lib/MC/WasmObjectWriter.cpp

   1 //===- lib/MC/WasmObjectWriter.cpp - Wasm File Writer ---------------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 // This file implements Wasm object file writer information.
  10 //
  11 //===----------------------------------------------------------------------===//
  12
  13 #include "llvm/ADT/STLExtras.h"
  14 #include "llvm/ADT/SmallPtrSet.h"
  15 #include "llvm/BinaryFormat/Wasm.h"
  16 #include "llvm/Config/llvm-config.h"
  17 #include "llvm/MC/MCAsmBackend.h"
  18 #include "llvm/MC/MCAsmLayout.h"
  19 #include "llvm/MC/MCAssembler.h"
  20 #include "llvm/MC/MCContext.h"
  21 #include "llvm/MC/MCExpr.h"
  22 #include "llvm/MC/MCFixupKindInfo.h"
  23 #include "llvm/MC/MCObjectWriter.h"
  24 #include "llvm/MC/MCSectionWasm.h"
  25 #include "llvm/MC/MCSymbolWasm.h"
  26 #include "llvm/MC/MCValue.h"
  27 #include "llvm/MC/MCWasmObjectWriter.h"
  28 #include "llvm/Support/Casting.h"
  29 #include "llvm/Support/Debug.h"
  30 #include "llvm/Support/ErrorHandling.h"
  31 #include "llvm/Support/LEB128.h"
  32 #include "llvm/Support/StringSaver.h"
  33 #include <vector>
  34
  35 using namespace llvm;
  36
  37 #define DEBUG_TYPE "mc"
  38
  39 namespace {
  40
  41 // Went we ceate the indirect function table we start at 1, so that there is
  42 // and emtpy slot at 0 and therefore calling a null function pointer will trap.
  43 static const uint32_t InitialTableOffset = 1;
  44
  45 // For patching purposes, we need to remember where each section starts, both
  46 // for patching up the section size field, and for patching up references to
  47 // locations within the section.
  48 struct SectionBookkeeping {
  49   // Where the size of the section is written.
  50   uint64_t SizeOffset;
  51   // Where the section header ends (without custom section name).
  52   uint64_t PayloadOffset;
  53   // Where the contents of the section starts.
  54   uint64_t ContentsOffset;
  55   uint32_t Index;
  56 };
  57
  58 // The signature of a wasm function or event, in a struct capable of being used
  59 // as a DenseMap key.
  60 // TODO: Consider using wasm::WasmSignature directly instead.
  61 struct WasmSignature {
  62   // Support empty and tombstone instances, needed by DenseMap.
  63   enum { Plain, Empty, Tombstone } State = Plain;
  64
  65   // The return types of the function.
  66   SmallVector<wasm::ValType, 1> Returns;
  67
  68   // The parameter types of the function.
  69   SmallVector<wasm::ValType, 4> Params;
  70
  71   bool operator==(const WasmSignature &Other) const {
  72     return State == Other.State && Returns == Other.Returns &&
  73            Params == Other.Params;
  74   }
  75 };
  76
  77 // Traits for using WasmSignature in a DenseMap.
  78 struct WasmSignatureDenseMapInfo {
  79   static WasmSignature getEmptyKey() {
  80     WasmSignature Sig;
  81     Sig.State = WasmSignature::Empty;
  82     return Sig;
  83   }
  84   static WasmSignature getTombstoneKey() {
  85     WasmSignature Sig;
  86     Sig.State = WasmSignature::Tombstone;
  87     return Sig;
  88   }
  89   static unsigned getHashValue(const WasmSignature &Sig) {
  90     uintptr_t Value = Sig.State;
  91     for (wasm::ValType Ret : Sig.Returns)
  92       Value += DenseMapInfo<uint32_t>::getHashValue(uint32_t(Ret));
  93     for (wasm::ValType Param : Sig.Params)
  94       Value += DenseMapInfo<uint32_t>::getHashValue(uint32_t(Param));
  95     return Value;
  96   }
  97   static bool isEqual(const WasmSignature &LHS, const WasmSignature &RHS) {
  98     return LHS == RHS;
  99   }
 100 };
 101
 102 // A wasm data segment.  A wasm binary contains only a single data section
 103 // but that can contain many segments, each with their own virtual location
 104 // in memory.  Each MCSection data created by llvm is modeled as its own
 105 // wasm data segment.
