE,
412     DK_BUNDLE_LOCK,
413     DK_BUNDLE_UNLOCK,
414     DK_ZERO,
415     DK_EXTERN,
416     DK_GLOBL,
417     DK_GLOBAL,
418     DK_LAZY_REFERENCE,
419     DK_NO_DEAD_STRIP,
420     DK_SYMBOL_RESOLVER,
421     DK_PRIVATE_EXTERN,
422     DK_REFERENCE,
423     DK_WEAK_DEFINITION,
424     DK_WEAK_REFERENCE,
425     DK_WEAK_DEF_CAN_BE_HIDDEN,
426     DK_COMM,
427     DK_COMMON,
428     DK_LCOMM,
429     DK_ABORT,
430     DK_INCLUDE,
431     DK_INCBIN,
432     DK_CODE16,
433     DK_CODE16GCC,
434     DK_REPT,
435     DK_IRP,
436     DK_IRPC,
437     DK_IF,
438     DK_IFEQ,
439     DK_IFGE,
440     DK_IFGT,
441     DK_IFLE,
442     DK_IFLT,
443     DK_IFNE,
444     DK_IFB,
445     DK_IFNB,
446     DK_IFC,
447     DK_IFEQS,
448     DK_IFNC,
449     DK_IFNES,
450     DK_IFDEF,
451     DK_IFNDEF,
452     DK_IFNOTDEF,
453     DK_ELSEIF,
454     DK_ELSE,
455     DK_ENDIF,
456     DK_SPACE,
457     DK_SKIP,
458     DK_FILE,
459     DK_LINE,
460     DK_LOC,
461     DK_STABS,
462     DK_CV_FILE,
463     DK_CV_FUNC_ID,
464     DK_CV_INLINE_SITE_ID,
465     DK_CV_LOC,
466     DK_CV_LINETABLE,
467     DK_CV_INLINE_LINETABLE,
468     DK_CV_DEF_RANGE,
469     DK_CV_STRINGTABLE,
470     DK_CV_FILECHECKSUMS,
471     DK_CV_FILECHECKSUM_OFFSET,
472     DK_CV_FPO_DATA,
473     DK_CFI_SECTIONS,
474     DK_CFI_STARTPROC,
475     DK_CFI_ENDPROC,
476     DK_CFI_DEF_CFA,
477     DK_CFI_DEF_CFA_OFFSET,
478     DK_CFI_ADJUST_CFA_OFFSET,
479     DK_CFI_DEF_CFA_REGISTER,
480     DK_CFI_OFFSET,
481     DK_CFI_REL_OFFSET,
482     DK_CFI_PERSONALITY,
483     DK_CFI_LSDA,
484     DK_CFI_REMEMBER_STATE,
485     DK_CFI_RESTORE_STATE,
486     DK_CFI_SAME_VALUE,
487     DK_CFI_RESTORE,
488     DK_CFI_ESCAPE,
489     DK_CFI_RETURN_COLUMN,
490     DK_CFI_SIGNAL_FRAME,
491     DK_CFI_UNDEFINED,
492     DK_CFI_REGISTER,
493     DK_CFI_WINDOW_SAVE,
494     DK_MACROS_ON,
495     DK_MACROS_OFF,
496     DK_ALTMACRO,
497     DK_NOALTMACRO,
498     DK_MACRO,
499     DK_EXITM,
500     DK_ENDM,
501     DK_ENDMACRO,
502     DK_PURGEM,
503     DK_SLEB128,
504     DK_ULEB128,
505     DK_ERR,
506     DK_ERROR,
507     DK_WARNING,
508     DK_PRINT,
509     DK_ADDRSIG,
510     DK_ADDRSIG_SYM,
511     DK_END
512   };
513 
514   /// Maps directive name --> DirectiveKind enum, for
515   /// directives parsed by this class.
516   StringMap<DirectiveKind> DirectiveKindMap;
517 
518   // ".ascii", ".asciz", ".string"
519   bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated);
520   bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc"
521   bool parseDirectiveValue(StringRef IDVal,
522                            unsigned Size);       // ".byte", ".long", ...
523   bool parseDirectiveOctaValue(StringRef IDVal); // ".octa", ...
524   bool parseDirectiveRealValue(StringRef IDVal,
525                                const fltSemantics &); // ".single", ...
526   bool parseDirectiveFill(); // ".fill"
527   bool parseDirectiveZero(); // ".zero"
528   // ".set", ".equ", ".equiv"
529   bool parseDirectiveSet(StringRef IDVal, bool allow_redef);
530   bool parseDirectiveOrg(); // ".org"
531   // ".align{,32}", ".p2align{,w,l}"
532   bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize);
533 
534   // ".file", ".line", ".loc", ".stabs"
535   bool parseDirectiveFile(SMLoc DirectiveLoc);
536   bool parseDirectiveLine();
537   bool parseDirectiveLoc();
538   bool parseDirectiveStabs();
539 
540   // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable",
541   // ".cv_inline_linetable", ".cv_def_range"
542   bool parseDirectiveCVFile();
543   bool parseDirectiveCVFuncId();
544   bool parseDirectiveCVInlineSiteId();
545   bool parseDirectiveCVLoc();
546   bool parseDirectiveCVLinetable();
547   bool parseDirectiveCVInlineLinetable();
548   bool parseDirectiveCVDefRange();
549   bool parseDirectiveCVStringTable();
550   bool parseDirectiveCVFileChecksums();
551   bool parseDirectiveCVFileChecksumOffset();
552   bool parseDirectiveCVFPOData();
553 
554   // .cfi directives
555   bool parseDirectiveCFIRegister(SMLoc DirectiveLoc);
556   bool parseDirectiveCFIWindowSave();
557   bool parseDirectiveCFISections();
558   bool parseDirectiveCFIStartProc();
559   bool parseDirectiveCFIEndProc();
560   bool parseDirectiveCFIDefCfaOffset();
561   bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc);
562   bool parseDirectiveCFIAdjustCfaOffset();
563   bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc);
564   bool parseDirectiveCFIOffset(SMLoc DirectiveLoc);
565   bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc);
566   bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality);
567   bool parseDirectiveCFIRememberState();
568   bool parseDirectiveCFIRestoreState();
569   bool parseDirectiveCFISameValue(SMLoc DirectiveLoc);
570   bool parseDirectiveCFIRestore(SMLoc DirectiveLoc);
571   bool parseDirectiveCFIEscape();
572   bool parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc);
573   bool parseDirectiveCFISignalFrame();
574   bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc);
575 
576   // macro directives
577   bool parseDirectivePurgeMacro(SMLoc DirectiveLoc);
578   bool parseDirectiveExitMacro(StringRef Directive);
579   bool parseDirectiveEndMacro(StringRef Directive);
580   bool parseDirectiveMacro(SMLoc DirectiveLoc);
581   bool parseDirectiveMacrosOnOff(StringRef Directive);
582   // alternate macro mode directives
583   bool parseDirectiveAltmacro(StringRef Directive);
584   // ".bundle_align_mode"
585   bool parseDirectiveBundleAlignMode();
586   // ".bundle_lock"
587   bool parseDirectiveBundleLock();
588   // ".bundle_unlock"
589   bool parseDirectiveBundleUnlock();
590 
591   // ".space", ".skip"
592   bool parseDirectiveSpace(StringRef IDVal);
593 
594   // ".dcb"
595   bool parseDirectiveDCB(StringRef IDVal, unsigned Size);
596   bool parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &);
597   // ".ds"
598   bool parseDirectiveDS(StringRef IDVal, unsigned Size);
599 
600   // .sleb128 (Signed=true) and .uleb128 (Signed=false)
601   bool parseDirectiveLEB128(bool Signed);
602 
603   /// Parse a directive like ".globl" which
604   /// accepts a single symbol (which should be a label or an external).
605   bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr);
606 
607   bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm"
608 
609   bool parseDirectiveAbort(); // ".abort"
610   bool parseDirectiveInclude(); // ".include"
611   bool parseDirectiveIncbin(); // ".incbin"
612 
613   // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne"
614   bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind);
615   // ".ifb" or ".ifnb", depending on ExpectBlank.
616   bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank);
617   // ".ifc" or ".ifnc", depending on ExpectEqual.
618   bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual);
619   // ".ifeqs" or ".ifnes", depending on ExpectEqual.
620   bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual);
621   // ".ifdef" or ".ifndef", depending on expect_defined
622   bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined);
623   bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif"
624   bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else"
625   bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif
626   bool parseEscapedString(std::string &Data) override;
627 
628   const MCExpr *applyModifierToExpr(const MCExpr *E,
629                                     MCSymbolRefExpr::VariantKind Variant);
630 
631   // Macro-like directives
632   MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc);
633   void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc,
634                                 raw_svector_ostream &OS);
635   bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive);
636   bool parseDirectiveIrp(SMLoc DirectiveLoc);  // ".irp"
637   bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc"
638   bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr"
639 
640   // "_emit" or "__emit"
641   bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info,
642                             size_t Len);
643 
644   // "align"
645   bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info);
646 
647   // "end"
648   bool parseDirectiveEnd(SMLoc DirectiveLoc);
649 
650   // ".err" or ".error"
651   bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage);
652 
653   // ".warning"
654   bool parseDirectiveWarning(SMLoc DirectiveLoc);
655 
656   // .print <double-quotes-string>
657   bool parseDirectivePrint(SMLoc DirectiveLoc);
658 
659   // Directives to support address-significance tables.
660   bool parseDirectiveAddrsig();
661   bool parseDirectiveAddrsigSym();
662 
663   void initializeDirectiveKindMap();
664 };
665 
666 } // end anonymous namespace
667 
668 namespace llvm {
669 
670 extern MCAsmParserExtension *createDarwinAsmParser();
671 extern MCAsmParserExtension *createELFAsmParser();
672 extern MCAsmParserExtension *createCOFFAsmParser();
673 
674 } // end namespace llvm
675 
676 enum { DEFAULT_ADDRSPACE = 0 };
677 
AsmParser(SourceMgr & SM,MCContext & Ctx,MCStreamer & Out,const MCAsmInfo & MAI,unsigned CB=0)678 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out,
679                      const MCAsmInfo &MAI, unsigned CB = 0)
680     : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM),
681       CurBuffer(CB ? CB : SM.getMainFileID()), MacrosEnabledFlag(true) {
682   HadError = false;
683   // Save the old handler.
684   SavedDiagHandler = SrcMgr.getDiagHandler();
685   SavedDiagContext = SrcMgr.getDiagContext();
686   // Set our own handler which calls the saved handler.
687   SrcMgr.setDiagHandler(DiagHandler, this);
688   Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer());
689 
690   // Initialize the platform / file format parser.
691   switch (Ctx.getObjectFileInfo()->getObjectFileType()) {
692   case MCObjectFileInfo::IsCOFF:
693     PlatformParser.reset(createCOFFAsmParser());
694     break;
695   case MCObjectFileInfo::IsMachO:
696     PlatformParser.reset(createDarwinAsmParser());
697     IsDarwin = true;
698     break;
699   case MCObjectFileInfo::IsELF:
700     PlatformParser.reset(createELFAsmParser());
701     break;
702   case MCObjectFileInfo::IsWasm:
703     // TODO: WASM will need its own MCAsmParserExtension implementation, but
704     // for now we can re-use the ELF one, since the directives can be the
705     // same for now.
706     PlatformParser.reset(createELFAsmParser());
707     break;
708   }
709 
710   PlatformParser->Initialize(*this);
711   initializeDirectiveKindMap();
712 
713   NumOfMacroInstantiations = 0;
714 }
715 
~AsmParser()716 AsmParser::~AsmParser() {
717   assert((HadError || ActiveMacros.empty()) &&
718          "Unexpected active macro instantiation!");
719 
720   // Restore the saved diagnostics handler and context for use during
721   // finalization.
722   SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext);
723 }
724 
printMacroInstantiations()725 void AsmParser::printMacroInstantiations() {
726   // Print the active macro instantiation stack.
727   for (std::vector<MacroInstantiation *>::const_reverse_iterator
728            it = ActiveMacros.rbegin(),
729            ie = ActiveMacros.rend();
730        it != ie; ++it)
731     printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note,
732                  "while in macro instantiation");
733 }
734 
Note(SMLoc L,const Twine & Msg,SMRange Range)735 void AsmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) {
736   printPendingErrors();
737   printMessage(L, SourceMgr::DK_Note, Msg, Range);
738   printMacroInstantiations();
739 }
740 
Warning(SMLoc L,const Twine & Msg,SMRange Range)741 bool AsmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) {
742   if(getTargetParser().getTargetOptions().MCNoWarn)
743     return false;
744   if (getTargetParser().getTargetOptions().MCFatalWarnings)
745     return Error(L, Msg, Range);
746   printMessage(L, SourceMgr::DK_Warning, Msg, Range);
747   printMacroInstantiations();
748   return false;
749 }
750 
printError(SMLoc L,const Twine & Msg,SMRange Range)751 bool AsmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) {
752   HadError = true;
753   printMessage(L, SourceMgr::DK_Error, Msg, Range);
754   printMacroInstantiations();
755   return true;
756 }
757 
enterIncludeFile(const std::string & Filename)758 bool AsmParser::enterIncludeFile(const std::string &Filename) {
759   std::string IncludedFile;
760   unsigned NewBuf =
761       SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile);
762   if (!NewBuf)
763     return true;
764 
765   CurBuffer = NewBuf;
766   Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer());
767   return false;
768 }
769 
770 /// Process the specified .incbin file by searching for it in the include paths
771 /// then just emitting the byte contents of the file to the streamer. This
772 /// returns true on failure.
processIncbinFile(const std::string & Filename,int64_t Skip,const MCExpr * Count,SMLoc Loc)773 bool AsmParser::processIncbinFile(const std::string &Filename, int64_t Skip,
774                                   const MCExpr *Count, SMLoc Loc) {
775   std::string IncludedFile;
776   unsigned NewBuf =
777       SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile);
778   if (!NewBuf)
779     return true;
780 
781   // Pick up the bytes from the file and emit them.
782   StringRef Bytes = SrcMgr.getMemoryBuffer(NewBuf)->getBuffer();
783   Bytes = Bytes.drop_front(Skip);
784   if (Count) {
785     int64_t Res;
786     if (!Count->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr()))
787       return Error(Loc, "expected absolute expression");
788     if (Res < 0)
789       return Warning(Loc, "negative count has no effect");
790     Bytes = Bytes.take_front(Res);
791   }
792   getStreamer().EmitBytes(Bytes);
793   return false;
794 }
795 
jumpToLoc(SMLoc Loc,unsigned InBuffer)796 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) {
797   CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc);
798   Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(),
799                   Loc.getPointer());
800 }
801 
Lex()802 const AsmToken &AsmParser::Lex() {
803   if (Lexer.getTok().is(AsmToken::Error))
804     Error(Lexer.getErrLoc(), Lexer.getErr());
805 
806   // if it's a end of statement with a comment in it
807   if (getTok().is(AsmToken::EndOfStatement)) {
808     // if this is a line comment output it.
809     if (!getTok().getString().empty() && getTok().getString().front() != '\n' &&
810         getTok().getString().front() != '\r' && MAI.preserveAsmComments())
811       Out.addExplicitComment(Twine(getTok().getString()));
812   }
813 
814   const AsmToken *tok = &Lexer.Lex();
815 
816   // Parse comments here to be deferred until end of next statement.
817   while (tok->is(AsmToken::Comment)) {
818     if (MAI.preserveAsmComments())
819       Out.addExplicitComment(Twine(tok->getString()));
820     tok = &Lexer.Lex();
821   }
822 
823   if (tok->is(AsmToken::Eof)) {
824     // If this is the end of an included file, pop the parent file off the
825     // include stack.
826     SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer);
827     if (ParentIncludeLoc != SMLoc()) {
828       jumpToLoc(ParentIncludeLoc);
829       return Lex();
830     }
831   }
832 
833   return *tok;
834 }
835 
enabledGenDwarfForAssembly()836 bool AsmParser::enabledGenDwarfForAssembly() {
837   // Check whether the user specified -g.
838   if (!getContext().getGenDwarfForAssembly())
839     return false;
840   // If we haven't encountered any .file directives (which would imply that
841   // the assembler source was produced with debug info already) then emit one
842   // describing the assembler source file itself.
843   if (getContext().getGenDwarfFileNumber() == 0)
844     getContext().setGenDwarfFileNumber(getStreamer().EmitDwarfFileDirective(
845         0, StringRef(), getContext().getMainFileName()));
846   return true;
847 }
848 
Run(bool NoInitialTextSection,bool NoFinalize)849 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) {
850   // Create the initial section, if requested.
851   if (!NoInitialTextSection)
852     Out.InitSections(false);
853 
854   // Prime the lexer.
855   Lex();
856 
857   HadError = false;
858   AsmCond StartingCondState = TheCondState;
859   SmallVector<AsmRewrite, 4> AsmStrRewrites;
860 
861   // If we are generating dwarf for assembly source files save the initial text
862   // section.  (Don't use enabledGenDwarfForAssembly() here, as we aren't
863   // emitting any actual debug info yet and haven't had a chance to parse any
864   // embedded .file directives.)
865   if (getContext().getGenDwarfForAssembly()) {
866     MCSection *Sec = getStreamer().getCurrentSectionOnly();
867     if (!Sec->getBeginSymbol()) {
868       MCSymbol *SectionStartSym = getContext().createTempSymbol();
869       getStreamer().EmitLabel(SectionStartSym);
870       Sec->setBeginSymbol(SectionStartSym);
871     }
872     bool InsertResult = getContext().addGenDwarfSection(Sec);
873     assert(InsertResult && ".text section should not have debug info yet");
874     (void)InsertResult;
875   }
876 
877   // While we have input, parse each statement.
878   while (Lexer.isNot(AsmToken::Eof)) {
879     ParseStatementInfo Info(&AsmStrRewrites);
880     if (!parseStatement(Info, nullptr))
881       continue;
882 
883     // If we have a Lexer Error we are on an Error Token. Load in Lexer Error
884     // for printing ErrMsg via Lex() only if no (presumably better) parser error
885     // exists.
886     if (!hasPendingError() && Lexer.getTok().is(AsmToken::Error)) {
887       Lex();
888     }
889 
890     // parseStatement returned true so may need to emit an error.
891     printPendingErrors();
892 
893     // Skipping to the next line if needed.
894     if (!getLexer().isAtStartOfStatement())
895       eatToEndOfStatement();
896   }
897 
898   // All errors should have been emitted.
899   assert(!hasPendingError() && "unexpected error from parseStatement");
900 
901   getTargetParser().flushPendingInstructions(getStreamer());
902 
903   if (TheCondState.TheCond != StartingCondState.TheCond ||
904       TheCondState.Ignore != StartingCondState.Ignore)
905     printError(getTok().getLoc(), "unmatched .ifs or .elses");
906   // Check to see there are no empty DwarfFile slots.
907   const auto &LineTables = getContext().getMCDwarfLineTables();
908   if (!LineTables.empty()) {
909     unsigned Index = 0;
910     for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) {
911       if (File.Name.empty() && Index != 0)
912         printError(getTok().getLoc(), "unassigned file number: " +
913                                           Twine(Index) +
914                                           " for .file directives");
915       ++Index;
916     }
917   }
918 
919   // Check to see that all assembler local symbols were actually defined.
920   // Targets that don't do subsections via symbols may not want this, though,
921   // so conservatively exclude them. Only do this if we're finalizing, though,
922   // as otherwise we won't necessarilly have seen everything yet.
923   if (!NoFinalize) {
924     if (MAI.hasSubsectionsViaSymbols()) {
925       for (const auto &TableEntry : getContext().getSymbols()) {
926         MCSymbol *Sym = TableEntry.getValue();
927         // Variable symbols may not be marked as defined, so check those
928         // explicitly. If we know it's a variable, we have a definition for
929         // the purposes of this check.
930         if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined())
931           // FIXME: We would really like to refer back to where the symbol was
932           // first referenced for a source location. We need to add something
933           // to track that. Currently, we just point to the end of the file.
934           printError(getTok().getLoc(), "assembler local symbol '" +
935                                             Sym->getName() + "' not defined");
936       }
937     }
938 
939     // Temporary symbols like the ones for directional jumps don't go in the
940     // symbol table. They also need to be diagnosed in all (final) cases.
941     for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) {
942       if (std::get<2>(LocSym)->isUndefined()) {
943         // Reset the state of any "# line file" directives we've seen to the
944         // context as it was at the diagnostic site.
945         CppHashInfo = std::get<1>(LocSym);
946         printError(std::get<0>(LocSym), "directional label undefined");
947       }
948     }
949   }
950 
951   // Finalize the output stream if there are no errors and if the client wants
952   // us to.
953   if (!HadError && !NoFinalize)
954     Out.Finish();
955 
956   return HadError || getContext().hadError();
957 }
958 
checkForValidSection()959 bool AsmParser::checkForValidSection() {
960   if (!ParsingInlineAsm && !getStreamer().getCurrentSectionOnly()) {
961     Out.InitSections(false);
962     return Error(getTok().getLoc(),
963                  "expected section directive before assembly directive");
964   }
965   return false;
966 }
967 
968 /// Throw away the rest of the line for testing purposes.
eatToEndOfStatement()969 void AsmParser::eatToEndOfStatement() {
970   while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof))
971     Lexer.Lex();
972 
973   // Eat EOL.
974   if (Lexer.is(AsmToken::EndOfStatement))
975     Lexer.Lex();
976 }
977 
parseStringToEndOfStatement()978 StringRef AsmParser::parseStringToEndOfStatement() {
979   const char *Start = getTok().getLoc().getPointer();
980 
981   while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof))
982     Lexer.Lex();
983 
984   const char *End = getTok().getLoc().getPointer();
985   return StringRef(Start, End - Start);
986 }
987 
parseStringToComma()988 StringRef AsmParser::parseStringToComma() {
989   const char *Start = getTok().getLoc().getPointer();
990 
991   while (Lexer.isNot(AsmToken::EndOfStatement) &&
992          Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof))
993     Lexer.Lex();
994 
995   const char *End = getTok().getLoc().getPointer();
996   return StringRef(Start, End - Start);
997 }
998 
999 /// Parse a paren expression and return it.
1000 /// NOTE: This assumes the leading '(' has already been consumed.
1001 ///
1002 /// parenexpr ::= expr)
1003 ///
parseParenExpr(const MCExpr * & Res,SMLoc & EndLoc)1004 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) {
1005   if (parseExpression(Res))
1006     return true;
1007   if (Lexer.isNot(AsmToken::RParen))
1008     return TokError("expected ')' in parentheses expression");
1009   EndLoc = Lexer.getTok().getEndLoc();
1010   Lex();
1011   return false;
1012 }
1013 
1014 /// Parse a bracket expression and return it.
1015 /// NOTE: This assumes the leading '[' has already been consumed.
1016 ///
1017 /// bracketexpr ::= expr]
1018 ///
parseBracketExpr(const MCExpr * & Res,SMLoc & EndLoc)1019 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) {
1020   if (parseExpression(Res))
1021     return true;
1022   EndLoc = getTok().getEndLoc();
1023   if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression"))
1024     return true;
1025   return false;
1026 }
1027 
1028 /// Parse a primary expression and return it.
1029 ///  primaryexpr ::= (parenexpr
1030 ///  primaryexpr ::= symbol
1031 ///  primaryexpr ::= number
1032 ///  primaryexpr ::= '.'
1033 ///  primaryexpr ::= ~,+,- primaryexpr
parsePrimaryExpr(const MCExpr * & Res,SMLoc & EndLoc)1034 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) {
1035   SMLoc FirstTokenLoc = getLexer().getLoc();
1036   AsmToken::TokenKind FirstTokenKind = Lexer.getKind();
1037   switch (FirstTokenKind) {
1038   default:
1039     return TokError("unknown token in expression");
1040   // If we have an error assume that we've already handled it.
1041   case AsmToken::Error:
1042     return true;
1043   case AsmToken::Exclaim:
1044     Lex(); // Eat the operator.
1045     if (parsePrimaryExpr(Res, EndLoc))
1046       return true;
1047     Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc);
1048     return false;
1049   case AsmToken::Dollar:
1050   case AsmToken::At:
1051   case AsmToken::String:
1052   case AsmToken::Identifier: {
1053     StringRef Identifier;
1054     if (parseIdentifier(Identifier)) {
1055       // We may have failed but $ may be a valid token.
1056       if (getTok().is(AsmToken::Dollar)) {
1057         if (Lexer.getMAI().getDollarIsPC()) {
1058           Lex();
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,
1064                                         getContext());
1065           EndLoc = FirstTokenLoc;
1066           return false;
1067         }
1068         return Error(FirstTokenLoc, "invalid token in expression");
1069       }
1070     }
1071     // Parse symbol variant
1072     std::pair<StringRef, StringRef> Split;
1073     if (!MAI.useParensForSymbolVariant()) {
1074       if (FirstTokenKind == AsmToken::String) {
1075         if (Lexer.is(AsmToken::At)) {
1076           Lex(); // eat @
1077           SMLoc AtLoc = getLexer().getLoc();
1078           StringRef VName;
1079           if (parseIdentifier(VName))
1080             return Error(AtLoc, "expected symbol variant after '@'");
1081 
1082           Split = std::make_pair(Identifier, VName);
1083         }
1084       } else {
1085         Split = Identifier.split('@');
1086       }
1087     } else if (Lexer.is(AsmToken::LParen)) {
1088       Lex(); // eat '('.
1089       StringRef VName;
1090       parseIdentifier(VName);
1091       // eat ')'.
1092       if (parseToken(AsmToken::RParen,
1093                      "unexpected token in variant, expected ')'"))
1094         return true;
1095       Split = std::make_pair(Identifier, VName);
1096     }
1097 
1098     EndLoc = SMLoc::getFromPointer(Identifier.end());
1099 
1100     // This is a symbol reference.
1101     StringRef SymbolName = Identifier;
1102     if (SymbolName.empty())
1103       return true;
1104 
1105     MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None;
1106 
1107     // Lookup the symbol variant if used.
1108     if (!Split.second.empty()) {
1109       Variant = MCSymbolRefExpr::getVariantKindForName(Split.second);
1110       if (Variant != MCSymbolRefExpr::VK_Invalid) {
1111         SymbolName = Split.first;
1112       } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) {
1113         Variant = MCSymbolRefExpr::VK_None;
1114       } else {
1115         return Error(SMLoc::getFromPointer(Split.second.begin()),
1116                      "invalid variant '" + Split.second + "'");
1117       }
1118     }
1119 
1120     MCSymbol *Sym = getContext().getOrCreateSymbol(SymbolName);
1121 
1122     // If this is an absolute variable reference, substitute it now to preserve
1123     // semantics in the face of reassignment.
