1 //===- ELFAsmParser.cpp - ELF Assembly Parser -----------------------------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
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/MC/MCSymbol.h"
20 #include "llvm/Support/ELF.h"
25 class ELFAsmParser : public MCAsmParserExtension {
26 template<bool (ELFAsmParser::*HandlerMethod)(StringRef, SMLoc)>
27 void addDirectiveHandler(StringRef Directive) {
28 MCAsmParser::ExtensionDirectiveHandler Handler = std::make_pair(
29 this, HandleDirective<ELFAsmParser, HandlerMethod>);
31 getParser().addDirectiveHandler(Directive, Handler);
34 bool ParseSectionSwitch(StringRef Section, unsigned Type, unsigned Flags,
38 ELFAsmParser() { BracketExpressionsSupported = true; }
40 virtual void Initialize(MCAsmParser &Parser) {
41 // Call the base implementation.
42 this->MCAsmParserExtension::Initialize(Parser);
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");
51 &ELFAsmParser::ParseSectionDirectiveDataRel>(".data.rel");
53 &ELFAsmParser::ParseSectionDirectiveDataRelRo>(".data.rel.ro");
55 &ELFAsmParser::ParseSectionDirectiveDataRelRoLocal>(".data.rel.ro.local");
57 &ELFAsmParser::ParseSectionDirectiveEhFrame>(".eh_frame");
58 addDirectiveHandler<&ELFAsmParser::ParseDirectiveSection>(".section");
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");
72 &ELFAsmParser::ParseDirectiveSymbolAttribute>(".protected");
74 &ELFAsmParser::ParseDirectiveSymbolAttribute>(".internal");
76 &ELFAsmParser::ParseDirectiveSymbolAttribute>(".hidden");
77 addDirectiveHandler<&ELFAsmParser::ParseDirectiveSubsection>(".subsection");
80 // FIXME: Part of this logic is duplicated in the MCELFStreamer. What is
81 // the best way for us to get access to it?
82 bool ParseSectionDirectiveData(StringRef, SMLoc) {
83 return ParseSectionSwitch(".data", ELF::SHT_PROGBITS,
84 ELF::SHF_WRITE |ELF::SHF_ALLOC,
85 SectionKind::getDataRel());
87 bool ParseSectionDirectiveText(StringRef, SMLoc) {
88 return ParseSectionSwitch(".text", ELF::SHT_PROGBITS,
90 ELF::SHF_ALLOC, SectionKind::getText());
92 bool ParseSectionDirectiveBSS(StringRef, SMLoc) {
93 return ParseSectionSwitch(".bss", ELF::SHT_NOBITS,
95 ELF::SHF_ALLOC, SectionKind::getBSS());
97 bool ParseSectionDirectiveRoData(StringRef, SMLoc) {
98 return ParseSectionSwitch(".rodata", ELF::SHT_PROGBITS,
100 SectionKind::getReadOnly());
102 bool ParseSectionDirectiveTData(StringRef, SMLoc) {
103 return ParseSectionSwitch(".tdata", ELF::SHT_PROGBITS,
105 ELF::SHF_TLS | ELF::SHF_WRITE,
106 SectionKind::getThreadData());
108 bool ParseSectionDirectiveTBSS(StringRef, SMLoc) {
109 return ParseSectionSwitch(".tbss", ELF::SHT_NOBITS,
111 ELF::SHF_TLS | ELF::SHF_WRITE,
112 SectionKind::getThreadBSS());
114 bool ParseSectionDirectiveDataRel(StringRef, SMLoc) {
115 return ParseSectionSwitch(".data.rel", ELF::SHT_PROGBITS,
118 SectionKind::getDataRel());
120 bool ParseSectionDirectiveDataRelRo(StringRef, SMLoc) {
121 return ParseSectionSwitch(".data.rel.ro", ELF::SHT_PROGBITS,
124 SectionKind::getReadOnlyWithRel());
126 bool ParseSectionDirectiveDataRelRoLocal(StringRef, SMLoc) {
127 return ParseSectionSwitch(".data.rel.ro.local", ELF::SHT_PROGBITS,
130 SectionKind::getReadOnlyWithRelLocal());
132 bool ParseSectionDirectiveEhFrame(StringRef, SMLoc) {
