AsmParser.cpp revision 202878 
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/MC/MCParser/AsmParser.h"
15#include "llvm/ADT/SmallString.h"
16#include "llvm/ADT/Twine.h"
17#include "llvm/MC/MCContext.h"
18#include "llvm/MC/MCExpr.h"
19#include "llvm/MC/MCInst.h"
20#include "llvm/MC/MCSectionMachO.h"
21#include "llvm/MC/MCStreamer.h"
22#include "llvm/MC/MCSymbol.h"
23#include "llvm/MC/MCValue.h"
24#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
25#include "llvm/Support/Compiler.h"
26#include "llvm/Support/SourceMgr.h"
27#include "llvm/Support/raw_ostream.h"
28#include "llvm/Target/TargetAsmParser.h"
29using namespace llvm;
30
31
32enum { DEFAULT_ADDRSPACE = 0 };
33
34// Mach-O section uniquing.
35//
36// FIXME: Figure out where this should live, it should be shared by
37// TargetLoweringObjectFile.
38typedef StringMap<const MCSectionMachO*> MachOUniqueMapTy;
39
40AsmParser::AsmParser(SourceMgr &_SM, MCContext &_Ctx, MCStreamer &_Out,
41                     const MCAsmInfo &_MAI)
42  : Lexer(_MAI), Ctx(_Ctx), Out(_Out), SrcMgr(_SM), TargetParser(0),
43    CurBuffer(0), SectionUniquingMap(0) {
44  Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer));
45
46  // Debugging directives.
47  AddDirectiveHandler(".file", &AsmParser::ParseDirectiveFile);
48  AddDirectiveHandler(".line", &AsmParser::ParseDirectiveLine);
49  AddDirectiveHandler(".loc", &AsmParser::ParseDirectiveLoc);
50}
51
52
53
54AsmParser::~AsmParser() {
55  // If we have the MachO uniquing map, free it.
56  delete (MachOUniqueMapTy*)SectionUniquingMap;
57}
58
59const MCSection *AsmParser::getMachOSection(const StringRef &Segment,
60                                            const StringRef &Section,
61                                            unsigned TypeAndAttributes,
62                                            unsigned Reserved2,
63                                            SectionKind Kind) const {
64  // We unique sections by their segment/section pair.  The returned section
65  // may not have the same flags as the requested section, if so this should be
66  // diagnosed by the client as an error.
67
68  // Create the map if it doesn't already exist.
69  if (SectionUniquingMap == 0)
70    SectionUniquingMap = new MachOUniqueMapTy();
71  MachOUniqueMapTy &Map = *(MachOUniqueMapTy*)SectionUniquingMap;
72
73  // Form the name to look up.
74  SmallString<64> Name;
75  Name += Segment;
76  Name.push_back(',');
77  Name += Section;
78
79  // Do the lookup, if we have a hit, return it.
80  const MCSectionMachO *&Entry = Map[Name.str()];
81
82  // FIXME: This should validate the type and attributes.
83  if (Entry) return Entry;
84
85  // Otherwise, return a new section.
86  return Entry = MCSectionMachO::Create(Segment, Section, TypeAndAttributes,
87                                        Reserved2, Kind, Ctx);
88}
89
90void AsmParser::Warning(SMLoc L, const Twine &Msg) {
91  PrintMessage(L, Msg.str(), "warning");
92}
93
94bool AsmParser::Error(SMLoc L, const Twine &Msg) {
95  PrintMessage(L, Msg.str(), "error");
96  return true;
97}
98
99bool AsmParser::TokError(const char *Msg) {
100  PrintMessage(Lexer.getLoc(), Msg, "error");
101  return true;
102}
103
104void AsmParser::PrintMessage(SMLoc Loc, const std::string &Msg,
105                             const char *Type) const {
106  SrcMgr.PrintMessage(Loc, Msg, Type);
107}
108
109bool AsmParser::EnterIncludeFile(const std::string &Filename) {
110  int NewBuf = SrcMgr.AddIncludeFile(Filename, Lexer.getLoc());
111  if (NewBuf == -1)
112    return true;
113
114  CurBuffer = NewBuf;
115
116  Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer));
117
118  return false;
119}
120
121const AsmToken &AsmParser::Lex() {
122  const AsmToken *tok = &Lexer.Lex();
123
124  if (tok->is(AsmToken::Eof)) {
125    // If this is the end of an included file, pop the parent file off the
126    // include stack.
127    SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer);
128    if (ParentIncludeLoc != SMLoc()) {
129      CurBuffer = SrcMgr.FindBufferContainingLoc(ParentIncludeLoc);
130      Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer),
131                      ParentIncludeLoc.getPointer());
132      tok = &Lexer.Lex();
133    }
134  }
135
136  if (tok->is(AsmToken::Error))
137    PrintMessage(Lexer.getErrLoc(), Lexer.getErr(), "error");
138
139  return *tok;
140}
141
142bool AsmParser::Run() {
143  // Create the initial section.
144  //
145  // FIXME: Support -n.
146  // FIXME: Target hook & command line option for initial section.
147  Out.SwitchSection(getMachOSection("__TEXT", "__text",
148                                    MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS,
149                                    0, SectionKind()));
150
151
152  // Prime the lexer.
153  Lex();
154
155  bool HadError = false;
156
157  AsmCond StartingCondState = TheCondState;
158
159  // While we have input, parse each statement.
160  while (Lexer.isNot(AsmToken::Eof)) {
161    // Handle conditional assembly here before calling ParseStatement()
162    if (Lexer.getKind() == AsmToken::Identifier) {
163      // If we have an identifier, handle it as the key symbol.
164      AsmToken ID = getTok();
165      SMLoc IDLoc = ID.getLoc();
166      StringRef IDVal = ID.getString();
167
168      if (IDVal == ".if" ||
169          IDVal == ".elseif" ||
170          IDVal == ".else" ||
171          IDVal == ".endif") {
172        if (!ParseConditionalAssemblyDirectives(IDVal, IDLoc))
173          continue;
174	HadError = true;
175	EatToEndOfStatement();
176	continue;
177      }
178    }
179    if (TheCondState.Ignore) {
180      EatToEndOfStatement();
181      continue;
182    }
183
184    if (!ParseStatement()) continue;
185
186    // We had an error, remember it and recover by skipping to the next line.
187    HadError = true;
188    EatToEndOfStatement();
189  }
190
191  if (TheCondState.TheCond != StartingCondState.TheCond ||
192      TheCondState.Ignore != StartingCondState.Ignore)
193    return TokError("unmatched .ifs or .elses");
194
195  if (!HadError)
196    Out.Finish();
197
198  return HadError;
199}
200
201/// ParseConditionalAssemblyDirectives - parse the conditional assembly
202/// directives
203bool AsmParser::ParseConditionalAssemblyDirectives(StringRef Directive,
204                                                   SMLoc DirectiveLoc) {
205  if (Directive == ".if")
206    return ParseDirectiveIf(DirectiveLoc);
207  if (Directive == ".elseif")
208    return ParseDirectiveElseIf(DirectiveLoc);
209  if (Directive == ".else")
210    return ParseDirectiveElse(DirectiveLoc);
211  if (Directive == ".endif")
212    return ParseDirectiveEndIf(DirectiveLoc);
213  return true;
214}
215
216/// EatToEndOfStatement - Throw away the rest of the line for testing purposes.
217void AsmParser::EatToEndOfStatement() {
218  while (Lexer.isNot(AsmToken::EndOfStatement) &&
219         Lexer.isNot(AsmToken::Eof))
220    Lex();
221
222  // Eat EOL.
223  if (Lexer.is(AsmToken::EndOfStatement))
224    Lex();
225}
226
227
228/// ParseParenExpr - Parse a paren expression and return it.
229/// NOTE: This assumes the leading '(' has already been consumed.
230///
231/// parenexpr ::= expr)
232///
233bool AsmParser::ParseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) {
234  if (ParseExpression(Res)) return true;
235  if (Lexer.isNot(AsmToken::RParen))
236    return TokError("expected ')' in parentheses expression");
237  EndLoc = Lexer.getLoc();
238  Lex();
239  return false;
240}
241
242MCSymbol *AsmParser::CreateSymbol(StringRef Name) {
243  if (MCSymbol *S = Ctx.LookupSymbol(Name))
244    return S;
245
246  // If the label starts with L it is an assembler temporary label.