 106 struct WasmDataSegment {
 107   MCSectionWasm *Section;
 108   StringRef Name;
 109   uint32_t Offset;
 110   uint32_t Alignment;
 111   uint32_t Flags;
 112   SmallVector<char, 4> Data;
 113 };
 114
 115 // A wasm function to be written into the function section.
 116 struct WasmFunction {
 117   uint32_t SigIndex;
 118   const MCSymbolWasm *Sym;
 119 };
 120
 121 // A wasm global to be written into the global section.
 122 struct WasmGlobal {
 123   wasm::WasmGlobalType Type;
 124   uint64_t InitialValue;
 125 };
 126
 127 // Information about a single item which is part of a COMDAT.  For each data
 128 // segment or function which is in the COMDAT, there is a corresponding
 129 // WasmComdatEntry.
 130 struct WasmComdatEntry {
 131   unsigned Kind;
 132   uint32_t Index;
 133 };
 134
 135 // Information about a single relocation.
 136 struct WasmRelocationEntry {
 137   uint64_t Offset;                   // Where is the relocation.
 138   const MCSymbolWasm *Symbol;        // The symbol to relocate with.
 139   int64_t Addend;                    // A value to add to the symbol.
 140   unsigned Type;                     // The type of the relocation.
 141   const MCSectionWasm *FixupSection; // The section the relocation is targeting.
 142
 143   WasmRelocationEntry(uint64_t Offset, const MCSymbolWasm *Symbol,
 144                       int64_t Addend, unsigned Type,
 145                       const MCSectionWasm *FixupSection)
 146       : Offset(Offset), Symbol(Symbol), Addend(Addend), Type(Type),
 147         FixupSection(FixupSection) {}
 148
 149   bool hasAddend() const {
 150     switch (Type) {
 151     case wasm::R_WASM_MEMORY_ADDR_LEB:
 152     case wasm::R_WASM_MEMORY_ADDR_SLEB:
 153     case wasm::R_WASM_MEMORY_ADDR_I32:
 154     case wasm::R_WASM_FUNCTION_OFFSET_I32:
 155     case wasm::R_WASM_SECTION_OFFSET_I32:
 156       return true;
 157     default:
 158       return false;
 159     }
 160   }
 161
 162   void print(raw_ostream &Out) const {
 163     Out << wasm::relocTypetoString(Type) << " Off=" << Offset
 164         << ", Sym=" << *Symbol << ", Addend=" << Addend
 165         << ", FixupSection=" << FixupSection->getSectionName();
 166   }
 167
 168 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
 169   LLVM_DUMP_METHOD void dump() const { print(dbgs()); }
 170 #endif
 171 };
 172
 173 static const uint32_t InvalidIndex = -1;
 174
 175 struct WasmCustomSection {
 176
 177   StringRef Name;
 178   MCSectionWasm *Section;
 179
 180   uint32_t OutputContentsOffset;
 181   uint32_t OutputIndex;
 182
 183   WasmCustomSection(StringRef Name, MCSectionWasm *Section)
 184       : Name(Name), Section(Section), OutputContentsOffset(0),
 185         OutputIndex(InvalidIndex) {}
 186 };
 187
 188 #if !defined(NDEBUG)
 189 raw_ostream &operator<<(raw_ostream &OS, const WasmRelocationEntry &Rel) {
 190   Rel.print(OS);
 191   return OS;
 192 }
 193 #endif
 194
 195 // Write X as an (unsigned) LEB value at offset Offset in Stream, padded
 196 // to allow patching.
 197 static void writePatchableLEB(raw_pwrite_stream &Stream, uint32_t X,
 198                               uint64_t Offset) {
 199   uint8_t Buffer[5];
 200   unsigned SizeLen = encodeULEB128(X, Buffer, 5);
 201   assert(SizeLen == 5);
 202   Stream.pwrite((char *)Buffer, SizeLen, Offset);
 203 }
 204
 205 // Write X as an signed LEB value at offset Offset in Stream, padded
 206 // to allow patching.
 207 static void writePatchableSLEB(raw_pwrite_stream &Stream, int32_t X,
 208                                uint64_t Offset) {
 209   uint8_t Buffer[5];
 210   unsigned SizeLen = encodeSLEB128(X, Buffer, 5);
 211   assert(SizeLen == 5);
 212   Stream.pwrite((char *)Buffer, SizeLen, Offset);
 213 }
 214
 215 // Write X as a plain integer value at offset Offset in Stream.