133 return ParseSectionSwitch(".eh_frame", ELF::SHT_PROGBITS,
136 SectionKind::getDataRel());
138 bool ParseDirectivePushSection(StringRef, SMLoc);
139 bool ParseDirectivePopSection(StringRef, SMLoc);
140 bool ParseDirectiveSection(StringRef, SMLoc);
141 bool ParseDirectiveSize(StringRef, SMLoc);
142 bool ParseDirectivePrevious(StringRef, SMLoc);
143 bool ParseDirectiveType(StringRef, SMLoc);
144 bool ParseDirectiveIdent(StringRef, SMLoc);
145 bool ParseDirectiveSymver(StringRef, SMLoc);
146 bool ParseDirectiveVersion(StringRef, SMLoc);
147 bool ParseDirectiveWeakref(StringRef, SMLoc);
148 bool ParseDirectiveSymbolAttribute(StringRef, SMLoc);
149 bool ParseDirectiveSubsection(StringRef, SMLoc);
152 bool ParseSectionName(StringRef &SectionName);
153 bool ParseSectionArguments(bool IsPush);
154 unsigned parseSunStyleSectionFlags();
159 /// ParseDirectiveSymbolAttribute
160 /// ::= { ".local", ".weak", ... } [ identifier ( , identifier )* ]
161 bool ELFAsmParser::ParseDirectiveSymbolAttribute(StringRef Directive, SMLoc) {
162 MCSymbolAttr Attr = StringSwitch<MCSymbolAttr>(Directive)
163 .Case(".weak", MCSA_Weak)
164 .Case(".local", MCSA_Local)
165 .Case(".hidden", MCSA_Hidden)
166 .Case(".internal", MCSA_Internal)
167 .Case(".protected", MCSA_Protected)
168 .Default(MCSA_Invalid);
169 assert(Attr != MCSA_Invalid && "unexpected symbol attribute directive!");
170 if (getLexer().isNot(AsmToken::EndOfStatement)) {
174 if (getParser().parseIdentifier(Name))
175 return TokError("expected identifier in directive");
177 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
179 getStreamer().EmitSymbolAttribute(Sym, Attr);
181 if (getLexer().is(AsmToken::EndOfStatement))
184 if (getLexer().isNot(AsmToken::Comma))
185 return TokError("unexpected token in directive");
194 bool ELFAsmParser::ParseSectionSwitch(StringRef Section, unsigned Type,
195 unsigned Flags, SectionKind Kind) {
196 const MCExpr *Subsection = 0;
197 if (getLexer().isNot(AsmToken::EndOfStatement)) {
198 if (getParser().parseExpression(Subsection))
202 getStreamer().SwitchSection(getContext().getELFSection(
203 Section, Type, Flags, Kind),
209 bool ELFAsmParser::ParseDirectiveSize(StringRef, SMLoc) {
211 if (getParser().parseIdentifier(Name))
212 return TokError("expected identifier in directive");
213 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
215 if (getLexer().isNot(AsmToken::Comma))
216 return TokError("unexpected token in directive");
220 if (getParser().parseExpression(Expr))
223 if (getLexer().isNot(AsmToken::EndOfStatement))
224 return TokError("unexpected token in directive");
226 getStreamer().EmitELFSize(Sym, Expr);
230 bool ELFAsmParser::ParseSectionName(StringRef &SectionName) {
231 // A section name can contain -, so we cannot just use
233 SMLoc FirstLoc = getLexer().getLoc();
236 if (getLexer().is(AsmToken::String)) {
237 SectionName = getTok().getIdentifier();
245 SMLoc PrevLoc = getLexer().getLoc();
246 if (getLexer().is(AsmToken::Minus)) {
248 Lex(); // Consume the "-".
249 } else if (getLexer().is(AsmToken::String)) {
250 CurSize = getTok().getIdentifier().size() + 2;
252 } else if (getLexer().is(AsmToken::Identifier)) {
253 CurSize = getTok().getIdentifier().size();
260 SectionName = StringRef(FirstLoc.getPointer(), Size);
262 // Make sure the following token is adjacent.