247  if (Name.startswith("L"))
248    return Ctx.CreateTemporarySymbol(Name);
249
250  return Ctx.CreateSymbol(Name);
251}
252
253/// ParsePrimaryExpr - Parse a primary expression and return it.
254///  primaryexpr ::= (parenexpr
255///  primaryexpr ::= symbol
256///  primaryexpr ::= number
257///  primaryexpr ::= ~,+,- primaryexpr
258bool AsmParser::ParsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) {
259  switch (Lexer.getKind()) {
260  default:
261    return TokError("unknown token in expression");
262  case AsmToken::Exclaim:
263    Lex(); // Eat the operator.
264    if (ParsePrimaryExpr(Res, EndLoc))
265      return true;
266    Res = MCUnaryExpr::CreateLNot(Res, getContext());
267    return false;
268  case AsmToken::String:
269  case AsmToken::Identifier: {
270    // This is a symbol reference.
271    MCSymbol *Sym = CreateSymbol(getTok().getIdentifier());
272    EndLoc = Lexer.getLoc();
273    Lex(); // Eat identifier.
274
275    // If this is an absolute variable reference, substitute it now to preserve
276    // semantics in the face of reassignment.
277    if (Sym->getValue() && isa<MCConstantExpr>(Sym->getValue())) {
278      Res = Sym->getValue();
279      return false;
280    }
281
282    // Otherwise create a symbol ref.
283    Res = MCSymbolRefExpr::Create(Sym, getContext());
284    return false;
285  }
286  case AsmToken::Integer:
287    Res = MCConstantExpr::Create(getTok().getIntVal(), getContext());
288    EndLoc = Lexer.getLoc();
289    Lex(); // Eat token.
290    return false;
291  case AsmToken::LParen:
292    Lex(); // Eat the '('.
293    return ParseParenExpr(Res, EndLoc);
294  case AsmToken::Minus:
295    Lex(); // Eat the operator.
296    if (ParsePrimaryExpr(Res, EndLoc))
297      return true;
298    Res = MCUnaryExpr::CreateMinus(Res, getContext());
299    return false;
300  case AsmToken::Plus:
301    Lex(); // Eat the operator.
302    if (ParsePrimaryExpr(Res, EndLoc))
303      return true;
304    Res = MCUnaryExpr::CreatePlus(Res, getContext());
305    return false;
306  case AsmToken::Tilde:
307    Lex(); // Eat the operator.
308    if (ParsePrimaryExpr(Res, EndLoc))
309      return true;
310    Res = MCUnaryExpr::CreateNot(Res, getContext());
311    return false;
312  }
313}
314
315bool AsmParser::ParseExpression(const MCExpr *&Res) {
316  SMLoc EndLoc;
317  return ParseExpression(Res, EndLoc);
318}
319
320/// ParseExpression - Parse an expression and return it.
321///
322///  expr ::= expr +,- expr          -> lowest.
323///  expr ::= expr |,^,&,! expr      -> middle.
324///  expr ::= expr *,/,%,<<,>> expr  -> highest.
325///  expr ::= primaryexpr
326///
327bool AsmParser::ParseExpression(const MCExpr *&Res, SMLoc &EndLoc) {
328  Res = 0;
329  return ParsePrimaryExpr(Res, EndLoc) ||
330         ParseBinOpRHS(1, Res, EndLoc);
331}
332
333bool AsmParser::ParseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) {
334  if (ParseParenExpr(Res, EndLoc))
335    return true;
336
337  return false;
338}
339
340bool AsmParser::ParseAbsoluteExpression(int64_t &Res) {
341  const MCExpr *Expr;
342
343  SMLoc StartLoc = Lexer.getLoc();
344  if (ParseExpression(Expr))
345    return true;
346
347  if (!Expr->EvaluateAsAbsolute(Res))
348    return Error(StartLoc, "expected absolute expression");
349
350  return false;
351}
352
353static unsigned getBinOpPrecedence(AsmToken::TokenKind K,
354                                   MCBinaryExpr::Opcode &Kind) {
355  switch (K) {
356  default:
357    return 0;    // not a binop.
358
359    // Lowest Precedence: &&, ||
360  case AsmToken::AmpAmp:
361    Kind = MCBinaryExpr::LAnd;
362    return 1;
363  case AsmToken::PipePipe:
364    Kind = MCBinaryExpr::LOr;
365    return 1;
366
367    // Low Precedence: +, -, ==, !=, <>, <, <=, >, >=
368  case AsmToken::Plus:
369    Kind = MCBinaryExpr::Add;
370    return 2;
371  case AsmToken::Minus:
372    Kind = MCBinaryExpr::Sub;
373    return 2;
374  case AsmToken::EqualEqual:
375    Kind = MCBinaryExpr::EQ;
376    return 2;
377  case AsmToken::ExclaimEqual:
378  case AsmToken::LessGreater:
379    Kind = MCBinaryExpr::NE;
380    return 2;
381  case AsmToken::Less:
382    Kind = MCBinaryExpr::LT;
383    return 2;
384  case AsmToken::LessEqual:
385    Kind = MCBinaryExpr::LTE;
386    return 2;
387  case AsmToken::Greater:
388    Kind = MCBinaryExpr::GT;
389    return 2;
390  case AsmToken::GreaterEqual:
391    Kind = MCBinaryExpr::GTE;
392    return 2;
393
394    // Intermediate Precedence: |, &, ^
395    //
396    // FIXME: gas seems to support '!' as an infix operator?
397  case AsmToken::Pipe:
398    Kind = MCBinaryExpr::Or;
399    return 3;
400  case AsmToken::Caret:
401    Kind = MCBinaryExpr::Xor;
402    return 3;
403  case AsmToken::Amp:
404    Kind = MCBinaryExpr::And;
405    return 3;
406
407    // Highest Precedence: *, /, %, <<, >>
408  case AsmToken::Star:
409    Kind = MCBinaryExpr::Mul;
410    return 4;
411  case AsmToken::Slash:
412    Kind = MCBinaryExpr::Div;
413    return 4;
414  case AsmToken::Percent:
415    Kind = MCBinaryExpr::Mod;
416    return 4;
417  case AsmToken::LessLess:
418    Kind = MCBinaryExpr::Shl;
419    return 4;
420  case AsmToken::GreaterGreater:
421    Kind = MCBinaryExpr::Shr;
422    return 4;
423  }
424}
425
426
427/// ParseBinOpRHS - Parse all binary operators with precedence >= 'Precedence'.
428/// Res contains the LHS of the expression on input.
429bool AsmParser::ParseBinOpRHS(unsigned Precedence, const MCExpr *&Res,
430                              SMLoc &EndLoc) {
431  while (1) {
432    MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add;
433    unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind);
434
435    // If the next token is lower precedence than we are allowed to eat, return
436    // successfully with what we ate already.
437    if (TokPrec < Precedence)
438      return false;
439
440    Lex();
441
442    // Eat the next primary expression.
443    const MCExpr *RHS;
444    if (ParsePrimaryExpr(RHS, EndLoc)) return true;
445
446    // If BinOp binds less tightly with RHS than the operator after RHS, let
447    // the pending operator take RHS as its LHS.
448    MCBinaryExpr::Opcode Dummy;
449    unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy);
450    if (TokPrec < NextTokPrec) {
451      if (ParseBinOpRHS(Precedence+1, RHS, EndLoc)) return true;
452    }
453
454    // Merge LHS and RHS according to operator.
455    Res = MCBinaryExpr::Create(Kind, Res, RHS, getContext());
456  }
457}
458
459
460
461
462/// ParseStatement:
463///   ::= EndOfStatement
464///   ::= Label* Directive ...Operands... EndOfStatement
465///   ::= Label* Identifier OperandList* EndOfStatement
466bool AsmParser::ParseStatement() {
467  if (Lexer.is(AsmToken::EndOfStatement)) {
468    Lex();
469    return false;
470  }
471
472  // Statements always start with an identifier.
473  AsmToken ID = getTok();
474  SMLoc IDLoc = ID.getLoc();
475  StringRef IDVal;
476  if (ParseIdentifier(IDVal))
477    return TokError("unexpected token at start of statement");
478
479  // FIXME: Recurse on local labels?