1124     if (Sym->isVariable()) {
1125       auto V = Sym->getVariableValue(/*SetUsed*/ false);
1126       bool DoInline = isa<MCConstantExpr>(V);
1127       if (auto TV = dyn_cast<MCTargetExpr>(V))
1128         DoInline = TV->inlineAssignedExpr();
1129       if (DoInline) {
1130         if (Variant)
1131           return Error(EndLoc, "unexpected modifier on variable reference");
1132         Res = Sym->getVariableValue(/*SetUsed*/ false);
1133         return false;
1134       }
1135     }
1136 
1137     // Otherwise create a symbol ref.
1138     Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc);
1139     return false;
1140   }
1141   case AsmToken::BigNum:
1142     return TokError("literal value out of range for directive");
1143   case AsmToken::Integer: {
1144     SMLoc Loc = getTok().getLoc();
1145     int64_t IntVal = getTok().getIntVal();
1146     Res = MCConstantExpr::create(IntVal, getContext());
1147     EndLoc = Lexer.getTok().getEndLoc();
1148     Lex(); // Eat token.
1149     // Look for 'b' or 'f' following an Integer as a directional label
1150     if (Lexer.getKind() == AsmToken::Identifier) {
1151       StringRef IDVal = getTok().getString();
1152       // Lookup the symbol variant if used.
1153       std::pair<StringRef, StringRef> Split = IDVal.split('@');
1154       MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None;
1155       if (Split.first.size() != IDVal.size()) {
1156         Variant = MCSymbolRefExpr::getVariantKindForName(Split.second);
1157         if (Variant == MCSymbolRefExpr::VK_Invalid)
1158           return TokError("invalid variant '" + Split.second + "'");
1159         IDVal = Split.first;
1160       }
1161       if (IDVal == "f" || IDVal == "b") {
1162         MCSymbol *Sym =
1163             Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b");
1164         Res = MCSymbolRefExpr::create(Sym, Variant, getContext());
1165         if (IDVal == "b" && Sym->isUndefined())
1166           return Error(Loc, "directional label undefined");
1167         DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym));
1168         EndLoc = Lexer.getTok().getEndLoc();
1169         Lex(); // Eat identifier.
1170       }
1171     }
1172     return false;
1173   }
1174   case AsmToken::Real: {
1175     APFloat RealVal(APFloat::IEEEdouble(), getTok().getString());
1176     uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue();
1177     Res = MCConstantExpr::create(IntVal, getContext());
1178     EndLoc = Lexer.getTok().getEndLoc();
1179     Lex(); // Eat token.
1180     return false;
1181   }
1182   case AsmToken::Dot: {
1183     // This is a '.' reference, which references the current PC.  Emit a
1184     // temporary label to the streamer and refer to it.
1185     MCSymbol *Sym = Ctx.createTempSymbol();
1186     Out.EmitLabel(Sym);
1187     Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext());
1188     EndLoc = Lexer.getTok().getEndLoc();
1189     Lex(); // Eat identifier.
1190     return false;
1191   }
1192   case AsmToken::LParen:
1193     Lex(); // Eat the '('.
1194     return parseParenExpr(Res, EndLoc);
1195   case AsmToken::LBrac:
1196     if (!PlatformParser->HasBracketExpressions())
1197       return TokError("brackets expression not supported on this target");
1198     Lex(); // Eat the '['.
1199     return parseBracketExpr(Res, EndLoc);
1200   case AsmToken::Minus:
1201     Lex(); // Eat the operator.
1202     if (parsePrimaryExpr(Res, EndLoc))
1203       return true;
1204     Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc);
1205     return false;
1206   case AsmToken::Plus:
1207     Lex(); // Eat the operator.
1208     if (parsePrimaryExpr(Res, EndLoc))
1209       return true;
1210     Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc);
1211     return false;
1212   case AsmToken::Tilde:
1213     Lex(); // Eat the operator.
1214     if (parsePrimaryExpr(Res, EndLoc))
1215       return true;
1216     Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc);
1217     return false;
1218   // MIPS unary expression operators. The lexer won't generate these tokens if
1219   // MCAsmInfo::HasMipsExpressions is false for the target.
1220   case AsmToken::PercentCall16:
1221   case AsmToken::PercentCall_Hi:
1222   case AsmToken::PercentCall_Lo:
1223   case AsmToken::PercentDtprel_Hi:
1224   case AsmToken::PercentDtprel_Lo:
1225   case AsmToken::PercentGot:
1226   case AsmToken::PercentGot_Disp:
1227   case AsmToken::PercentGot_Hi:
1228   case AsmToken::PercentGot_Lo:
1229   case AsmToken::PercentGot_Ofst:
1230   case AsmToken::PercentGot_Page:
1231   case AsmToken::PercentGottprel:
1232   case AsmToken::PercentGp_Rel:
1233   case AsmToken::PercentHi:
1234   case AsmToken::PercentHigher:
1235   case AsmToken::PercentHighest:
1236   case AsmToken::PercentLo:
1237   case AsmToken::PercentNeg:
1238   case AsmToken::PercentPcrel_Hi:
1239   case AsmToken::PercentPcrel_Lo:
1240   case AsmToken::PercentTlsgd:
1241   case AsmToken::PercentTlsldm:
1242   case AsmToken::PercentTprel_Hi:
1243   case AsmToken::PercentTprel_Lo:
1244     Lex(); // Eat the operator.
1245     if (Lexer.isNot(AsmToken::LParen))
1246       return TokError("expected '(' after operator");
1247     Lex(); // Eat the operator.
1248     if (parseExpression(Res, EndLoc))
1249       return true;
1250     if (Lexer.isNot(AsmToken::RParen))
1251       return TokError("expected ')'");
1252     Lex(); // Eat the operator.
1253     Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx);
1254     return !Res;
1255   }
1256 }
1257 
parseExpression(const MCExpr * & Res)1258 bool AsmParser::parseExpression(const MCExpr *&Res) {
1259   SMLoc EndLoc;
1260   return parseExpression(Res, EndLoc);
1261 }
1262 
1263 const MCExpr *
applyModifierToExpr(const MCExpr * E,MCSymbolRefExpr::VariantKind Variant)1264 AsmParser::applyModifierToExpr(const MCExpr *E,
1265                                MCSymbolRefExpr::VariantKind Variant) {
1266   // Ask the target implementation about this expression first.
1267   const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx);
1268   if (NewE)
1269     return NewE;
1270   // Recurse over the given expression, rebuilding it to apply the given variant
1271   // if there is exactly one symbol.
1272   switch (E->getKind()) {
1273   case MCExpr::Target:
1274   case MCExpr::Constant:
1275     return nullptr;
1276 
1277   case MCExpr::SymbolRef: {
1278     const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E);
1279 
1280     if (SRE->getKind() != MCSymbolRefExpr::VK_None) {
1281       TokError("invalid variant on expression '" + getTok().getIdentifier() +
1282                "' (already modified)");
1283       return E;
1284     }
1285 
1286     return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext());
1287   }
1288 
1289   case MCExpr::Unary: {
1290     const MCUnaryExpr *UE = cast<MCUnaryExpr>(E);
1291     const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant);
1292     if (!Sub)
1293       return nullptr;
1294     return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext());
1295   }
1296 
1297   case MCExpr::Binary: {
1298     const MCBinaryExpr *BE = cast<MCBinaryExpr>(E);
1299     const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant);
1300     const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant);
1301 
1302     if (!LHS && !RHS)
1303       return nullptr;
1304 
1305     if (!LHS)
1306       LHS = BE->getLHS();
1307     if (!RHS)
1308       RHS = BE->getRHS();
1309 
1310     return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext());
1311   }
1312   }
1313 
1314   llvm_unreachable("Invalid expression kind!");
1315 }
1316 
1317 /// This function checks if the next token is <string> type or arithmetic.
1318 /// string that begin with character '<' must end with character '>'.
1319 /// otherwise it is arithmetics.
1320 /// If the function returns a 'true' value,
1321 /// the End argument will be filled with the last location pointed to the '>'
1322 /// character.
1323 
1324 /// There is a gap between the AltMacro's documentation and the single quote implementation.
1325 /// GCC does not fully support this feature and so we will not support it.
1326 /// TODO: Adding single quote as a string.
isAltmacroString(SMLoc & StrLoc,SMLoc & EndLoc)1327 bool AsmParser::isAltmacroString(SMLoc &StrLoc, SMLoc &EndLoc) {
1328   assert((StrLoc.getPointer() != NULL) &&
1329          "Argument to the function cannot be a NULL value");
1330   const char *CharPtr = StrLoc.getPointer();
1331   while ((*CharPtr != '>') && (*CharPtr != '\n') && (*CharPtr != '\r') &&
1332          (*CharPtr != '\0')) {
1333     if (*CharPtr == '!')
1334       CharPtr++;
1335     CharPtr++;
1336   }
1337   if (*CharPtr == '>') {
1338     EndLoc = StrLoc.getFromPointer(CharPtr + 1);
1339     return true;
1340   }
1341   return false;
1342 }
1343 
1344 /// creating a string without the escape characters '!'.
altMacroString(StringRef AltMacroStr,std::string & Res)1345 void AsmParser::altMacroString(StringRef AltMacroStr,std::string &Res) {
1346   for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) {
1347     if (AltMacroStr[Pos] == '!')
1348       Pos++;
1349     Res += AltMacroStr[Pos];
1350   }
1351 }
1352 
1353 /// Parse an expression and return it.
1354 ///
1355 ///  expr ::= expr &&,|| expr               -> lowest.
1356 ///  expr ::= expr |,^,&,! expr
1357 ///  expr ::= expr ==,!=,<>,<,<=,>,>= expr
1358 ///  expr ::= expr <<,>> expr
1359 ///  expr ::= expr +,- expr
1360 ///  expr ::= expr *,/,% expr               -> highest.
1361 ///  expr ::= primaryexpr
1362 ///
parseExpression(const MCExpr * & Res,SMLoc & EndLoc)1363 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) {
1364   // Parse the expression.
1365   Res = nullptr;
1366   if (getTargetParser().parsePrimaryExpr(Res, EndLoc) ||
1367       parseBinOpRHS(1, Res, EndLoc))
1368     return true;
1369 
1370   // As a special case, we support 'a op b @ modifier' by rewriting the
1371   // expression to include the modifier. This is inefficient, but in general we
1372   // expect users to use 'a@modifier op b'.
1373   if (Lexer.getKind() == AsmToken::At) {
1374     Lex();
1375 
1376     if (Lexer.isNot(AsmToken::Identifier))
1377       return TokError("unexpected symbol modifier following '@'");
1378 
1379     MCSymbolRefExpr::VariantKind Variant =
1380         MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier());
1381     if (Variant == MCSymbolRefExpr::VK_Invalid)
1382       return TokError("invalid variant '" + getTok().getIdentifier() + "'");
1383 
1384     const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant);
1385     if (!ModifiedRes) {
1386       return TokError("invalid modifier '" + getTok().getIdentifier() +
1387                       "' (no symbols present)");
1388     }
1389 
1390     Res = ModifiedRes;
1391     Lex();
1392   }
1393 
1394   // Try to constant fold it up front, if possible. Do not exploit
1395   // assembler here.
1396   int64_t Value;
1397   if (Res->evaluateAsAbsolute(Value))
1398     Res = MCConstantExpr::create(Value, getContext());
1399 
1400   return false;
1401 }
1402 
parseParenExpression(const MCExpr * & Res,SMLoc & EndLoc)1403 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) {
1404   Res = nullptr;
1405   return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc);
1406 }
1407 
parseParenExprOfDepth(unsigned ParenDepth,const MCExpr * & Res,SMLoc & EndLoc)1408 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res,
1409                                       SMLoc &EndLoc) {
1410   if (parseParenExpr(Res, EndLoc))
1411     return true;
1412 
1413   for (; ParenDepth > 0; --ParenDepth) {
1414     if (parseBinOpRHS(1, Res, EndLoc))
1415       return true;
1416 
1417     // We don't Lex() the last RParen.
1418     // This is the same behavior as parseParenExpression().
1419     if (ParenDepth - 1 > 0) {
1420       EndLoc = getTok().getEndLoc();
1421       if (parseToken(AsmToken::RParen,
1422                      "expected ')' in parentheses expression"))
1423         return true;
1424     }
1425   }
1426   return false;
1427 }
1428 
parseAbsoluteExpression(int64_t & Res)1429 bool AsmParser::parseAbsoluteExpression(int64_t &Res) {
1430   const MCExpr *Expr;
1431 
1432   SMLoc StartLoc = Lexer.getLoc();
1433   if (parseExpression(Expr))
1434     return true;
1435 
1436   if (!Expr->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr()))
1437     return Error(StartLoc, "expected absolute expression");
1438 
1439   return false;
1440 }
1441 
getDarwinBinOpPrecedence(AsmToken::TokenKind K,MCBinaryExpr::Opcode & Kind,bool ShouldUseLogicalShr)1442 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K,
1443                                          MCBinaryExpr::Opcode &Kind,
1444                                          bool ShouldUseLogicalShr) {
1445   switch (K) {
1446   default:
1447     return 0; // not a binop.
1448 
1449   // Lowest Precedence: &&, ||
1450   case AsmToken::AmpAmp:
1451     Kind = MCBinaryExpr::LAnd;
1452     return 1;
1453   case AsmToken::PipePipe:
1454     Kind = MCBinaryExpr::LOr;
1455     return 1;
1456 
1457   // Low Precedence: |, &, ^
1458   //
1459   // FIXME: gas seems to support '!' as an infix operator?
1460   case AsmToken::Pipe:
1461     Kind = MCBinaryExpr::Or;
1462     return 2;
1463   case AsmToken::Caret:
1464     Kind = MCBinaryExpr::Xor;
1465     return 2;
1466   case AsmToken::Amp:
1467     Kind = MCBinaryExpr::And;
1468     return 2;
1469 
1470   // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >=
1471   case AsmToken::EqualEqual:
1472     Kind = MCBinaryExpr::EQ;
1473     return 3;
1474   case AsmToken::ExclaimEqual:
1475   case AsmToken::LessGreater:
1476     Kind = MCBinaryExpr::NE;
1477     return 3;
1478   case AsmToken::Less:
1479     Kind = MCBinaryExpr::LT;
1480     return 3;
1481   case AsmToken::LessEqual:
1482     Kind = MCBinaryExpr::LTE;
1483     return 3;
1484   case AsmToken::Greater:
1485     Kind = MCBinaryExpr::GT;
1486     return 3;
1487   case AsmToken::GreaterEqual:
1488     Kind = MCBinaryExpr::GTE;
1489     return 3;
1490 
1491   // Intermediate Precedence: <<, >>
1492   case AsmToken::LessLess:
1493     Kind = MCBinaryExpr::Shl;
1494     return 4;
1495   case AsmToken::GreaterGreater:
1496     Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr;
1497     return 4;
1498 
1499   // High Intermediate Precedence: +, -
1500   case AsmToken::Plus:
1501     Kind = MCBinaryExpr::Add;
1502     return 5;
1503   case AsmToken::Minus:
1504     Kind = MCBinaryExpr::Sub;
1505     return 5;
1506 
1507   // Highest Precedence: *, /, %
1508   case AsmToken::Star:
1509     Kind = MCBinaryExpr::Mul;
1510     return 6;
1511   case AsmToken::Slash:
1512     Kind = MCBinaryExpr::Div;
1513     return 6;
1514   case AsmToken::Percent:
1515     Kind = MCBinaryExpr::Mod;
1516     return 6;
1517   }
1518 }
1519 
getGNUBinOpPrecedence(AsmToken::TokenKind K,MCBinaryExpr::Opcode & Kind,bool ShouldUseLogicalShr)1520 static unsigned getGNUBinOpPrecedence(AsmToken::TokenKind K,
1521                                       MCBinaryExpr::Opcode &Kind,
1522                                       bool ShouldUseLogicalShr) {
1523   switch (K) {
1524   default:
1525     return 0; // not a binop.
1526 
1527   // Lowest Precedence: &&, ||
1528   case AsmToken::AmpAmp:
1529     Kind = MCBinaryExpr::LAnd;
1530     return 2;
1531   case AsmToken::PipePipe:
1532     Kind = MCBinaryExpr::LOr;
1533     return 1;
1534 
1535   // Low Precedence: ==, !=, <>, <, <=, >, >=
1536   case AsmToken::EqualEqual:
1537     Kind = MCBinaryExpr::EQ;
1538     return 3;
1539   case AsmToken::ExclaimEqual:
1540   case AsmToken::LessGreater:
1541     Kind = MCBinaryExpr::NE;
1542     return 3;
1543   case AsmToken::Less:
1544     Kind = MCBinaryExpr::LT;
1545     return 3;
1546   case AsmToken::LessEqual:
1547     Kind = MCBinaryExpr::LTE;
1548     return 3;
1549   case AsmToken::Greater:
1550     Kind = MCBinaryExpr::GT;
1551     return 3;
1552   case AsmToken::GreaterEqual:
1553     Kind = MCBinaryExpr::GTE;
1554     return 3;
1555 
1556   // Low Intermediate Precedence: +, -
1557   case AsmToken::Plus:
1558     Kind = MCBinaryExpr::Add;
1559     return 4;
1560   case AsmToken::Minus:
1561     Kind = MCBinaryExpr::Sub;
1562     return 4;
1563 
1564   // High Intermediate Precedence: |, &, ^
1565   //
1566   // FIXME: gas seems to support '!' as an infix operator?
1567   case AsmToken::Pipe:
1568     Kind = MCBinaryExpr::Or;
1569     return 5;
1570   case AsmToken::Caret:
1571     Kind = MCBinaryExpr::Xor;
1572     return 5;
1573   case AsmToken::Amp:
1574     Kind = MCBinaryExpr::And;
1575     return 5;
1576 
1577   // Highest Precedence: *, /, %, <<, >>
1578   case AsmToken::Star:
1579     Kind = MCBinaryExpr::Mul;
1580     return 6;
1581   case AsmToken::Slash:
1582     Kind = MCBinaryExpr::Div;
1583     return 6;
1584   case AsmToken::Percent:
1585     Kind = MCBinaryExpr::Mod;
1586     return 6;
1587   case AsmToken::LessLess:
1588     Kind = MCBinaryExpr::Shl;
1589     return 6;
1590   case AsmToken::GreaterGreater:
1591     Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr;
1592     return 6;
1593   }
1594 }
1595 
getBinOpPrecedence(AsmToken::TokenKind K,MCBinaryExpr::Opcode & Kind)1596 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K,
1597                                        MCBinaryExpr::Opcode &Kind) {
1598   bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr();
1599   return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr)
1600                   : getGNUBinOpPrecedence(K, Kind, ShouldUseLogicalShr);
1601 }
1602 
1603 /// Parse all binary operators with precedence >= 'Precedence'.
1604 /// Res contains the LHS of the expression on input.
parseBinOpRHS(unsigned Precedence,const MCExpr * & Res,SMLoc & EndLoc)1605 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res,
1606                               SMLoc &EndLoc) {
1607   SMLoc StartLoc = Lexer.getLoc();
1608   while (true) {
1609     MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add;
1610     unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind);
1611 
1612     // If the next token is lower precedence than we are allowed to eat, return
1613     // successfully with what we ate already.
1614     if (TokPrec < Precedence)
1615       return false;
1616 
1617     Lex();
1618 
1619     // Eat the next primary expression.
1620     const MCExpr *RHS;
1621     if (getTargetParser().parsePrimaryExpr(RHS, EndLoc))
1622       return true;
1623 
1624     // If BinOp binds less tightly with RHS than the operator after RHS, let
1625     // the pending operator take RHS as its LHS.
1626     MCBinaryExpr::Opcode Dummy;
1627     unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy);
1628     if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc))
1629       return true;
1630 
1631     // Merge LHS and RHS according to operator.
1632     Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc);
1633   }
1634 }
1635 
1636 /// ParseStatement:
1637 ///   ::= EndOfStatement
1638 ///   ::= Label* Directive ...Operands... EndOfStatement
1639 ///   ::= Label* Identifier OperandList* EndOfStatement
parseStatement(ParseStatementInfo & Info,MCAsmParserSemaCallback * SI)1640 bool AsmParser::parseStatement(ParseStatementInfo &Info,
1641                                MCAsmParserSemaCallback *SI) {
1642   assert(!hasPendingError() && "parseStatement started with pending error");
1643   // Eat initial spaces and comments
1644   while (Lexer.is(AsmToken::Space))
1645     Lex();
1646   if (Lexer.is(AsmToken::EndOfStatement)) {
1647     // if this is a line comment we can drop it safely
1648     if (getTok().getString().empty() || getTok().getString().front() == '\r' ||
1649         getTok().getString().front() == '\n')
1650       Out.AddBlankLine();
1651     Lex();
1652     return false;
1653   }
1654   // Statements always start with an identifier.
1655   AsmToken ID = getTok();
1656   SMLoc IDLoc = ID.getLoc();
1657   StringRef IDVal;
1658   int64_t LocalLabelVal = -1;
1659   if (Lexer.is(AsmToken::HashDirective))
1660     return parseCppHashLineFilenameComment(IDLoc);
1661   // Allow an integer followed by a ':' as a directional local label.
1662   if (Lexer.is(AsmToken::Integer)) {
1663     LocalLabelVal = getTok().getIntVal();
1664     if (LocalLabelVal < 0) {
1665       if (!TheCondState.Ignore) {
1666         Lex(); // always eat a token
1667         return Error(IDLoc, "unexpected token at start of statement");
1668       }
1669       IDVal = "";
1670     } else {
1671       IDVal = getTok().getString();
1672       Lex(); // Consume the integer token to be used as an identifier token.
1673       if (Lexer.getKind() != AsmToken::Colon) {
1674         if (!TheCondState.Ignore) {
1675           Lex(); // always eat a token
1676           return Error(IDLoc, "unexpected token at start of statement");
1677         }
1678       }
1679     }
1680   } else if (Lexer.is(AsmToken::Dot)) {
1681     // Treat '.' as a valid identifier in this context.
1682     Lex();
1683     IDVal = ".";
1684   } else if (Lexer.is(AsmToken::LCurly)) {
1685     // Treat '{' as a valid identifier in this context.
1686     Lex();
1687     IDVal = "{";
1688 
1689   } else if (Lexer.is(AsmToken::RCurly)) {
1690     // Treat '}' as a valid identifier in this context.
1691     Lex();
1692     IDVal = "}";
1693   } else if (Lexer.is(AsmToken::Star) &&
1694              getTargetParser().starIsStartOfStatement()) {
1695     // Accept '*' as a valid start of statement.
1696     Lex();
1697     IDVal = "*";
1698   } else if (parseIdentifier(IDVal)) {
1699     if (!TheCondState.Ignore) {
1700       Lex(); // always eat a token
1701       return Error(IDLoc, "unexpected token at start of statement");
1702     }
1703     IDVal = "";
1704   }
1705 
1706   // Handle conditional assembly here before checking for skipping.  We
1707   // have to do this so that .endif isn't skipped in a ".if 0" block for
1708   // example.
1709   StringMap<DirectiveKind>::const_iterator DirKindIt =
1710       DirectiveKindMap.find(IDVal);
1711   DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end())
1712                               ? DK_NO_DIRECTIVE
1713                               : DirKindIt->getValue();
1714   switch (DirKind) {
1715   default:
1716     break;
1717   case DK_IF:
1718   case DK_IFEQ:
1719   case DK_IFGE:
1720   case DK_IFGT:
1721   case DK_IFLE:
1722   case DK_IFLT:
1723   case DK_IFNE:
1724     return parseDirectiveIf(IDLoc, DirKind);
1725   case DK_IFB:
1726     return parseDirectiveIfb(IDLoc, true);
1727   case DK_IFNB:
1728     return parseDirectiveIfb(IDLoc, false);
1729   case DK_IFC:
1730     return parseDirectiveIfc(IDLoc, true);
1731   case DK_IFEQS:
1732     return parseDirectiveIfeqs(IDLoc, true);
1733   case DK_IFNC:
1734     return parseDirectiveIfc(IDLoc, false);
1735   case DK_IFNES:
1736     return parseDirectiveIfeqs(IDLoc, false);
1737   case DK_IFDEF:
1738     return parseDirectiveIfdef(IDLoc, true);
1739   case DK_IFNDEF:
1740   case DK_IFNOTDEF:
1741     return parseDirectiveIfdef(IDLoc, false);
1742   case DK_ELSEIF:
1743     return parseDirectiveElseIf(IDLoc);
1744   case DK_ELSE:
1745     return parseDirectiveElse(IDLoc);
1746   case DK_ENDIF:
1747     return parseDirectiveEndIf(IDLoc);
1748   }
1749 
1750   // Ignore the statement if in the middle of inactive conditional
1751   // (e.g. ".if 0").
1752   if (TheCondState.Ignore) {
1753     eatToEndOfStatement();
1754     return false;
1755   }
1756 
1757   // FIXME: Recurse on local labels?
1758 
1759   // See what kind of statement we have.
1760   switch (Lexer.getKind()) {
1761   case AsmToken::Colon: {
1762     if (!getTargetParser().isLabel(ID))
1763       break;
1764     if (checkForValidSection())
1765       return true;
1766 
1767     // identifier ':'   -> Label.
1768     Lex();
1769 
1770     // Diagnose attempt to use '.' as a label.
1771     if (IDVal == ".")
1772       return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label");
1773 
1774     // Diagnose attempt to use a variable as a label.
1775     //
1776     // FIXME: Diagnostics. Note the location of the definition as a label.
1777     // FIXME: This doesn't diagnose assignment to a symbol which has been
1778     // implicitly marked as external.
1779     MCSymbol *Sym;
1780     if (LocalLabelVal == -1) {
1781       if (ParsingInlineAsm && SI) {
1782         StringRef RewrittenLabel =
1783             SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true);
1784         assert(!RewrittenLabel.empty() &&
1785                "We should have an internal name here.");
1786         Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(),
1787                                        RewrittenLabel);
1788         IDVal = RewrittenLabel;
1789       }
1790       Sym = getContext().getOrCreateSymbol(IDVal);
1791     } else
1792       Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal);
1793     // End of Labels should be treated as end of line for lexing
1794     // purposes but that information is not available to the Lexer who
1795     // does not understand Labels. This may cause us to see a Hash
1796     // here instead of a preprocessor line comment.
1797     if (getTok().is(AsmToken::Hash)) {
1798       StringRef CommentStr = parseStringToEndOfStatement();
1799       Lexer.Lex();
1800       Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr));
1801     }
1802 
1803     // Consume any end of statement token, if present, to avoid spurious
1804     // AddBlankLine calls().
1805     if (getTok().is(AsmToken::EndOfStatement)) {
1806       Lex();
1807     }
1808 
1809     // Emit the label.
1810     if (!getTargetParser().isParsingInlineAsm())
1811       Out.EmitLabel(Sym, IDLoc);
1812 
1813     // If we are generating dwarf for assembly source files then gather the
1814     // info to make a dwarf label entry for this label if needed.
