1//===- ELFAsmParser.cpp - ELF Assembly Parser -----------------------------===//
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#include "llvm/MC/MCParser/MCAsmParserExtension.h"
11#include "llvm/ADT/StringSwitch.h"
12#include "llvm/ADT/Twine.h"
13#include "llvm/MC/MCAsmInfo.h"
14#include "llvm/MC/MCContext.h"
15#include "llvm/MC/MCExpr.h"
16#include "llvm/MC/MCParser/MCAsmLexer.h"
17#include "llvm/MC/MCSectionELF.h"
18#include "llvm/MC/MCStreamer.h"
19#include "llvm/Support/ELF.h"
20using namespace llvm;
21
22namespace {
23
24class ELFAsmParser : public MCAsmParserExtension {
25  template<bool (ELFAsmParser::*Handler)(StringRef, SMLoc)>
26  void AddDirectiveHandler(StringRef Directive) {
27    getParser().AddDirectiveHandler(this, Directive,
28                                    HandleDirective<ELFAsmParser, Handler>);
29  }
30
31  bool ParseSectionSwitch(StringRef Section, unsigned Type,
32                          unsigned Flags, SectionKind Kind);
33  bool SeenIdent;
34
35public:
36  ELFAsmParser() : SeenIdent(false) {
37    BracketExpressionsSupported = true;
38  }
39
40  virtual void Initialize(MCAsmParser &Parser) {
41    // Call the base implementation.
42    this->MCAsmParserExtension::Initialize(Parser);
43
44    AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveData>(".data");
45    AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveText>(".text");
46    AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveBSS>(".bss");
47    AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveRoData>(".rodata");
48    AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveTData>(".tdata");
49    AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveTBSS>(".tbss");
50    AddDirectiveHandler<
51      &ELFAsmParser::ParseSectionDirectiveDataRel>(".data.rel");
52    AddDirectiveHandler<
53      &ELFAsmParser::ParseSectionDirectiveDataRelRo>(".data.rel.ro");
54    AddDirectiveHandler<
55      &ELFAsmParser::ParseSectionDirectiveDataRelRoLocal>(".data.rel.ro.local");
56    AddDirectiveHandler<
57      &ELFAsmParser::ParseSectionDirectiveEhFrame>(".eh_frame");
58    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSection>(".section");
59    AddDirectiveHandler<
60      &ELFAsmParser::ParseDirectivePushSection>(".pushsection");
61    AddDirectiveHandler<&ELFAsmParser::ParseDirectivePopSection>(".popsection");
62    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSize>(".size");
63    AddDirectiveHandler<&ELFAsmParser::ParseDirectivePrevious>(".previous");
64    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveType>(".type");
65    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveIdent>(".ident");
66    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSymver>(".symver");
67    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveVersion>(".version");
68    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveWeakref>(".weakref");
69    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSymbolAttribute>(".weak");
70    AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSymbolAttribute>(".local");
71    AddDirectiveHandler<
72      &ELFAsmParser::ParseDirectiveSymbolAttribute>(".protected");
73    AddDirectiveHandler<
74      &ELFAsmParser::ParseDirectiveSymbolAttribute>(".internal");
75    AddDirectiveHandler<
76      &ELFAsmParser::ParseDirectiveSymbolAttribute>(".hidden");
77  }
78
79  // FIXME: Part of this logic is duplicated in the MCELFStreamer. What is
80  // the best way for us to get access to it?