480
481  // See what kind of statement we have.
482  switch (Lexer.getKind()) {
483  case AsmToken::Colon: {
484    // identifier ':'   -> Label.
485    Lex();
486
487    // Diagnose attempt to use a variable as a label.
488    //
489    // FIXME: Diagnostics. Note the location of the definition as a label.
490    // FIXME: This doesn't diagnose assignment to a symbol which has been
491    // implicitly marked as external.
492    MCSymbol *Sym = CreateSymbol(IDVal);
493    if (!Sym->isUndefined())
494      return Error(IDLoc, "invalid symbol redefinition");
495
496    // Emit the label.
497    Out.EmitLabel(Sym);
498
499    return ParseStatement();
500  }
501
502  case AsmToken::Equal:
503    // identifier '=' ... -> assignment statement
504    Lex();
505
506    return ParseAssignment(IDVal);
507
508  default: // Normal instruction or directive.
509    break;
510  }
511
512  // Otherwise, we have a normal instruction or directive.
513  if (IDVal[0] == '.') {
514    // FIXME: This should be driven based on a hash lookup and callback.
515    if (IDVal == ".section")
516      return ParseDirectiveDarwinSection();
517    if (IDVal == ".text")
518      // FIXME: This changes behavior based on the -static flag to the
519      // assembler.
520      return ParseDirectiveSectionSwitch("__TEXT", "__text",
521                                     MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS);
522    if (IDVal == ".const")
523      return ParseDirectiveSectionSwitch("__TEXT", "__const");
524    if (IDVal == ".static_const")
525      return ParseDirectiveSectionSwitch("__TEXT", "__static_const");
526    if (IDVal == ".cstring")
527      return ParseDirectiveSectionSwitch("__TEXT","__cstring",
528                                         MCSectionMachO::S_CSTRING_LITERALS);
529    if (IDVal == ".literal4")
530      return ParseDirectiveSectionSwitch("__TEXT", "__literal4",
531                                         MCSectionMachO::S_4BYTE_LITERALS,
532                                         4);
533    if (IDVal == ".literal8")
534      return ParseDirectiveSectionSwitch("__TEXT", "__literal8",
535                                         MCSectionMachO::S_8BYTE_LITERALS,
536                                         8);
537    if (IDVal == ".literal16")
538      return ParseDirectiveSectionSwitch("__TEXT","__literal16",
539                                         MCSectionMachO::S_16BYTE_LITERALS,
540                                         16);
541    if (IDVal == ".constructor")
542      return ParseDirectiveSectionSwitch("__TEXT","__constructor");
543    if (IDVal == ".destructor")
544      return ParseDirectiveSectionSwitch("__TEXT","__destructor");
545    if (IDVal == ".fvmlib_init0")
546      return ParseDirectiveSectionSwitch("__TEXT","__fvmlib_init0");
547    if (IDVal == ".fvmlib_init1")
548      return ParseDirectiveSectionSwitch("__TEXT","__fvmlib_init1");
549
550    // FIXME: The assembler manual claims that this has the self modify code
551    // flag, at least on x86-32, but that does not appear to be correct.
552    if (IDVal == ".symbol_stub")
553      return ParseDirectiveSectionSwitch("__TEXT","__symbol_stub",
554                                         MCSectionMachO::S_SYMBOL_STUBS |
555                                       MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS,
556                                          // FIXME: Different on PPC and ARM.
557                                         0, 16);
558    // FIXME: PowerPC only?
559    if (IDVal == ".picsymbol_stub")
560      return ParseDirectiveSectionSwitch("__TEXT","__picsymbol_stub",
561                                         MCSectionMachO::S_SYMBOL_STUBS |
562                                       MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS,
563                                         0, 26);
564    if (IDVal == ".data")
565      return ParseDirectiveSectionSwitch("__DATA", "__data");
566    if (IDVal == ".static_data")
567      return ParseDirectiveSectionSwitch("__DATA", "__static_data");
568
569    // FIXME: The section names of these two are misspelled in the assembler
570    // manual.
571    if (IDVal == ".non_lazy_symbol_pointer")
572      return ParseDirectiveSectionSwitch("__DATA", "__nl_symbol_ptr",
573                                     MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS,
574                                         4);
575    if (IDVal == ".lazy_symbol_pointer")
576      return ParseDirectiveSectionSwitch("__DATA", "__la_symbol_ptr",
577                                         MCSectionMachO::S_LAZY_SYMBOL_POINTERS,
578                                         4);
579
580    if (IDVal == ".dyld")
581      return ParseDirectiveSectionSwitch("__DATA", "__dyld");
582    if (IDVal == ".mod_init_func")
583      return ParseDirectiveSectionSwitch("__DATA", "__mod_init_func",
584                                       MCSectionMachO::S_MOD_INIT_FUNC_POINTERS,
585                                         4);
586    if (IDVal == ".mod_term_func")
587      return ParseDirectiveSectionSwitch("__DATA", "__mod_term_func",
588                                       MCSectionMachO::S_MOD_TERM_FUNC_POINTERS,
589                                         4);
590    if (IDVal == ".const_data")
591      return ParseDirectiveSectionSwitch("__DATA", "__const");
592
593
594    if (IDVal == ".objc_class")
595      return ParseDirectiveSectionSwitch("__OBJC", "__class",
596                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
597    if (IDVal == ".objc_meta_class")
598      return ParseDirectiveSectionSwitch("__OBJC", "__meta_class",
599                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
600    if (IDVal == ".objc_cat_cls_meth")
601      return ParseDirectiveSectionSwitch("__OBJC", "__cat_cls_meth",
602                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
603    if (IDVal == ".objc_cat_inst_meth")
604      return ParseDirectiveSectionSwitch("__OBJC", "__cat_inst_meth",
605                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
606    if (IDVal == ".objc_protocol")
607      return ParseDirectiveSectionSwitch("__OBJC", "__protocol",
608                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
609    if (IDVal == ".objc_string_object")
610      return ParseDirectiveSectionSwitch("__OBJC", "__string_object",
611                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
612    if (IDVal == ".objc_cls_meth")
613      return ParseDirectiveSectionSwitch("__OBJC", "__cls_meth",
614                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
615    if (IDVal == ".objc_inst_meth")
616      return ParseDirectiveSectionSwitch("__OBJC", "__inst_meth",
617                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
618    if (IDVal == ".objc_cls_refs")
619      return ParseDirectiveSectionSwitch("__OBJC", "__cls_refs",
620                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP |
621                                         MCSectionMachO::S_LITERAL_POINTERS,
622                                         4);
623    if (IDVal == ".objc_message_refs")
624      return ParseDirectiveSectionSwitch("__OBJC", "__message_refs",
625                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP |
626                                         MCSectionMachO::S_LITERAL_POINTERS,
627                                         4);
628    if (IDVal == ".objc_symbols")
629      return ParseDirectiveSectionSwitch("__OBJC", "__symbols",
630                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
631    if (IDVal == ".objc_category")
632      return ParseDirectiveSectionSwitch("__OBJC", "__category",
633                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
634    if (IDVal == ".objc_class_vars")
635      return ParseDirectiveSectionSwitch("__OBJC", "__class_vars",
636                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
637    if (IDVal == ".objc_instance_vars")
638      return ParseDirectiveSectionSwitch("__OBJC", "__instance_vars",
639                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
640    if (IDVal == ".objc_module_info")
641      return ParseDirectiveSectionSwitch("__OBJC", "__module_info",
642                                         MCSectionMachO::S_ATTR_NO_DEAD_STRIP);
643    if (IDVal == ".objc_class_names")
644      return ParseDirectiveSectionSwitch("__TEXT", "__cstring",
645                                         MCSectionMachO::S_CSTRING_LITERALS);
646    if (IDVal == ".objc_meth_var_types")
647      return ParseDirectiveSectionSwitch("__TEXT", "__cstring",
648                                         MCSectionMachO::S_CSTRING_LITERALS);
649    if (IDVal == ".objc_meth_var_names")
650      return ParseDirectiveSectionSwitch("__TEXT", "__cstring",
651                                         MCSectionMachO::S_CSTRING_LITERALS);
652    if (IDVal == ".objc_selector_strs")
653      return ParseDirectiveSectionSwitch("__OBJC", "__selector_strs",
654                                         MCSectionMachO::S_CSTRING_LITERALS);
655
656    // Assembler features
657    if (IDVal == ".set")
658      return ParseDirectiveSet();
659
660    // Data directives
661
662    if (IDVal == ".ascii")
663      return ParseDirectiveAscii(false);
664    if (IDVal == ".asciz")
665      return ParseDirectiveAscii(true);
666
667    if (IDVal == ".byte")
668      return ParseDirectiveValue(1);
669    if (IDVal == ".short")
670      return ParseDirectiveValue(2);
671    if (IDVal == ".long")
672      return ParseDirectiveValue(4);
673    if (IDVal == ".quad")
674      return ParseDirectiveValue(8);
675
676    // FIXME: Target hooks for IsPow2.