263 if (PrevLoc.getPointer() + CurSize != getTok().getLoc().getPointer())
272 static SectionKind computeSectionKind(unsigned Flags) {
273 if (Flags & ELF::SHF_EXECINSTR)
274 return SectionKind::getText();
275 if (Flags & ELF::SHF_TLS)
276 return SectionKind::getThreadData();
277 return SectionKind::getDataRel();
280 static unsigned parseSectionFlags(StringRef flagsStr, bool *UseLastGroup) {
283 for (unsigned i = 0; i < flagsStr.size(); i++) {
284 switch (flagsStr[i]) {
286 flags |= ELF::SHF_ALLOC;
289 flags |= ELF::SHF_EXCLUDE;
292 flags |= ELF::SHF_EXECINSTR;
295 flags |= ELF::SHF_WRITE;
298 flags |= ELF::SHF_MERGE;
301 flags |= ELF::SHF_STRINGS;
304 flags |= ELF::SHF_TLS;
307 flags |= ELF::XCORE_SHF_CP_SECTION;
310 flags |= ELF::XCORE_SHF_DP_SECTION;
313 flags |= ELF::SHF_GROUP;
316 *UseLastGroup = true;
326 unsigned ELFAsmParser::parseSunStyleSectionFlags() {
328 while (getLexer().is(AsmToken::Hash)) {
331 if (!getLexer().is(AsmToken::Identifier))
334 StringRef flagId = getTok().getIdentifier();
335 if (flagId == "alloc")
336 flags |= ELF::SHF_ALLOC;
337 else if (flagId == "execinstr")
338 flags |= ELF::SHF_EXECINSTR;
339 else if (flagId == "write")
340 flags |= ELF::SHF_WRITE;
341 else if (flagId == "tls")
342 flags |= ELF::SHF_TLS;
346 Lex(); // Eat the flag.
348 if (!getLexer().is(AsmToken::Comma))
350 Lex(); // Eat the comma.
356 bool ELFAsmParser::ParseDirectivePushSection(StringRef s, SMLoc loc) {
357 getStreamer().PushSection();
359 if (ParseSectionArguments(/*IsPush=*/true)) {
360 getStreamer().PopSection();
367 bool ELFAsmParser::ParseDirectivePopSection(StringRef, SMLoc) {
368 if (!getStreamer().PopSection())
369 return TokError(".popsection without corresponding .pushsection");
373 // FIXME: This is a work in progress.
374 bool ELFAsmParser::ParseDirectiveSection(StringRef, SMLoc) {
375 return ParseSectionArguments(/*IsPush=*/false);
378 bool ELFAsmParser::ParseSectionArguments(bool IsPush) {
379 StringRef SectionName;
381 if (ParseSectionName(SectionName))
382 return TokError("expected identifier in directive");
388 const MCExpr *Subsection = 0;
389 bool UseLastGroup = false;
391 // Set the defaults first.
392 if (SectionName == ".fini" || SectionName == ".init" ||
393 SectionName == ".rodata")
394 Flags |= ELF::SHF_ALLOC;
395 if (SectionName == ".fini" || SectionName == ".init")
396 Flags |= ELF::SHF_EXECINSTR;
398 if (getLexer().is(AsmToken::Comma)) {
401 if (IsPush && getLexer().isNot(AsmToken::String)) {
402 if (getParser().parseExpression(Subsection))
404 if (getLexer().isNot(AsmToken::Comma))
411 if (getLexer().isNot(AsmToken::String)) {
412 if (!getContext().getAsmInfo()->usesSunStyleELFSectionSwitchSyntax()
413 || getLexer().isNot(AsmToken::Hash))
414 return TokError("expected string in directive");
415 extraFlags = parseSunStyleSectionFlags();
417 StringRef FlagsStr = getTok().getStringContents();
419 extraFlags = parseSectionFlags(FlagsStr, &UseLastGroup);
422 if (extraFlags == -1U)
423 return TokError("unknown flag");
426 bool Mergeable = Flags & ELF::SHF_MERGE;
427 bool Group = Flags & ELF::SHF_GROUP;
428 if (Group && UseLastGroup)
429 return TokError("Section cannot specifiy a group name while also acting "
430 "as a member of the last group");
432 if (getLexer().isNot(AsmToken::Comma)) {
434 return TokError("Mergeable section must specify the type");
436 return TokError("Group section must specify the type");
439 if (getLexer().is(AsmToken::At) || getLexer().is(AsmToken::Percent) ||
440 getLexer().is(AsmToken::String)) {
441 if (!getLexer().is(AsmToken::String))
444 return TokError("expected '@<type>', '%<type>' or \"<type>\"");
446 if (getParser().parseIdentifier(TypeName))
447 return TokError("expected identifier in directive");
450 if (getLexer().isNot(AsmToken::Comma))
451 return TokError("expected the entry size");
453 if (getParser().parseAbsoluteExpression(Size))