81  bool ParseSectionDirectiveData(StringRef, SMLoc) {
82    return ParseSectionSwitch(".data", ELF::SHT_PROGBITS,
83                              ELF::SHF_WRITE |ELF::SHF_ALLOC,
84                              SectionKind::getDataRel());
85  }
86  bool ParseSectionDirectiveText(StringRef, SMLoc) {
87    return ParseSectionSwitch(".text", ELF::SHT_PROGBITS,
88                              ELF::SHF_EXECINSTR |
89                              ELF::SHF_ALLOC, SectionKind::getText());
90  }
91  bool ParseSectionDirectiveBSS(StringRef, SMLoc) {
92    return ParseSectionSwitch(".bss", ELF::SHT_NOBITS,
93                              ELF::SHF_WRITE |
94                              ELF::SHF_ALLOC, SectionKind::getBSS());
95  }
96  bool ParseSectionDirectiveRoData(StringRef, SMLoc) {
97    return ParseSectionSwitch(".rodata", ELF::SHT_PROGBITS,
98                              ELF::SHF_ALLOC,
99                              SectionKind::getReadOnly());
100  }
101  bool ParseSectionDirectiveTData(StringRef, SMLoc) {
102    return ParseSectionSwitch(".tdata", ELF::SHT_PROGBITS,
103                              ELF::SHF_ALLOC |
104                              ELF::SHF_TLS | ELF::SHF_WRITE,
105                              SectionKind::getThreadData());
106  }
107  bool ParseSectionDirectiveTBSS(StringRef, SMLoc) {
108    return ParseSectionSwitch(".tbss", ELF::SHT_NOBITS,
109                              ELF::SHF_ALLOC |
110                              ELF::SHF_TLS | ELF::SHF_WRITE,
111                              SectionKind::getThreadBSS());
112  }
113  bool ParseSectionDirectiveDataRel(StringRef, SMLoc) {
114    return ParseSectionSwitch(".data.rel", ELF::SHT_PROGBITS,
115                              ELF::SHF_ALLOC |
116                              ELF::SHF_WRITE,
117                              SectionKind::getDataRel());
118  }
119  bool ParseSectionDirectiveDataRelRo(StringRef, SMLoc) {
120    return ParseSectionSwitch(".data.rel.ro", ELF::SHT_PROGBITS,
121                              ELF::SHF_ALLOC |
122                              ELF::SHF_WRITE,
123                              SectionKind::getReadOnlyWithRel());
124  }
125  bool ParseSectionDirectiveDataRelRoLocal(StringRef, SMLoc) {
126    return ParseSectionSwitch(".data.rel.ro.local", ELF::SHT_PROGBITS,
127                              ELF::SHF_ALLOC |
128                              ELF::SHF_WRITE,
129                              SectionKind::getReadOnlyWithRelLocal());
130  }
131  bool ParseSectionDirectiveEhFrame(StringRef, SMLoc) {
132    return ParseSectionSwitch(".eh_frame", ELF::SHT_PROGBITS,
133                              ELF::SHF_ALLOC |
134                              ELF::SHF_WRITE,
135                              SectionKind::getDataRel());
136  }
137  bool ParseDirectivePushSection(StringRef, SMLoc);
138  bool ParseDirectivePopSection(StringRef, SMLoc);
139  bool ParseDirectiveSection(StringRef, SMLoc);
140  bool ParseDirectiveSize(StringRef, SMLoc);
141  bool ParseDirectivePrevious(StringRef, SMLoc);
142  bool ParseDirectiveType(StringRef, SMLoc);
143  bool ParseDirectiveIdent(StringRef, SMLoc);
144  bool ParseDirectiveSymver(StringRef, SMLoc);
145  bool ParseDirectiveVersion(StringRef, SMLoc);
146  bool ParseDirectiveWeakref(StringRef, SMLoc);
147  bool ParseDirectiveSymbolAttribute(StringRef, SMLoc);
148