1815     if (enabledGenDwarfForAssembly())
1816       MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(),
1817                                  IDLoc);
1818 
1819     getTargetParser().onLabelParsed(Sym);
1820 
1821     return false;
1822   }
1823 
1824   case AsmToken::Equal:
1825     if (!getTargetParser().equalIsAsmAssignment())
1826       break;
1827     // identifier '=' ... -> assignment statement
1828     Lex();
1829 
1830     return parseAssignment(IDVal, true);
1831 
1832   default: // Normal instruction or directive.
1833     break;
1834   }
1835 
1836   // If macros are enabled, check to see if this is a macro instantiation.
1837   if (areMacrosEnabled())
1838     if (const MCAsmMacro *M = getContext().lookupMacro(IDVal)) {
1839       return handleMacroEntry(M, IDLoc);
1840     }
1841 
1842   // Otherwise, we have a normal instruction or directive.
1843 
1844   // Directives start with "."
1845   if (IDVal[0] == '.' && IDVal != ".") {
1846     // There are several entities interested in parsing directives:
1847     //
1848     // 1. The target-specific assembly parser. Some directives are target
1849     //    specific or may potentially behave differently on certain targets.
1850     // 2. Asm parser extensions. For example, platform-specific parsers
1851     //    (like the ELF parser) register themselves as extensions.
1852     // 3. The generic directive parser implemented by this class. These are
1853     //    all the directives that behave in a target and platform independent
1854     //    manner, or at least have a default behavior that's shared between
1855     //    all targets and platforms.
1856 
1857     getTargetParser().flushPendingInstructions(getStreamer());
1858 
1859     SMLoc StartTokLoc = getTok().getLoc();
1860     bool TPDirectiveReturn = getTargetParser().ParseDirective(ID);
1861 
1862     if (hasPendingError())
1863       return true;
1864     // Currently the return value should be true if we are
1865     // uninterested but as this is at odds with the standard parsing
1866     // convention (return true = error) we have instances of a parsed
1867     // directive that fails returning true as an error. Catch these
1868     // cases as best as possible errors here.
1869     if (TPDirectiveReturn && StartTokLoc != getTok().getLoc())
1870       return true;
1871     // Return if we did some parsing or believe we succeeded.
1872     if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc())
1873       return false;
1874 
1875     // Next, check the extension directive map to see if any extension has
1876     // registered itself to parse this directive.
1877     std::pair<MCAsmParserExtension *, DirectiveHandler> Handler =
1878         ExtensionDirectiveMap.lookup(IDVal);
1879     if (Handler.first)
1880       return (*Handler.second)(Handler.first, IDVal, IDLoc);
1881 
1882     // Finally, if no one else is interested in this directive, it must be
1883     // generic and familiar to this class.
1884     switch (DirKind) {
1885     default:
1886       break;
1887     case DK_SET:
1888     case DK_EQU:
1889       return parseDirectiveSet(IDVal, true);
1890     case DK_EQUIV:
1891       return parseDirectiveSet(IDVal, false);
1892     case DK_ASCII:
1893       return parseDirectiveAscii(IDVal, false);
1894     case DK_ASCIZ:
1895     case DK_STRING:
1896       return parseDirectiveAscii(IDVal, true);
1897     case DK_BYTE:
1898     case DK_DC_B:
1899       return parseDirectiveValue(IDVal, 1);
1900     case DK_DC:
1901     case DK_DC_W:
1902     case DK_SHORT:
1903     case DK_VALUE:
1904     case DK_2BYTE:
1905       return parseDirectiveValue(IDVal, 2);
1906     case DK_LONG:
1907     case DK_INT:
1908     case DK_4BYTE:
1909     case DK_DC_L:
1910       return parseDirectiveValue(IDVal, 4);
1911     case DK_QUAD:
1912     case DK_8BYTE:
1913       return parseDirectiveValue(IDVal, 8);
1914     case DK_DC_A:
1915       return parseDirectiveValue(
1916           IDVal, getContext().getAsmInfo()->getCodePointerSize());
1917     case DK_OCTA:
1918       return parseDirectiveOctaValue(IDVal);
1919     case DK_SINGLE:
1920     case DK_FLOAT:
1921     case DK_DC_S:
1922       return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle());
1923     case DK_DOUBLE:
1924     case DK_DC_D:
1925       return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble());
1926     case DK_ALIGN: {
1927       bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes();
1928       return parseDirectiveAlign(IsPow2, /*ExprSize=*/1);
1929     }
1930     case DK_ALIGN32: {
1931       bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes();
1932       return parseDirectiveAlign(IsPow2, /*ExprSize=*/4);
1933     }
1934     case DK_BALIGN:
1935       return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1);
1936     case DK_BALIGNW:
1937       return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2);
1938     case DK_BALIGNL:
1939       return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4);
1940     case DK_P2ALIGN:
1941       return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1);
1942     case DK_P2ALIGNW:
1943       return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2);
1944     case DK_P2ALIGNL:
1945       return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4);
1946     case DK_ORG:
1947       return parseDirectiveOrg();
1948     case DK_FILL:
1949       return parseDirectiveFill();
1950     case DK_ZERO:
1951       return parseDirectiveZero();
1952     case DK_EXTERN:
1953       eatToEndOfStatement(); // .extern is the default, ignore it.
1954       return false;
1955     case DK_GLOBL:
1956     case DK_GLOBAL:
1957       return parseDirectiveSymbolAttribute(MCSA_Global);
1958     case DK_LAZY_REFERENCE:
1959       return parseDirectiveSymbolAttribute(MCSA_LazyReference);
1960     case DK_NO_DEAD_STRIP:
1961       return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip);
1962     case DK_SYMBOL_RESOLVER:
1963       return parseDirectiveSymbolAttribute(MCSA_SymbolResolver);
1964     case DK_PRIVATE_EXTERN:
1965       return parseDirectiveSymbolAttribute(MCSA_PrivateExtern);
1966     case DK_REFERENCE:
1967       return parseDirectiveSymbolAttribute(MCSA_Reference);
1968     case DK_WEAK_DEFINITION:
1969       return parseDirectiveSymbolAttribute(MCSA_WeakDefinition);
1970     case DK_WEAK_REFERENCE:
1971       return parseDirectiveSymbolAttribute(MCSA_WeakReference);
1972     case DK_WEAK_DEF_CAN_BE_HIDDEN:
1973       return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate);
1974     case DK_COMM:
1975     case DK_COMMON:
1976       return parseDirectiveComm(/*IsLocal=*/false);
1977     case DK_LCOMM:
1978       return parseDirectiveComm(/*IsLocal=*/true);
1979     case DK_ABORT:
1980       return parseDirectiveAbort();
1981     case DK_INCLUDE:
1982       return parseDirectiveInclude();
1983     case DK_INCBIN:
1984       return parseDirectiveIncbin();
1985     case DK_CODE16:
1986     case DK_CODE16GCC:
1987       return TokError(Twine(IDVal) +
1988                       " not currently supported for this target");
1989     case DK_REPT:
1990       return parseDirectiveRept(IDLoc, IDVal);
1991     case DK_IRP:
1992       return parseDirectiveIrp(IDLoc);
1993     case DK_IRPC:
1994       return parseDirectiveIrpc(IDLoc);
1995     case DK_ENDR:
1996       return parseDirectiveEndr(IDLoc);
1997     case DK_BUNDLE_ALIGN_MODE:
1998       return parseDirectiveBundleAlignMode();
1999     case DK_BUNDLE_LOCK:
2000       return parseDirectiveBundleLock();
2001     case DK_BUNDLE_UNLOCK:
2002       return parseDirectiveBundleUnlock();
2003     case DK_SLEB128:
2004       return parseDirectiveLEB128(true);
2005     case DK_ULEB128:
2006       return parseDirectiveLEB128(false);
2007     case DK_SPACE:
2008     case DK_SKIP:
2009       return parseDirectiveSpace(IDVal);
2010     case DK_FILE:
2011       return parseDirectiveFile(IDLoc);
2012     case DK_LINE:
2013       return parseDirectiveLine();
2014     case DK_LOC:
2015       return parseDirectiveLoc();
2016     case DK_STABS:
2017       return parseDirectiveStabs();
2018     case DK_CV_FILE:
2019       return parseDirectiveCVFile();
2020     case DK_CV_FUNC_ID:
2021       return parseDirectiveCVFuncId();
2022     case DK_CV_INLINE_SITE_ID:
2023       return parseDirectiveCVInlineSiteId();
2024     case DK_CV_LOC:
2025       return parseDirectiveCVLoc();
2026     case DK_CV_LINETABLE:
2027       return parseDirectiveCVLinetable();
2028     case DK_CV_INLINE_LINETABLE:
2029       return parseDirectiveCVInlineLinetable();
2030     case DK_CV_DEF_RANGE:
2031       return parseDirectiveCVDefRange();
2032     case DK_CV_STRINGTABLE:
2033       return parseDirectiveCVStringTable();
2034     case DK_CV_FILECHECKSUMS:
2035       return parseDirectiveCVFileChecksums();
2036     case DK_CV_FILECHECKSUM_OFFSET:
2037       return parseDirectiveCVFileChecksumOffset();
2038     case DK_CV_FPO_DATA:
2039       return parseDirectiveCVFPOData();
2040     case DK_CFI_SECTIONS:
2041       return parseDirectiveCFISections();
2042     case DK_CFI_STARTPROC:
2043       return parseDirectiveCFIStartProc();
2044     case DK_CFI_ENDPROC:
2045       return parseDirectiveCFIEndProc();
2046     case DK_CFI_DEF_CFA:
2047       return parseDirectiveCFIDefCfa(IDLoc);
2048     case DK_CFI_DEF_CFA_OFFSET:
2049       return parseDirectiveCFIDefCfaOffset();
2050     case DK_CFI_ADJUST_CFA_OFFSET:
2051       return parseDirectiveCFIAdjustCfaOffset();
2052     case DK_CFI_DEF_CFA_REGISTER:
2053       return parseDirectiveCFIDefCfaRegister(IDLoc);
2054     case DK_CFI_OFFSET:
2055       return parseDirectiveCFIOffset(IDLoc);
2056     case DK_CFI_REL_OFFSET:
2057       return parseDirectiveCFIRelOffset(IDLoc);
2058     case DK_CFI_PERSONALITY:
2059       return parseDirectiveCFIPersonalityOrLsda(true);
2060     case DK_CFI_LSDA:
2061       return parseDirectiveCFIPersonalityOrLsda(false);
2062     case DK_CFI_REMEMBER_STATE:
2063       return parseDirectiveCFIRememberState();
2064     case DK_CFI_RESTORE_STATE:
2065       return parseDirectiveCFIRestoreState();
2066     case DK_CFI_SAME_VALUE:
2067       return parseDirectiveCFISameValue(IDLoc);
2068     case DK_CFI_RESTORE:
2069       return parseDirectiveCFIRestore(IDLoc);
2070     case DK_CFI_ESCAPE:
2071       return parseDirectiveCFIEscape();
2072     case DK_CFI_RETURN_COLUMN:
2073       return parseDirectiveCFIReturnColumn(IDLoc);
2074     case DK_CFI_SIGNAL_FRAME:
2075       return parseDirectiveCFISignalFrame();
2076     case DK_CFI_UNDEFINED:
2077       return parseDirectiveCFIUndefined(IDLoc);
2078     case DK_CFI_REGISTER:
2079       return parseDirectiveCFIRegister(IDLoc);
2080     case DK_CFI_WINDOW_SAVE:
2081       return parseDirectiveCFIWindowSave();
2082     case DK_MACROS_ON:
2083     case DK_MACROS_OFF:
2084       return parseDirectiveMacrosOnOff(IDVal);
2085     case DK_MACRO:
2086       return parseDirectiveMacro(IDLoc);
2087     case DK_ALTMACRO:
2088     case DK_NOALTMACRO:
2089       return parseDirectiveAltmacro(IDVal);
2090     case DK_EXITM:
2091       return parseDirectiveExitMacro(IDVal);
2092     case DK_ENDM:
2093     case DK_ENDMACRO:
2094       return parseDirectiveEndMacro(IDVal);
2095     case DK_PURGEM:
2096       return parseDirectivePurgeMacro(IDLoc);
2097     case DK_END:
2098       return parseDirectiveEnd(IDLoc);
2099     case DK_ERR:
2100       return parseDirectiveError(IDLoc, false);
2101     case DK_ERROR:
2102       return parseDirectiveError(IDLoc, true);
2103     case DK_WARNING:
2104       return parseDirectiveWarning(IDLoc);
2105     case DK_RELOC:
2106       return parseDirectiveReloc(IDLoc);
2107     case DK_DCB:
2108     case DK_DCB_W:
2109       return parseDirectiveDCB(IDVal, 2);
2110     case DK_DCB_B:
2111       return parseDirectiveDCB(IDVal, 1);
2112     case DK_DCB_D:
2113       return parseDirectiveRealDCB(IDVal, APFloat::IEEEdouble());
2114     case DK_DCB_L:
2115       return parseDirectiveDCB(IDVal, 4);
2116     case DK_DCB_S:
2117       return parseDirectiveRealDCB(IDVal, APFloat::IEEEsingle());
2118     case DK_DC_X:
2119     case DK_DCB_X:
2120       return TokError(Twine(IDVal) +
2121                       " not currently supported for this target");
2122     case DK_DS:
2123     case DK_DS_W:
2124       return parseDirectiveDS(IDVal, 2);
2125     case DK_DS_B:
2126       return parseDirectiveDS(IDVal, 1);
2127     case DK_DS_D:
2128       return parseDirectiveDS(IDVal, 8);
2129     case DK_DS_L:
2130     case DK_DS_S:
2131       return parseDirectiveDS(IDVal, 4);
2132     case DK_DS_P:
2133     case DK_DS_X:
2134       return parseDirectiveDS(IDVal, 12);
2135     case DK_PRINT:
2136       return parseDirectivePrint(IDLoc);
2137     case DK_ADDRSIG:
2138       return parseDirectiveAddrsig();
2139     case DK_ADDRSIG_SYM:
2140       return parseDirectiveAddrsigSym();
2141     }
2142 
2143     return Error(IDLoc, "unknown directive");
2144   }
2145 
2146   // __asm _emit or __asm __emit
2147   if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" ||
2148                            IDVal == "_EMIT" || IDVal == "__EMIT"))
2149     return parseDirectiveMSEmit(IDLoc, Info, IDVal.size());
2150 
2151   // __asm align
2152   if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN"))
2153     return parseDirectiveMSAlign(IDLoc, Info);
2154 
2155   if (ParsingInlineAsm && (IDVal == "even" || IDVal == "EVEN"))
2156     Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4);
2157   if (checkForValidSection())
2158     return true;
2159 
2160   // Canonicalize the opcode to lower case.
2161   std::string OpcodeStr = IDVal.lower();
2162   ParseInstructionInfo IInfo(Info.AsmRewrites);
2163   bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID,
2164                                                           Info.ParsedOperands);
2165   Info.ParseError = ParseHadError;
2166 
2167   // Dump the parsed representation, if requested.
2168   if (getShowParsedOperands()) {
2169     SmallString<256> Str;
2170     raw_svector_ostream OS(Str);
2171     OS << "parsed instruction: [";
2172     for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) {
2173       if (i != 0)
2174         OS << ", ";
2175       Info.ParsedOperands[i]->print(OS);
2176     }
2177     OS << "]";
2178 
2179     printMessage(IDLoc, SourceMgr::DK_Note, OS.str());
2180   }
2181 
2182   // Fail even if ParseInstruction erroneously returns false.
2183   if (hasPendingError() || ParseHadError)
2184     return true;
2185 
2186   // If we are generating dwarf for the current section then generate a .loc
2187   // directive for the instruction.
2188   if (!ParseHadError && enabledGenDwarfForAssembly() &&
2189       getContext().getGenDwarfSectionSyms().count(
2190           getStreamer().getCurrentSectionOnly())) {
2191     unsigned Line;
2192     if (ActiveMacros.empty())
2193       Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer);
2194     else
2195       Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc,
2196                                    ActiveMacros.front()->ExitBuffer);
2197 
2198     // If we previously parsed a cpp hash file line comment then make sure the
2199     // current Dwarf File is for the CppHashFilename if not then emit the
2200     // Dwarf File table for it and adjust the line number for the .loc.
2201     if (!CppHashInfo.Filename.empty()) {
2202       unsigned FileNumber = getStreamer().EmitDwarfFileDirective(
2203           0, StringRef(), CppHashInfo.Filename);
2204       getContext().setGenDwarfFileNumber(FileNumber);
2205 
2206       unsigned CppHashLocLineNo =
2207         SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf);
2208       Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo);
2209     }
2210 
2211     getStreamer().EmitDwarfLocDirective(
2212         getContext().getGenDwarfFileNumber(), Line, 0,
2213         DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0,
2214         StringRef());
2215   }
2216 
2217   // If parsing succeeded, match the instruction.
2218   if (!ParseHadError) {
2219     uint64_t ErrorInfo;
2220     if (getTargetParser().MatchAndEmitInstruction(
2221             IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo,
2222             getTargetParser().isParsingInlineAsm()))
2223       return true;
2224   }
2225   return false;
2226 }
2227 
2228 // Parse and erase curly braces marking block start/end
2229 bool
parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> & AsmStrRewrites)2230 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) {
2231   // Identify curly brace marking block start/end
2232   if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly))
2233     return false;
2234 
2235   SMLoc StartLoc = Lexer.getLoc();
2236   Lex(); // Eat the brace
2237   if (Lexer.is(AsmToken::EndOfStatement))
2238     Lex(); // Eat EndOfStatement following the brace
2239 
2240   // Erase the block start/end brace from the output asm string
2241   AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() -
2242                                                   StartLoc.getPointer());
2243   return true;
2244 }
2245 
2246 /// parseCppHashLineFilenameComment as this:
2247 ///   ::= # number "filename"
parseCppHashLineFilenameComment(SMLoc L)2248 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L) {
2249   Lex(); // Eat the hash token.
2250   // Lexer only ever emits HashDirective if it fully formed if it's
2251   // done the checking already so this is an internal error.
2252   assert(getTok().is(AsmToken::Integer) &&
2253          "Lexing Cpp line comment: Expected Integer");
2254   int64_t LineNumber = getTok().getIntVal();
2255   Lex();
2256   assert(getTok().is(AsmToken::String) &&
2257          "Lexing Cpp line comment: Expected String");
2258   StringRef Filename = getTok().getString();
2259   Lex();
2260 
2261   // Get rid of the enclosing quotes.
2262   Filename = Filename.substr(1, Filename.size() - 2);
2263 
2264   // Save the SMLoc, Filename and LineNumber for later use by diagnostics.
2265   CppHashInfo.Loc = L;
2266   CppHashInfo.Filename = Filename;
2267   CppHashInfo.LineNumber = LineNumber;
2268   CppHashInfo.Buf = CurBuffer;
2269   return false;
2270 }
2271 
2272 /// will use the last parsed cpp hash line filename comment
2273 /// for the Filename and LineNo if any in the diagnostic.
DiagHandler(const SMDiagnostic & Diag,void * Context)2274 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) {
2275   const AsmParser *Parser = static_cast<const AsmParser *>(Context);
2276   raw_ostream &OS = errs();
2277 
2278   const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr();
2279   SMLoc DiagLoc = Diag.getLoc();
2280   unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc);
2281   unsigned CppHashBuf =
2282       Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc);
2283 
2284   // Like SourceMgr::printMessage() we need to print the include stack if any
2285   // before printing the message.
2286   unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc);
2287   if (!Parser->SavedDiagHandler && DiagCurBuffer &&
2288       DiagCurBuffer != DiagSrcMgr.getMainFileID()) {
2289     SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer);
2290     DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS);
2291   }
2292 
2293   // If we have not parsed a cpp hash line filename comment or the source
2294   // manager changed or buffer changed (like in a nested include) then just
2295   // print the normal diagnostic using its Filename and LineNo.
2296   if (!Parser->CppHashInfo.LineNumber || &DiagSrcMgr != &Parser->SrcMgr ||
2297       DiagBuf != CppHashBuf) {
2298     if (Parser->SavedDiagHandler)
2299       Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext);
2300     else
2301       Diag.print(nullptr, OS);
2302     return;
2303   }
2304 
2305   // Use the CppHashFilename and calculate a line number based on the
2306   // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc
2307   // for the diagnostic.
2308   const std::string &Filename = Parser->CppHashInfo.Filename;
2309 
2310   int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf);
2311   int CppHashLocLineNo =
2312       Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf);
2313   int LineNo =
2314       Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo);
2315 
2316   SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo,
2317                        Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(),
2318                        Diag.getLineContents(), Diag.getRanges());
2319 
2320   if (Parser->SavedDiagHandler)
2321     Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext);
2322   else
2323     NewDiag.print(nullptr, OS);
2324 }
2325 
2326 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The
2327 // difference being that that function accepts '@' as part of identifiers and
2328 // we can't do that. AsmLexer.cpp should probably be changed to handle
2329 // '@' as a special case when needed.
isIdentifierChar(char c)2330 static bool isIdentifierChar(char c) {
2331   return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' ||
2332          c == '.';
2333 }
2334 
expandMacro(raw_svector_ostream & OS,StringRef Body,ArrayRef<MCAsmMacroParameter> Parameters,ArrayRef<MCAsmMacroArgument> A,bool EnableAtPseudoVariable,SMLoc L)2335 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body,
2336                             ArrayRef<MCAsmMacroParameter> Parameters,
2337                             ArrayRef<MCAsmMacroArgument> A,
2338                             bool EnableAtPseudoVariable, SMLoc L) {
2339   unsigned NParameters = Parameters.size();
2340   bool HasVararg = NParameters ? Parameters.back().Vararg : false;
2341   if ((!IsDarwin || NParameters != 0) && NParameters != A.size())
2342     return Error(L, "Wrong number of arguments");
2343 
2344   // A macro without parameters is handled differently on Darwin:
2345   // gas accepts no arguments and does no substitutions
2346   while (!Body.empty()) {
2347     // Scan for the next substitution.
2348     std::size_t End = Body.size(), Pos = 0;
2349     for (; Pos != End; ++Pos) {
2350       // Check for a substitution or escape.
2351       if (IsDarwin && !NParameters) {
2352         // This macro has no parameters, look for $0, $1, etc.
2353         if (Body[Pos] != '$' || Pos + 1 == End)
2354           continue;
2355 
2356         char Next = Body[Pos + 1];
2357         if (Next == '$' || Next == 'n' ||
2358             isdigit(static_cast<unsigned char>(Next)))
2359           break;
2360       } else {
2361         // This macro has parameters, look for \foo, \bar, etc.
2362         if (Body[Pos] == '\\' && Pos + 1 != End)
2363           break;
2364       }
2365     }
2366 
2367     // Add the prefix.
2368     OS << Body.slice(0, Pos);
2369 
2370     // Check if we reached the end.
2371     if (Pos == End)
2372       break;
2373 
2374     if (IsDarwin && !NParameters) {
2375       switch (Body[Pos + 1]) {
2376       // $$ => $
2377       case '$':
2378         OS << '$';
2379         break;
2380 
2381       // $n => number of arguments
2382       case 'n':
2383         OS << A.size();
2384         break;
2385 
2386       // $[0-9] => argument
2387       default: {
2388         // Missing arguments are ignored.
2389         unsigned Index = Body[Pos + 1] - '0';
2390         if (Index >= A.size())
2391           break;
2392 
2393         // Otherwise substitute with the token values, with spaces eliminated.
2394         for (const AsmToken &Token : A[Index])
2395           OS << Token.getString();
2396         break;
2397       }
2398       }
2399       Pos += 2;
2400     } else {
2401       unsigned I = Pos + 1;
2402 
2403       // Check for the \@ pseudo-variable.
2404       if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End)
2405         ++I;
2406       else
2407         while (isIdentifierChar(Body[I]) && I + 1 != End)
2408           ++I;
2409 
2410       const char *Begin = Body.data() + Pos + 1;
2411       StringRef Argument(Begin, I - (Pos + 1));
2412       unsigned Index = 0;
2413 
2414       if (Argument == "@") {
2415         OS << NumOfMacroInstantiations;
2416         Pos += 2;
2417       } else {
2418         for (; Index < NParameters; ++Index)
2419           if (Parameters[Index].Name == Argument)
2420             break;
2421 
2422         if (Index == NParameters) {
2423           if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')')
2424             Pos += 3;
2425           else {
2426             OS << '\\' << Argument;
2427             Pos = I;
2428           }
2429         } else {
2430           bool VarargParameter = HasVararg && Index == (NParameters - 1);
2431           for (const AsmToken &Token : A[Index])
2432             // For altmacro mode, you can write '%expr'.
2433             // The prefix '%' evaluates the expression 'expr'
2434             // and uses the result as a string (e.g. replace %(1+2) with the string "3").
2435             // Here, we identify the integer token which is the result of the
2436             // absolute expression evaluation and replace it with its string representation.
2437             if ((Lexer.IsaAltMacroMode()) &&
2438                  (*(Token.getString().begin()) == '%') && Token.is(AsmToken::Integer))
2439               // Emit an integer value to the buffer.
2440               OS << Token.getIntVal();
2441             // Only Token that was validated as a string and begins with '<'
2442             // is considered altMacroString!!!
2443             else if ((Lexer.IsaAltMacroMode()) &&
2444                      (*(Token.getString().begin()) == '<') &&
2445                      Token.is(AsmToken::String)) {
2446               std::string Res;
2447               altMacroString(Token.getStringContents(), Res);
2448               OS << Res;
2449             }
2450             // We expect no quotes around the string's contents when
2451             // parsing for varargs.
2452             else if (Token.isNot(AsmToken::String) || VarargParameter)
2453               OS << Token.getString();
2454             else
2455               OS << Token.getStringContents();
2456 
2457           Pos += 1 + Argument.size();
2458         }
2459       }
2460     }
2461     // Update the scan point.
