lib/MC/MCParser/ELFAsmParser.cpp

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