149private:
150  bool ParseSectionName(StringRef &SectionName);
151};
152
153}
154
155/// ParseDirectiveSymbolAttribute
156///  ::= { ".local", ".weak", ... } [ identifier ( , identifier )* ]
157bool ELFAsmParser::ParseDirectiveSymbolAttribute(StringRef Directive, SMLoc) {
158  MCSymbolAttr Attr = StringSwitch<MCSymbolAttr>(Directive)
159    .Case(".weak", MCSA_Weak)
160    .Case(".local", MCSA_Local)
161    .Case(".hidden", MCSA_Hidden)
162    .Case(".internal", MCSA_Internal)
163    .Case(".protected", MCSA_Protected)
164    .Default(MCSA_Invalid);
165  assert(Attr != MCSA_Invalid && "unexpected symbol attribute directive!");
166  if (getLexer().isNot(AsmToken::EndOfStatement)) {
167    for (;;) {
168      StringRef Name;
169
170      if (getParser().ParseIdentifier(Name))
171        return TokError("expected identifier in directive");
172
173      MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
174
175      getStreamer().EmitSymbolAttribute(Sym, Attr);
176
177      if (getLexer().is(AsmToken::EndOfStatement))
178        break;
179
180      if (getLexer().isNot(AsmToken::Comma))
181        return TokError("unexpected token in directive");
182      Lex();
183    }
184  }
185
186  Lex();
187  return false;
188}
189
190bool ELFAsmParser::ParseSectionSwitch(StringRef Section, unsigned Type,
191                                      unsigned Flags, SectionKind Kind) {
192  if (getLexer().isNot(AsmToken::EndOfStatement))
193    return TokError("unexpected token in section switching directive");
194  Lex();
195
196  getStreamer().SwitchSection(getContext().getELFSection(
197                                Section, Type, Flags, Kind));
198
199  return false;
200}
201
202bool ELFAsmParser::ParseDirectiveSize(StringRef, SMLoc) {
203  StringRef Name;
204  if (getParser().ParseIdentifier(Name))
205    return TokError("expected identifier in directive");
206  MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
207
208  if (getLexer().isNot(AsmToken::Comma))
209    return TokError("unexpected token in directive");
210  Lex();
211
212  const MCExpr *Expr;
213  if (getParser().ParseExpression(Expr))
214    return true;
215
216  if (getLexer().isNot(AsmToken::EndOfStatement))
217    return TokError("unexpected token in directive");
218
219  getStreamer().EmitELFSize(Sym, Expr);
220  return false;
221}
222
223bool ELFAsmParser::ParseSectionName(StringRef &SectionName) {
224  // A section name can contain -, so we cannot just use
225  // ParseIdentifier.
226  SMLoc FirstLoc = getLexer().getLoc();
227  unsigned Size = 0;
228
229  if (getLexer().is(AsmToken::String)) {
230    SectionName = getTok().getIdentifier();
231    Lex();
232    return false;
233  }
234
235  for (;;) {
236    StringRef Tmp;
237    unsigned CurSize;
238
239    SMLoc PrevLoc = getLexer().getLoc();
240    if (getLexer().is(AsmToken::Minus)) {
241      CurSize = 1;
242      Lex(); // Consume the "-".
243    } else if (getLexer().is(AsmToken::String)) {
244      CurSize = getTok().getIdentifier().size() + 2;
245      Lex();
246    } else if (getLexer().is(AsmToken::Identifier)) {
247      CurSize = getTok().getIdentifier().size();
248      Lex();
249    } else {
250      break;
251    }
252
253    Size += CurSize;
254    SectionName = StringRef(FirstLoc.getPointer(), Size);
255
256    // Make sure the following token is adjacent.