677    if (IDVal == ".align")
678      return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1);
679    if (IDVal == ".align32")
680      return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4);
681    if (IDVal == ".balign")
682      return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1);
683    if (IDVal == ".balignw")
684      return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2);
685    if (IDVal == ".balignl")
686      return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4);
687    if (IDVal == ".p2align")
688      return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1);
689    if (IDVal == ".p2alignw")
690      return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2);
691    if (IDVal == ".p2alignl")
692      return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4);
693
694    if (IDVal == ".org")
695      return ParseDirectiveOrg();
696
697    if (IDVal == ".fill")
698      return ParseDirectiveFill();
699    if (IDVal == ".space")
700      return ParseDirectiveSpace();
701
702    // Symbol attribute directives
703
704    if (IDVal == ".globl" || IDVal == ".global")
705      return ParseDirectiveSymbolAttribute(MCSA_Global);
706    if (IDVal == ".hidden")
707      return ParseDirectiveSymbolAttribute(MCSA_Hidden);
708    if (IDVal == ".indirect_symbol")
709      return ParseDirectiveSymbolAttribute(MCSA_IndirectSymbol);
710    if (IDVal == ".internal")
711      return ParseDirectiveSymbolAttribute(MCSA_Internal);
712    if (IDVal == ".lazy_reference")
713      return ParseDirectiveSymbolAttribute(MCSA_LazyReference);
714    if (IDVal == ".no_dead_strip")
715      return ParseDirectiveSymbolAttribute(MCSA_NoDeadStrip);
716    if (IDVal == ".private_extern")
717      return ParseDirectiveSymbolAttribute(MCSA_PrivateExtern);
718    if (IDVal == ".protected")
719      return ParseDirectiveSymbolAttribute(MCSA_Protected);
720    if (IDVal == ".reference")
721      return ParseDirectiveSymbolAttribute(MCSA_Reference);
722    if (IDVal == ".weak")
723      return ParseDirectiveSymbolAttribute(MCSA_Weak);
724    if (IDVal == ".weak_definition")
725      return ParseDirectiveSymbolAttribute(MCSA_WeakDefinition);
726    if (IDVal == ".weak_reference")
727      return ParseDirectiveSymbolAttribute(MCSA_WeakReference);
728
729    if (IDVal == ".comm")
730      return ParseDirectiveComm(/*IsLocal=*/false);
731    if (IDVal == ".lcomm")
732      return ParseDirectiveComm(/*IsLocal=*/true);
733    if (IDVal == ".zerofill")
734      return ParseDirectiveDarwinZerofill();
735    if (IDVal == ".desc")
736      return ParseDirectiveDarwinSymbolDesc();
737    if (IDVal == ".lsym")
738      return ParseDirectiveDarwinLsym();
739
740    if (IDVal == ".subsections_via_symbols")
741      return ParseDirectiveDarwinSubsectionsViaSymbols();
742    if (IDVal == ".abort")
743      return ParseDirectiveAbort();
744    if (IDVal == ".include")
745      return ParseDirectiveInclude();
746    if (IDVal == ".dump")
747      return ParseDirectiveDarwinDumpOrLoad(IDLoc, /*IsDump=*/true);
748    if (IDVal == ".load")
749      return ParseDirectiveDarwinDumpOrLoad(IDLoc, /*IsLoad=*/false);
750
751    // Look up the handler in the handler table,
752    bool(AsmParser::*Handler)(StringRef, SMLoc) = DirectiveMap[IDVal];
753    if (Handler)
754      return (this->*Handler)(IDVal, IDLoc);
755
756    // Target hook for parsing target specific directives.
757    if (!getTargetParser().ParseDirective(ID))
758      return false;
759
760    Warning(IDLoc, "ignoring directive for now");
761    EatToEndOfStatement();
762    return false;
763  }
764
765
766  SmallVector<MCParsedAsmOperand*, 8> ParsedOperands;
767  if (getTargetParser().ParseInstruction(IDVal, IDLoc, ParsedOperands))
768    // FIXME: Leaking ParsedOperands on failure.
769    return true;
770
771  if (Lexer.isNot(AsmToken::EndOfStatement))
772    // FIXME: Leaking ParsedOperands on failure.
773    return TokError("unexpected token in argument list");
774
775  // Eat the end of statement marker.
776  Lex();
777
778
779  MCInst Inst;
780
781  bool MatchFail = getTargetParser().MatchInstruction(ParsedOperands, Inst);
782
783  // Free any parsed operands.
784  for (unsigned i = 0, e = ParsedOperands.size(); i != e; ++i)
785    delete ParsedOperands[i];
786
787  if (MatchFail) {
788    // FIXME: We should give nicer diagnostics about the exact failure.
789    Error(IDLoc, "unrecognized instruction");
790    return true;
791  }
792
793  // Instruction is good, process it.
794  Out.EmitInstruction(Inst);
795
796  // Skip to end of line for now.
797  return false;
798}
799
800bool AsmParser::ParseAssignment(const StringRef &Name) {
801  // FIXME: Use better location, we should use proper tokens.
802  SMLoc EqualLoc = Lexer.getLoc();
803
804  const MCExpr *Value;
805  SMLoc StartLoc = Lexer.getLoc();
806  if (ParseExpression(Value))
807    return true;
808
809  if (Lexer.isNot(AsmToken::EndOfStatement))
810    return TokError("unexpected token in assignment");
811
812  // Eat the end of statement marker.
813  Lex();
814
815  // Validate that the LHS is allowed to be a variable (either it has not been
816  // used as a symbol, or it is an absolute symbol).
817  MCSymbol *Sym = getContext().LookupSymbol(Name);
818  if (Sym) {
819    // Diagnose assignment to a label.
820    //
821    // FIXME: Diagnostics. Note the location of the definition as a label.
822    // FIXME: Diagnose assignment to protected identifier (e.g., register name).
823    if (!Sym->isUndefined() && !Sym->isAbsolute())
824      return Error(EqualLoc, "redefinition of '" + Name + "'");
825    else if (!Sym->isVariable())
826      return Error(EqualLoc, "invalid assignment to '" + Name + "'");
827    else if (!isa<MCConstantExpr>(Sym->getValue()))
828      return Error(EqualLoc, "invalid reassignment of non-absolute variable '" +
829                   Name + "'");
830  } else
831    Sym = CreateSymbol(Name);
832
833  // FIXME: Handle '.'.
834
835  // Do the assignment.
836  Out.EmitAssignment(Sym, Value);
837
838  return false;
839}
840
841/// ParseIdentifier:
842///   ::= identifier
843///   ::= string
844bool AsmParser::ParseIdentifier(StringRef &Res) {
845  if (Lexer.isNot(AsmToken::Identifier) &&
846      Lexer.isNot(AsmToken::String))
847    return true;
848
849  Res = getTok().getIdentifier();
850
851  Lex(); // Consume the identifier token.
852
853  return false;
854}
855
856/// ParseDirectiveSet:
857///   ::= .set identifier ',' expression
858bool AsmParser::ParseDirectiveSet() {
859  StringRef Name;
860
861  if (ParseIdentifier(Name))
862    return TokError("expected identifier after '.set' directive");
863
864  if (Lexer.isNot(AsmToken::Comma))
865    return TokError("unexpected token in '.set'");
866  Lex();
867
868  return ParseAssignment(Name);
869}
870
871/// ParseDirectiveSection:
872///   ::= .section identifier (',' identifier)*
873/// FIXME: This should actually parse out the segment, section, attributes and
874/// sizeof_stub fields.