2462     Body = Body.substr(Pos);
2463   }
2464 
2465   return false;
2466 }
2467 
MacroInstantiation(SMLoc IL,int EB,SMLoc EL,size_t CondStackDepth)2468 MacroInstantiation::MacroInstantiation(SMLoc IL, int EB, SMLoc EL,
2469                                        size_t CondStackDepth)
2470     : InstantiationLoc(IL), ExitBuffer(EB), ExitLoc(EL),
2471       CondStackDepth(CondStackDepth) {}
2472 
isOperator(AsmToken::TokenKind kind)2473 static bool isOperator(AsmToken::TokenKind kind) {
2474   switch (kind) {
2475   default:
2476     return false;
2477   case AsmToken::Plus:
2478   case AsmToken::Minus:
2479   case AsmToken::Tilde:
2480   case AsmToken::Slash:
2481   case AsmToken::Star:
2482   case AsmToken::Dot:
2483   case AsmToken::Equal:
2484   case AsmToken::EqualEqual:
2485   case AsmToken::Pipe:
2486   case AsmToken::PipePipe:
2487   case AsmToken::Caret:
2488   case AsmToken::Amp:
2489   case AsmToken::AmpAmp:
2490   case AsmToken::Exclaim:
2491   case AsmToken::ExclaimEqual:
2492   case AsmToken::Less:
2493   case AsmToken::LessEqual:
2494   case AsmToken::LessLess:
2495   case AsmToken::LessGreater:
2496   case AsmToken::Greater:
2497   case AsmToken::GreaterEqual:
2498   case AsmToken::GreaterGreater:
2499     return true;
2500   }
2501 }
2502 
2503 namespace {
2504 
2505 class AsmLexerSkipSpaceRAII {
2506 public:
AsmLexerSkipSpaceRAII(AsmLexer & Lexer,bool SkipSpace)2507   AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) {
2508     Lexer.setSkipSpace(SkipSpace);
2509   }
2510 
~AsmLexerSkipSpaceRAII()2511   ~AsmLexerSkipSpaceRAII() {
2512     Lexer.setSkipSpace(true);
2513   }
2514 
2515 private:
2516   AsmLexer &Lexer;
2517 };
2518 
2519 } // end anonymous namespace
2520 
parseMacroArgument(MCAsmMacroArgument & MA,bool Vararg)2521 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) {
2522 
2523   if (Vararg) {
2524     if (Lexer.isNot(AsmToken::EndOfStatement)) {
2525       StringRef Str = parseStringToEndOfStatement();
2526       MA.emplace_back(AsmToken::String, Str);
2527     }
2528     return false;
2529   }
2530 
2531   unsigned ParenLevel = 0;
2532 
2533   // Darwin doesn't use spaces to delmit arguments.
2534   AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin);
2535 
2536   bool SpaceEaten;
2537 
2538   while (true) {
2539     SpaceEaten = false;
2540     if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal))
2541       return TokError("unexpected token in macro instantiation");
2542 
2543     if (ParenLevel == 0) {
2544 
2545       if (Lexer.is(AsmToken::Comma))
2546         break;
2547 
2548       if (Lexer.is(AsmToken::Space)) {
2549         SpaceEaten = true;
2550         Lexer.Lex(); // Eat spaces
2551       }
2552 
2553       // Spaces can delimit parameters, but could also be part an expression.
2554       // If the token after a space is an operator, add the token and the next
2555       // one into this argument
2556       if (!IsDarwin) {
2557         if (isOperator(Lexer.getKind())) {
2558           MA.push_back(getTok());
2559           Lexer.Lex();
2560 
2561           // Whitespace after an operator can be ignored.
2562           if (Lexer.is(AsmToken::Space))
2563             Lexer.Lex();
2564 
2565           continue;
2566         }
2567       }
2568       if (SpaceEaten)
2569         break;
2570     }
2571 
2572     // handleMacroEntry relies on not advancing the lexer here
2573     // to be able to fill in the remaining default parameter values
2574     if (Lexer.is(AsmToken::EndOfStatement))
2575       break;
2576 
2577     // Adjust the current parentheses level.
2578     if (Lexer.is(AsmToken::LParen))
2579       ++ParenLevel;
2580     else if (Lexer.is(AsmToken::RParen) && ParenLevel)
2581       --ParenLevel;
2582 
2583     // Append the token to the current argument list.
2584     MA.push_back(getTok());
2585     Lexer.Lex();
2586   }
2587 
2588   if (ParenLevel != 0)
2589     return TokError("unbalanced parentheses in macro argument");
2590   return false;
2591 }
2592 
2593 // Parse the macro instantiation arguments.
parseMacroArguments(const MCAsmMacro * M,MCAsmMacroArguments & A)2594 bool AsmParser::parseMacroArguments(const MCAsmMacro *M,
2595                                     MCAsmMacroArguments &A) {
2596   const unsigned NParameters = M ? M->Parameters.size() : 0;
2597   bool NamedParametersFound = false;
2598   SmallVector<SMLoc, 4> FALocs;
2599 
2600   A.resize(NParameters);
2601   FALocs.resize(NParameters);
2602 
2603   // Parse two kinds of macro invocations:
2604   // - macros defined without any parameters accept an arbitrary number of them
2605   // - macros defined with parameters accept at most that many of them
2606   bool HasVararg = NParameters ? M->Parameters.back().Vararg : false;
2607   for (unsigned Parameter = 0; !NParameters || Parameter < NParameters;
2608        ++Parameter) {
2609     SMLoc IDLoc = Lexer.getLoc();
2610     MCAsmMacroParameter FA;
2611 
2612     if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) {
2613       if (parseIdentifier(FA.Name))
2614         return Error(IDLoc, "invalid argument identifier for formal argument");
2615 
2616       if (Lexer.isNot(AsmToken::Equal))
2617         return TokError("expected '=' after formal parameter identifier");
2618 
2619       Lex();
2620 
2621       NamedParametersFound = true;
2622     }
2623     bool Vararg = HasVararg && Parameter == (NParameters - 1);
2624 
2625     if (NamedParametersFound && FA.Name.empty())
2626       return Error(IDLoc, "cannot mix positional and keyword arguments");
2627 
2628     SMLoc StrLoc = Lexer.getLoc();
2629     SMLoc EndLoc;
2630     if (Lexer.IsaAltMacroMode() && Lexer.is(AsmToken::Percent)) {
2631         const MCExpr *AbsoluteExp;
2632         int64_t Value;
2633         /// Eat '%'
2634         Lex();
2635         if (parseExpression(AbsoluteExp, EndLoc))
2636           return false;
2637         if (!AbsoluteExp->evaluateAsAbsolute(Value,
2638                                              getStreamer().getAssemblerPtr()))
2639           return Error(StrLoc, "expected absolute expression");
2640         const char *StrChar = StrLoc.getPointer();
2641         const char *EndChar = EndLoc.getPointer();
2642         AsmToken newToken(AsmToken::Integer, StringRef(StrChar , EndChar - StrChar), Value);
2643         FA.Value.push_back(newToken);
2644     } else if (Lexer.IsaAltMacroMode() && Lexer.is(AsmToken::Less) &&
2645                isAltmacroString(StrLoc, EndLoc)) {
2646         const char *StrChar = StrLoc.getPointer();
2647         const char *EndChar = EndLoc.getPointer();
2648         jumpToLoc(EndLoc, CurBuffer);
2649         /// Eat from '<' to '>'
2650         Lex();
2651         AsmToken newToken(AsmToken::String, StringRef(StrChar, EndChar - StrChar));
2652         FA.Value.push_back(newToken);
2653     } else if(parseMacroArgument(FA.Value, Vararg))
2654         return true;
2655 
2656     unsigned PI = Parameter;
2657     if (!FA.Name.empty()) {
2658       unsigned FAI = 0;
2659       for (FAI = 0; FAI < NParameters; ++FAI)
2660         if (M->Parameters[FAI].Name == FA.Name)
2661           break;
2662 
2663       if (FAI >= NParameters) {
2664         assert(M && "expected macro to be defined");
2665         return Error(IDLoc, "parameter named '" + FA.Name +
2666                                 "' does not exist for macro '" + M->Name + "'");
2667       }
2668       PI = FAI;
2669     }
2670 
2671     if (!FA.Value.empty()) {
2672       if (A.size() <= PI)
2673         A.resize(PI + 1);
2674       A[PI] = FA.Value;
2675 
2676       if (FALocs.size() <= PI)
2677         FALocs.resize(PI + 1);
2678 
2679       FALocs[PI] = Lexer.getLoc();
2680     }
2681 
2682     // At the end of the statement, fill in remaining arguments that have
2683     // default values. If there aren't any, then the next argument is
2684     // required but missing
2685     if (Lexer.is(AsmToken::EndOfStatement)) {
2686       bool Failure = false;
2687       for (unsigned FAI = 0; FAI < NParameters; ++FAI) {
2688         if (A[FAI].empty()) {
2689           if (M->Parameters[FAI].Required) {
2690             Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(),
2691                   "missing value for required parameter "
2692                   "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'");
2693             Failure = true;
2694           }
2695 
2696           if (!M->Parameters[FAI].Value.empty())
2697             A[FAI] = M->Parameters[FAI].Value;
2698         }
2699       }
2700       return Failure;
2701     }
2702 
2703     if (Lexer.is(AsmToken::Comma))
2704       Lex();
2705   }
2706 
2707   return TokError("too many positional arguments");
2708 }
2709 
handleMacroEntry(const MCAsmMacro * M,SMLoc NameLoc)2710 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) {
2711   // Arbitrarily limit macro nesting depth (default matches 'as'). We can
2712   // eliminate this, although we should protect against infinite loops.
2713   unsigned MaxNestingDepth = AsmMacroMaxNestingDepth;
2714   if (ActiveMacros.size() == MaxNestingDepth) {
2715     std::ostringstream MaxNestingDepthError;
2716     MaxNestingDepthError << "macros cannot be nested more than "
2717                          << MaxNestingDepth << " levels deep."
2718                          << " Use -asm-macro-max-nesting-depth to increase "
2719                             "this limit.";
2720     return TokError(MaxNestingDepthError.str());
2721   }
2722 
2723   MCAsmMacroArguments A;
2724   if (parseMacroArguments(M, A))
2725     return true;
2726 
2727   // Macro instantiation is lexical, unfortunately. We construct a new buffer
2728   // to hold the macro body with substitutions.
2729   SmallString<256> Buf;
2730   StringRef Body = M->Body;
2731   raw_svector_ostream OS(Buf);
2732 
2733   if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc()))
2734     return true;
2735 
2736   // We include the .endmacro in the buffer as our cue to exit the macro
2737   // instantiation.
2738   OS << ".endmacro\n";
2739 
2740   std::unique_ptr<MemoryBuffer> Instantiation =
2741       MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>");
2742 
2743   // Create the macro instantiation object and add to the current macro
2744   // instantiation stack.
2745   MacroInstantiation *MI = new MacroInstantiation(
2746       NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size());
2747   ActiveMacros.push_back(MI);
2748 
2749   ++NumOfMacroInstantiations;
2750 
2751   // Jump to the macro instantiation and prime the lexer.
2752   CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc());
2753   Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer());
2754   Lex();
2755 
2756   return false;
2757 }
2758 
handleMacroExit()2759 void AsmParser::handleMacroExit() {
2760   // Jump to the EndOfStatement we should return to, and consume it.
2761   jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer);
2762   Lex();
2763 
2764   // Pop the instantiation entry.
2765   delete ActiveMacros.back();
2766   ActiveMacros.pop_back();
2767 }
2768 
parseAssignment(StringRef Name,bool allow_redef,bool NoDeadStrip)2769 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef,
2770                                 bool NoDeadStrip) {
2771   MCSymbol *Sym;
2772   const MCExpr *Value;
2773   if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym,
2774                                                Value))
2775     return true;
2776 
2777   if (!Sym) {
2778     // In the case where we parse an expression starting with a '.', we will
2779     // not generate an error, nor will we create a symbol.  In this case we
2780     // should just return out.
2781     return false;
2782   }
2783 
2784   // Do the assignment.
2785   Out.EmitAssignment(Sym, Value);
2786   if (NoDeadStrip)
2787     Out.EmitSymbolAttribute(Sym, MCSA_NoDeadStrip);
2788 
2789   return false;
2790 }
2791 
2792 /// parseIdentifier:
2793 ///   ::= identifier
2794 ///   ::= string
parseIdentifier(StringRef & Res)2795 bool AsmParser::parseIdentifier(StringRef &Res) {
2796   // The assembler has relaxed rules for accepting identifiers, in particular we
2797   // allow things like '.globl $foo' and '.def @feat.00', which would normally be
2798   // separate tokens. At this level, we have already lexed so we cannot (currently)
2799   // handle this as a context dependent token, instead we detect adjacent tokens
2800   // and return the combined identifier.
2801   if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) {
2802     SMLoc PrefixLoc = getLexer().getLoc();
2803 
2804     // Consume the prefix character, and check for a following identifier.
2805 
2806     AsmToken Buf[1];
2807     Lexer.peekTokens(Buf, false);
2808 
2809     if (Buf[0].isNot(AsmToken::Identifier))
2810       return true;
2811 
2812     // We have a '$' or '@' followed by an identifier, make sure they are adjacent.
2813     if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer())
2814       return true;
2815 
2816     // eat $ or @
2817     Lexer.Lex(); // Lexer's Lex guarantees consecutive token.
2818     // Construct the joined identifier and consume the token.
2819     Res =
2820         StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1);
2821     Lex(); // Parser Lex to maintain invariants.
2822     return false;
2823   }
2824 
2825   if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String))
2826     return true;
2827 
2828   Res = getTok().getIdentifier();
2829 
2830   Lex(); // Consume the identifier token.
2831 
2832   return false;
2833 }
2834 
2835 /// parseDirectiveSet:
2836 ///   ::= .equ identifier ',' expression
2837 ///   ::= .equiv identifier ',' expression
2838 ///   ::= .set identifier ',' expression
parseDirectiveSet(StringRef IDVal,bool allow_redef)2839 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) {
2840   StringRef Name;
2841   if (check(parseIdentifier(Name), "expected identifier") ||
2842       parseToken(AsmToken::Comma) || parseAssignment(Name, allow_redef, true))
2843     return addErrorSuffix(" in '" + Twine(IDVal) + "' directive");
2844   return false;
2845 }
2846 
parseEscapedString(std::string & Data)2847 bool AsmParser::parseEscapedString(std::string &Data) {
2848   if (check(getTok().isNot(AsmToken::String), "expected string"))
2849     return true;
2850 
2851   Data = "";
2852   StringRef Str = getTok().getStringContents();
2853   for (unsigned i = 0, e = Str.size(); i != e; ++i) {
2854     if (Str[i] != '\\') {
2855       Data += Str[i];
2856       continue;
2857     }
2858 
2859     // Recognize escaped characters. Note that this escape semantics currently
2860     // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes.
2861     ++i;
2862     if (i == e)
2863       return TokError("unexpected backslash at end of string");
2864 
2865     // Recognize octal sequences.
2866     if ((unsigned)(Str[i] - '0') <= 7) {
2867       // Consume up to three octal characters.
2868       unsigned Value = Str[i] - '0';
2869 
2870       if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) {
2871         ++i;
2872         Value = Value * 8 + (Str[i] - '0');
2873 
2874         if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) {
2875           ++i;
2876           Value = Value * 8 + (Str[i] - '0');
2877         }
2878       }
2879 
2880       if (Value > 255)
2881         return TokError("invalid octal escape sequence (out of range)");
2882 
2883       Data += (unsigned char)Value;
2884       continue;
2885     }
2886 
2887     // Otherwise recognize individual escapes.
2888     switch (Str[i]) {
2889     default:
2890       // Just reject invalid escape sequences for now.
2891       return TokError("invalid escape sequence (unrecognized character)");
2892 
2893     case 'b': Data += '\b'; break;
2894     case 'f': Data += '\f'; break;
2895     case 'n': Data += '\n'; break;
2896     case 'r': Data += '\r'; break;
2897     case 't': Data += '\t'; break;
2898     case '"': Data += '"'; break;
2899     case '\\': Data += '\\'; break;
2900     }
2901   }
2902 
2903   Lex();
2904   return false;
2905 }
2906 
2907 /// parseDirectiveAscii:
2908 ///   ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ]
parseDirectiveAscii(StringRef IDVal,bool ZeroTerminated)2909 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) {
2910   auto parseOp = [&]() -> bool {
2911     std::string Data;
2912     if (checkForValidSection() || parseEscapedString(Data))
2913       return true;
2914     getStreamer().EmitBytes(Data);
2915     if (ZeroTerminated)
2916       getStreamer().EmitBytes(StringRef("\0", 1));
2917     return false;
2918   };
2919 
2920   if (parseMany(parseOp))
2921     return addErrorSuffix(" in '" + Twine(IDVal) + "' directive");
2922   return false;
2923 }
2924 
2925 /// parseDirectiveReloc
2926 ///  ::= .reloc expression , identifier [ , expression ]
parseDirectiveReloc(SMLoc DirectiveLoc)2927 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) {
2928   const MCExpr *Offset;
2929   const MCExpr *Expr = nullptr;
2930 
2931   SMLoc OffsetLoc = Lexer.getTok().getLoc();
2932   int64_t OffsetValue;
2933   // We can only deal with constant expressions at the moment.
2934 
2935   if (parseExpression(Offset))
2936     return true;
2937 
2938   if (check(!Offset->evaluateAsAbsolute(OffsetValue,
2939                                         getStreamer().getAssemblerPtr()),
2940             OffsetLoc, "expression is not a constant value") ||
2941       check(OffsetValue < 0, OffsetLoc, "expression is negative") ||
2942       parseToken(AsmToken::Comma, "expected comma") ||
2943       check(getTok().isNot(AsmToken::Identifier), "expected relocation name"))
2944     return true;
2945 
2946   SMLoc NameLoc = Lexer.getTok().getLoc();
2947   StringRef Name = Lexer.getTok().getIdentifier();
2948   Lex();
2949 
2950   if (Lexer.is(AsmToken::Comma)) {
2951     Lex();
2952     SMLoc ExprLoc = Lexer.getLoc();
2953     if (parseExpression(Expr))
2954       return true;
2955 
2956     MCValue Value;
2957     if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr))
2958       return Error(ExprLoc, "expression must be relocatable");
2959   }
2960 
2961   if (parseToken(AsmToken::EndOfStatement,
2962                  "unexpected token in .reloc directive"))
2963       return true;
2964 
2965   const MCTargetAsmParser &MCT = getTargetParser();
2966   const MCSubtargetInfo &STI = MCT.getSTI();
2967   if (getStreamer().EmitRelocDirective(*Offset, Name, Expr, DirectiveLoc, STI))
2968     return Error(NameLoc, "unknown relocation name");
2969 
2970   return false;
2971 }
2972 
2973 /// parseDirectiveValue
2974 ///  ::= (.byte | .short | ... ) [ expression (, expression)* ]
parseDirectiveValue(StringRef IDVal,unsigned Size)2975 bool AsmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) {
2976   auto parseOp = [&]() -> bool {
2977     const MCExpr *Value;
2978     SMLoc ExprLoc = getLexer().getLoc();
2979     if (checkForValidSection() || parseExpression(Value))
2980       return true;
2981     // Special case constant expressions to match code generator.
2982     if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) {
2983       assert(Size <= 8 && "Invalid size");
2984       uint64_t IntValue = MCE->getValue();
2985       if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue))
2986         return Error(ExprLoc, "out of range literal value");
2987       getStreamer().EmitIntValue(IntValue, Size);
2988     } else
2989       getStreamer().EmitValue(Value, Size, ExprLoc);
2990     return false;
2991   };
2992 
2993   if (parseMany(parseOp))
2994     return addErrorSuffix(" in '" + Twine(IDVal) + "' directive");
2995   return false;
2996 }
2997 
parseHexOcta(AsmParser & Asm,uint64_t & hi,uint64_t & lo)2998 static bool parseHexOcta(AsmParser &Asm, uint64_t &hi, uint64_t &lo) {
2999   if (Asm.getTok().isNot(AsmToken::Integer) &&
3000       Asm.getTok().isNot(AsmToken::BigNum))
3001     return Asm.TokError("unknown token in expression");
3002   SMLoc ExprLoc = Asm.getTok().getLoc();
3003   APInt IntValue = Asm.getTok().getAPIntVal();
3004   Asm.Lex();
3005   if (!IntValue.isIntN(128))
3006     return Asm.Error(ExprLoc, "out of range literal value");
3007   if (!IntValue.isIntN(64)) {
3008     hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue();
3009     lo = IntValue.getLoBits(64).getZExtValue();
3010   } else {
3011     hi = 0;
3012     lo = IntValue.getZExtValue();
3013   }
3014   return false;
3015 }
3016 
3017 /// ParseDirectiveOctaValue
3018 ///  ::= .octa [ hexconstant (, hexconstant)* ]
3019 
parseDirectiveOctaValue(StringRef IDVal)3020 bool AsmParser::parseDirectiveOctaValue(StringRef IDVal) {
3021   auto parseOp = [&]() -> bool {
3022     if (checkForValidSection())
3023       return true;
3024     uint64_t hi, lo;
3025     if (parseHexOcta(*this, hi, lo))
3026       return true;
3027     if (MAI.isLittleEndian()) {
3028       getStreamer().EmitIntValue(lo, 8);
3029       getStreamer().EmitIntValue(hi, 8);
3030     } else {
3031       getStreamer().EmitIntValue(hi, 8);
3032       getStreamer().EmitIntValue(lo, 8);
3033     }
3034     return false;
3035   };
3036 
3037   if (parseMany(parseOp))
3038     return addErrorSuffix(" in '" + Twine(IDVal) + "' directive");
3039   return false;
3040 }
3041 
parseRealValue(const fltSemantics & Semantics,APInt & Res)3042 bool AsmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) {
3043   // We don't truly support arithmetic on floating point expressions, so we
3044   // have to manually parse unary prefixes.
3045   bool IsNeg = false;
3046   if (getLexer().is(AsmToken::Minus)) {
3047     Lexer.Lex();
3048     IsNeg = true;
3049   } else if (getLexer().is(AsmToken::Plus))
3050     Lexer.Lex();
3051 
3052   if (Lexer.is(AsmToken::Error))
3053     return TokError(Lexer.getErr());
3054   if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) &&
3055       Lexer.isNot(AsmToken::Identifier))
3056     return TokError("unexpected token in directive");
3057 
3058   // Convert to an APFloat.
3059   APFloat Value(Semantics);
3060   StringRef IDVal = getTok().getString();
3061   if (getLexer().is(AsmToken::Identifier)) {
3062     if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf"))
3063       Value = APFloat::getInf(Semantics);
3064     else if (!IDVal.compare_lower("nan"))
3065       Value = APFloat::getNaN(Semantics, false, ~0);
3066     else
3067       return TokError("invalid floating point literal");
3068   } else if (Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) ==
3069              APFloat::opInvalidOp)
3070     return TokError("invalid floating point literal");
3071   if (IsNeg)
3072     Value.changeSign();
3073 
3074   // Consume the numeric token.
3075   Lex();
3076 
3077   Res = Value.bitcastToAPInt();
3078 
3079   return false;
3080 }
3081 
3082 /// parseDirectiveRealValue
3083 ///  ::= (.single | .double) [ expression (, expression)* ]
parseDirectiveRealValue(StringRef IDVal,const fltSemantics & Semantics)3084 bool AsmParser::parseDirectiveRealValue(StringRef IDVal,
3085                                         const fltSemantics &Semantics) {
3086   auto parseOp = [&]() -> bool {
3087     APInt AsInt;
3088     if (checkForValidSection() || parseRealValue(Semantics, AsInt))
3089       return true;
3090     getStreamer().EmitIntValue(AsInt.getLimitedValue(),
3091                                AsInt.getBitWidth() / 8);
3092     return false;
3093   };
3094 
3095   if (parseMany(parseOp))
3096     return addErrorSuffix(" in '" + Twine(IDVal) + "' directive");
3097   return false;
3098 }
3099 
3100 /// parseDirectiveZero
3101 ///  ::= .zero expression
parseDirectiveZero()3102 bool AsmParser::parseDirectiveZero() {
3103   SMLoc NumBytesLoc = Lexer.getLoc();
3104   const MCExpr *NumBytes;
3105   if (checkForValidSection() || parseExpression(NumBytes))
3106     return true;
3107 
3108   int64_t Val = 0;
3109   if (getLexer().is(AsmToken::Comma)) {
3110     Lex();
3111     if (parseAbsoluteExpression(Val))
3112       return true;
3113   }
3114 
3115   if (parseToken(AsmToken::EndOfStatement,
3116                  "unexpected token in '.zero' directive"))
3117     return true;
3118   getStreamer().emitFill(*NumBytes, Val, NumBytesLoc);
3119 
3120   return false;
3121 }
3122 
3123 /// parseDirectiveFill
3124 ///  ::= .fill expression [ , expression [ , expression ] ]
parseDirectiveFill()3125 bool AsmParser::parseDirectiveFill() {
3126   SMLoc NumValuesLoc = Lexer.getLoc();
3127   const MCExpr *NumValues;
3128   if (checkForValidSection() || parseExpression(NumValues))
3129     return true;
3130 
3131   int64_t FillSize = 1;
3132   int64_t FillExpr = 0;
3133 
3134   SMLoc SizeLoc, ExprLoc;
3135 
3136   if (parseOptionalToken(AsmToken::Comma)) {
3137     SizeLoc = getTok().getLoc();
3138     if (parseAbsoluteExpression(FillSize))
3139       return true;
3140     if (parseOptionalToken(AsmToken::Comma)) {
3141       ExprLoc = getTok().getLoc();
3142       if (parseAbsoluteExpression(FillExpr))
3143         return true;
3144     }
3145   }
3146   if (parseToken(AsmToken::EndOfStatement,
3147                  "unexpected token in '.fill' directive"))
3148     return true;
3149 
3150   if (FillSize < 0) {
3151     Warning(SizeLoc, "'.fill' directive with negative size has no effect");
3152     return false;
3153   }
3154   if (FillSize > 8) {
3155     Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8");
3156     FillSize = 8;
3157   }
3158 
3159   if (!isUInt<32>(FillExpr) && FillSize > 4)
3160     Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits");
3161 
3162   getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc);
3163 
3164   return false;
3165 }
3166 
3167 /// parseDirectiveOrg
3168 ///  ::= .org expression [ , expression ]
parseDirectiveOrg()3169 bool AsmParser::parseDirectiveOrg() {
3170   const MCExpr *Offset;
3171   SMLoc OffsetLoc = Lexer.getLoc();
3172   if (checkForValidSection() || parseExpression(Offset))
3173     return true;
3174 
3175   // Parse optional fill expression.
3176   int64_t FillExpr = 0;
3177   if (parseOptionalToken(AsmToken::Comma))
3178     if (parseAbsoluteExpression(FillExpr))
3179       return addErrorSuffix(" in '.org' directive");
3180   if (parseToken(AsmToken::EndOfStatement))
3181     return addErrorSuffix(" in '.org' directive");
3182 
3183   getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc);
3184   return false;
3185 }
3186 
3187 /// parseDirectiveAlign
3188 ///  ::= {.align, ...} expression [ , expression [ , expression ]]
parseDirectiveAlign(bool IsPow2,unsigned ValueSize)3189 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) {
3190   SMLoc AlignmentLoc = getLexer().getLoc();
3191   int64_t Alignment;
3192   SMLoc MaxBytesLoc;
3193   bool HasFillExpr = false;
3194   int64_t FillExpr = 0;
3195   int64_t MaxBytesToFill = 0;
3196 
3197   auto parseAlign = [&]() -> bool {
3198     if (parseAbsoluteExpression(Alignment))
3199       return true;
3200     if (parseOptionalToken(AsmToken::Comma)) {
3201       // The fill expression can be omitted while specifying a maximum number of
3202       // alignment bytes, e.g:
3203       //  .align 3,,4
3204       if (getTok().isNot(AsmToken::Comma)) {
3205         HasFillExpr = true;
3206         if (parseAbsoluteExpression(FillExpr))
3207           return true;
3208       }
3209       if (parseOptionalToken(AsmToken::Comma))
3210         if (parseTokenLoc(MaxBytesLoc) ||
3211             parseAbsoluteExpression(MaxBytesToFill))
3212           return true;
3213     }
3214     return parseToken(AsmToken::EndOfStatement);
3215   };
3216 
3217   if (checkForValidSection())
3218     return addErrorSuffix(" in directive");
3219   // Ignore empty '.p2align' directives for GNU-as compatibility
3220   if (IsPow2 && (ValueSize == 1) && getTok().is(AsmToken::EndOfStatement)) {
3221     Warning(AlignmentLoc, "p2align directive with no operand(s) is ignored");
3222     return parseToken(AsmToken::EndOfStatement);
3223   }
3224   if (parseAlign())
3225     return addErrorSuffix(" in directive");
3226 
3227   // Always emit an alignment here even if we thrown an error.