257    if (PrevLoc.getPointer() + CurSize != getTok().getLoc().getPointer())
258      break;
259  }
260  if (Size == 0)
261    return true;
262
263  return false;
264}
265
266static SectionKind computeSectionKind(unsigned Flags) {
267  if (Flags & ELF::SHF_EXECINSTR)
268    return SectionKind::getText();
269  if (Flags & ELF::SHF_TLS)
270    return SectionKind::getThreadData();
271  return SectionKind::getDataRel();
272}
273
274static int parseSectionFlags(StringRef flagsStr) {
275  int flags = 0;
276
277  for (unsigned i = 0; i < flagsStr.size(); i++) {
278    switch (flagsStr[i]) {
279    case 'a':
280      flags |= ELF::SHF_ALLOC;
281      break;
282    case 'x':
283      flags |= ELF::SHF_EXECINSTR;
284      break;
285    case 'w':
286      flags |= ELF::SHF_WRITE;
287      break;
288    case 'M':
289      flags |= ELF::SHF_MERGE;
290      break;
291    case 'S':
292      flags |= ELF::SHF_STRINGS;
293      break;
294    case 'T':
295      flags |= ELF::SHF_TLS;
296      break;
297    case 'c':
298      flags |= ELF::XCORE_SHF_CP_SECTION;
299      break;
300    case 'd':
301      flags |= ELF::XCORE_SHF_DP_SECTION;
302      break;
303    case 'G':
304      flags |= ELF::SHF_GROUP;
305      break;
306    default:
307      return -1;
308    }
309  }
310
311  return flags;
312}
313
314bool ELFAsmParser::ParseDirectivePushSection(StringRef s, SMLoc loc) {
315  getStreamer().PushSection();
316
317  if (ParseDirectiveSection(s, loc)) {
318    getStreamer().PopSection();
319    return true;
320  }
321
322  return false;
323}
324
325bool ELFAsmParser::ParseDirectivePopSection(StringRef, SMLoc) {
326  if (!getStreamer().PopSection())
327    return TokError(".popsection without corresponding .pushsection");
328  return false;
329}
330
331// FIXME: This is a work in progress.
332bool ELFAsmParser::ParseDirectiveSection(StringRef, SMLoc) {
333  StringRef SectionName;
334
335  if (ParseSectionName(SectionName))
336    return TokError("expected identifier in directive");
337
338  StringRef TypeName;
339  int64_t Size = 0;
340  StringRef GroupName;
341  unsigned Flags = 0;
342
343  // Set the defaults first.
344  if (SectionName == ".fini" || SectionName == ".init" ||
345      SectionName == ".rodata")
346    Flags |= ELF::SHF_ALLOC;
347  if (SectionName == ".fini" || SectionName == ".init")
348    Flags |= ELF::SHF_EXECINSTR;
349
350  if (getLexer().is(AsmToken::Comma)) {
351    Lex();
352
353    if (getLexer().isNot(AsmToken::String))
354      return TokError("expected string in directive");
355
356    StringRef FlagsStr = getTok().getStringContents();
357    Lex();
358
359    int extraFlags = parseSectionFlags(FlagsStr);
360    if (extraFlags < 0)
361      return TokError("unknown flag");
362    Flags |= extraFlags;
363
364    bool Mergeable = Flags & ELF::SHF_MERGE;
365    bool Group = Flags & ELF::SHF_GROUP;
366
367    if (getLexer().isNot(AsmToken::Comma)) {
368      if (Mergeable)
369        return TokError("Mergeable section must specify the type");
370      if (Group)
371        return TokError("Group section must specify the type");
372    } else {
373      Lex();
374      if (getLexer().isNot(AsmToken::Percent) && getLexer().isNot(AsmToken::At))
375        return TokError("expected '@' or '%' before type");
376
377      Lex();
378      if (getParser().ParseIdentifier(TypeName))
379        return TokError("expected identifier in directive");
380
381      if (Mergeable) {
382        if (getLexer().isNot(AsmToken::Comma))
383          return TokError("expected the entry size");
384        Lex();
385        if (getParser().ParseAbsoluteExpression(Size))
386          return true;
387        if (Size <= 0)