875bool AsmParser::ParseDirectiveDarwinSection() {
876  SMLoc Loc = Lexer.getLoc();
877
878  StringRef SectionName;
879  if (ParseIdentifier(SectionName))
880    return Error(Loc, "expected identifier after '.section' directive");
881
882  // Verify there is a following comma.
883  if (!Lexer.is(AsmToken::Comma))
884    return TokError("unexpected token in '.section' directive");
885
886  std::string SectionSpec = SectionName;
887  SectionSpec += ",";
888
889  // Add all the tokens until the end of the line, ParseSectionSpecifier will
890  // handle this.
891  StringRef EOL = Lexer.LexUntilEndOfStatement();
892  SectionSpec.append(EOL.begin(), EOL.end());
893
894  Lex();
895  if (Lexer.isNot(AsmToken::EndOfStatement))
896    return TokError("unexpected token in '.section' directive");
897  Lex();
898
899
900  StringRef Segment, Section;
901  unsigned TAA, StubSize;
902  std::string ErrorStr =
903    MCSectionMachO::ParseSectionSpecifier(SectionSpec, Segment, Section,
904                                          TAA, StubSize);
905
906  if (!ErrorStr.empty())
907    return Error(Loc, ErrorStr.c_str());
908
909  // FIXME: Arch specific.
910  Out.SwitchSection(getMachOSection(Segment, Section, TAA, StubSize,
911                                    SectionKind()));
912  return false;
913}
914
915/// ParseDirectiveSectionSwitch -
916bool AsmParser::ParseDirectiveSectionSwitch(const char *Segment,
917                                            const char *Section,
918                                            unsigned TAA, unsigned Align,
919                                            unsigned StubSize) {
920  if (Lexer.isNot(AsmToken::EndOfStatement))
921    return TokError("unexpected token in section switching directive");
922  Lex();
923
924  // FIXME: Arch specific.
925  Out.SwitchSection(getMachOSection(Segment, Section, TAA, StubSize,
926                                    SectionKind()));
927
928  // Set the implicit alignment, if any.
929  //
930  // FIXME: This isn't really what 'as' does; I think it just uses the implicit
931  // alignment on the section (e.g., if one manually inserts bytes into the
932  // section, then just issueing the section switch directive will not realign
933  // the section. However, this is arguably more reasonable behavior, and there
934  // is no good reason for someone to intentionally emit incorrectly sized
935  // values into the implicitly aligned sections.
936  if (Align)
937    Out.EmitValueToAlignment(Align, 0, 1, 0);
938
939  return false;
940}
941
942bool AsmParser::ParseEscapedString(std::string &Data) {
943  assert(Lexer.is(AsmToken::String) && "Unexpected current token!");
944
945  Data = "";
946  StringRef Str = getTok().getStringContents();
947  for (unsigned i = 0, e = Str.size(); i != e; ++i) {
948    if (Str[i] != '\\') {
949      Data += Str[i];
950      continue;
951    }
952
953    // Recognize escaped characters. Note that this escape semantics currently
954    // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes.
955    ++i;
956    if (i == e)
957      return TokError("unexpected backslash at end of string");
958
959    // Recognize octal sequences.
960    if ((unsigned) (Str[i] - '0') <= 7) {
961      // Consume up to three octal characters.
962      unsigned Value = Str[i] - '0';
963
964      if (i + 1 != e && ((unsigned) (Str[i + 1] - '0')) <= 7) {
965        ++i;
966        Value = Value * 8 + (Str[i] - '0');
967
968        if (i + 1 != e && ((unsigned) (Str[i + 1] - '0')) <= 7) {
969          ++i;
970          Value = Value * 8 + (Str[i] - '0');
971        }
972      }
973
974      if (Value > 255)
975        return TokError("invalid octal escape sequence (out of range)");
976
977      Data += (unsigned char) Value;
978      continue;
979    }
980
981    // Otherwise recognize individual escapes.
982    switch (Str[i]) {
983    default:
984      // Just reject invalid escape sequences for now.
985      return TokError("invalid escape sequence (unrecognized character)");
986
987    case 'b': Data += '\b'; break;
988    case 'f': Data += '\f'; break;
989    case 'n': Data += '\n'; break;
990    case 'r': Data += '\r'; break;
991    case 't': Data += '\t'; break;
992    case '"': Data += '"'; break;
993    case '\\': Data += '\\'; break;
994    }
995  }
996
997  return false;
998}
999
1000/// ParseDirectiveAscii:
1001///   ::= ( .ascii | .asciz ) [ "string" ( , "string" )* ]
1002bool AsmParser::ParseDirectiveAscii(bool ZeroTerminated) {
1003  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1004    for (;;) {
1005      if (Lexer.isNot(AsmToken::String))
1006        return TokError("expected string in '.ascii' or '.asciz' directive");
1007
1008      std::string Data;
1009      if (ParseEscapedString(Data))
1010        return true;
1011
1012      Out.EmitBytes(Data, DEFAULT_ADDRSPACE);
1013      if (ZeroTerminated)
1014        Out.EmitBytes(StringRef("\0", 1), DEFAULT_ADDRSPACE);
1015
1016      Lex();
1017
1018      if (Lexer.is(AsmToken::EndOfStatement))
1019        break;
1020
1021      if (Lexer.isNot(AsmToken::Comma))
1022        return TokError("unexpected token in '.ascii' or '.asciz' directive");
1023      Lex();
1024    }
1025  }
1026
1027  Lex();
1028  return false;
1029}
1030
1031/// ParseDirectiveValue
1032///  ::= (.byte | .short | ... ) [ expression (, expression)* ]
1033bool AsmParser::ParseDirectiveValue(unsigned Size) {
1034  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1035    for (;;) {
1036      const MCExpr *Value;
1037      SMLoc ATTRIBUTE_UNUSED StartLoc = Lexer.getLoc();
1038      if (ParseExpression(Value))
1039        return true;
1040
1041      Out.EmitValue(Value, Size, DEFAULT_ADDRSPACE);
1042
1043      if (Lexer.is(AsmToken::EndOfStatement))
1044        break;
1045
1046      // FIXME: Improve diagnostic.
1047      if (Lexer.isNot(AsmToken::Comma))
1048        return TokError("unexpected token in directive");
1049      Lex();
1050    }
1051  }
1052
1053  Lex();
1054  return false;
1055}
1056
1057/// ParseDirectiveSpace
1058///  ::= .space expression [ , expression ]
1059bool AsmParser::ParseDirectiveSpace() {
1060  int64_t NumBytes;
1061  if (ParseAbsoluteExpression(NumBytes))
1062    return true;
1063
1064  int64_t FillExpr = 0;
1065  bool HasFillExpr = false;
1066  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1067    if (Lexer.isNot(AsmToken::Comma))
1068      return TokError("unexpected token in '.space' directive");
1069    Lex();
1070
1071    if (ParseAbsoluteExpression(FillExpr))
1072      return true;
1073
1074    HasFillExpr = true;
1075
1076    if (Lexer.isNot(AsmToken::EndOfStatement))
1077      return TokError("unexpected token in '.space' directive");
1078  }
1079
1080  Lex();
1081
1082  if (NumBytes <= 0)
1083    return TokError("invalid number of bytes in '.space' directive");
1084
1085  // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0.