3228   bool ReturnVal = false;
3229 
3230   // Compute alignment in bytes.
3231   if (IsPow2) {
3232     // FIXME: Diagnose overflow.
3233     if (Alignment >= 32) {
3234       ReturnVal |= Error(AlignmentLoc, "invalid alignment value");
3235       Alignment = 31;
3236     }
3237 
3238     Alignment = 1ULL << Alignment;
3239   } else {
3240     // Reject alignments that aren't either a power of two or zero,
3241     // for gas compatibility. Alignment of zero is silently rounded
3242     // up to one.
3243     if (Alignment == 0)
3244       Alignment = 1;
3245     if (!isPowerOf2_64(Alignment))
3246       ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2");
3247   }
3248 
3249   // Diagnose non-sensical max bytes to align.
3250   if (MaxBytesLoc.isValid()) {
3251     if (MaxBytesToFill < 1) {
3252       ReturnVal |= Error(MaxBytesLoc,
3253                          "alignment directive can never be satisfied in this "
3254                          "many bytes, ignoring maximum bytes expression");
3255       MaxBytesToFill = 0;
3256     }
3257 
3258     if (MaxBytesToFill >= Alignment) {
3259       Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and "
3260                            "has no effect");
3261       MaxBytesToFill = 0;
3262     }
3263   }
3264 
3265   // Check whether we should use optimal code alignment for this .align
3266   // directive.
3267   const MCSection *Section = getStreamer().getCurrentSectionOnly();
3268   assert(Section && "must have section to emit alignment");
3269   bool UseCodeAlign = Section->UseCodeAlign();
3270   if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) &&
3271       ValueSize == 1 && UseCodeAlign) {
3272     getStreamer().EmitCodeAlignment(Alignment, MaxBytesToFill);
3273   } else {
3274     // FIXME: Target specific behavior about how the "extra" bytes are filled.
3275     getStreamer().EmitValueToAlignment(Alignment, FillExpr, ValueSize,
3276                                        MaxBytesToFill);
3277   }
3278 
3279   return ReturnVal;
3280 }
3281 
3282 /// parseDirectiveFile
3283 /// ::= .file filename
3284 /// ::= .file number [directory] filename [md5 checksum] [source source-text]
parseDirectiveFile(SMLoc DirectiveLoc)3285 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) {
3286   // FIXME: I'm not sure what this is.
3287   int64_t FileNumber = -1;
3288   if (getLexer().is(AsmToken::Integer)) {
3289     FileNumber = getTok().getIntVal();
3290     Lex();
3291 
3292     if (FileNumber < 0)
3293       return TokError("negative file number");
3294   }
3295 
3296   std::string Path;
3297 
3298   // Usually the directory and filename together, otherwise just the directory.
3299   // Allow the strings to have escaped octal character sequence.
3300   if (check(getTok().isNot(AsmToken::String),
3301             "unexpected token in '.file' directive") ||
3302       parseEscapedString(Path))
3303     return true;
3304 
3305   StringRef Directory;
3306   StringRef Filename;
3307   std::string FilenameData;
3308   if (getLexer().is(AsmToken::String)) {
3309     if (check(FileNumber == -1,
3310               "explicit path specified, but no file number") ||
3311         parseEscapedString(FilenameData))
3312       return true;
3313     Filename = FilenameData;
3314     Directory = Path;
3315   } else {
3316     Filename = Path;
3317   }
3318 
3319   uint64_t MD5Hi, MD5Lo;
3320   bool HasMD5 = false;
3321 
3322   Optional<StringRef> Source;
3323   bool HasSource = false;
3324   std::string SourceString;
3325 
3326   while (!parseOptionalToken(AsmToken::EndOfStatement)) {
3327     StringRef Keyword;
3328     if (check(getTok().isNot(AsmToken::Identifier),
3329               "unexpected token in '.file' directive") ||
3330         parseIdentifier(Keyword))
3331       return true;
3332     if (Keyword == "md5") {
3333       HasMD5 = true;
3334       if (check(FileNumber == -1,
3335                 "MD5 checksum specified, but no file number") ||
3336           parseHexOcta(*this, MD5Hi, MD5Lo))
3337         return true;
3338     } else if (Keyword == "source") {
3339       HasSource = true;
3340       if (check(FileNumber == -1,
3341                 "source specified, but no file number") ||
3342           check(getTok().isNot(AsmToken::String),
3343                 "unexpected token in '.file' directive") ||
3344           parseEscapedString(SourceString))
3345         return true;
3346     } else {
3347       return TokError("unexpected token in '.file' directive");
3348     }
3349   }
3350 
3351   if (FileNumber == -1)
3352     getStreamer().EmitFileDirective(Filename);
3353   else {
3354     // In case there is a -g option as well as debug info from directive .file,
3355     // we turn off the -g option, directly use the existing debug info instead.
3356     // Also reset any implicit ".file 0" for the assembler source.
3357     if (Ctx.getGenDwarfForAssembly()) {
3358       Ctx.getMCDwarfLineTable(0).resetRootFile();
3359       Ctx.setGenDwarfForAssembly(false);
3360     }
3361 
3362     MD5::MD5Result *CKMem = nullptr;
3363     if (HasMD5) {
3364       CKMem = (MD5::MD5Result *)Ctx.allocate(sizeof(MD5::MD5Result), 1);
3365       for (unsigned i = 0; i != 8; ++i) {
3366         CKMem->Bytes[i] = uint8_t(MD5Hi >> ((7 - i) * 8));
3367         CKMem->Bytes[i + 8] = uint8_t(MD5Lo >> ((7 - i) * 8));
3368       }
3369     }
3370     if (HasSource) {
3371       char *SourceBuf = static_cast<char *>(Ctx.allocate(SourceString.size()));
3372       memcpy(SourceBuf, SourceString.data(), SourceString.size());
3373       Source = StringRef(SourceBuf, SourceString.size());
3374     }
3375     if (FileNumber == 0) {
3376       if (Ctx.getDwarfVersion() < 5)
3377         return Warning(DirectiveLoc, "file 0 not supported prior to DWARF-5");
3378       getStreamer().emitDwarfFile0Directive(Directory, Filename, CKMem, Source);
3379     } else {
3380       Expected<unsigned> FileNumOrErr = getStreamer().tryEmitDwarfFileDirective(
3381           FileNumber, Directory, Filename, CKMem, Source);
3382       if (!FileNumOrErr)
3383         return Error(DirectiveLoc, toString(FileNumOrErr.takeError()));
3384       FileNumber = FileNumOrErr.get();
3385     }
3386     // Alert the user if there are some .file directives with MD5 and some not.
3387     // But only do that once.
3388     if (!ReportedInconsistentMD5 && !Ctx.isDwarfMD5UsageConsistent(0)) {
3389       ReportedInconsistentMD5 = true;
3390       return Warning(DirectiveLoc, "inconsistent use of MD5 checksums");
3391     }
3392   }
3393 
3394   return false;
3395 }
3396 
3397 /// parseDirectiveLine
3398 /// ::= .line [number]
parseDirectiveLine()3399 bool AsmParser::parseDirectiveLine() {
3400   int64_t LineNumber;
3401   if (getLexer().is(AsmToken::Integer)) {
3402     if (parseIntToken(LineNumber, "unexpected token in '.line' directive"))
3403       return true;
3404     (void)LineNumber;
3405     // FIXME: Do something with the .line.
3406   }
3407   if (parseToken(AsmToken::EndOfStatement,
3408                  "unexpected token in '.line' directive"))
3409     return true;
3410 
3411   return false;
3412 }
3413 
3414 /// parseDirectiveLoc
3415 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end]
3416 ///                                [epilogue_begin] [is_stmt VALUE] [isa VALUE]
3417 /// The first number is a file number, must have been previously assigned with
3418 /// a .file directive, the second number is the line number and optionally the
3419 /// third number is a column position (zero if not specified).  The remaining
3420 /// optional items are .loc sub-directives.
parseDirectiveLoc()3421 bool AsmParser::parseDirectiveLoc() {
3422   int64_t FileNumber = 0, LineNumber = 0;
3423   SMLoc Loc = getTok().getLoc();
3424   if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") ||
3425       check(FileNumber < 1 && Ctx.getDwarfVersion() < 5, Loc,
3426             "file number less than one in '.loc' directive") ||
3427       check(!getContext().isValidDwarfFileNumber(FileNumber), Loc,
3428             "unassigned file number in '.loc' directive"))
3429     return true;
3430 
3431   // optional
3432   if (getLexer().is(AsmToken::Integer)) {
3433     LineNumber = getTok().getIntVal();
3434     if (LineNumber < 0)
3435       return TokError("line number less than zero in '.loc' directive");
3436     Lex();
3437   }
3438 
3439   int64_t ColumnPos = 0;
3440   if (getLexer().is(AsmToken::Integer)) {
3441     ColumnPos = getTok().getIntVal();
3442     if (ColumnPos < 0)
3443       return TokError("column position less than zero in '.loc' directive");
3444     Lex();
3445   }
3446 
3447   unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0;
3448   unsigned Isa = 0;
3449   int64_t Discriminator = 0;
3450 
3451   auto parseLocOp = [&]() -> bool {
3452     StringRef Name;
3453     SMLoc Loc = getTok().getLoc();
3454     if (parseIdentifier(Name))
3455       return TokError("unexpected token in '.loc' directive");
3456 
3457     if (Name == "basic_block")
3458       Flags |= DWARF2_FLAG_BASIC_BLOCK;
3459     else if (Name == "prologue_end")
3460       Flags |= DWARF2_FLAG_PROLOGUE_END;
3461     else if (Name == "epilogue_begin")
3462       Flags |= DWARF2_FLAG_EPILOGUE_BEGIN;
3463     else if (Name == "is_stmt") {
3464       Loc = getTok().getLoc();
3465       const MCExpr *Value;
3466       if (parseExpression(Value))
3467         return true;
3468       // The expression must be the constant 0 or 1.
3469       if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) {
3470         int Value = MCE->getValue();
3471         if (Value == 0)
3472           Flags &= ~DWARF2_FLAG_IS_STMT;
3473         else if (Value == 1)
3474           Flags |= DWARF2_FLAG_IS_STMT;
3475         else
3476           return Error(Loc, "is_stmt value not 0 or 1");
3477       } else {
3478         return Error(Loc, "is_stmt value not the constant value of 0 or 1");
3479       }
3480     } else if (Name == "isa") {
3481       Loc = getTok().getLoc();
3482       const MCExpr *Value;
3483       if (parseExpression(Value))
3484         return true;
3485       // The expression must be a constant greater or equal to 0.
3486       if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) {
3487         int Value = MCE->getValue();
3488         if (Value < 0)
3489           return Error(Loc, "isa number less than zero");
3490         Isa = Value;
3491       } else {
3492         return Error(Loc, "isa number not a constant value");
3493       }
3494     } else if (Name == "discriminator") {
3495       if (parseAbsoluteExpression(Discriminator))
3496         return true;
3497     } else {
3498       return Error(Loc, "unknown sub-directive in '.loc' directive");
3499     }
3500     return false;
3501   };
3502 
3503   if (parseMany(parseLocOp, false /*hasComma*/))
3504     return true;
3505 
3506   getStreamer().EmitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags,
3507                                       Isa, Discriminator, StringRef());
3508 
3509   return false;
3510 }
3511 
3512 /// parseDirectiveStabs
3513 /// ::= .stabs string, number, number, number
parseDirectiveStabs()3514 bool AsmParser::parseDirectiveStabs() {
3515   return TokError("unsupported directive '.stabs'");
3516 }
3517 
3518 /// parseDirectiveCVFile
3519 /// ::= .cv_file number filename [checksum] [checksumkind]
parseDirectiveCVFile()3520 bool AsmParser::parseDirectiveCVFile() {
3521   SMLoc FileNumberLoc = getTok().getLoc();
3522   int64_t FileNumber;
3523   std::string Filename;
3524   std::string Checksum;
3525   int64_t ChecksumKind = 0;
3526 
3527   if (parseIntToken(FileNumber,
3528                     "expected file number in '.cv_file' directive") ||
3529       check(FileNumber < 1, FileNumberLoc, "file number less than one") ||
3530       check(getTok().isNot(AsmToken::String),
3531             "unexpected token in '.cv_file' directive") ||
3532       parseEscapedString(Filename))
3533     return true;
3534   if (!parseOptionalToken(AsmToken::EndOfStatement)) {
3535     if (check(getTok().isNot(AsmToken::String),
3536               "unexpected token in '.cv_file' directive") ||
3537         parseEscapedString(Checksum) ||
3538         parseIntToken(ChecksumKind,
3539                       "expected checksum kind in '.cv_file' directive") ||
3540         parseToken(AsmToken::EndOfStatement,
3541                    "unexpected token in '.cv_file' directive"))
3542       return true;
3543   }
3544 
3545   Checksum = fromHex(Checksum);
3546   void *CKMem = Ctx.allocate(Checksum.size(), 1);
3547   memcpy(CKMem, Checksum.data(), Checksum.size());
3548   ArrayRef<uint8_t> ChecksumAsBytes(reinterpret_cast<const uint8_t *>(CKMem),
3549                                     Checksum.size());
3550 
3551   if (!getStreamer().EmitCVFileDirective(FileNumber, Filename, ChecksumAsBytes,
3552                                          static_cast<uint8_t>(ChecksumKind)))
3553     return Error(FileNumberLoc, "file number already allocated");
3554 
3555   return false;
3556 }
3557 
parseCVFunctionId(int64_t & FunctionId,StringRef DirectiveName)3558 bool AsmParser::parseCVFunctionId(int64_t &FunctionId,
3559                                   StringRef DirectiveName) {
3560   SMLoc Loc;
3561   return parseTokenLoc(Loc) ||
3562          parseIntToken(FunctionId, "expected function id in '" + DirectiveName +
3563                                        "' directive") ||
3564          check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc,
3565                "expected function id within range [0, UINT_MAX)");
3566 }
3567 
parseCVFileId(int64_t & FileNumber,StringRef DirectiveName)3568 bool AsmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) {
3569   SMLoc Loc;
3570   return parseTokenLoc(Loc) ||
3571          parseIntToken(FileNumber, "expected integer in '" + DirectiveName +
3572                                        "' directive") ||
3573          check(FileNumber < 1, Loc, "file number less than one in '" +
3574                                         DirectiveName + "' directive") ||
3575          check(!getCVContext().isValidFileNumber(FileNumber), Loc,
3576                "unassigned file number in '" + DirectiveName + "' directive");
3577 }
3578 
3579 /// parseDirectiveCVFuncId
3580 /// ::= .cv_func_id FunctionId
3581 ///
3582 /// Introduces a function ID that can be used with .cv_loc.
parseDirectiveCVFuncId()3583 bool AsmParser::parseDirectiveCVFuncId() {
3584   SMLoc FunctionIdLoc = getTok().getLoc();
3585   int64_t FunctionId;
3586 
3587   if (parseCVFunctionId(FunctionId, ".cv_func_id") ||
3588       parseToken(AsmToken::EndOfStatement,
3589                  "unexpected token in '.cv_func_id' directive"))
3590     return true;
3591 
3592   if (!getStreamer().EmitCVFuncIdDirective(FunctionId))
3593     return Error(FunctionIdLoc, "function id already allocated");
3594 
3595   return false;
3596 }
3597 
3598 /// parseDirectiveCVInlineSiteId
3599 /// ::= .cv_inline_site_id FunctionId
3600 ///         "within" IAFunc
3601 ///         "inlined_at" IAFile IALine [IACol]
3602 ///
3603 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined
3604 /// at" source location information for use in the line table of the caller,
3605 /// whether the caller is a real function or another inlined call site.
parseDirectiveCVInlineSiteId()3606 bool AsmParser::parseDirectiveCVInlineSiteId() {
3607   SMLoc FunctionIdLoc = getTok().getLoc();
3608   int64_t FunctionId;
3609   int64_t IAFunc;
3610   int64_t IAFile;
3611   int64_t IALine;
3612   int64_t IACol = 0;
3613 
3614   // FunctionId
3615   if (parseCVFunctionId(FunctionId, ".cv_inline_site_id"))
3616     return true;
3617 
3618   // "within"
3619   if (check((getLexer().isNot(AsmToken::Identifier) ||
3620              getTok().getIdentifier() != "within"),
3621             "expected 'within' identifier in '.cv_inline_site_id' directive"))
3622     return true;
3623   Lex();
3624 
3625   // IAFunc
3626   if (parseCVFunctionId(IAFunc, ".cv_inline_site_id"))
3627     return true;
3628 
3629   // "inlined_at"
3630   if (check((getLexer().isNot(AsmToken::Identifier) ||
3631              getTok().getIdentifier() != "inlined_at"),
3632             "expected 'inlined_at' identifier in '.cv_inline_site_id' "
3633             "directive") )
3634     return true;
3635   Lex();
3636 
3637   // IAFile IALine
3638   if (parseCVFileId(IAFile, ".cv_inline_site_id") ||
3639       parseIntToken(IALine, "expected line number after 'inlined_at'"))
3640     return true;
3641 
3642   // [IACol]
3643   if (getLexer().is(AsmToken::Integer)) {
3644     IACol = getTok().getIntVal();
3645     Lex();
3646   }
3647 
3648   if (parseToken(AsmToken::EndOfStatement,
3649                  "unexpected token in '.cv_inline_site_id' directive"))
3650     return true;
3651 
3652   if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile,
3653                                                  IALine, IACol, FunctionIdLoc))
3654     return Error(FunctionIdLoc, "function id already allocated");
3655 
3656   return false;
3657 }
3658 
3659 /// parseDirectiveCVLoc
3660 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end]
3661 ///                                [is_stmt VALUE]
3662 /// The first number is a file number, must have been previously assigned with
3663 /// a .file directive, the second number is the line number and optionally the
3664 /// third number is a column position (zero if not specified).  The remaining
3665 /// optional items are .loc sub-directives.
parseDirectiveCVLoc()3666 bool AsmParser::parseDirectiveCVLoc() {
3667   SMLoc DirectiveLoc = getTok().getLoc();
3668   int64_t FunctionId, FileNumber;
3669   if (parseCVFunctionId(FunctionId, ".cv_loc") ||
3670       parseCVFileId(FileNumber, ".cv_loc"))
3671     return true;
3672 
3673   int64_t LineNumber = 0;
3674   if (getLexer().is(AsmToken::Integer)) {
3675     LineNumber = getTok().getIntVal();
3676     if (LineNumber < 0)
3677       return TokError("line number less than zero in '.cv_loc' directive");
3678     Lex();
3679   }
3680 
3681   int64_t ColumnPos = 0;
3682   if (getLexer().is(AsmToken::Integer)) {
3683     ColumnPos = getTok().getIntVal();
3684     if (ColumnPos < 0)
3685       return TokError("column position less than zero in '.cv_loc' directive");
3686     Lex();
3687   }
3688 
3689   bool PrologueEnd = false;
3690   uint64_t IsStmt = 0;
3691 
3692   auto parseOp = [&]() -> bool {
3693     StringRef Name;
3694     SMLoc Loc = getTok().getLoc();
3695     if (parseIdentifier(Name))
3696       return TokError("unexpected token in '.cv_loc' directive");
3697     if (Name == "prologue_end")
3698       PrologueEnd = true;
3699     else if (Name == "is_stmt") {
3700       Loc = getTok().getLoc();
3701       const MCExpr *Value;
3702       if (parseExpression(Value))
3703         return true;
3704       // The expression must be the constant 0 or 1.
3705       IsStmt = ~0ULL;
3706       if (const auto *MCE = dyn_cast<MCConstantExpr>(Value))
3707         IsStmt = MCE->getValue();
3708 
3709       if (IsStmt > 1)
3710         return Error(Loc, "is_stmt value not 0 or 1");
3711     } else {
3712       return Error(Loc, "unknown sub-directive in '.cv_loc' directive");
3713     }
3714     return false;
3715   };
3716 
3717   if (parseMany(parseOp, false /*hasComma*/))
3718     return true;
3719 
3720   getStreamer().EmitCVLocDirective(FunctionId, FileNumber, LineNumber,
3721                                    ColumnPos, PrologueEnd, IsStmt, StringRef(),
3722                                    DirectiveLoc);
3723   return false;
3724 }
3725 
3726 /// parseDirectiveCVLinetable
3727 /// ::= .cv_linetable FunctionId, FnStart, FnEnd
parseDirectiveCVLinetable()3728 bool AsmParser::parseDirectiveCVLinetable() {
3729   int64_t FunctionId;
3730   StringRef FnStartName, FnEndName;
3731   SMLoc Loc = getTok().getLoc();
3732   if (parseCVFunctionId(FunctionId, ".cv_linetable") ||
3733       parseToken(AsmToken::Comma,
3734                  "unexpected token in '.cv_linetable' directive") ||
3735       parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc,
3736                                   "expected identifier in directive") ||
3737       parseToken(AsmToken::Comma,
3738                  "unexpected token in '.cv_linetable' directive") ||
3739       parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc,
3740                                   "expected identifier in directive"))
3741     return true;
3742 
3743   MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName);
3744   MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName);
3745 
3746   getStreamer().EmitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym);
3747   return false;
3748 }
3749 
3750 /// parseDirectiveCVInlineLinetable
3751 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd
parseDirectiveCVInlineLinetable()3752 bool AsmParser::parseDirectiveCVInlineLinetable() {
3753   int64_t PrimaryFunctionId, SourceFileId, SourceLineNum;
3754   StringRef FnStartName, FnEndName;
3755   SMLoc Loc = getTok().getLoc();
3756   if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") ||
3757       parseTokenLoc(Loc) ||
3758       parseIntToken(
3759           SourceFileId,
3760           "expected SourceField in '.cv_inline_linetable' directive") ||
3761       check(SourceFileId <= 0, Loc,
3762             "File id less than zero in '.cv_inline_linetable' directive") ||
3763       parseTokenLoc(Loc) ||
3764       parseIntToken(
3765           SourceLineNum,
3766           "expected SourceLineNum in '.cv_inline_linetable' directive") ||
3767       check(SourceLineNum < 0, Loc,
3768             "Line number less than zero in '.cv_inline_linetable' directive") ||
3769       parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc,
3770                                   "expected identifier in directive") ||
3771       parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc,
3772                                   "expected identifier in directive"))
3773     return true;
3774 
3775   if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement"))
3776     return true;
3777 
3778   MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName);
3779   MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName);
3780   getStreamer().EmitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId,
3781                                                SourceLineNum, FnStartSym,
3782                                                FnEndSym);
3783   return false;
3784 }
3785 
3786 /// parseDirectiveCVDefRange
3787 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes*
parseDirectiveCVDefRange()3788 bool AsmParser::parseDirectiveCVDefRange() {
3789   SMLoc Loc;
3790   std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges;
3791   while (getLexer().is(AsmToken::Identifier)) {
3792     Loc = getLexer().getLoc();
3793     StringRef GapStartName;
3794     if (parseIdentifier(GapStartName))
3795       return Error(Loc, "expected identifier in directive");
3796     MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName);
3797 
3798     Loc = getLexer().getLoc();
3799     StringRef GapEndName;
3800     if (parseIdentifier(GapEndName))
3801       return Error(Loc, "expected identifier in directive");
3802     MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName);
3803 
3804     Ranges.push_back({GapStartSym, GapEndSym});
3805   }
3806 
3807   std::string FixedSizePortion;
3808   if (parseToken(AsmToken::Comma, "unexpected token in directive") ||
3809       parseEscapedString(FixedSizePortion))
3810     return true;
3811 
3812   getStreamer().EmitCVDefRangeDirective(Ranges, FixedSizePortion);
3813   return false;
3814 }
3815 
3816 /// parseDirectiveCVStringTable
3817 /// ::= .cv_stringtable
parseDirectiveCVStringTable()3818 bool AsmParser::parseDirectiveCVStringTable() {
3819   getStreamer().EmitCVStringTableDirective();
3820   return false;
3821 }
3822 
3823 /// parseDirectiveCVFileChecksums
3824 /// ::= .cv_filechecksums
parseDirectiveCVFileChecksums()3825 bool AsmParser::parseDirectiveCVFileChecksums() {
3826   getStreamer().EmitCVFileChecksumsDirective();
3827   return false;
3828 }
3829 
3830 /// parseDirectiveCVFileChecksumOffset
3831 /// ::= .cv_filechecksumoffset fileno
parseDirectiveCVFileChecksumOffset()3832 bool AsmParser::parseDirectiveCVFileChecksumOffset() {
3833   int64_t FileNo;
3834   if (parseIntToken(FileNo, "expected identifier in directive"))
3835     return true;
3836   if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement"))
3837     return true;
3838   getStreamer().EmitCVFileChecksumOffsetDirective(FileNo);
3839   return false;
3840 }
3841 
3842 /// parseDirectiveCVFPOData
3843 /// ::= .cv_fpo_data procsym
parseDirectiveCVFPOData()3844 bool AsmParser::parseDirectiveCVFPOData() {
3845   SMLoc DirLoc = getLexer().getLoc();
3846   StringRef ProcName;
3847   if (parseIdentifier(ProcName))
3848     return TokError("expected symbol name");
3849   if (parseEOL("unexpected tokens"))
3850     return addErrorSuffix(" in '.cv_fpo_data' directive");
3851   MCSymbol *ProcSym = getContext().getOrCreateSymbol(ProcName);
3852   getStreamer().EmitCVFPOData(ProcSym, DirLoc);
3853   return false;
3854 }
3855 
3856 /// parseDirectiveCFISections
3857 /// ::= .cfi_sections section [, section]
parseDirectiveCFISections()3858 bool AsmParser::parseDirectiveCFISections() {
3859   StringRef Name;
3860   bool EH = false;
3861   bool Debug = false;
3862 
3863   if (parseIdentifier(Name))
3864     return TokError("Expected an identifier");
3865 
3866   if (Name == ".eh_frame")
3867     EH = true;
3868   else if (Name == ".debug_frame")
3869     Debug = true;
3870 
3871   if (getLexer().is(AsmToken::Comma)) {
3872     Lex();
3873 
3874     if (parseIdentifier(Name))
3875       return TokError("Expected an identifier");
3876 
3877     if (Name == ".eh_frame")
3878       EH = true;
3879     else if (Name == ".debug_frame")
3880       Debug = true;
3881   }
3882 
3883   getStreamer().EmitCFISections(EH, Debug);
3884   return false;
3885 }
3886 
3887 /// parseDirectiveCFIStartProc
3888 /// ::= .cfi_startproc [simple]
parseDirectiveCFIStartProc()3889 bool AsmParser::parseDirectiveCFIStartProc() {
3890   StringRef Simple;
3891   if (!parseOptionalToken(AsmToken::EndOfStatement)) {
3892     if (check(parseIdentifier(Simple) || Simple != "simple",
3893               "unexpected token") ||
3894         parseToken(AsmToken::EndOfStatement))
3895       return addErrorSuffix(" in '.cfi_startproc' directive");
3896   }
3897 
3898   getStreamer().EmitCFIStartProc(!Simple.empty());
3899   return false;
3900 }
3901 
3902 /// parseDirectiveCFIEndProc
3903 /// ::= .cfi_endproc
parseDirectiveCFIEndProc()3904 bool AsmParser::parseDirectiveCFIEndProc() {
3905   getStreamer().EmitCFIEndProc();
3906   return false;
3907 }
3908 
3909 /// parse register name or number.