388          return TokError("entry size must be positive");
389      }
390
391      if (Group) {
392        if (getLexer().isNot(AsmToken::Comma))
393          return TokError("expected group name");
394        Lex();
395        if (getParser().ParseIdentifier(GroupName))
396          return true;
397        if (getLexer().is(AsmToken::Comma)) {
398          Lex();
399          StringRef Linkage;
400          if (getParser().ParseIdentifier(Linkage))
401            return true;
402          if (Linkage != "comdat")
403            return TokError("Linkage must be 'comdat'");
404        }
405      }
406    }
407  }
408
409  if (getLexer().isNot(AsmToken::EndOfStatement))
410    return TokError("unexpected token in directive");
411
412  unsigned Type = ELF::SHT_PROGBITS;
413
414  if (!TypeName.empty()) {
415    if (TypeName == "init_array")
416      Type = ELF::SHT_INIT_ARRAY;
417    else if (TypeName == "fini_array")
418      Type = ELF::SHT_FINI_ARRAY;
419    else if (TypeName == "preinit_array")
420      Type = ELF::SHT_PREINIT_ARRAY;
421    else if (TypeName == "nobits")
422      Type = ELF::SHT_NOBITS;
423    else if (TypeName == "progbits")
424      Type = ELF::SHT_PROGBITS;
425    else if (TypeName == "note")
426      Type = ELF::SHT_NOTE;
427    else if (TypeName == "unwind")
428      Type = ELF::SHT_X86_64_UNWIND;
429    else
430      return TokError("unknown section type");
431  }
432
433  SectionKind Kind = computeSectionKind(Flags);
434  getStreamer().SwitchSection(getContext().getELFSection(SectionName, Type,
435                                                         Flags, Kind, Size,
436                                                         GroupName));
437  return false;
438}
439
440bool ELFAsmParser::ParseDirectivePrevious(StringRef DirName, SMLoc) {
441  const MCSection *PreviousSection = getStreamer().getPreviousSection();
442  if (PreviousSection == NULL)
443      return TokError(".previous without corresponding .section");
444  getStreamer().SwitchSection(PreviousSection);
445
446  return false;
447}
448
449/// ParseDirectiveELFType
450///  ::= .type identifier , @attribute
451bool ELFAsmParser::ParseDirectiveType(StringRef, SMLoc) {
452  StringRef Name;
453  if (getParser().ParseIdentifier(Name))
454    return TokError("expected identifier in directive");
455
456  // Handle the identifier as the key symbol.
457  MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
458
459  if (getLexer().isNot(AsmToken::Comma))
460    return TokError("unexpected token in '.type' directive");
461  Lex();
462
463  if (getLexer().isNot(AsmToken::Percent) && getLexer().isNot(AsmToken::At))
464    return TokError("expected '@' or '%' before type");
465  Lex();
466
467  StringRef Type;
468  SMLoc TypeLoc;
469
470  TypeLoc = getLexer().getLoc();
471  if (getParser().ParseIdentifier(Type))
472    return TokError("expected symbol type in directive");
473
474  MCSymbolAttr Attr = StringSwitch<MCSymbolAttr>(Type)
475    .Case("function", MCSA_ELF_TypeFunction)
476    .Case("object", MCSA_ELF_TypeObject)
477    .Case("tls_object", MCSA_ELF_TypeTLS)
478    .Case("common", MCSA_ELF_TypeCommon)
479    .Case("notype", MCSA_ELF_TypeNoType)
480    .Case("gnu_unique_object", MCSA_ELF_TypeGnuUniqueObject)
481    .Case("gnu_indirect_function", MCSA_ELF_TypeIndFunction)
482    .Default(MCSA_Invalid);
483
484  if (Attr == MCSA_Invalid)
485    return Error(TypeLoc, "unsupported attribute in '.type' directive");
486
487  if (getLexer().isNot(AsmToken::EndOfStatement))
488    return TokError("unexpected token in '.type' directive");
489
490  Lex();
491
492  getStreamer().EmitSymbolAttribute(Sym, Attr);
493
494  return false;
495}
496
497/// ParseDirectiveIdent
498///  ::= .ident string