1086  Out.EmitFill(NumBytes, FillExpr, DEFAULT_ADDRSPACE);
1087
1088  return false;
1089}
1090
1091/// ParseDirectiveFill
1092///  ::= .fill expression , expression , expression
1093bool AsmParser::ParseDirectiveFill() {
1094  int64_t NumValues;
1095  if (ParseAbsoluteExpression(NumValues))
1096    return true;
1097
1098  if (Lexer.isNot(AsmToken::Comma))
1099    return TokError("unexpected token in '.fill' directive");
1100  Lex();
1101
1102  int64_t FillSize;
1103  if (ParseAbsoluteExpression(FillSize))
1104    return true;
1105
1106  if (Lexer.isNot(AsmToken::Comma))
1107    return TokError("unexpected token in '.fill' directive");
1108  Lex();
1109
1110  int64_t FillExpr;
1111  if (ParseAbsoluteExpression(FillExpr))
1112    return true;
1113
1114  if (Lexer.isNot(AsmToken::EndOfStatement))
1115    return TokError("unexpected token in '.fill' directive");
1116
1117  Lex();
1118
1119  if (FillSize != 1 && FillSize != 2 && FillSize != 4 && FillSize != 8)
1120    return TokError("invalid '.fill' size, expected 1, 2, 4, or 8");
1121
1122  for (uint64_t i = 0, e = NumValues; i != e; ++i)
1123    Out.EmitValue(MCConstantExpr::Create(FillExpr, getContext()), FillSize,
1124                  DEFAULT_ADDRSPACE);
1125
1126  return false;
1127}
1128
1129/// ParseDirectiveOrg
1130///  ::= .org expression [ , expression ]
1131bool AsmParser::ParseDirectiveOrg() {
1132  const MCExpr *Offset;
1133  SMLoc StartLoc = Lexer.getLoc();
1134  if (ParseExpression(Offset))
1135    return true;
1136
1137  // Parse optional fill expression.
1138  int64_t FillExpr = 0;
1139  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1140    if (Lexer.isNot(AsmToken::Comma))
1141      return TokError("unexpected token in '.org' directive");
1142    Lex();
1143
1144    if (ParseAbsoluteExpression(FillExpr))
1145      return true;
1146
1147    if (Lexer.isNot(AsmToken::EndOfStatement))
1148      return TokError("unexpected token in '.org' directive");
1149  }
1150
1151  Lex();
1152
1153  // FIXME: Only limited forms of relocatable expressions are accepted here, it
1154  // has to be relative to the current section.
1155  Out.EmitValueToOffset(Offset, FillExpr);
1156
1157  return false;
1158}
1159
1160/// ParseDirectiveAlign
1161///  ::= {.align, ...} expression [ , expression [ , expression ]]
1162bool AsmParser::ParseDirectiveAlign(bool IsPow2, unsigned ValueSize) {
1163  SMLoc AlignmentLoc = Lexer.getLoc();
1164  int64_t Alignment;
1165  if (ParseAbsoluteExpression(Alignment))
1166    return true;
1167
1168  SMLoc MaxBytesLoc;
1169  bool HasFillExpr = false;
1170  int64_t FillExpr = 0;
1171  int64_t MaxBytesToFill = 0;
1172  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1173    if (Lexer.isNot(AsmToken::Comma))
1174      return TokError("unexpected token in directive");
1175    Lex();
1176
1177    // The fill expression can be omitted while specifying a maximum number of
1178    // alignment bytes, e.g:
1179    //  .align 3,,4
1180    if (Lexer.isNot(AsmToken::Comma)) {
1181      HasFillExpr = true;
1182      if (ParseAbsoluteExpression(FillExpr))
1183        return true;
1184    }
1185
1186    if (Lexer.isNot(AsmToken::EndOfStatement)) {
1187      if (Lexer.isNot(AsmToken::Comma))
1188        return TokError("unexpected token in directive");
1189      Lex();
1190
1191      MaxBytesLoc = Lexer.getLoc();
1192      if (ParseAbsoluteExpression(MaxBytesToFill))
1193        return true;
1194
1195      if (Lexer.isNot(AsmToken::EndOfStatement))
1196        return TokError("unexpected token in directive");
1197    }
1198  }
1199
1200  Lex();
1201
1202  if (!HasFillExpr) {
1203    // FIXME: Sometimes fill with nop.
1204    FillExpr = 0;
1205  }
1206
1207  // Compute alignment in bytes.
1208  if (IsPow2) {
1209    // FIXME: Diagnose overflow.
1210    if (Alignment >= 32) {
1211      Error(AlignmentLoc, "invalid alignment value");
1212      Alignment = 31;
1213    }
1214
1215    Alignment = 1ULL << Alignment;
1216  }
1217
1218  // Diagnose non-sensical max bytes to align.
1219  if (MaxBytesLoc.isValid()) {
1220    if (MaxBytesToFill < 1) {
1221      Error(MaxBytesLoc, "alignment directive can never be satisfied in this "
1222            "many bytes, ignoring maximum bytes expression");
1223      MaxBytesToFill = 0;
1224    }
1225
1226    if (MaxBytesToFill >= Alignment) {
1227      Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and "
1228              "has no effect");
1229      MaxBytesToFill = 0;
1230    }
1231  }
1232
1233  // FIXME: Target specific behavior about how the "extra" bytes are filled.
1234  Out.EmitValueToAlignment(Alignment, FillExpr, ValueSize, MaxBytesToFill);
1235
1236  return false;
1237}
1238
1239/// ParseDirectiveSymbolAttribute
1240///  ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ]
1241bool AsmParser::ParseDirectiveSymbolAttribute(MCSymbolAttr Attr) {
1242  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1243    for (;;) {
1244      StringRef Name;
1245
1246      if (ParseIdentifier(Name))
1247        return TokError("expected identifier in directive");
1248
1249      MCSymbol *Sym = CreateSymbol(Name);
1250
1251      Out.EmitSymbolAttribute(Sym, Attr);
1252
1253      if (Lexer.is(AsmToken::EndOfStatement))
1254        break;
1255
1256      if (Lexer.isNot(AsmToken::Comma))
1257        return TokError("unexpected token in directive");
1258      Lex();
1259    }
1260  }
1261
1262  Lex();
1263  return false;
1264}
1265
1266/// ParseDirectiveDarwinSymbolDesc
1267///  ::= .desc identifier , expression
1268bool AsmParser::ParseDirectiveDarwinSymbolDesc() {
1269  StringRef Name;
1270  if (ParseIdentifier(Name))
1271    return TokError("expected identifier in directive");
1272
1273  // Handle the identifier as the key symbol.
1274  MCSymbol *Sym = CreateSymbol(Name);
1275
1276  if (Lexer.isNot(AsmToken::Comma))
1277    return TokError("unexpected token in '.desc' directive");
1278  Lex();
1279
1280  SMLoc DescLoc = Lexer.getLoc();
1281  int64_t DescValue;
1282  if (ParseAbsoluteExpression(DescValue))
1283    return true;
1284
1285  if (Lexer.isNot(AsmToken::EndOfStatement))
1286    return TokError("unexpected token in '.desc' directive");
1287
1288  Lex();
1289
1290  // Set the n_desc field of this Symbol to this DescValue
1291  Out.EmitSymbolDesc(Sym, DescValue);
1292
1293  return false;
1294}
1295
1296/// ParseDirectiveComm
1297///  ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ]
1298bool AsmParser::ParseDirectiveComm(bool IsLocal) {
1299  SMLoc IDLoc = Lexer.getLoc();
1300  StringRef Name;
1301  if (ParseIdentifier(Name))
1302    return TokError("expected identifier in directive");
1303
1304  // Handle the identifier as the key symbol.
1305  MCSymbol *Sym = CreateSymbol(Name);
1306
1307  if (Lexer.isNot(AsmToken::Comma))
1308    return TokError("unexpected token in directive");
1309  Lex();
1310
1311  int64_t Size;
1312  SMLoc SizeLoc = Lexer.getLoc();
1313  if (ParseAbsoluteExpression(Size))
1314    return true;
1315
1316  int64_t Pow2Alignment = 0;
1317  SMLoc Pow2AlignmentLoc;
1318  if (Lexer.is(AsmToken::Comma)) {
1319    Lex();
1320    Pow2AlignmentLoc = Lexer.getLoc();
1321    if (ParseAbsoluteExpression(Pow2Alignment))
1322      return true;
1323
1324    // If this target takes alignments in bytes (not log) validate and convert.
1325    if (Lexer.getMAI().getAlignmentIsInBytes()) {
1326      if (!isPowerOf2_64(Pow2Alignment))
1327        return Error(Pow2AlignmentLoc, "alignment must be a power of 2");
1328      Pow2Alignment = Log2_64(Pow2Alignment);
1329    }
1330  }
1331
1332  if (Lexer.isNot(AsmToken::EndOfStatement))
1333    return TokError("unexpected token in '.comm' or '.lcomm' directive");
1334
1335  Lex();
1336
1337  // NOTE: a size of zero for a .comm should create a undefined symbol
1338  // but a size of .lcomm creates a bss symbol of size zero.
