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