 216 static void writeI32(raw_pwrite_stream &Stream, uint32_t X, uint64_t Offset) {
 217   uint8_t Buffer[4];
 218   support::endian::write32le(Buffer, X);
 219   Stream.pwrite((char *)Buffer, sizeof(Buffer), Offset);
 220 }
 221
 222 class WasmObjectWriter : public MCObjectWriter {
 223   support::endian::Writer W;
 224
 225   /// The target specific Wasm writer instance.
 226   std::unique_ptr<MCWasmObjectTargetWriter> TargetObjectWriter;
 227
 228   // Relocations for fixing up references in the code section.
 229   std::vector<WasmRelocationEntry> CodeRelocations;
 230   uint32_t CodeSectionIndex;
 231
 232   // Relocations for fixing up references in the data section.
 233   std::vector<WasmRelocationEntry> DataRelocations;
 234   uint32_t DataSectionIndex;
 235
 236   // Index values to use for fixing up call_indirect type indices.
 237   // Maps function symbols to the index of the type of the function
 238   DenseMap<const MCSymbolWasm *, uint32_t> TypeIndices;
 239   // Maps function symbols to the table element index space. Used
 240   // for TABLE_INDEX relocation types (i.e. address taken functions).
 241   DenseMap<const MCSymbolWasm *, uint32_t> TableIndices;
 242   // Maps function/global symbols to the function/global/event/section index
 243   // space.
 244   DenseMap<const MCSymbolWasm *, uint32_t> WasmIndices;
 245   // Maps data symbols to the Wasm segment and offset/size with the segment.
 246   DenseMap<const MCSymbolWasm *, wasm::WasmDataReference> DataLocations;
 247
 248   // Stores output data (index, relocations, content offset) for custom
 249   // section.
 250   std::vector<WasmCustomSection> CustomSections;
 251   std::unique_ptr<WasmCustomSection> ProducersSection;
 252   // Relocations for fixing up references in the custom sections.
 253   DenseMap<const MCSectionWasm *, std::vector<WasmRelocationEntry>>
 254       CustomSectionsRelocations;
 255
 256   // Map from section to defining function symbol.
 257   DenseMap<const MCSection *, const MCSymbol *> SectionFunctions;
 258
 259   DenseMap<WasmSignature, uint32_t, WasmSignatureDenseMapInfo> SignatureIndices;
 260   SmallVector<WasmSignature, 4> Signatures;
 261   SmallVector<WasmGlobal, 4> Globals;
 262   SmallVector<WasmDataSegment, 4> DataSegments;
 263   unsigned NumFunctionImports = 0;
 264   unsigned NumGlobalImports = 0;
 265   unsigned NumEventImports = 0;
 266   uint32_t SectionCount = 0;
 267
 268   // TargetObjectWriter wrappers.
 269   bool is64Bit() const { return TargetObjectWriter->is64Bit(); }
 270   unsigned getRelocType(const MCValue &Target, const MCFixup &Fixup) const {
 271     return TargetObjectWriter->getRelocType(Target, Fixup);
 272   }
 273
 274   void startSection(SectionBookkeeping &Section, unsigned SectionId);
 275   void startCustomSection(SectionBookkeeping &Section, StringRef Name);
 276   void endSection(SectionBookkeeping &Section);
 277
 278 public:
 279   WasmObjectWriter(std::unique_ptr<MCWasmObjectTargetWriter> MOTW,
 280                    raw_pwrite_stream &OS)
 281       : W(OS, support::little), TargetObjectWriter(std::move(MOTW)) {}
 282
 283 private:
 284   void reset() override {
 285     CodeRelocations.clear();
 286     DataRelocations.clear();
 287     TypeIndices.clear();
 288     WasmIndices.clear();
 289     TableIndices.clear();
 290     DataLocations.clear();
 291     CustomSections.clear();
 292     ProducersSection.reset();
 293     CustomSectionsRelocations.clear();
 294     SignatureIndices.clear();
 295     Signatures.clear();
 296     Globals.clear();
 297     DataSegments.clear();
 298     SectionFunctions.clear();
 299     NumFunctionImports = 0;
 300     NumGlobalImports = 0;
 301     MCObjectWriter::reset();
 302   }
 303
 304   void writeHeader(const MCAssembler &Asm);
 305
 306   void recordRelocation(MCAssembler &Asm, const MCAsmLayout &Layout,
 307                         const MCFragment *Fragment, const MCFixup &Fixup,
 308                         MCValue Target, uint64_t &FixedValue) override;
 309
 310   void executePostLayoutBinding(MCAssembler &Asm,