parseRegisterOrRegisterNumber(int64_t & Register,SMLoc DirectiveLoc)3910 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register,
3911                                               SMLoc DirectiveLoc) {
3912   unsigned RegNo;
3913 
3914   if (getLexer().isNot(AsmToken::Integer)) {
3915     if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc))
3916       return true;
3917     Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true);
3918   } else
3919     return parseAbsoluteExpression(Register);
3920 
3921   return false;
3922 }
3923 
3924 /// parseDirectiveCFIDefCfa
3925 /// ::= .cfi_def_cfa register,  offset
parseDirectiveCFIDefCfa(SMLoc DirectiveLoc)3926 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) {
3927   int64_t Register = 0, Offset = 0;
3928   if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) ||
3929       parseToken(AsmToken::Comma, "unexpected token in directive") ||
3930       parseAbsoluteExpression(Offset))
3931     return true;
3932 
3933   getStreamer().EmitCFIDefCfa(Register, Offset);
3934   return false;
3935 }
3936 
3937 /// parseDirectiveCFIDefCfaOffset
3938 /// ::= .cfi_def_cfa_offset offset
parseDirectiveCFIDefCfaOffset()3939 bool AsmParser::parseDirectiveCFIDefCfaOffset() {
3940   int64_t Offset = 0;
3941   if (parseAbsoluteExpression(Offset))
3942     return true;
3943 
3944   getStreamer().EmitCFIDefCfaOffset(Offset);
3945   return false;
3946 }
3947 
3948 /// parseDirectiveCFIRegister
3949 /// ::= .cfi_register register, register
parseDirectiveCFIRegister(SMLoc DirectiveLoc)3950 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) {
3951   int64_t Register1 = 0, Register2 = 0;
3952   if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) ||
3953       parseToken(AsmToken::Comma, "unexpected token in directive") ||
3954       parseRegisterOrRegisterNumber(Register2, DirectiveLoc))
3955     return true;
3956 
3957   getStreamer().EmitCFIRegister(Register1, Register2);
3958   return false;
3959 }
3960 
3961 /// parseDirectiveCFIWindowSave
3962 /// ::= .cfi_window_save
parseDirectiveCFIWindowSave()3963 bool AsmParser::parseDirectiveCFIWindowSave() {
3964   getStreamer().EmitCFIWindowSave();
3965   return false;
3966 }
3967 
3968 /// parseDirectiveCFIAdjustCfaOffset
3969 /// ::= .cfi_adjust_cfa_offset adjustment
parseDirectiveCFIAdjustCfaOffset()3970 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() {
3971   int64_t Adjustment = 0;
3972   if (parseAbsoluteExpression(Adjustment))
3973     return true;
3974 
3975   getStreamer().EmitCFIAdjustCfaOffset(Adjustment);
3976   return false;
3977 }
3978 
3979 /// parseDirectiveCFIDefCfaRegister
3980 /// ::= .cfi_def_cfa_register register
parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc)3981 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) {
3982   int64_t Register = 0;
3983   if (parseRegisterOrRegisterNumber(Register, DirectiveLoc))
3984     return true;
3985 
3986   getStreamer().EmitCFIDefCfaRegister(Register);
3987   return false;
3988 }
3989 
3990 /// parseDirectiveCFIOffset
3991 /// ::= .cfi_offset register, offset
parseDirectiveCFIOffset(SMLoc DirectiveLoc)3992 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) {
3993   int64_t Register = 0;
3994   int64_t Offset = 0;
3995 
3996   if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) ||
3997       parseToken(AsmToken::Comma, "unexpected token in directive") ||
3998       parseAbsoluteExpression(Offset))
3999     return true;
4000 
4001   getStreamer().EmitCFIOffset(Register, Offset);
4002   return false;
4003 }
4004 
4005 /// parseDirectiveCFIRelOffset
4006 /// ::= .cfi_rel_offset register, offset
parseDirectiveCFIRelOffset(SMLoc DirectiveLoc)4007 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) {
4008   int64_t Register = 0, Offset = 0;
4009 
4010   if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) ||
4011       parseToken(AsmToken::Comma, "unexpected token in directive") ||
4012       parseAbsoluteExpression(Offset))
4013     return true;
4014 
4015   getStreamer().EmitCFIRelOffset(Register, Offset);
4016   return false;
4017 }
4018 
isValidEncoding(int64_t Encoding)4019 static bool isValidEncoding(int64_t Encoding) {
4020   if (Encoding & ~0xff)
4021     return false;
4022 
4023   if (Encoding == dwarf::DW_EH_PE_omit)
4024     return true;
4025 
4026   const unsigned Format = Encoding & 0xf;
4027   if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 &&
4028       Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 &&
4029       Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 &&
4030       Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed)
4031     return false;
4032 
4033   const unsigned Application = Encoding & 0x70;
4034   if (Application != dwarf::DW_EH_PE_absptr &&
4035       Application != dwarf::DW_EH_PE_pcrel)
4036     return false;
4037 
4038   return true;
4039 }
4040 
4041 /// parseDirectiveCFIPersonalityOrLsda
4042 /// IsPersonality true for cfi_personality, false for cfi_lsda
4043 /// ::= .cfi_personality encoding, [symbol_name]
4044 /// ::= .cfi_lsda encoding, [symbol_name]
parseDirectiveCFIPersonalityOrLsda(bool IsPersonality)4045 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) {
4046   int64_t Encoding = 0;
4047   if (parseAbsoluteExpression(Encoding))
4048     return true;
4049   if (Encoding == dwarf::DW_EH_PE_omit)
4050     return false;
4051 
4052   StringRef Name;
4053   if (check(!isValidEncoding(Encoding), "unsupported encoding.") ||
4054       parseToken(AsmToken::Comma, "unexpected token in directive") ||
4055       check(parseIdentifier(Name), "expected identifier in directive"))
4056     return true;
4057 
4058   MCSymbol *Sym = getContext().getOrCreateSymbol(Name);
4059 
4060   if (IsPersonality)
4061     getStreamer().EmitCFIPersonality(Sym, Encoding);
4062   else
4063     getStreamer().EmitCFILsda(Sym, Encoding);
4064   return false;
4065 }
4066 
4067 /// parseDirectiveCFIRememberState
4068 /// ::= .cfi_remember_state
parseDirectiveCFIRememberState()4069 bool AsmParser::parseDirectiveCFIRememberState() {
4070   getStreamer().EmitCFIRememberState();
4071   return false;
4072 }
4073 
4074 /// parseDirectiveCFIRestoreState
4075 /// ::= .cfi_remember_state
parseDirectiveCFIRestoreState()4076 bool AsmParser::parseDirectiveCFIRestoreState() {
4077   getStreamer().EmitCFIRestoreState();
4078   return false;
4079 }
4080 
4081 /// parseDirectiveCFISameValue
4082 /// ::= .cfi_same_value register
parseDirectiveCFISameValue(SMLoc DirectiveLoc)4083 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) {
4084   int64_t Register = 0;
4085 
4086   if (parseRegisterOrRegisterNumber(Register, DirectiveLoc))
4087     return true;
4088 
4089   getStreamer().EmitCFISameValue(Register);
4090   return false;
4091 }
4092 
4093 /// parseDirectiveCFIRestore
4094 /// ::= .cfi_restore register
parseDirectiveCFIRestore(SMLoc DirectiveLoc)4095 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) {
4096   int64_t Register = 0;
4097   if (parseRegisterOrRegisterNumber(Register, DirectiveLoc))
4098     return true;
4099 
4100   getStreamer().EmitCFIRestore(Register);
4101   return false;
4102 }
4103 
4104 /// parseDirectiveCFIEscape
4105 /// ::= .cfi_escape expression[,...]
parseDirectiveCFIEscape()4106 bool AsmParser::parseDirectiveCFIEscape() {
4107   std::string Values;
4108   int64_t CurrValue;
4109   if (parseAbsoluteExpression(CurrValue))
4110     return true;
4111 
4112   Values.push_back((uint8_t)CurrValue);
4113 
4114   while (getLexer().is(AsmToken::Comma)) {
4115     Lex();
4116 
4117     if (parseAbsoluteExpression(CurrValue))
4118       return true;
4119 
4120     Values.push_back((uint8_t)CurrValue);
4121   }
4122 
4123   getStreamer().EmitCFIEscape(Values);
4124   return false;
4125 }
4126 
4127 /// parseDirectiveCFIReturnColumn
4128 /// ::= .cfi_return_column register
parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc)4129 bool AsmParser::parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc) {
4130   int64_t Register = 0;
4131   if (parseRegisterOrRegisterNumber(Register, DirectiveLoc))
4132     return true;
4133   getStreamer().EmitCFIReturnColumn(Register);
4134   return false;
4135 }
4136 
4137 /// parseDirectiveCFISignalFrame
4138 /// ::= .cfi_signal_frame
parseDirectiveCFISignalFrame()4139 bool AsmParser::parseDirectiveCFISignalFrame() {
4140   if (parseToken(AsmToken::EndOfStatement,
4141                  "unexpected token in '.cfi_signal_frame'"))
4142     return true;
4143 
4144   getStreamer().EmitCFISignalFrame();
4145   return false;
4146 }
4147 
4148 /// parseDirectiveCFIUndefined
4149 /// ::= .cfi_undefined register
parseDirectiveCFIUndefined(SMLoc DirectiveLoc)4150 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) {
4151   int64_t Register = 0;
4152 
4153   if (parseRegisterOrRegisterNumber(Register, DirectiveLoc))
4154     return true;
4155 
4156   getStreamer().EmitCFIUndefined(Register);
4157   return false;
4158 }
4159 
4160 /// parseDirectiveAltmacro
4161 /// ::= .altmacro
4162 /// ::= .noaltmacro
parseDirectiveAltmacro(StringRef Directive)4163 bool AsmParser::parseDirectiveAltmacro(StringRef Directive) {
4164   if (getLexer().isNot(AsmToken::EndOfStatement))
4165     return TokError("unexpected token in '" + Directive + "' directive");
4166   if (Directive == ".altmacro")
4167     getLexer().SetAltMacroMode(true);
4168   else
4169     getLexer().SetAltMacroMode(false);
4170   return false;
4171 }
4172 
4173 /// parseDirectiveMacrosOnOff
4174 /// ::= .macros_on
4175 /// ::= .macros_off
parseDirectiveMacrosOnOff(StringRef Directive)4176 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) {
4177   if (parseToken(AsmToken::EndOfStatement,
4178                  "unexpected token in '" + Directive + "' directive"))
4179     return true;
4180 
4181   setMacrosEnabled(Directive == ".macros_on");
4182   return false;
4183 }
4184 
4185 /// parseDirectiveMacro
4186 /// ::= .macro name[,] [parameters]
parseDirectiveMacro(SMLoc DirectiveLoc)4187 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) {
4188   StringRef Name;
4189   if (parseIdentifier(Name))
4190     return TokError("expected identifier in '.macro' directive");
4191 
4192   if (getLexer().is(AsmToken::Comma))
4193     Lex();
4194 
4195   MCAsmMacroParameters Parameters;
4196   while (getLexer().isNot(AsmToken::EndOfStatement)) {
4197 
4198     if (!Parameters.empty() && Parameters.back().Vararg)
4199       return Error(Lexer.getLoc(),
4200                    "Vararg parameter '" + Parameters.back().Name +
4201                    "' should be last one in the list of parameters.");
4202 
4203     MCAsmMacroParameter Parameter;
4204     if (parseIdentifier(Parameter.Name))
4205       return TokError("expected identifier in '.macro' directive");
4206 
4207     // Emit an error if two (or more) named parameters share the same name
4208     for (const MCAsmMacroParameter& CurrParam : Parameters)
4209       if (CurrParam.Name.equals(Parameter.Name))
4210         return TokError("macro '" + Name + "' has multiple parameters"
4211                         " named '" + Parameter.Name + "'");
4212 
4213     if (Lexer.is(AsmToken::Colon)) {
4214       Lex();  // consume ':'
4215 
4216       SMLoc QualLoc;
4217       StringRef Qualifier;
4218 
4219       QualLoc = Lexer.getLoc();
4220       if (parseIdentifier(Qualifier))
4221         return Error(QualLoc, "missing parameter qualifier for "
4222                      "'" + Parameter.Name + "' in macro '" + Name + "'");
4223 
4224       if (Qualifier == "req")
4225         Parameter.Required = true;
4226       else if (Qualifier == "vararg")
4227         Parameter.Vararg = true;
4228       else
4229         return Error(QualLoc, Qualifier + " is not a valid parameter qualifier "
4230                      "for '" + Parameter.Name + "' in macro '" + Name + "'");
4231     }
4232 
4233     if (getLexer().is(AsmToken::Equal)) {
4234       Lex();
4235 
4236       SMLoc ParamLoc;
4237 
4238       ParamLoc = Lexer.getLoc();
4239       if (parseMacroArgument(Parameter.Value, /*Vararg=*/false ))
4240         return true;
4241 
4242       if (Parameter.Required)
4243         Warning(ParamLoc, "pointless default value for required parameter "
4244                 "'" + Parameter.Name + "' in macro '" + Name + "'");
4245     }
4246 
4247     Parameters.push_back(std::move(Parameter));
4248 
4249     if (getLexer().is(AsmToken::Comma))
4250       Lex();
4251   }
4252 
4253   // Eat just the end of statement.
4254   Lexer.Lex();
4255 
4256   // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors
4257   AsmToken EndToken, StartToken = getTok();
4258   unsigned MacroDepth = 0;
4259   // Lex the macro definition.
4260   while (true) {
4261     // Ignore Lexing errors in macros.
4262     while (Lexer.is(AsmToken::Error)) {
4263       Lexer.Lex();
4264     }
4265 
4266     // Check whether we have reached the end of the file.
4267     if (getLexer().is(AsmToken::Eof))
4268       return Error(DirectiveLoc, "no matching '.endmacro' in definition");
4269 
4270     // Otherwise, check whether we have reach the .endmacro.
4271     if (getLexer().is(AsmToken::Identifier)) {
4272       if (getTok().getIdentifier() == ".endm" ||
4273           getTok().getIdentifier() == ".endmacro") {
4274         if (MacroDepth == 0) { // Outermost macro.
4275           EndToken = getTok();
4276           Lexer.Lex();
4277           if (getLexer().isNot(AsmToken::EndOfStatement))
4278             return TokError("unexpected token in '" + EndToken.getIdentifier() +
4279                             "' directive");
4280           break;
4281         } else {
4282           // Otherwise we just found the end of an inner macro.
4283           --MacroDepth;
4284         }
4285       } else if (getTok().getIdentifier() == ".macro") {
4286         // We allow nested macros. Those aren't instantiated until the outermost
4287         // macro is expanded so just ignore them for now.
4288         ++MacroDepth;
4289       }
4290     }
4291 
4292     // Otherwise, scan til the end of the statement.
4293     eatToEndOfStatement();
4294   }
4295 
4296   if (getContext().lookupMacro(Name)) {
4297     return Error(DirectiveLoc, "macro '" + Name + "' is already defined");
4298   }
4299 
4300   const char *BodyStart = StartToken.getLoc().getPointer();
4301   const char *BodyEnd = EndToken.getLoc().getPointer();
4302   StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart);
4303   checkForBadMacro(DirectiveLoc, Name, Body, Parameters);
4304   MCAsmMacro Macro(Name, Body, std::move(Parameters));
4305   DEBUG_WITH_TYPE("asm-macros", dbgs() << "Defining new macro:\n";
4306                   Macro.dump());
4307   getContext().defineMacro(Name, std::move(Macro));
4308   return false;
4309 }
4310 
4311 /// checkForBadMacro
4312 ///
4313 /// With the support added for named parameters there may be code out there that
4314 /// is transitioning from positional parameters.  In versions of gas that did
4315 /// not support named parameters they would be ignored on the macro definition.
4316 /// But to support both styles of parameters this is not possible so if a macro
4317 /// definition has named parameters but does not use them and has what appears
4318 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a
4319 /// warning that the positional parameter found in body which have no effect.
4320 /// Hoping the developer will either remove the named parameters from the macro
4321 /// definition so the positional parameters get used if that was what was
4322 /// intended or change the macro to use the named parameters.  It is possible
4323 /// this warning will trigger when the none of the named parameters are used
4324 /// and the strings like $1 are infact to simply to be passed trough unchanged.
checkForBadMacro(SMLoc DirectiveLoc,StringRef Name,StringRef Body,ArrayRef<MCAsmMacroParameter> Parameters)4325 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name,
4326                                  StringRef Body,
4327                                  ArrayRef<MCAsmMacroParameter> Parameters) {
4328   // If this macro is not defined with named parameters the warning we are
4329   // checking for here doesn't apply.
4330   unsigned NParameters = Parameters.size();
4331   if (NParameters == 0)
4332     return;
4333 
4334   bool NamedParametersFound = false;
4335   bool PositionalParametersFound = false;
4336 
4337   // Look at the body of the macro for use of both the named parameters and what
4338   // are likely to be positional parameters.  This is what expandMacro() is
4339   // doing when it finds the parameters in the body.
4340   while (!Body.empty()) {
4341     // Scan for the next possible parameter.
4342     std::size_t End = Body.size(), Pos = 0;
4343     for (; Pos != End; ++Pos) {
4344       // Check for a substitution or escape.
4345       // This macro is defined with parameters, look for \foo, \bar, etc.
4346       if (Body[Pos] == '\\' && Pos + 1 != End)
4347         break;
4348 
4349       // This macro should have parameters, but look for $0, $1, ..., $n too.
4350       if (Body[Pos] != '$' || Pos + 1 == End)
4351         continue;
4352       char Next = Body[Pos + 1];
4353       if (Next == '$' || Next == 'n' ||
4354           isdigit(static_cast<unsigned char>(Next)))
4355         break;
4356     }
4357 
4358     // Check if we reached the end.
4359     if (Pos == End)
4360       break;
4361 
4362     if (Body[Pos] == '$') {
4363       switch (Body[Pos + 1]) {
4364       // $$ => $
4365       case '$':
4366         break;
4367 
4368       // $n => number of arguments
4369       case 'n':
4370         PositionalParametersFound = true;
4371         break;
4372 
4373       // $[0-9] => argument
4374       default: {
4375         PositionalParametersFound = true;
4376         break;
4377       }
4378       }
4379       Pos += 2;
4380     } else {
4381       unsigned I = Pos + 1;
4382       while (isIdentifierChar(Body[I]) && I + 1 != End)
4383         ++I;
4384 
4385       const char *Begin = Body.data() + Pos + 1;
4386       StringRef Argument(Begin, I - (Pos + 1));
4387       unsigned Index = 0;
4388       for (; Index < NParameters; ++Index)
4389         if (Parameters[Index].Name == Argument)
4390           break;
4391 
4392       if (Index == NParameters) {
4393         if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')')
4394           Pos += 3;
4395         else {
4396           Pos = I;
4397         }
4398       } else {
4399         NamedParametersFound = true;
4400         Pos += 1 + Argument.size();
4401       }
4402     }
4403     // Update the scan point.
4404     Body = Body.substr(Pos);
4405   }
4406 
4407   if (!NamedParametersFound && PositionalParametersFound)
4408     Warning(DirectiveLoc, "macro defined with named parameters which are not "
4409                           "used in macro body, possible positional parameter "
4410                           "found in body which will have no effect");
4411 }
4412 
4413 /// parseDirectiveExitMacro
4414 /// ::= .exitm
parseDirectiveExitMacro(StringRef Directive)4415 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) {
4416   if (parseToken(AsmToken::EndOfStatement,
4417                  "unexpected token in '" + Directive + "' directive"))
4418     return true;
4419 
4420   if (!isInsideMacroInstantiation())
4421     return TokError("unexpected '" + Directive + "' in file, "
4422                                                  "no current macro definition");
4423 
4424   // Exit all conditionals that are active in the current macro.
4425   while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) {
4426     TheCondState = TheCondStack.back();
4427     TheCondStack.pop_back();
4428   }
4429 
4430   handleMacroExit();
4431   return false;
4432 }
4433 
4434 /// parseDirectiveEndMacro
4435 /// ::= .endm
4436 /// ::= .endmacro
parseDirectiveEndMacro(StringRef Directive)4437 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) {
4438   if (getLexer().isNot(AsmToken::EndOfStatement))
4439     return TokError("unexpected token in '" + Directive + "' directive");
4440 
4441   // If we are inside a macro instantiation, terminate the current
4442   // instantiation.
4443   if (isInsideMacroInstantiation()) {
4444     handleMacroExit();
4445     return false;
4446   }
4447 
4448   // Otherwise, this .endmacro is a stray entry in the file; well formed
4449   // .endmacro directives are handled during the macro definition parsing.
4450   return TokError("unexpected '" + Directive + "' in file, "
4451                                                "no current macro definition");
4452 }
4453 
4454 /// parseDirectivePurgeMacro
4455 /// ::= .purgem
parseDirectivePurgeMacro(SMLoc DirectiveLoc)4456 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) {
4457   StringRef Name;
4458   SMLoc Loc;
4459   if (parseTokenLoc(Loc) ||
4460       check(parseIdentifier(Name), Loc,
4461             "expected identifier in '.purgem' directive") ||
4462       parseToken(AsmToken::EndOfStatement,
4463                  "unexpected token in '.purgem' directive"))
4464     return true;
4465 
4466   if (!getContext().lookupMacro(Name))
4467     return Error(DirectiveLoc, "macro '" + Name + "' is not defined");
4468 
4469   getContext().undefineMacro(Name);
4470   DEBUG_WITH_TYPE("asm-macros", dbgs()
4471                                     << "Un-defining macro: " << Name << "\n");
4472   return false;
4473 }
4474 
4475 /// parseDirectiveBundleAlignMode
4476 /// ::= {.bundle_align_mode} expression
parseDirectiveBundleAlignMode()4477 bool AsmParser::parseDirectiveBundleAlignMode() {
4478   // Expect a single argument: an expression that evaluates to a constant
4479   // in the inclusive range 0-30.
4480   SMLoc ExprLoc = getLexer().getLoc();
4481   int64_t AlignSizePow2;
4482   if (checkForValidSection() || parseAbsoluteExpression(AlignSizePow2) ||
4483       parseToken(AsmToken::EndOfStatement, "unexpected token after expression "
4484                                            "in '.bundle_align_mode' "
4485                                            "directive") ||
4486       check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc,
4487             "invalid bundle alignment size (expected between 0 and 30)"))
4488     return true;
4489 
4490   // Because of AlignSizePow2's verified range we can safely truncate it to
4491   // unsigned.
4492   getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2));
4493   return false;
4494 }
4495 
4496 /// parseDirectiveBundleLock
4497 /// ::= {.bundle_lock} [align_to_end]
parseDirectiveBundleLock()4498 bool AsmParser::parseDirectiveBundleLock() {
4499   if (checkForValidSection())
4500     return true;
4501   bool AlignToEnd = false;
4502 
4503   StringRef Option;
4504   SMLoc Loc = getTok().getLoc();
4505   const char *kInvalidOptionError =
4506       "invalid option for '.bundle_lock' directive";
4507 
4508   if (!parseOptionalToken(AsmToken::EndOfStatement)) {
4509     if (check(parseIdentifier(Option), Loc, kInvalidOptionError) ||
4510         check(Option != "align_to_end", Loc, kInvalidOptionError) ||
4511         parseToken(AsmToken::EndOfStatement,
4512                    "unexpected token after '.bundle_lock' directive option"))
4513       return true;
4514     AlignToEnd = true;
4515   }
4516 
4517   getStreamer().EmitBundleLock(AlignToEnd);
4518   return false;
4519 }
4520 
4521 /// parseDirectiveBundleLock
4522 /// ::= {.bundle_lock}
parseDirectiveBundleUnlock()4523 bool AsmParser::parseDirectiveBundleUnlock() {
4524   if (checkForValidSection() ||
4525       parseToken(AsmToken::EndOfStatement,
4526                  "unexpected token in '.bundle_unlock' directive"))
4527     return true;
4528 
4529   getStreamer().EmitBundleUnlock();
4530   return false;
4531 }
4532 
4533 /// parseDirectiveSpace
4534 /// ::= (.skip | .space) expression [ , expression ]
parseDirectiveSpace(StringRef IDVal)4535 bool AsmParser::parseDirectiveSpace(StringRef IDVal) {
4536   SMLoc NumBytesLoc = Lexer.getLoc();
4537   const MCExpr *NumBytes;
4538   if (checkForValidSection() || parseExpression(NumBytes))
4539     return true;
4540 
4541   int64_t FillExpr = 0;
4542   if (parseOptionalToken(AsmToken::Comma))
4543     if (parseAbsoluteExpression(FillExpr))
4544       return addErrorSuffix("in '" + Twine(IDVal) + "' directive");
4545   if (parseToken(AsmToken::EndOfStatement))
4546     return addErrorSuffix("in '" + Twine(IDVal) + "' directive");
4547 
4548   // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0.
4549   getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc);
4550 
4551   return false;
4552 }
4553 
4554 /// parseDirectiveDCB
4555 /// ::= .dcb.{b, l, w} expression, expression
parseDirectiveDCB(StringRef IDVal,unsigned Size)4556 bool AsmParser::parseDirectiveDCB(StringRef IDVal, unsigned Size) {
4557   SMLoc NumValuesLoc = Lexer.getLoc();
4558   int64_t NumValues;
4559   if (checkForValidSection() || parseAbsoluteExpression(NumValues))
4560     return true;
4561 
4562   if (NumValues < 0) {
4563     Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect");
4564     return false;
4565   }
4566 
4567   if (parseToken(AsmToken::Comma,
4568                  "unexpected token in '" + Twine(IDVal) + "' directive"))
4569     return true;
4570 
4571   const MCExpr *Value;
4572   SMLoc ExprLoc = getLexer().getLoc();
4573   if (parseExpression(Value))
4574     return true;
4575 
4576   // Special case constant expressions to match code generator.