499bool ELFAsmParser::ParseDirectiveIdent(StringRef, SMLoc) {
500  if (getLexer().isNot(AsmToken::String))
501    return TokError("unexpected token in '.ident' directive");
502
503  StringRef Data = getTok().getIdentifier();
504
505  Lex();
506
507  const MCSection *Comment =
508    getContext().getELFSection(".comment", ELF::SHT_PROGBITS,
509                               ELF::SHF_MERGE |
510                               ELF::SHF_STRINGS,
511                               SectionKind::getReadOnly(),
512                               1, "");
513
514  getStreamer().PushSection();
515  getStreamer().SwitchSection(Comment);
516  if (!SeenIdent) {
517    getStreamer().EmitIntValue(0, 1);
518    SeenIdent = true;
519  }
520  getStreamer().EmitBytes(Data, 0);
521  getStreamer().EmitIntValue(0, 1);
522  getStreamer().PopSection();
523  return false;
524}
525
526/// ParseDirectiveSymver
527///  ::= .symver foo, bar2@zed
528bool ELFAsmParser::ParseDirectiveSymver(StringRef, SMLoc) {
529  StringRef Name;
530  if (getParser().ParseIdentifier(Name))
531    return TokError("expected identifier in directive");
532
533  if (getLexer().isNot(AsmToken::Comma))
534    return TokError("expected a comma");
535
536  Lex();
537
538  StringRef AliasName;
539  if (getParser().ParseIdentifier(AliasName))
540    return TokError("expected identifier in directive");
541
542  if (AliasName.find('@') == StringRef::npos)
543    return TokError("expected a '@' in the name");
544
545  MCSymbol *Alias = getContext().GetOrCreateSymbol(AliasName);
546  MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
547  const MCExpr *Value = MCSymbolRefExpr::Create(Sym, getContext());
548
549  getStreamer().EmitAssignment(Alias, Value);
550  return false;
551}
552
553/// ParseDirectiveVersion
554///  ::= .version string
555bool ELFAsmParser::ParseDirectiveVersion(StringRef, SMLoc) {
556  if (getLexer().isNot(AsmToken::String))
557    return TokError("unexpected token in '.version' directive");
558
559  StringRef Data = getTok().getIdentifier();
560
561  Lex();
562
563  const MCSection *Note =
564    getContext().getELFSection(".note", ELF::SHT_NOTE, 0,
565                               SectionKind::getReadOnly());
566
567  getStreamer().PushSection();
568  getStreamer().SwitchSection(Note);
569  getStreamer().EmitIntValue(Data.size()+1, 4); // namesz.
570  getStreamer().EmitIntValue(0, 4);             // descsz = 0 (no description).
571  getStreamer().EmitIntValue(1, 4);             // type = NT_VERSION.
572  getStreamer().EmitBytes(Data, 0);             // name.
573  getStreamer().EmitIntValue(0, 1);             // terminate the string.
574  getStreamer().EmitValueToAlignment(4);        // ensure 4 byte alignment.
575  getStreamer().PopSection();
576  return false;
577}
578
579/// ParseDirectiveWeakref
580///  ::= .weakref foo, bar
581bool ELFAsmParser::ParseDirectiveWeakref(StringRef, SMLoc) {
582  // FIXME: Share code with the other alias building directives.
583
584  StringRef AliasName;
585  if (getParser().ParseIdentifier(AliasName))
586    return TokError("expected identifier in directive");
587
588  if (getLexer().isNot(AsmToken::Comma))
589    return TokError("expected a comma");
590
591  Lex();
592
593  StringRef Name;
594  if (getParser().ParseIdentifier(Name))
595    return TokError("expected identifier in directive");
596
597  MCSymbol *Alias = getContext().GetOrCreateSymbol(AliasName);
598
599  MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
600
601  getStreamer().EmitWeakReference(Alias, Sym);
602  return false;
603}
604
605namespace llvm {
606
607MCAsmParserExtension *createELFAsmParser() {
608  return new ELFAsmParser;
609}
610
611}
612