1339  if (Size < 0)
1340    return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't "
1341                 "be less than zero");
1342
1343  // NOTE: The alignment in the directive is a power of 2 value, the assember
1344  // may internally end up wanting an alignment in bytes.
1345  // FIXME: Diagnose overflow.
1346  if (Pow2Alignment < 0)
1347    return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive "
1348                 "alignment, can't be less than zero");
1349
1350  if (!Sym->isUndefined())
1351    return Error(IDLoc, "invalid symbol redefinition");
1352
1353  // '.lcomm' is equivalent to '.zerofill'.
1354  // Create the Symbol as a common or local common with Size and Pow2Alignment
1355  if (IsLocal) {
1356    Out.EmitZerofill(getMachOSection("__DATA", "__bss",
1357                                     MCSectionMachO::S_ZEROFILL, 0,
1358                                     SectionKind()),
1359                     Sym, Size, 1 << Pow2Alignment);
1360    return false;
1361  }
1362
1363  Out.EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment);
1364  return false;
1365}
1366
1367/// ParseDirectiveDarwinZerofill
1368///  ::= .zerofill segname , sectname [, identifier , size_expression [
1369///      , align_expression ]]
1370bool AsmParser::ParseDirectiveDarwinZerofill() {
1371  // FIXME: Handle quoted names here.
1372
1373  if (Lexer.isNot(AsmToken::Identifier))
1374    return TokError("expected segment name after '.zerofill' directive");
1375  StringRef Segment = getTok().getString();
1376  Lex();
1377
1378  if (Lexer.isNot(AsmToken::Comma))
1379    return TokError("unexpected token in directive");
1380  Lex();
1381
1382  if (Lexer.isNot(AsmToken::Identifier))
1383    return TokError("expected section name after comma in '.zerofill' "
1384                    "directive");
1385  StringRef Section = getTok().getString();
1386  Lex();
1387
1388  // If this is the end of the line all that was wanted was to create the
1389  // the section but with no symbol.
1390  if (Lexer.is(AsmToken::EndOfStatement)) {
1391    // Create the zerofill section but no symbol
1392    Out.EmitZerofill(getMachOSection(Segment, Section,
1393                                     MCSectionMachO::S_ZEROFILL, 0,
1394                                     SectionKind()));
1395    return false;
1396  }
1397
1398  if (Lexer.isNot(AsmToken::Comma))
1399    return TokError("unexpected token in directive");
1400  Lex();
1401
1402  if (Lexer.isNot(AsmToken::Identifier))
1403    return TokError("expected identifier in directive");
1404
1405  // handle the identifier as the key symbol.
1406  SMLoc IDLoc = Lexer.getLoc();
1407  MCSymbol *Sym = CreateSymbol(getTok().getString());
1408  Lex();
1409
1410  if (Lexer.isNot(AsmToken::Comma))
1411    return TokError("unexpected token in directive");
1412  Lex();
1413
1414  int64_t Size;
1415  SMLoc SizeLoc = Lexer.getLoc();
1416  if (ParseAbsoluteExpression(Size))
1417    return true;
1418
1419  int64_t Pow2Alignment = 0;
1420  SMLoc Pow2AlignmentLoc;
1421  if (Lexer.is(AsmToken::Comma)) {
1422    Lex();
1423    Pow2AlignmentLoc = Lexer.getLoc();
1424    if (ParseAbsoluteExpression(Pow2Alignment))
1425      return true;
1426  }
1427
1428  if (Lexer.isNot(AsmToken::EndOfStatement))
1429    return TokError("unexpected token in '.zerofill' directive");
1430
1431  Lex();
1432
1433  if (Size < 0)
1434    return Error(SizeLoc, "invalid '.zerofill' directive size, can't be less "
1435                 "than zero");
1436
1437  // NOTE: The alignment in the directive is a power of 2 value, the assember
1438  // may internally end up wanting an alignment in bytes.
1439  // FIXME: Diagnose overflow.
1440  if (Pow2Alignment < 0)
1441    return Error(Pow2AlignmentLoc, "invalid '.zerofill' directive alignment, "
1442                 "can't be less than zero");
1443
1444  if (!Sym->isUndefined())
1445    return Error(IDLoc, "invalid symbol redefinition");
1446
1447  // Create the zerofill Symbol with Size and Pow2Alignment
1448  //
1449  // FIXME: Arch specific.
1450  Out.EmitZerofill(getMachOSection(Segment, Section,
1451                                 MCSectionMachO::S_ZEROFILL, 0,
1452                                 SectionKind()),
1453                   Sym, Size, 1 << Pow2Alignment);
1454
1455  return false;
1456}
1457
1458/// ParseDirectiveDarwinSubsectionsViaSymbols
1459///  ::= .subsections_via_symbols
1460bool AsmParser::ParseDirectiveDarwinSubsectionsViaSymbols() {
1461  if (Lexer.isNot(AsmToken::EndOfStatement))
1462    return TokError("unexpected token in '.subsections_via_symbols' directive");
1463
1464  Lex();
1465
1466  Out.EmitAssemblerFlag(MCAF_SubsectionsViaSymbols);
1467
1468  return false;
1469}
1470
1471/// ParseDirectiveAbort
1472///  ::= .abort [ "abort_string" ]
1473bool AsmParser::ParseDirectiveAbort() {
1474  // FIXME: Use loc from directive.
1475  SMLoc Loc = Lexer.getLoc();
1476
1477  StringRef Str = "";
1478  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1479    if (Lexer.isNot(AsmToken::String))
1480      return TokError("expected string in '.abort' directive");
1481
1482    Str = getTok().getString();
1483
1484    Lex();
1485  }
1486
1487  if (Lexer.isNot(AsmToken::EndOfStatement))
1488    return TokError("unexpected token in '.abort' directive");
1489
1490  Lex();
1491
1492  // FIXME: Handle here.
1493  if (Str.empty())
1494    Error(Loc, ".abort detected. Assembly stopping.");
1495  else
1496    Error(Loc, ".abort '" + Str + "' detected. Assembly stopping.");
1497
1498  return false;
1499}
1500
1501/// ParseDirectiveLsym
1502///  ::= .lsym identifier , expression
1503bool AsmParser::ParseDirectiveDarwinLsym() {
1504  StringRef Name;
1505  if (ParseIdentifier(Name))
1506    return TokError("expected identifier in directive");
1507
1508  // Handle the identifier as the key symbol.
1509  MCSymbol *Sym = CreateSymbol(Name);
1510
1511  if (Lexer.isNot(AsmToken::Comma))
1512    return TokError("unexpected token in '.lsym' directive");
1513  Lex();
1514
1515  const MCExpr *Value;
1516  SMLoc StartLoc = Lexer.getLoc();
1517  if (ParseExpression(Value))
1518    return true;
1519
1520  if (Lexer.isNot(AsmToken::EndOfStatement))
1521    return TokError("unexpected token in '.lsym' directive");
1522
1523  Lex();
1524
1525  // We don't currently support this directive.
1526  //
1527  // FIXME: Diagnostic location!
1528  (void) Sym;
1529  return TokError("directive '.lsym' is unsupported");
1530}
1531
1532/// ParseDirectiveInclude
1533///  ::= .include "filename"
1534bool AsmParser::ParseDirectiveInclude() {
1535  if (Lexer.isNot(AsmToken::String))
1536    return TokError("expected string in '.include' directive");
1537
1538  std::string Filename = getTok().getString();
1539  SMLoc IncludeLoc = Lexer.getLoc();
1540  Lex();
1541
1542  if (Lexer.isNot(AsmToken::EndOfStatement))
1543    return TokError("unexpected token in '.include' directive");
1544
1545  // Strip the quotes.
1546  Filename = Filename.substr(1, Filename.size()-2);
1547
1548  // Attempt to switch the lexer to the included file before consuming the end
1549  // of statement to avoid losing it when we switch.