4577   if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) {
4578     assert(Size <= 8 && "Invalid size");
4579     uint64_t IntValue = MCE->getValue();
4580     if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue))
4581       return Error(ExprLoc, "literal value out of range for directive");
4582     for (uint64_t i = 0, e = NumValues; i != e; ++i)
4583       getStreamer().EmitIntValue(IntValue, Size);
4584   } else {
4585     for (uint64_t i = 0, e = NumValues; i != e; ++i)
4586       getStreamer().EmitValue(Value, Size, ExprLoc);
4587   }
4588 
4589   if (parseToken(AsmToken::EndOfStatement,
4590                  "unexpected token in '" + Twine(IDVal) + "' directive"))
4591     return true;
4592 
4593   return false;
4594 }
4595 
4596 /// parseDirectiveRealDCB
4597 /// ::= .dcb.{d, s} expression, expression
parseDirectiveRealDCB(StringRef IDVal,const fltSemantics & Semantics)4598 bool AsmParser::parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &Semantics) {
4599   SMLoc NumValuesLoc = Lexer.getLoc();
4600   int64_t NumValues;
4601   if (checkForValidSection() || parseAbsoluteExpression(NumValues))
4602     return true;
4603 
4604   if (NumValues < 0) {
4605     Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect");
4606     return false;
4607   }
4608 
4609   if (parseToken(AsmToken::Comma,
4610                  "unexpected token in '" + Twine(IDVal) + "' directive"))
4611     return true;
4612 
4613   APInt AsInt;
4614   if (parseRealValue(Semantics, AsInt))
4615     return true;
4616 
4617   if (parseToken(AsmToken::EndOfStatement,
4618                  "unexpected token in '" + Twine(IDVal) + "' directive"))
4619     return true;
4620 
4621   for (uint64_t i = 0, e = NumValues; i != e; ++i)
4622     getStreamer().EmitIntValue(AsInt.getLimitedValue(),
4623                                AsInt.getBitWidth() / 8);
4624 
4625   return false;
4626 }
4627 
4628 /// parseDirectiveDS
4629 /// ::= .ds.{b, d, l, p, s, w, x} expression
parseDirectiveDS(StringRef IDVal,unsigned Size)4630 bool AsmParser::parseDirectiveDS(StringRef IDVal, unsigned Size) {
4631   SMLoc NumValuesLoc = Lexer.getLoc();
4632   int64_t NumValues;
4633   if (checkForValidSection() || parseAbsoluteExpression(NumValues))
4634     return true;
4635 
4636   if (NumValues < 0) {
4637     Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect");
4638     return false;
4639   }
4640 
4641   if (parseToken(AsmToken::EndOfStatement,
4642                  "unexpected token in '" + Twine(IDVal) + "' directive"))
4643     return true;
4644 
4645   for (uint64_t i = 0, e = NumValues; i != e; ++i)
4646     getStreamer().emitFill(Size, 0);
4647 
4648   return false;
4649 }
4650 
4651 /// parseDirectiveLEB128
4652 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ]
parseDirectiveLEB128(bool Signed)4653 bool AsmParser::parseDirectiveLEB128(bool Signed) {
4654   if (checkForValidSection())
4655     return true;
4656 
4657   auto parseOp = [&]() -> bool {
4658     const MCExpr *Value;
4659     if (parseExpression(Value))
4660       return true;
4661     if (Signed)
4662       getStreamer().EmitSLEB128Value(Value);
4663     else
4664       getStreamer().EmitULEB128Value(Value);
4665     return false;
4666   };
4667 
4668   if (parseMany(parseOp))
4669     return addErrorSuffix(" in directive");
4670 
4671   return false;
4672 }
4673 
4674 /// parseDirectiveSymbolAttribute
4675 ///  ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ]
parseDirectiveSymbolAttribute(MCSymbolAttr Attr)4676 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) {
4677   auto parseOp = [&]() -> bool {
4678     StringRef Name;
4679     SMLoc Loc = getTok().getLoc();
4680     if (parseIdentifier(Name))
4681       return Error(Loc, "expected identifier");
4682     MCSymbol *Sym = getContext().getOrCreateSymbol(Name);
4683 
4684     // Assembler local symbols don't make any sense here. Complain loudly.
4685     if (Sym->isTemporary())
4686       return Error(Loc, "non-local symbol required");
4687 
4688     if (!getStreamer().EmitSymbolAttribute(Sym, Attr))
4689       return Error(Loc, "unable to emit symbol attribute");
4690     return false;
4691   };
4692 
4693   if (parseMany(parseOp))
4694     return addErrorSuffix(" in directive");
4695   return false;
4696 }
4697 
4698 /// parseDirectiveComm
4699 ///  ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ]
parseDirectiveComm(bool IsLocal)4700 bool AsmParser::parseDirectiveComm(bool IsLocal) {
4701   if (checkForValidSection())
4702     return true;
4703 
4704   SMLoc IDLoc = getLexer().getLoc();
4705   StringRef Name;
4706   if (parseIdentifier(Name))
4707     return TokError("expected identifier in directive");
4708 
4709   // Handle the identifier as the key symbol.
4710   MCSymbol *Sym = getContext().getOrCreateSymbol(Name);
4711 
4712   if (getLexer().isNot(AsmToken::Comma))
4713     return TokError("unexpected token in directive");
4714   Lex();
4715 
4716   int64_t Size;
4717   SMLoc SizeLoc = getLexer().getLoc();
4718   if (parseAbsoluteExpression(Size))
4719     return true;
4720 
4721   int64_t Pow2Alignment = 0;
4722   SMLoc Pow2AlignmentLoc;
4723   if (getLexer().is(AsmToken::Comma)) {
4724     Lex();
4725     Pow2AlignmentLoc = getLexer().getLoc();
4726     if (parseAbsoluteExpression(Pow2Alignment))
4727       return true;
4728 
4729     LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType();
4730     if (IsLocal && LCOMM == LCOMM::NoAlignment)
4731       return Error(Pow2AlignmentLoc, "alignment not supported on this target");
4732 
4733     // If this target takes alignments in bytes (not log) validate and convert.
4734     if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) ||
4735         (IsLocal && LCOMM == LCOMM::ByteAlignment)) {
4736       if (!isPowerOf2_64(Pow2Alignment))
4737         return Error(Pow2AlignmentLoc, "alignment must be a power of 2");
4738       Pow2Alignment = Log2_64(Pow2Alignment);
4739     }
4740   }
4741 
4742   if (parseToken(AsmToken::EndOfStatement,
4743                  "unexpected token in '.comm' or '.lcomm' directive"))
4744     return true;
4745 
4746   // NOTE: a size of zero for a .comm should create a undefined symbol
4747   // but a size of .lcomm creates a bss symbol of size zero.
4748   if (Size < 0)
4749     return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't "
4750                           "be less than zero");
4751 
4752   // NOTE: The alignment in the directive is a power of 2 value, the assembler
4753   // may internally end up wanting an alignment in bytes.
4754   // FIXME: Diagnose overflow.
4755   if (Pow2Alignment < 0)
4756     return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive "
4757                                    "alignment, can't be less than zero");
4758 
4759   Sym->redefineIfPossible();
4760   if (!Sym->isUndefined())
4761     return Error(IDLoc, "invalid symbol redefinition");
4762 
4763   // Create the Symbol as a common or local common with Size and Pow2Alignment
4764   if (IsLocal) {
4765     getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment);
4766     return false;
4767   }
4768 
4769   getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment);
4770   return false;
4771 }
4772 
4773 /// parseDirectiveAbort
4774 ///  ::= .abort [... message ...]
parseDirectiveAbort()4775 bool AsmParser::parseDirectiveAbort() {
4776   // FIXME: Use loc from directive.
4777   SMLoc Loc = getLexer().getLoc();
4778 
4779   StringRef Str = parseStringToEndOfStatement();
4780   if (parseToken(AsmToken::EndOfStatement,
4781                  "unexpected token in '.abort' directive"))
4782     return true;
4783 
4784   if (Str.empty())
4785     return Error(Loc, ".abort detected. Assembly stopping.");
4786   else
4787     return Error(Loc, ".abort '" + Str + "' detected. Assembly stopping.");
4788   // FIXME: Actually abort assembly here.
4789 
4790   return false;
4791 }
4792 
4793 /// parseDirectiveInclude
4794 ///  ::= .include "filename"
parseDirectiveInclude()4795 bool AsmParser::parseDirectiveInclude() {
4796   // Allow the strings to have escaped octal character sequence.
4797   std::string Filename;
4798   SMLoc IncludeLoc = getTok().getLoc();
4799 
4800   if (check(getTok().isNot(AsmToken::String),
4801             "expected string in '.include' directive") ||
4802       parseEscapedString(Filename) ||
4803       check(getTok().isNot(AsmToken::EndOfStatement),
4804             "unexpected token in '.include' directive") ||
4805       // Attempt to switch the lexer to the included file before consuming the
4806       // end of statement to avoid losing it when we switch.
4807       check(enterIncludeFile(Filename), IncludeLoc,
4808             "Could not find include file '" + Filename + "'"))
4809     return true;
4810 
4811   return false;
4812 }
4813 
4814 /// parseDirectiveIncbin
4815 ///  ::= .incbin "filename" [ , skip [ , count ] ]
parseDirectiveIncbin()4816 bool AsmParser::parseDirectiveIncbin() {
4817   // Allow the strings to have escaped octal character sequence.
4818   std::string Filename;
4819   SMLoc IncbinLoc = getTok().getLoc();
4820   if (check(getTok().isNot(AsmToken::String),
4821             "expected string in '.incbin' directive") ||
4822       parseEscapedString(Filename))
4823     return true;
4824 
4825   int64_t Skip = 0;
4826   const MCExpr *Count = nullptr;
4827   SMLoc SkipLoc, CountLoc;
4828   if (parseOptionalToken(AsmToken::Comma)) {
4829     // The skip expression can be omitted while specifying the count, e.g:
4830     //  .incbin "filename",,4
4831     if (getTok().isNot(AsmToken::Comma)) {
4832       if (parseTokenLoc(SkipLoc) || parseAbsoluteExpression(Skip))
4833         return true;
4834     }
4835     if (parseOptionalToken(AsmToken::Comma)) {
4836       CountLoc = getTok().getLoc();
4837       if (parseExpression(Count))
4838         return true;
4839     }
4840   }
4841 
4842   if (parseToken(AsmToken::EndOfStatement,
4843                  "unexpected token in '.incbin' directive"))
4844     return true;
4845 
4846   if (check(Skip < 0, SkipLoc, "skip is negative"))
4847     return true;
4848 
4849   // Attempt to process the included file.
4850   if (processIncbinFile(Filename, Skip, Count, CountLoc))
4851     return Error(IncbinLoc, "Could not find incbin file '" + Filename + "'");
4852   return false;
4853 }
4854 
4855 /// parseDirectiveIf
4856 /// ::= .if{,eq,ge,gt,le,lt,ne} expression
parseDirectiveIf(SMLoc DirectiveLoc,DirectiveKind DirKind)4857 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) {
4858   TheCondStack.push_back(TheCondState);
4859   TheCondState.TheCond = AsmCond::IfCond;
4860   if (TheCondState.Ignore) {
4861     eatToEndOfStatement();
4862   } else {
4863     int64_t ExprValue;
4864     if (parseAbsoluteExpression(ExprValue) ||
4865         parseToken(AsmToken::EndOfStatement,
4866                    "unexpected token in '.if' directive"))
4867       return true;
4868 
4869     switch (DirKind) {
4870     default:
4871       llvm_unreachable("unsupported directive");
4872     case DK_IF:
4873     case DK_IFNE:
4874       break;
4875     case DK_IFEQ:
4876       ExprValue = ExprValue == 0;
4877       break;
4878     case DK_IFGE:
4879       ExprValue = ExprValue >= 0;
4880       break;
4881     case DK_IFGT:
4882       ExprValue = ExprValue > 0;
4883       break;
4884     case DK_IFLE:
4885       ExprValue = ExprValue <= 0;
4886       break;
4887     case DK_IFLT:
4888       ExprValue = ExprValue < 0;
4889       break;
4890     }
4891 
4892     TheCondState.CondMet = ExprValue;
4893     TheCondState.Ignore = !TheCondState.CondMet;
4894   }
4895 
4896   return false;
4897 }
4898 
4899 /// parseDirectiveIfb
4900 /// ::= .ifb string
parseDirectiveIfb(SMLoc DirectiveLoc,bool ExpectBlank)4901 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) {
4902   TheCondStack.push_back(TheCondState);
4903   TheCondState.TheCond = AsmCond::IfCond;
4904 
4905   if (TheCondState.Ignore) {
4906     eatToEndOfStatement();
4907   } else {
4908     StringRef Str = parseStringToEndOfStatement();
4909 
4910     if (parseToken(AsmToken::EndOfStatement,
4911                    "unexpected token in '.ifb' directive"))
4912       return true;
4913 
4914     TheCondState.CondMet = ExpectBlank == Str.empty();
4915     TheCondState.Ignore = !TheCondState.CondMet;
4916   }
4917 
4918   return false;
4919 }
4920 
4921 /// parseDirectiveIfc
4922 /// ::= .ifc string1, string2
4923 /// ::= .ifnc string1, string2
parseDirectiveIfc(SMLoc DirectiveLoc,bool ExpectEqual)4924 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) {
4925   TheCondStack.push_back(TheCondState);
4926   TheCondState.TheCond = AsmCond::IfCond;
4927 
4928   if (TheCondState.Ignore) {
4929     eatToEndOfStatement();
4930   } else {
4931     StringRef Str1 = parseStringToComma();
4932 
4933     if (parseToken(AsmToken::Comma, "unexpected token in '.ifc' directive"))
4934       return true;
4935 
4936     StringRef Str2 = parseStringToEndOfStatement();
4937 
4938     if (parseToken(AsmToken::EndOfStatement,
4939                    "unexpected token in '.ifc' directive"))
4940       return true;
4941 
4942     TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim());
4943     TheCondState.Ignore = !TheCondState.CondMet;
4944   }
4945 
4946   return false;
4947 }
4948 
4949 /// parseDirectiveIfeqs
4950 ///   ::= .ifeqs string1, string2
parseDirectiveIfeqs(SMLoc DirectiveLoc,bool ExpectEqual)4951 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) {
4952   if (Lexer.isNot(AsmToken::String)) {
4953     if (ExpectEqual)
4954       return TokError("expected string parameter for '.ifeqs' directive");
4955     return TokError("expected string parameter for '.ifnes' directive");
4956   }
4957 
4958   StringRef String1 = getTok().getStringContents();
4959   Lex();
4960 
4961   if (Lexer.isNot(AsmToken::Comma)) {
4962     if (ExpectEqual)
4963       return TokError(
4964           "expected comma after first string for '.ifeqs' directive");
4965     return TokError("expected comma after first string for '.ifnes' directive");
4966   }
4967 
4968   Lex();
4969 
4970   if (Lexer.isNot(AsmToken::String)) {
4971     if (ExpectEqual)
4972       return TokError("expected string parameter for '.ifeqs' directive");
4973     return TokError("expected string parameter for '.ifnes' directive");
4974   }
4975 
4976   StringRef String2 = getTok().getStringContents();
4977   Lex();
4978 
4979   TheCondStack.push_back(TheCondState);
4980   TheCondState.TheCond = AsmCond::IfCond;
4981   TheCondState.CondMet = ExpectEqual == (String1 == String2);
4982   TheCondState.Ignore = !TheCondState.CondMet;
4983 
4984   return false;
4985 }
4986 
4987 /// parseDirectiveIfdef
4988 /// ::= .ifdef symbol
parseDirectiveIfdef(SMLoc DirectiveLoc,bool expect_defined)4989 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) {
4990   StringRef Name;
4991   TheCondStack.push_back(TheCondState);
4992   TheCondState.TheCond = AsmCond::IfCond;
4993 
4994   if (TheCondState.Ignore) {
4995     eatToEndOfStatement();
4996   } else {
4997     if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") ||
4998         parseToken(AsmToken::EndOfStatement, "unexpected token in '.ifdef'"))
4999       return true;
5000 
5001     MCSymbol *Sym = getContext().lookupSymbol(Name);
5002 
5003     if (expect_defined)
5004       TheCondState.CondMet = (Sym && !Sym->isUndefined());
5005     else
5006       TheCondState.CondMet = (!Sym || Sym->isUndefined());
5007     TheCondState.Ignore = !TheCondState.CondMet;
5008   }
5009 
5010   return false;
5011 }
5012 
5013 /// parseDirectiveElseIf
5014 /// ::= .elseif expression
parseDirectiveElseIf(SMLoc DirectiveLoc)5015 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) {
5016   if (TheCondState.TheCond != AsmCond::IfCond &&
5017       TheCondState.TheCond != AsmCond::ElseIfCond)
5018     return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an"
5019                                " .if or  an .elseif");
5020   TheCondState.TheCond = AsmCond::ElseIfCond;
5021 
5022   bool LastIgnoreState = false;
5023   if (!TheCondStack.empty())
5024     LastIgnoreState = TheCondStack.back().Ignore;
5025   if (LastIgnoreState || TheCondState.CondMet) {
5026     TheCondState.Ignore = true;
5027     eatToEndOfStatement();
5028   } else {
5029     int64_t ExprValue;
5030     if (parseAbsoluteExpression(ExprValue))
5031       return true;
5032 
5033     if (parseToken(AsmToken::EndOfStatement,
5034                    "unexpected token in '.elseif' directive"))
5035       return true;
5036 
5037     TheCondState.CondMet = ExprValue;
5038     TheCondState.Ignore = !TheCondState.CondMet;
5039   }
5040 
5041   return false;
5042 }
5043 
5044 /// parseDirectiveElse
5045 /// ::= .else
parseDirectiveElse(SMLoc DirectiveLoc)5046 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) {
5047   if (parseToken(AsmToken::EndOfStatement,
5048                  "unexpected token in '.else' directive"))
5049     return true;
5050 
5051   if (TheCondState.TheCond != AsmCond::IfCond &&
5052       TheCondState.TheCond != AsmCond::ElseIfCond)
5053     return Error(DirectiveLoc, "Encountered a .else that doesn't follow "
5054                                " an .if or an .elseif");
5055   TheCondState.TheCond = AsmCond::ElseCond;
5056   bool LastIgnoreState = false;
5057   if (!TheCondStack.empty())
5058     LastIgnoreState = TheCondStack.back().Ignore;
5059   if (LastIgnoreState || TheCondState.CondMet)
5060     TheCondState.Ignore = true;
5061   else
5062     TheCondState.Ignore = false;
5063 
5064   return false;
5065 }
5066 
5067 /// parseDirectiveEnd
5068 /// ::= .end
parseDirectiveEnd(SMLoc DirectiveLoc)5069 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) {
5070   if (parseToken(AsmToken::EndOfStatement,
5071                  "unexpected token in '.end' directive"))
5072     return true;
5073 
5074   while (Lexer.isNot(AsmToken::Eof))
5075     Lexer.Lex();
5076 
5077   return false;
5078 }
5079 
5080 /// parseDirectiveError
5081 ///   ::= .err
5082 ///   ::= .error [string]
parseDirectiveError(SMLoc L,bool WithMessage)5083 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) {
5084   if (!TheCondStack.empty()) {
5085     if (TheCondStack.back().Ignore) {
5086       eatToEndOfStatement();
5087       return false;
5088     }
5089   }
5090 
5091   if (!WithMessage)
5092     return Error(L, ".err encountered");
5093 
5094   StringRef Message = ".error directive invoked in source file";
5095   if (Lexer.isNot(AsmToken::EndOfStatement)) {
5096     if (Lexer.isNot(AsmToken::String))
5097       return TokError(".error argument must be a string");
5098 
5099     Message = getTok().getStringContents();
5100     Lex();
5101   }
5102 
5103   return Error(L, Message);
5104 }
5105 
5106 /// parseDirectiveWarning
5107 ///   ::= .warning [string]
parseDirectiveWarning(SMLoc L)5108 bool AsmParser::parseDirectiveWarning(SMLoc L) {
5109   if (!TheCondStack.empty()) {
5110     if (TheCondStack.back().Ignore) {
5111       eatToEndOfStatement();
5112       return false;
5113     }
5114   }
5115 
5116   StringRef Message = ".warning directive invoked in source file";
5117 
5118   if (!parseOptionalToken(AsmToken::EndOfStatement)) {
5119     if (Lexer.isNot(AsmToken::String))
5120       return TokError(".warning argument must be a string");
5121 
5122     Message = getTok().getStringContents();
5123     Lex();
5124     if (parseToken(AsmToken::EndOfStatement,
5125                    "expected end of statement in '.warning' directive"))
5126       return true;
5127   }
5128 
5129   return Warning(L, Message);
5130 }
5131 
5132 /// parseDirectiveEndIf
5133 /// ::= .endif
parseDirectiveEndIf(SMLoc DirectiveLoc)5134 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) {
5135   if (parseToken(AsmToken::EndOfStatement,
5136                  "unexpected token in '.endif' directive"))
5137     return true;
5138 
5139   if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty())
5140     return Error(DirectiveLoc, "Encountered a .endif that doesn't follow "
5141                                "an .if or .else");
5142   if (!TheCondStack.empty()) {
5143     TheCondState = TheCondStack.back();
5144     TheCondStack.pop_back();
5145   }
5146 
5147   return false;
5148 }
5149 
initializeDirectiveKindMap()5150 void AsmParser::initializeDirectiveKindMap() {
5151   DirectiveKindMap[".set"] = DK_SET;
5152   DirectiveKindMap[".equ"] = DK_EQU;
5153   DirectiveKindMap[".equiv"] = DK_EQUIV;
5154   DirectiveKindMap[".ascii"] = DK_ASCII;
5155   DirectiveKindMap[".asciz"] = DK_ASCIZ;
5156   DirectiveKindMap[".string"] = DK_STRING;
5157   DirectiveKindMap[".byte"] = DK_BYTE;
5158   DirectiveKindMap[".short"] = DK_SHORT;
5159   DirectiveKindMap[".value"] = DK_VALUE;
5160   DirectiveKindMap[".2byte"] = DK_2BYTE;
5161   DirectiveKindMap[".long"] = DK_LONG;
5162   DirectiveKindMap[".int"] = DK_INT;
5163   DirectiveKindMap[".4byte"] = DK_4BYTE;
5164   DirectiveKindMap[".quad"] = DK_QUAD;
5165   DirectiveKindMap[".8byte"] = DK_8BYTE;
5166   DirectiveKindMap[".octa"] = DK_OCTA;
5167   DirectiveKindMap[".single"] = DK_SINGLE;
5168   DirectiveKindMap[".float"] = DK_FLOAT;
5169   DirectiveKindMap[".double"] = DK_DOUBLE;
5170   DirectiveKindMap[".align"] = DK_ALIGN;
5171   DirectiveKindMap[".align32"] = DK_ALIGN32;
5172   DirectiveKindMap[".balign"] = DK_BALIGN;
5173   DirectiveKindMap[".balignw"] = DK_BALIGNW;
5174   DirectiveKindMap[".balignl"] = DK_BALIGNL;
5175   DirectiveKindMap[".p2align"] = DK_P2ALIGN;
5176   DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW;
5177   DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL;
5178   DirectiveKindMap[".org"] = DK_ORG;
5179   DirectiveKindMap[".fill"] = DK_FILL;
5180   DirectiveKindMap[".zero"] = DK_ZERO;
5181   DirectiveKindMap[".extern"] = DK_EXTERN;
5182   DirectiveKindMap[".globl"] = DK_GLOBL;
5183   DirectiveKindMap[".global"] = DK_GLOBAL;
5184   DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE;
5185   DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP;
5186   DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER;
5187   DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN;
5188   DirectiveKindMap[".reference"] = DK_REFERENCE;
5189   DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION;
5190   DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE;
5191   DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN;
5192   DirectiveKindMap[".comm"] = DK_COMM;
5193   DirectiveKindMap[".common"] = DK_COMMON;
5194   DirectiveKindMap[".lcomm"] = DK_LCOMM;
5195   DirectiveKindMap[".abort"] = DK_ABORT;
5196   DirectiveKindMap[".include"] = DK_INCLUDE;
5197   DirectiveKindMap[".incbin"] = DK_INCBIN;
5198   DirectiveKindMap[".code16"] = DK_CODE16;
5199   DirectiveKindMap[".code16gcc"] = DK_CODE16GCC;
5200   DirectiveKindMap[".rept"] = DK_REPT;
5201   DirectiveKindMap[".rep"] = DK_REPT;
5202   DirectiveKindMap[".irp"] = DK_IRP;
5203   DirectiveKindMap[".irpc"] = DK_IRPC;
5204   DirectiveKindMap[".endr"] = DK_ENDR;
5205   DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE;
5206   DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK;
5207   DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK;
5208   DirectiveKindMap[".if"] = DK_IF;
5209   DirectiveKindMap[".ifeq"] = DK_IFEQ;
5210   DirectiveKindMap[".ifge"] = DK_IFGE;
5211   DirectiveKindMap[".ifgt"] = DK_IFGT;
5212   DirectiveKindMap[".ifle"] = DK_IFLE;
5213   DirectiveKindMap[".iflt"] = DK_IFLT;
5214   DirectiveKindMap[".ifne"] = DK_IFNE;
5215   DirectiveKindMap[".ifb"] = DK_IFB;
5216   DirectiveKindMap[".ifnb"] = DK_IFNB;
5217   DirectiveKindMap[".ifc"] = DK_IFC;
5218   DirectiveKindMap[".ifeqs"] = DK_IFEQS;
5219   DirectiveKindMap[".ifnc"] = DK_IFNC;
5220   DirectiveKindMap[".ifnes"] = DK_IFNES;
5221   DirectiveKindMap[".ifdef"] = DK_IFDEF;
5222   DirectiveKindMap[".ifndef"] = DK_IFNDEF;
5223   DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF;
5224   DirectiveKindMap[".elseif"] = DK_ELSEIF;
5225   DirectiveKindMap[".else"] = DK_ELSE;
5226   DirectiveKindMap[".end"] = DK_END;
5227   DirectiveKindMap[".endif"] = DK_ENDIF;
5228   DirectiveKindMap[".skip"] = DK_SKIP;
5229   DirectiveKindMap[".space"] = DK_SPACE;
5230   DirectiveKindMap[".file"] = DK_FILE;
5231   DirectiveKindMap[".line"] = DK_LINE;
5232   DirectiveKindMap[".loc"] = DK_LOC;
5233   DirectiveKindMap[".stabs"] = DK_STABS;
5234   DirectiveKindMap[".cv_file"] = DK_CV_FILE;
5235   DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID;
5236   DirectiveKindMap[".cv_loc"] = DK_CV_LOC;
5237   DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE;
5238   DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE;
5239   DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID;
5240   DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE;
5241   DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE;
5242   DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS;
5243   DirectiveKindMap[".cv_filechecksumoffset"] = DK_CV_FILECHECKSUM_OFFSET;
5244   DirectiveKindMap[".cv_fpo_data"] = DK_CV_FPO_DATA;
5245   DirectiveKindMap[".sleb128"] = DK_SLEB128;
5246   DirectiveKindMap[".uleb128"] = DK_ULEB128;
5247   DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS;
5248   DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC;
5249   DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC;
5250   DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA;
5251   DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET;
5252   DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET;
5253   DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER;
5254   DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET;
5255   DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET;
5256   DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY;
5257   DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA;
5258   DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE;
5259   DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE;
5260   DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE;
5261   DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE;
5262   DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE;
5263   DirectiveKindMap[".cfi_return_column"] = DK_CFI_RETURN_COLUMN;
5264   DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME;
5265   DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED;
5266   DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER;
5267   DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE;
5268   DirectiveKindMap[".macros_on"] = DK_MACROS_ON;
5269   DirectiveKindMap[".macros_off"] = DK_MACROS_OFF;
5270   DirectiveKindMap[".macro"] = DK_MACRO;
5271   DirectiveKindMap[".exitm"] = DK_EXITM;
5272   DirectiveKindMap[".endm"] = DK_ENDM;
5273   DirectiveKindMap[".endmacro"] = DK_ENDMACRO;
5274   DirectiveKindMap[".purgem"] = DK_PURGEM;
5275   DirectiveKindMap[".err"] = DK_ERR;
5276   DirectiveKindMap[".error"] = DK_ERROR;
5277   DirectiveKindMap[".warning"] = DK_WARNING;
5278   DirectiveKindMap[".altmacro"] = DK_ALTMACRO;
5279   DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO;
5280   DirectiveKindMap[".reloc"] = DK_RELOC;
5281   DirectiveKindMap[".dc"] = DK_DC;
5282   DirectiveKindMap[".dc.a"] = DK_DC_A;
5283   DirectiveKindMap[".dc.b"] = DK_DC_B;
5284   DirectiveKindMap[".dc.d"] = DK_DC_D;
5285   DirectiveKindMap[".dc.l"] = DK_DC_L;
5286   DirectiveKindMap[".dc.s"] = DK_DC_S;
5287   DirectiveKindMap[".dc.w"] = DK_DC_W;
5288   DirectiveKindMap[".dc.x"] = DK_DC_X;
5289   DirectiveKindMap[".dcb"] = DK_DCB;
5290   DirectiveKindMap[".dcb.b"] = DK_DCB_B;
5291   DirectiveKindMap[".dcb.d"] = DK_DCB_D;
5292   DirectiveKindMap[".dcb.l"] = DK_DCB_L;
5293   DirectiveKindMap[".dcb.s"] = DK_DCB_S;
5294   DirectiveKindMap[".dcb.w"] = DK_DCB_W;
5295   DirectiveKindMap[".dcb.x"] = DK_DCB_X;
5296   DirectiveKindMap[".ds"] = DK_DS;
5297   DirectiveKindMap[".ds.b"] = DK_DS_B;
5298   DirectiveKindMap[".ds.d"] = DK_DS_D;
5299   DirectiveKindMap[".ds.l"] = DK_DS_L;
5300   DirectiveKindMap[".ds.p"] = DK_DS_P;
5301   DirectiveKindMap[".ds.s"] = DK_DS_S;
5302   DirectiveKindMap[".ds.w"] = DK_DS_W;
5303   DirectiveKindMap[".ds.x"] = DK_DS_X;
5304   DirectiveKindMap[".print"] = DK_PRINT;
5305   DirectiveKindMap[".addrsig"] = DK_ADDRSIG;
5306   DirectiveKindMap[".addrsig_sym"] = DK_ADDRSIG_SYM;
5307 }
5308 
parseMacroLikeBody(SMLoc DirectiveLoc)5309 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) {
5310   AsmToken EndToken, StartToken = getTok();
5311 
5312   unsigned NestLevel = 0;
5313   while (true) {
5314     // Check whether we have reached the end of the file.