456 return TokError("entry size must be positive");
460 if (getLexer().isNot(AsmToken::Comma))
461 return TokError("expected group name");
463 if (getParser().parseIdentifier(GroupName))
465 if (getLexer().is(AsmToken::Comma)) {
468 if (getParser().parseIdentifier(Linkage))
470 if (Linkage != "comdat")
471 return TokError("Linkage must be 'comdat'");
478 if (getLexer().isNot(AsmToken::EndOfStatement))
479 return TokError("unexpected token in directive");
481 unsigned Type = ELF::SHT_PROGBITS;
483 if (TypeName.empty()) {
484 if (SectionName.startswith(".note"))
485 Type = ELF::SHT_NOTE;
486 else if (SectionName == ".init_array")
487 Type = ELF::SHT_INIT_ARRAY;
488 else if (SectionName == ".fini_array")
489 Type = ELF::SHT_FINI_ARRAY;
490 else if (SectionName == ".preinit_array")
491 Type = ELF::SHT_PREINIT_ARRAY;
493 if (TypeName == "init_array")
494 Type = ELF::SHT_INIT_ARRAY;
495 else if (TypeName == "fini_array")
496 Type = ELF::SHT_FINI_ARRAY;
497 else if (TypeName == "preinit_array")
498 Type = ELF::SHT_PREINIT_ARRAY;
499 else if (TypeName == "nobits")
500 Type = ELF::SHT_NOBITS;
501 else if (TypeName == "progbits")
502 Type = ELF::SHT_PROGBITS;
503 else if (TypeName == "note")
504 Type = ELF::SHT_NOTE;
505 else if (TypeName == "unwind")
506 Type = ELF::SHT_X86_64_UNWIND;
508 return TokError("unknown section type");
512 MCSectionSubPair CurrentSection = getStreamer().getCurrentSection();
513 if (const MCSectionELF *Section =
514 cast_or_null<MCSectionELF>(CurrentSection.first))
515 if (const MCSymbol *Group = Section->getGroup()) {
516 GroupName = Group->getName();
517 Flags |= ELF::SHF_GROUP;
521 SectionKind Kind = computeSectionKind(Flags);
522 getStreamer().SwitchSection(getContext().getELFSection(SectionName, Type,
529 bool ELFAsmParser::ParseDirectivePrevious(StringRef DirName, SMLoc) {
530 MCSectionSubPair PreviousSection = getStreamer().getPreviousSection();
531 if (PreviousSection.first == NULL)
532 return TokError(".previous without corresponding .section");
533 getStreamer().SwitchSection(PreviousSection.first, PreviousSection.second);
538 /// ParseDirectiveELFType
539 /// ::= .type identifier , STT_<TYPE_IN_UPPER_CASE>
540 /// ::= .type identifier , #attribute
541 /// ::= .type identifier , @attribute
542 /// ::= .type identifier , %attribute
543 /// ::= .type identifier , "attribute"
544 bool ELFAsmParser::ParseDirectiveType(StringRef, SMLoc) {
546 if (getParser().parseIdentifier(Name))
547 return TokError("expected identifier in directive");
549 // Handle the identifier as the key symbol.
550 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
552 if (getLexer().isNot(AsmToken::Comma))
553 return TokError("unexpected token in '.type' directive");
559 if (getLexer().is(AsmToken::Identifier)) {
560 TypeLoc = getLexer().getLoc();
561 if (getParser().parseIdentifier(Type))
562 return TokError("expected symbol type in directive");
563 Attr = StringSwitch<MCSymbolAttr>(Type)
564 .Case("STT_FUNC", MCSA_ELF_TypeFunction)
565 .Case("STT_OBJECT", MCSA_ELF_TypeObject)
566 .Case("STT_TLS", MCSA_ELF_TypeTLS)
567 .Case("STT_COMMON", MCSA_ELF_TypeCommon)
568 .Case("STT_NOTYPE", MCSA_ELF_TypeNoType)
569 .Case("STT_GNU_IFUNC", MCSA_ELF_TypeIndFunction)
570 .Default(MCSA_Invalid);
571 } else if (getLexer().is(AsmToken::Hash) || getLexer().is(AsmToken::At) ||
572 getLexer().is(AsmToken::Percent) ||
573 getLexer().is(AsmToken::String)) {
574 if (!getLexer().is(AsmToken::String))
577 TypeLoc = getLexer().getLoc();
578 if (getParser().parseIdentifier(Type))
579 return TokError("expected symbol type in directive");
580 Attr = StringSwitch<MCSymbolAttr>(Type)
581 .Case("function", MCSA_ELF_TypeFunction)
582 .Case("object", MCSA_ELF_TypeObject)
583 .Case("tls_object", MCSA_ELF_TypeTLS)
584 .Case("common", MCSA_ELF_TypeCommon)
585 .Case("notype", MCSA_ELF_TypeNoType)
586 .Case("gnu_unique_object", MCSA_ELF_TypeGnuUniqueObject)