 311                                 const MCAsmLayout &Layout) override;
 312
 313   uint64_t writeObject(MCAssembler &Asm, const MCAsmLayout &Layout) override;
 314
 315   void writeString(const StringRef Str) {
 316     encodeULEB128(Str.size(), W.OS);
 317     W.OS << Str;
 318   }
 319
 320   void writeValueType(wasm::ValType Ty) { W.OS << static_cast<char>(Ty); }
 321
 322   void writeTypeSection(ArrayRef<WasmSignature> Signatures);
 323   void writeImportSection(ArrayRef<wasm::WasmImport> Imports, uint32_t DataSize,
 324                           uint32_t NumElements);
 325   void writeFunctionSection(ArrayRef<WasmFunction> Functions);
 326   void writeGlobalSection();
 327   void writeExportSection(ArrayRef<wasm::WasmExport> Exports);
 328   void writeElemSection(ArrayRef<uint32_t> TableElems);
 329   void writeCodeSection(const MCAssembler &Asm, const MCAsmLayout &Layout,
 330                         ArrayRef<WasmFunction> Functions);
 331   void writeDataSection();
 332   void writeEventSection(ArrayRef<wasm::WasmEventType> Events);
 333   void writeRelocSection(uint32_t SectionIndex, StringRef Name,
 334                          std::vector<WasmRelocationEntry> &Relocations);
 335   void writeLinkingMetaDataSection(
 336       ArrayRef<wasm::WasmSymbolInfo> SymbolInfos,
 337       ArrayRef<std::pair<uint16_t, uint32_t>> InitFuncs,
 338       const std::map<StringRef, std::vector<WasmComdatEntry>> &Comdats);
 339   void writeCustomSection(WasmCustomSection &CustomSection,
 340                           const MCAssembler &Asm, const MCAsmLayout &Layout);
 341   void writeCustomRelocSections();
 342   void
 343   updateCustomSectionRelocations(const SmallVector<WasmFunction, 4> &Functions,
 344                                  const MCAsmLayout &Layout);
 345
 346   uint32_t getProvisionalValue(const WasmRelocationEntry &RelEntry);
 347   void applyRelocations(ArrayRef<WasmRelocationEntry> Relocations,
 348                         uint64_t ContentsOffset);
 349
 350   uint32_t getRelocationIndexValue(const WasmRelocationEntry &RelEntry);
 351   uint32_t getFunctionType(const MCSymbolWasm &Symbol);
 352   uint32_t getEventType(const MCSymbolWasm &Symbol);
 353   void registerFunctionType(const MCSymbolWasm &Symbol);
 354   void registerEventType(const MCSymbolWasm &Symbol);
 355 };
 356
 357 } // end anonymous namespace
 358
 359 // Write out a section header and a patchable section size field.
 360 void WasmObjectWriter::startSection(SectionBookkeeping &Section,
 361                                     unsigned SectionId) {
 362   LLVM_DEBUG(dbgs() << "startSection " << SectionId << "\n");
 363   W.OS << char(SectionId);
 364
 365   Section.SizeOffset = W.OS.tell();
 366
 367   // The section size. We don't know the size yet, so reserve enough space
 368   // for any 32-bit value; we'll patch it later.
 369   encodeULEB128(0, W.OS, 5);
 370
 371   // The position where the section starts, for measuring its size.
 372   Section.ContentsOffset = W.OS.tell();
 373   Section.PayloadOffset = W.OS.tell();
 374   Section.Index = SectionCount++;
 375 }
 376
 377 void WasmObjectWriter::startCustomSection(SectionBookkeeping &Section,
 378                                           StringRef Name) {
 379   LLVM_DEBUG(dbgs() << "startCustomSection " << Name << "\n");
 380   startSection(Section, wasm::WASM_SEC_CUSTOM);
 381
 382   // The position where the section header ends, for measuring its size.
 383   Section.PayloadOffset = W.OS.tell();
 384
 385   // Custom sections in wasm also have a string identifier.
 386   writeString(Name);
 387
 388   // The position where the custom section starts.
 389   Section.ContentsOffset = W.OS.tell();
 390 }
 391
 392 // Now that the section is complete and we know how big it is, patch up the
 393 // section size field at the start of the section.
 394 void WasmObjectWriter::endSection(SectionBookkeeping &Section) {
 395   uint64_t Size = W.OS.tell();
 396   // /dev/null doesn't support seek/tell and can report offset of 0.
 397   // Simply skip this patching in that case.