5315     if (getLexer().is(AsmToken::Eof)) {
5316       printError(DirectiveLoc, "no matching '.endr' in definition");
5317       return nullptr;
5318     }
5319 
5320     if (Lexer.is(AsmToken::Identifier) &&
5321         (getTok().getIdentifier() == ".rep" ||
5322          getTok().getIdentifier() == ".rept" ||
5323          getTok().getIdentifier() == ".irp" ||
5324          getTok().getIdentifier() == ".irpc")) {
5325       ++NestLevel;
5326     }
5327 
5328     // Otherwise, check whether we have reached the .endr.
5329     if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") {
5330       if (NestLevel == 0) {
5331         EndToken = getTok();
5332         Lex();
5333         if (Lexer.isNot(AsmToken::EndOfStatement)) {
5334           printError(getTok().getLoc(),
5335                      "unexpected token in '.endr' directive");
5336           return nullptr;
5337         }
5338         break;
5339       }
5340       --NestLevel;
5341     }
5342 
5343     // Otherwise, scan till the end of the statement.
5344     eatToEndOfStatement();
5345   }
5346 
5347   const char *BodyStart = StartToken.getLoc().getPointer();
5348   const char *BodyEnd = EndToken.getLoc().getPointer();
5349   StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart);
5350 
5351   // We Are Anonymous.
5352   MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters());
5353   return &MacroLikeBodies.back();
5354 }
5355 
instantiateMacroLikeBody(MCAsmMacro * M,SMLoc DirectiveLoc,raw_svector_ostream & OS)5356 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc,
5357                                          raw_svector_ostream &OS) {
5358   OS << ".endr\n";
5359 
5360   std::unique_ptr<MemoryBuffer> Instantiation =
5361       MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>");
5362 
5363   // Create the macro instantiation object and add to the current macro
5364   // instantiation stack.
5365   MacroInstantiation *MI = new MacroInstantiation(
5366       DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size());
5367   ActiveMacros.push_back(MI);
5368 
5369   // Jump to the macro instantiation and prime the lexer.
5370   CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc());
5371   Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer());
5372   Lex();
5373 }
5374 
5375 /// parseDirectiveRept
5376 ///   ::= .rep | .rept count
parseDirectiveRept(SMLoc DirectiveLoc,StringRef Dir)5377 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) {
5378   const MCExpr *CountExpr;
5379   SMLoc CountLoc = getTok().getLoc();
5380   if (parseExpression(CountExpr))
5381     return true;
5382 
5383   int64_t Count;
5384   if (!CountExpr->evaluateAsAbsolute(Count, getStreamer().getAssemblerPtr())) {
5385     return Error(CountLoc, "unexpected token in '" + Dir + "' directive");
5386   }
5387 
5388   if (check(Count < 0, CountLoc, "Count is negative") ||
5389       parseToken(AsmToken::EndOfStatement,
5390                  "unexpected token in '" + Dir + "' directive"))
5391     return true;
5392 
5393   // Lex the rept definition.
5394   MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc);
5395   if (!M)
5396     return true;
5397 
5398   // Macro instantiation is lexical, unfortunately. We construct a new buffer
5399   // to hold the macro body with substitutions.
5400   SmallString<256> Buf;
5401   raw_svector_ostream OS(Buf);
5402   while (Count--) {
5403     // Note that the AtPseudoVariable is disabled for instantiations of .rep(t).
5404     if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc()))
5405       return true;
5406   }
5407   instantiateMacroLikeBody(M, DirectiveLoc, OS);
5408 
5409   return false;
5410 }
5411 
5412 /// parseDirectiveIrp
5413 /// ::= .irp symbol,values
parseDirectiveIrp(SMLoc DirectiveLoc)5414 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) {
5415   MCAsmMacroParameter Parameter;
5416   MCAsmMacroArguments A;
5417   if (check(parseIdentifier(Parameter.Name),
5418             "expected identifier in '.irp' directive") ||
5419       parseToken(AsmToken::Comma, "expected comma in '.irp' directive") ||
5420       parseMacroArguments(nullptr, A) ||
5421       parseToken(AsmToken::EndOfStatement, "expected End of Statement"))
5422     return true;
5423 
5424   // Lex the irp definition.
5425   MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc);
5426   if (!M)
5427     return true;
5428 
5429   // Macro instantiation is lexical, unfortunately. We construct a new buffer
5430   // to hold the macro body with substitutions.
5431   SmallString<256> Buf;
5432   raw_svector_ostream OS(Buf);
5433 
5434   for (const MCAsmMacroArgument &Arg : A) {
5435     // Note that the AtPseudoVariable is enabled for instantiations of .irp.
5436     // This is undocumented, but GAS seems to support it.
5437     if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc()))
5438       return true;
5439   }
5440 
5441   instantiateMacroLikeBody(M, DirectiveLoc, OS);
5442 
5443   return false;
5444 }
5445 
5446 /// parseDirectiveIrpc
5447 /// ::= .irpc symbol,values
parseDirectiveIrpc(SMLoc DirectiveLoc)5448 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) {
5449   MCAsmMacroParameter Parameter;
5450   MCAsmMacroArguments A;
5451 
5452   if (check(parseIdentifier(Parameter.Name),
5453             "expected identifier in '.irpc' directive") ||
5454       parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") ||
5455       parseMacroArguments(nullptr, A))
5456     return true;
5457 
5458   if (A.size() != 1 || A.front().size() != 1)
5459     return TokError("unexpected token in '.irpc' directive");
5460 
5461   // Eat the end of statement.
5462   if (parseToken(AsmToken::EndOfStatement, "expected end of statement"))
5463     return true;
5464 
5465   // Lex the irpc definition.
5466   MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc);
5467   if (!M)
5468     return true;
5469 
5470   // Macro instantiation is lexical, unfortunately. We construct a new buffer
5471   // to hold the macro body with substitutions.
5472   SmallString<256> Buf;
5473   raw_svector_ostream OS(Buf);
5474 
5475   StringRef Values = A.front().front().getString();
5476   for (std::size_t I = 0, End = Values.size(); I != End; ++I) {
5477     MCAsmMacroArgument Arg;
5478     Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1));
5479 
5480     // Note that the AtPseudoVariable is enabled for instantiations of .irpc.
5481     // This is undocumented, but GAS seems to support it.
5482     if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc()))
5483       return true;
5484   }
5485 
5486   instantiateMacroLikeBody(M, DirectiveLoc, OS);
5487 
5488   return false;
5489 }
5490 
parseDirectiveEndr(SMLoc DirectiveLoc)5491 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) {
5492   if (ActiveMacros.empty())
5493     return TokError("unmatched '.endr' directive");
5494 
5495   // The only .repl that should get here are the ones created by
5496   // instantiateMacroLikeBody.
5497   assert(getLexer().is(AsmToken::EndOfStatement));
5498 
5499   handleMacroExit();
5500   return false;
5501 }
5502 
parseDirectiveMSEmit(SMLoc IDLoc,ParseStatementInfo & Info,size_t Len)5503 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info,
5504                                      size_t Len) {
5505   const MCExpr *Value;
5506   SMLoc ExprLoc = getLexer().getLoc();
5507   if (parseExpression(Value))
5508     return true;
5509   const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value);
5510   if (!MCE)
5511     return Error(ExprLoc, "unexpected expression in _emit");
5512   uint64_t IntValue = MCE->getValue();
5513   if (!isUInt<8>(IntValue) && !isInt<8>(IntValue))
5514     return Error(ExprLoc, "literal value out of range for directive");
5515 
5516   Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len);
5517   return false;
5518 }
5519 
parseDirectiveMSAlign(SMLoc IDLoc,ParseStatementInfo & Info)5520 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) {
5521   const MCExpr *Value;
5522   SMLoc ExprLoc = getLexer().getLoc();
5523   if (parseExpression(Value))
5524     return true;
5525   const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value);
5526   if (!MCE)
5527     return Error(ExprLoc, "unexpected expression in align");
5528   uint64_t IntValue = MCE->getValue();
5529   if (!isPowerOf2_64(IntValue))
5530     return Error(ExprLoc, "literal value not a power of two greater then zero");
5531 
5532   Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue));
5533   return false;
5534 }
5535 
parseDirectivePrint(SMLoc DirectiveLoc)5536 bool AsmParser::parseDirectivePrint(SMLoc DirectiveLoc) {
5537   const AsmToken StrTok = getTok();
5538   Lex();
5539   if (StrTok.isNot(AsmToken::String) || StrTok.getString().front() != '"')
5540     return Error(DirectiveLoc, "expected double quoted string after .print");
5541   if (parseToken(AsmToken::EndOfStatement, "expected end of statement"))
5542     return true;
5543   llvm::outs() << StrTok.getStringContents() << '\n';
5544   return false;
5545 }
5546 
parseDirectiveAddrsig()5547 bool AsmParser::parseDirectiveAddrsig() {
5548   getStreamer().EmitAddrsig();
5549   return false;
5550 }
5551 
parseDirectiveAddrsigSym()5552 bool AsmParser::parseDirectiveAddrsigSym() {
5553   StringRef Name;
5554   if (check(parseIdentifier(Name),
5555             "expected identifier in '.addrsig_sym' directive"))
5556     return true;
5557   MCSymbol *Sym = getContext().getOrCreateSymbol(Name);
5558   getStreamer().EmitAddrsigSym(Sym);
5559   return false;
5560 }
5561 
5562 // We are comparing pointers, but the pointers are relative to a single string.
5563 // Thus, this should always be deterministic.
rewritesSort(const AsmRewrite * AsmRewriteA,const AsmRewrite * AsmRewriteB)5564 static int rewritesSort(const AsmRewrite *AsmRewriteA,
5565                         const AsmRewrite *AsmRewriteB) {
5566   if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer())
5567     return -1;
5568   if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer())
5569     return 1;
5570 
5571   // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output
5572   // rewrite to the same location.  Make sure the SizeDirective rewrite is
5573   // performed first, then the Imm/ImmPrefix and finally the Input/Output.  This
5574   // ensures the sort algorithm is stable.
5575   if (AsmRewritePrecedence[AsmRewriteA->Kind] >
5576       AsmRewritePrecedence[AsmRewriteB->Kind])
5577     return -1;
5578 
5579   if (AsmRewritePrecedence[AsmRewriteA->Kind] <
5580       AsmRewritePrecedence[AsmRewriteB->Kind])
5581     return 1;
5582   llvm_unreachable("Unstable rewrite sort.");
5583 }
5584 
parseMSInlineAsm(void * AsmLoc,std::string & AsmString,unsigned & NumOutputs,unsigned & NumInputs,SmallVectorImpl<std::pair<void *,bool>> & OpDecls,SmallVectorImpl<std::string> & Constraints,SmallVectorImpl<std::string> & Clobbers,const MCInstrInfo * MII,const MCInstPrinter * IP,MCAsmParserSemaCallback & SI)5585 bool AsmParser::parseMSInlineAsm(
5586     void *AsmLoc, std::string &AsmString, unsigned &NumOutputs,
5587     unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls,
5588     SmallVectorImpl<std::string> &Constraints,
5589     SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII,
5590     const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) {
5591   SmallVector<void *, 4> InputDecls;
5592   SmallVector<void *, 4> OutputDecls;
5593   SmallVector<bool, 4> InputDeclsAddressOf;
5594   SmallVector<bool, 4> OutputDeclsAddressOf;
5595   SmallVector<std::string, 4> InputConstraints;
5596   SmallVector<std::string, 4> OutputConstraints;
5597   SmallVector<unsigned, 4> ClobberRegs;
5598 
5599   SmallVector<AsmRewrite, 4> AsmStrRewrites;
5600 
5601   // Prime the lexer.
5602   Lex();
5603 
5604   // While we have input, parse each statement.
5605   unsigned InputIdx = 0;
5606   unsigned OutputIdx = 0;
5607   while (getLexer().isNot(AsmToken::Eof)) {
5608     // Parse curly braces marking block start/end
5609     if (parseCurlyBlockScope(AsmStrRewrites))
5610       continue;
5611 
5612     ParseStatementInfo Info(&AsmStrRewrites);
5613     bool StatementErr = parseStatement(Info, &SI);
5614 
5615     if (StatementErr || Info.ParseError) {
5616       // Emit pending errors if any exist.
5617       printPendingErrors();
5618       return true;
5619     }
5620 
5621     // No pending error should exist here.
5622     assert(!hasPendingError() && "unexpected error from parseStatement");
5623 
5624     if (Info.Opcode == ~0U)
5625       continue;
5626 
5627     const MCInstrDesc &Desc = MII->get(Info.Opcode);
5628 
5629     // Build the list of clobbers, outputs and inputs.
5630     for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) {
5631       MCParsedAsmOperand &Operand = *Info.ParsedOperands[i];
5632 
5633       // Immediate.
5634       if (Operand.isImm())
5635         continue;
5636 
5637       // Register operand.
5638       if (Operand.isReg() && !Operand.needAddressOf() &&
5639           !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) {
5640         unsigned NumDefs = Desc.getNumDefs();
5641         // Clobber.
5642         if (NumDefs && Operand.getMCOperandNum() < NumDefs)
5643           ClobberRegs.push_back(Operand.getReg());
5644         continue;
5645       }
5646 
5647       // Expr/Input or Output.
5648       StringRef SymName = Operand.getSymName();
5649       if (SymName.empty())
5650         continue;
5651 
5652       void *OpDecl = Operand.getOpDecl();
5653       if (!OpDecl)
5654         continue;
5655 
5656       bool isOutput = (i == 1) && Desc.mayStore();
5657       SMLoc Start = SMLoc::getFromPointer(SymName.data());
5658       if (isOutput) {
5659         ++InputIdx;
5660         OutputDecls.push_back(OpDecl);
5661         OutputDeclsAddressOf.push_back(Operand.needAddressOf());
5662         OutputConstraints.push_back(("=" + Operand.getConstraint()).str());
5663         AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size());
5664       } else {
5665         InputDecls.push_back(OpDecl);
5666         InputDeclsAddressOf.push_back(Operand.needAddressOf());
5667         InputConstraints.push_back(Operand.getConstraint().str());
5668         AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size());
5669       }
5670     }
5671 
5672     // Consider implicit defs to be clobbers.  Think of cpuid and push.
5673     ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(),
5674                                 Desc.getNumImplicitDefs());
5675     ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end());
5676   }
5677 
5678   // Set the number of Outputs and Inputs.
5679   NumOutputs = OutputDecls.size();
5680   NumInputs = InputDecls.size();
5681 
5682   // Set the unique clobbers.
5683   array_pod_sort(ClobberRegs.begin(), ClobberRegs.end());
5684   ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()),
5685                     ClobberRegs.end());
5686   Clobbers.assign(ClobberRegs.size(), std::string());
5687   for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) {
5688     raw_string_ostream OS(Clobbers[I]);
5689     IP->printRegName(OS, ClobberRegs[I]);
5690   }
5691 
5692   // Merge the various outputs and inputs.  Output are expected first.
5693   if (NumOutputs || NumInputs) {
5694     unsigned NumExprs = NumOutputs + NumInputs;
5695     OpDecls.resize(NumExprs);
5696     Constraints.resize(NumExprs);
5697     for (unsigned i = 0; i < NumOutputs; ++i) {
5698       OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]);
5699       Constraints[i] = OutputConstraints[i];
5700     }
5701     for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) {
5702       OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]);
5703       Constraints[j] = InputConstraints[i];
5704     }
5705   }
5706 
5707   // Build the IR assembly string.
5708   std::string AsmStringIR;
5709   raw_string_ostream OS(AsmStringIR);
5710   StringRef ASMString =
5711       SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer();
5712   const char *AsmStart = ASMString.begin();
5713   const char *AsmEnd = ASMString.end();
5714   array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort);
5715   for (const AsmRewrite &AR : AsmStrRewrites) {
5716     AsmRewriteKind Kind = AR.Kind;
5717 
5718     const char *Loc = AR.Loc.getPointer();
5719     assert(Loc >= AsmStart && "Expected Loc to be at or after Start!");
5720 
5721     // Emit everything up to the immediate/expression.
5722     if (unsigned Len = Loc - AsmStart)
5723       OS << StringRef(AsmStart, Len);
5724 
5725     // Skip the original expression.
5726     if (Kind == AOK_Skip) {
5727       AsmStart = Loc + AR.Len;
5728       continue;
5729     }
5730 
5731     unsigned AdditionalSkip = 0;
5732     // Rewrite expressions in $N notation.
5733     switch (Kind) {
5734     default:
5735       break;
5736     case AOK_IntelExpr:
5737       assert(AR.IntelExp.isValid() && "cannot write invalid intel expression");
5738       if (AR.IntelExp.NeedBracs)
5739         OS << "[";
5740       if (AR.IntelExp.hasBaseReg())
5741         OS << AR.IntelExp.BaseReg;
5742       if (AR.IntelExp.hasIndexReg())
5743         OS << (AR.IntelExp.hasBaseReg() ? " + " : "")
5744            << AR.IntelExp.IndexReg;
5745       if (AR.IntelExp.Scale > 1)
5746           OS << " * $$" << AR.IntelExp.Scale;
5747       if (AR.IntelExp.Imm || !AR.IntelExp.hasRegs())
5748         OS << (AR.IntelExp.hasRegs() ? " + $$" : "$$") << AR.IntelExp.Imm;
5749       if (AR.IntelExp.NeedBracs)
5750         OS << "]";
5751       break;
5752     case AOK_Label:
5753       OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label;
5754       break;
5755     case AOK_Input:
5756       OS << '$' << InputIdx++;
5757       break;
5758     case AOK_Output:
5759       OS << '$' << OutputIdx++;
5760       break;
5761     case AOK_SizeDirective:
5762       switch (AR.Val) {
5763       default: break;
5764       case 8:  OS << "byte ptr "; break;
5765       case 16: OS << "word ptr "; break;
5766       case 32: OS << "dword ptr "; break;
5767       case 64: OS << "qword ptr "; break;
5768       case 80: OS << "xword ptr "; break;
5769       case 128: OS << "xmmword ptr "; break;
5770       case 256: OS << "ymmword ptr "; break;
5771       }
5772       break;
5773     case AOK_Emit:
5774       OS << ".byte";
5775       break;
5776     case AOK_Align: {
5777       // MS alignment directives are measured in bytes. If the native assembler
5778       // measures alignment in bytes, we can pass it straight through.
5779       OS << ".align";
5780       if (getContext().getAsmInfo()->getAlignmentIsInBytes())
5781         break;
5782 
5783       // Alignment is in log2 form, so print that instead and skip the original
5784       // immediate.
5785       unsigned Val = AR.Val;
5786       OS << ' ' << Val;
5787       assert(Val < 10 && "Expected alignment less then 2^10.");
5788       AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4;
5789       break;
5790     }
5791     case AOK_EVEN:
5792       OS << ".even";
5793       break;
5794     case AOK_EndOfStatement:
5795       OS << "\n\t";
5796       break;
5797     }
5798 
5799     // Skip the original expression.
5800     AsmStart = Loc + AR.Len + AdditionalSkip;
5801   }
5802 
5803   // Emit the remainder of the asm string.
5804   if (AsmStart != AsmEnd)
5805     OS << StringRef(AsmStart, AsmEnd - AsmStart);
5806 
5807   AsmString = OS.str();
5808   return false;
5809 }
5810 
5811 namespace llvm {
5812 namespace MCParserUtils {
5813 
5814 /// Returns whether the given symbol is used anywhere in the given expression,
5815 /// or subexpressions.
isSymbolUsedInExpression(const MCSymbol * Sym,const MCExpr * Value)5816 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) {
5817   switch (Value->getKind()) {
5818   case MCExpr::Binary: {
5819     const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value);
5820     return isSymbolUsedInExpression(Sym, BE->getLHS()) ||
5821            isSymbolUsedInExpression(Sym, BE->getRHS());
5822   }
5823   case MCExpr::Target:
5824   case MCExpr::Constant:
5825     return false;
5826   case MCExpr::SymbolRef: {
5827     const MCSymbol &S =
5828         static_cast<const MCSymbolRefExpr *>(Value)->getSymbol();
5829     if (S.isVariable())
5830       return isSymbolUsedInExpression(Sym, S.getVariableValue());
5831     return &S == Sym;
5832   }
5833   case MCExpr::Unary:
5834     return isSymbolUsedInExpression(
5835         Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr());
5836   }
5837 
5838   llvm_unreachable("Unknown expr kind!");
5839 }
5840 
parseAssignmentExpression(StringRef Name,bool allow_redef,MCAsmParser & Parser,MCSymbol * & Sym,const MCExpr * & Value)5841 bool parseAssignmentExpression(StringRef Name, bool allow_redef,
5842                                MCAsmParser &Parser, MCSymbol *&Sym,
5843                                const MCExpr *&Value) {
5844 
5845   // FIXME: Use better location, we should use proper tokens.
5846   SMLoc EqualLoc = Parser.getTok().getLoc();
5847   if (Parser.parseExpression(Value))
5848     return Parser.TokError("missing expression");
5849 
5850   // Note: we don't count b as used in "a = b". This is to allow
5851   // a = b
5852   // b = c
5853 
5854   if (Parser.parseToken(AsmToken::EndOfStatement))
5855     return true;
5856 
5857   // Validate that the LHS is allowed to be a variable (either it has not been
5858   // used as a symbol, or it is an absolute symbol).
5859   Sym = Parser.getContext().lookupSymbol(Name);
5860   if (Sym) {
5861     // Diagnose assignment to a label.
5862     //
5863     // FIXME: Diagnostics. Note the location of the definition as a label.
5864     // FIXME: Diagnose assignment to protected identifier (e.g., register name).
5865     if (isSymbolUsedInExpression(Sym, Value))
5866       return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'");
5867     else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() &&
5868              !Sym->isVariable())
5869       ; // Allow redefinitions of undefined symbols only used in directives.
5870     else if (Sym->isVariable() && !Sym->isUsed() && allow_redef)
5871       ; // Allow redefinitions of variables that haven't yet been used.
5872     else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef))
5873       return Parser.Error(EqualLoc, "redefinition of '" + Name + "'");
5874     else if (!Sym->isVariable())
5875       return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'");
5876     else if (!isa<MCConstantExpr>(Sym->getVariableValue()))
5877       return Parser.Error(EqualLoc,
5878                           "invalid reassignment of non-absolute variable '" +
5879                               Name + "'");
5880   } else if (Name == ".") {
5881     Parser.getStreamer().emitValueToOffset(Value, 0, EqualLoc);
5882     return false;
5883   } else
5884     Sym = Parser.getContext().getOrCreateSymbol(Name);
5885 
5886   Sym->setRedefinable(allow_redef);
5887 
5888   return false;
5889 }
5890 
5891 } // end namespace MCParserUtils
5892 } // end namespace llvm
5893 
5894 /// Create an MCAsmParser instance.
createMCAsmParser(SourceMgr & SM,MCContext & C,MCStreamer & Out,const MCAsmInfo & MAI,unsigned CB)5895 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C,
5896                                      MCStreamer &Out, const MCAsmInfo &MAI,
5897                                      unsigned CB) {
5898   return new AsmParser(SM, C, Out, MAI, CB);
5899 }
5900