587 .Case("gnu_indirect_function", MCSA_ELF_TypeIndFunction)
588 .Default(MCSA_Invalid);
590 return TokError("expected STT_<TYPE_IN_UPPER_CASE>, '#<type>', '@<type>', "
591 "'%<type>' or \"<type>\"");
593 if (Attr == MCSA_Invalid)
594 return Error(TypeLoc, "unsupported attribute in '.type' directive");
596 if (getLexer().isNot(AsmToken::EndOfStatement))
597 return TokError("unexpected token in '.type' directive");
601 getStreamer().EmitSymbolAttribute(Sym, Attr);
606 /// ParseDirectiveIdent
607 /// ::= .ident string
608 bool ELFAsmParser::ParseDirectiveIdent(StringRef, SMLoc) {
609 if (getLexer().isNot(AsmToken::String))
610 return TokError("unexpected token in '.ident' directive");
612 StringRef Data = getTok().getIdentifier();
616 getStreamer().EmitIdent(Data);
620 /// ParseDirectiveSymver
621 /// ::= .symver foo, bar2@zed
622 bool ELFAsmParser::ParseDirectiveSymver(StringRef, SMLoc) {
624 if (getParser().parseIdentifier(Name))
625 return TokError("expected identifier in directive");
627 if (getLexer().isNot(AsmToken::Comma))
628 return TokError("expected a comma");
630 // ARM assembly uses @ for a comment...
631 // except when parsing the second parameter of the .symver directive.
632 // Force the next symbol to allow @ in the identifier, which is
633 // required for this directive and then reset it to its initial state.
634 const bool AllowAtInIdentifier = getLexer().getAllowAtInIdentifier();
635 getLexer().setAllowAtInIdentifier(true);
637 getLexer().setAllowAtInIdentifier(AllowAtInIdentifier);
640 if (getParser().parseIdentifier(AliasName))
641 return TokError("expected identifier in directive");
643 if (AliasName.find('@') == StringRef::npos)
644 return TokError("expected a '@' in the name");
646 MCSymbol *Alias = getContext().GetOrCreateSymbol(AliasName);
647 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
648 const MCExpr *Value = MCSymbolRefExpr::Create(Sym, getContext());
650 getStreamer().EmitAssignment(Alias, Value);
654 /// ParseDirectiveVersion
655 /// ::= .version string
656 bool ELFAsmParser::ParseDirectiveVersion(StringRef, SMLoc) {
657 if (getLexer().isNot(AsmToken::String))
658 return TokError("unexpected token in '.version' directive");
660 StringRef Data = getTok().getIdentifier();
664 const MCSection *Note =
665 getContext().getELFSection(".note", ELF::SHT_NOTE, 0,
666 SectionKind::getReadOnly());
668 getStreamer().PushSection();
669 getStreamer().SwitchSection(Note);
670 getStreamer().EmitIntValue(Data.size()+1, 4); // namesz.
671 getStreamer().EmitIntValue(0, 4); // descsz = 0 (no description).
672 getStreamer().EmitIntValue(1, 4); // type = NT_VERSION.
673 getStreamer().EmitBytes(Data); // name.
674 getStreamer().EmitIntValue(0, 1); // terminate the string.
675 getStreamer().EmitValueToAlignment(4); // ensure 4 byte alignment.
676 getStreamer().PopSection();
680 /// ParseDirectiveWeakref
681 /// ::= .weakref foo, bar
682 bool ELFAsmParser::ParseDirectiveWeakref(StringRef, SMLoc) {
683 // FIXME: Share code with the other alias building directives.
686 if (getParser().parseIdentifier(AliasName))
687 return TokError("expected identifier in directive");
689 if (getLexer().isNot(AsmToken::Comma))
690 return TokError("expected a comma");
695 if (getParser().parseIdentifier(Name))
696 return TokError("expected identifier in directive");
698 MCSymbol *Alias = getContext().GetOrCreateSymbol(AliasName);
700 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
702 getStreamer().EmitWeakReference(Alias, Sym);
706 bool ELFAsmParser::ParseDirectiveSubsection(StringRef, SMLoc) {
707 const MCExpr *Subsection = 0;
708 if (getLexer().isNot(AsmToken::EndOfStatement)) {
709 if (getParser().parseExpression(Subsection))
713 if (getLexer().isNot(AsmToken::EndOfStatement))
714 return TokError("unexpected token in directive");
716 getStreamer().SubSection(Subsection);
722 MCAsmParserExtension *createELFAsmParser() {
723 return new ELFAsmParser;