 398   if (!Size)
 399     return;
 400
 401   Size -= Section.PayloadOffset;
 402   if (uint32_t(Size) != Size)
 403     report_fatal_error("section size does not fit in a uint32_t");
 404
 405   LLVM_DEBUG(dbgs() << "endSection size=" << Size << "\n");
 406
 407   // Write the final section size to the payload_len field, which follows
 408   // the section id byte.
 409   writePatchableLEB(static_cast<raw_pwrite_stream &>(W.OS), Size,
 410                     Section.SizeOffset);
 411 }
 412
 413 // Emit the Wasm header.
 414 void WasmObjectWriter::writeHeader(const MCAssembler &Asm) {
 415   W.OS.write(wasm::WasmMagic, sizeof(wasm::WasmMagic));
 416   W.write<uint32_t>(wasm::WasmVersion);
 417 }
 418
 419 void WasmObjectWriter::executePostLayoutBinding(MCAssembler &Asm,
 420                                                 const MCAsmLayout &Layout) {
 421   // Build a map of sections to the function that defines them, for use
 422   // in recordRelocation.
 423   for (const MCSymbol &S : Asm.symbols()) {
 424     const auto &WS = static_cast<const MCSymbolWasm &>(S);
 425     if (WS.isDefined() && WS.isFunction() && !WS.isVariable()) {
 426       const auto &Sec = static_cast<const MCSectionWasm &>(S.getSection());
 427       auto Pair = SectionFunctions.insert(std::make_pair(&Sec, &S));
 428       if (!Pair.second)
 429         report_fatal_error("section already has a defining function: " +
 430                            Sec.getSectionName());
 431     }
 432   }
 433 }
 434
 435 void WasmObjectWriter::recordRelocation(MCAssembler &Asm,
 436                                         const MCAsmLayout &Layout,
 437                                         const MCFragment *Fragment,
 438                                         const MCFixup &Fixup, MCValue Target,
 439                                         uint64_t &FixedValue) {
 440   MCAsmBackend &Backend = Asm.getBackend();
 441   bool IsPCRel = Backend.getFixupKindInfo(Fixup.getKind()).Flags &
 442                  MCFixupKindInfo::FKF_IsPCRel;
 443   const auto &FixupSection = cast<MCSectionWasm>(*Fragment->getParent());
 444   uint64_t C = Target.getConstant();
 445   uint64_t FixupOffset = Layout.getFragmentOffset(Fragment) + Fixup.getOffset();
 446   MCContext &Ctx = Asm.getContext();
 447
 448   // The .init_array isn't translated as data, so don't do relocations in it.
 449   if (FixupSection.getSectionName().startswith(".init_array"))
 450     return;
 451
 452   if (const MCSymbolRefExpr *RefB = Target.getSymB()) {
 453     assert(RefB->getKind() == MCSymbolRefExpr::VK_None &&
 454            "Should not have constructed this");
 455
 456     // Let A, B and C being the components of Target and R be the location of
 457     // the fixup. If the fixup is not pcrel, we want to compute (A - B + C).
 458     // If it is pcrel, we want to compute (A - B + C - R).
 459
 460     // In general, Wasm has no relocations for -B. It can only represent (A + C)
 461     // or (A + C - R). If B = R + K and the relocation is not pcrel, we can
 462     // replace B to implement it: (A - R - K + C)
 463     if (IsPCRel) {
 464       Ctx.reportError(
 465           Fixup.getLoc(),
 466           "No relocation available to represent this relative expression");
 467       return;
 468     }
 469
 470     const auto &SymB = cast<MCSymbolWasm>(RefB->getSymbol());
 471
 472     if (SymB.isUndefined()) {
 473       Ctx.reportError(Fixup.getLoc(),
 474                       Twine("symbol '") + SymB.getName() +
 475                           "' can not be undefined in a subtraction expression");
 476       return;
 477     }
 478
 479     assert(!SymB.isAbsolute() && "Should have been folded");
 480     const MCSection &SecB = SymB.getSection();
 481     if (&SecB != &FixupSection) {
 482       Ctx.reportError(Fixup.getLoc(),
 483                       "Cannot represent a difference across sections");
 484       return;
 485     }
 486
 487     uint64_t SymBOffset = Layout.getSymbolOffset(SymB);
 488     uint64_t K = SymBOffset - FixupOffset;
 489     IsPCRel = true;
 490     C -= K;
 491   }
 492
 493   // We either rejected the fixup or folded B into C at this point.