1550  if (EnterIncludeFile(Filename)) {
1551    PrintMessage(IncludeLoc,
1552                 "Could not find include file '" + Filename + "'",
1553                 "error");
1554    return true;
1555  }
1556
1557  return false;
1558}
1559
1560/// ParseDirectiveDarwinDumpOrLoad
1561///  ::= ( .dump | .load ) "filename"
1562bool AsmParser::ParseDirectiveDarwinDumpOrLoad(SMLoc IDLoc, bool IsDump) {
1563  if (Lexer.isNot(AsmToken::String))
1564    return TokError("expected string in '.dump' or '.load' directive");
1565
1566  Lex();
1567
1568  if (Lexer.isNot(AsmToken::EndOfStatement))
1569    return TokError("unexpected token in '.dump' or '.load' directive");
1570
1571  Lex();
1572
1573  // FIXME: If/when .dump and .load are implemented they will be done in the
1574  // the assembly parser and not have any need for an MCStreamer API.
1575  if (IsDump)
1576    Warning(IDLoc, "ignoring directive .dump for now");
1577  else
1578    Warning(IDLoc, "ignoring directive .load for now");
1579
1580  return false;
1581}
1582
1583/// ParseDirectiveIf
1584/// ::= .if expression
1585bool AsmParser::ParseDirectiveIf(SMLoc DirectiveLoc) {
1586  // Consume the identifier that was the .if directive
1587  Lex();
1588
1589  TheCondStack.push_back(TheCondState);
1590  TheCondState.TheCond = AsmCond::IfCond;
1591  if(TheCondState.Ignore) {
1592    EatToEndOfStatement();
1593  }
1594  else {
1595    int64_t ExprValue;
1596    if (ParseAbsoluteExpression(ExprValue))
1597      return true;
1598
1599    if (Lexer.isNot(AsmToken::EndOfStatement))
1600      return TokError("unexpected token in '.if' directive");
1601
1602    Lex();
1603
1604    TheCondState.CondMet = ExprValue;
1605    TheCondState.Ignore = !TheCondState.CondMet;
1606  }
1607
1608  return false;
1609}
1610
1611/// ParseDirectiveElseIf
1612/// ::= .elseif expression
1613bool AsmParser::ParseDirectiveElseIf(SMLoc DirectiveLoc) {
1614  if (TheCondState.TheCond != AsmCond::IfCond &&
1615      TheCondState.TheCond != AsmCond::ElseIfCond)
1616      Error(DirectiveLoc, "Encountered a .elseif that doesn't follow a .if or "
1617                          " an .elseif");
1618  TheCondState.TheCond = AsmCond::ElseIfCond;
1619
1620  // Consume the identifier that was the .elseif directive
1621  Lex();
1622
1623  bool LastIgnoreState = false;
1624  if (!TheCondStack.empty())
1625      LastIgnoreState = TheCondStack.back().Ignore;
1626  if (LastIgnoreState || TheCondState.CondMet) {
1627    TheCondState.Ignore = true;
1628    EatToEndOfStatement();
1629  }
1630  else {
1631    int64_t ExprValue;
1632    if (ParseAbsoluteExpression(ExprValue))
1633      return true;
1634
1635    if (Lexer.isNot(AsmToken::EndOfStatement))
1636      return TokError("unexpected token in '.elseif' directive");
1637
1638    Lex();
1639    TheCondState.CondMet = ExprValue;
1640    TheCondState.Ignore = !TheCondState.CondMet;
1641  }
1642
1643  return false;
1644}
1645
1646/// ParseDirectiveElse
1647/// ::= .else
1648bool AsmParser::ParseDirectiveElse(SMLoc DirectiveLoc) {
1649  // Consume the identifier that was the .else directive
1650  Lex();
1651
1652  if (Lexer.isNot(AsmToken::EndOfStatement))
1653    return TokError("unexpected token in '.else' directive");
1654
1655  Lex();
1656
1657  if (TheCondState.TheCond != AsmCond::IfCond &&
1658      TheCondState.TheCond != AsmCond::ElseIfCond)
1659      Error(DirectiveLoc, "Encountered a .else that doesn't follow a .if or an "
1660                          ".elseif");
1661  TheCondState.TheCond = AsmCond::ElseCond;
1662  bool LastIgnoreState = false;
1663  if (!TheCondStack.empty())
1664    LastIgnoreState = TheCondStack.back().Ignore;
1665  if (LastIgnoreState || TheCondState.CondMet)
1666    TheCondState.Ignore = true;
1667  else
1668    TheCondState.Ignore = false;
1669
1670  return false;
1671}
1672
1673/// ParseDirectiveEndIf
1674/// ::= .endif
1675bool AsmParser::ParseDirectiveEndIf(SMLoc DirectiveLoc) {
1676  // Consume the identifier that was the .endif directive
1677  Lex();
1678
1679  if (Lexer.isNot(AsmToken::EndOfStatement))
1680    return TokError("unexpected token in '.endif' directive");
1681
1682  Lex();
1683
1684  if ((TheCondState.TheCond == AsmCond::NoCond) ||
1685      TheCondStack.empty())
1686    Error(DirectiveLoc, "Encountered a .endif that doesn't follow a .if or "
1687                        ".else");
1688  if (!TheCondStack.empty()) {
1689    TheCondState = TheCondStack.back();
1690    TheCondStack.pop_back();
1691  }
1692
1693  return false;
1694}
1695
1696/// ParseDirectiveFile
1697/// ::= .file [number] string
1698bool AsmParser::ParseDirectiveFile(StringRef, SMLoc DirectiveLoc) {
1699  // FIXME: I'm not sure what this is.
1700  int64_t FileNumber = -1;
1701  if (Lexer.is(AsmToken::Integer)) {
1702    FileNumber = getTok().getIntVal();
1703    Lex();
1704
1705    if (FileNumber < 1)
1706      return TokError("file number less than one");
1707  }
1708
1709  if (Lexer.isNot(AsmToken::String))
1710    return TokError("unexpected token in '.file' directive");
1711
1712  StringRef ATTRIBUTE_UNUSED FileName = getTok().getString();
1713  Lex();
1714
1715  if (Lexer.isNot(AsmToken::EndOfStatement))
1716    return TokError("unexpected token in '.file' directive");
1717
1718  // FIXME: Do something with the .file.
1719
1720  return false;
1721}
1722
1723/// ParseDirectiveLine
1724/// ::= .line [number]
1725bool AsmParser::ParseDirectiveLine(StringRef, SMLoc DirectiveLoc) {
1726  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1727    if (Lexer.isNot(AsmToken::Integer))
1728      return TokError("unexpected token in '.line' directive");
1729
1730    int64_t LineNumber = getTok().getIntVal();
1731    (void) LineNumber;
1732    Lex();
1733
1734    // FIXME: Do something with the .line.
1735  }
1736
1737  if (Lexer.isNot(AsmToken::EndOfStatement))
1738    return TokError("unexpected token in '.file' directive");
1739
1740  return false;
1741}
1742
1743
1744/// ParseDirectiveLoc
1745/// ::= .loc number [number [number]]
1746bool AsmParser::ParseDirectiveLoc(StringRef, SMLoc DirectiveLoc) {
1747  if (Lexer.isNot(AsmToken::Integer))
1748    return TokError("unexpected token in '.loc' directive");
1749
1750  // FIXME: What are these fields?
1751  int64_t FileNumber = getTok().getIntVal();
1752  (void) FileNumber;
1753  // FIXME: Validate file.
1754
1755  Lex();
1756  if (Lexer.isNot(AsmToken::EndOfStatement)) {
1757    if (Lexer.isNot(AsmToken::Integer))
1758      return TokError("unexpected token in '.loc' directive");
1759
1760    int64_t Param2 = getTok().getIntVal();
1761    (void) Param2;
1762    Lex();
1763
1764    if (Lexer.isNot(AsmToken::EndOfStatement)) {
1765      if (Lexer.isNot(AsmToken::Integer))
1766        return TokError("unexpected token in '.loc' directive");
1767
1768      int64_t Param3 = getTok().getIntVal();
1769      (void) Param3;
1770      Lex();
1771
1772      // FIXME: Do something with the .loc.
1773    }
1774  }
1775
1776  if (Lexer.isNot(AsmToken::EndOfStatement))
1777    return TokError("unexpected token in '.file' directive");
1778
1779  return false;
1780}
1781
1782