 494   const MCSymbolRefExpr *RefA = Target.getSymA();
 495   const auto *SymA = RefA ? cast<MCSymbolWasm>(&RefA->getSymbol()) : nullptr;
 496
 497   if (SymA && SymA->isVariable()) {
 498     const MCExpr *Expr = SymA->getVariableValue();
 499     const auto *Inner = cast<MCSymbolRefExpr>(Expr);
 500     if (Inner->getKind() == MCSymbolRefExpr::VK_WEAKREF)
 501       llvm_unreachable("weakref used in reloc not yet implemented");
 502   }
 503
 504   // Put any constant offset in an addend. Offsets can be negative, and
 505   // LLVM expects wrapping, in contrast to wasm's immediates which can't
 506   // be negative and don't wrap.
 507   FixedValue = 0;
 508
 509   unsigned Type = getRelocType(Target, Fixup);
 510   assert(!IsPCRel);
 511   assert(SymA);
 512
 513   // Absolute offset within a section or a function.
 514   // Currently only supported for for metadata sections.
 515   // See: test/MC/WebAssembly/blockaddress.ll
 516   if (Type == wasm::R_WASM_FUNCTION_OFFSET_I32 ||
 517       Type == wasm::R_WASM_SECTION_OFFSET_I32) {
 518     if (!FixupSection.getKind().isMetadata())
 519       report_fatal_error("relocations for function or section offsets are "
 520                          "only supported in metadata sections");
 521
 522     const MCSymbol *SectionSymbol = nullptr;
 523     const MCSection &SecA = SymA->getSection();
 524     if (SecA.getKind().isText())
 525       SectionSymbol = SectionFunctions.find(&SecA)->second;
 526     else
 527       SectionSymbol = SecA.getBeginSymbol();
 528     if (!SectionSymbol)
 529       report_fatal_error("section symbol is required for relocation");
 530
 531     C += Layout.getSymbolOffset(*SymA);
 532     SymA = cast<MCSymbolWasm>(SectionSymbol);
 533   }
 534
 535   // Relocation other than R_WASM_TYPE_INDEX_LEB are required to be
 536   // against a named symbol.
 537   if (Type != wasm::R_WASM_TYPE_INDEX_LEB) {
 538     if (SymA->getName().empty())
 539       report_fatal_error("relocations against un-named temporaries are not yet "
 540                          "supported by wasm");
 541
 542     SymA->setUsedInReloc();
 543   }
 544
 545   WasmRelocationEntry Rec(FixupOffset, SymA, C, Type, &FixupSection);
 546   LLVM_DEBUG(dbgs() << "WasmReloc: " << Rec << "\n");
 547
 548   if (FixupSection.isWasmData()) {
 549     DataRelocations.push_back(Rec);
 550   } else if (FixupSection.getKind().isText()) {
 551     CodeRelocations.push_back(Rec);
 552   } else if (FixupSection.getKind().isMetadata()) {
 553     CustomSectionsRelocations[&FixupSection].push_back(Rec);
 554   } else {
 555     llvm_unreachable("unexpected section type");
 556   }
 557 }
 558
 559 static const MCSymbolWasm *resolveSymbol(const MCSymbolWasm &Symbol) {
 560   if (Symbol.isVariable()) {
 561     const MCExpr *Expr = Symbol.getVariableValue();
 562     auto *Inner = cast<MCSymbolRefExpr>(Expr);
 563     return cast<MCSymbolWasm>(&Inner->getSymbol());
 564   }
 565   return &Symbol;
 566 }
 567
 568 // Compute a value to write into the code at the location covered
 569 // by RelEntry. This value isn't used by the static linker; it just serves
 570 // to make the object format more readable and more likely to be directly
 571 // useable.
 572 uint32_t
 573 WasmObjectWriter::getProvisionalValue(const WasmRelocationEntry &RelEntry) {
 574   switch (RelEntry.Type) {
 575   case wasm::R_WASM_TABLE_INDEX_SLEB:
 576   case wasm::R_WASM_TABLE_INDEX_I32: {
 577     // Provisional value is table address of the resolved symbol itself
 578     const MCSymbolWasm *Sym = resolveSymbol(*RelEntry.Symbol);
 579     assert(Sym->isFunction());
 580     return TableIndices[Sym];
 581   }
 582   case wasm::R_WASM_TYPE_INDEX_LEB:
 583     // Provisional value is same as the index
 584     return getRelocationIndexValue(RelEntry);
 585   case wasm::R_WASM_FUNCTION_INDEX_LEB:
 586   case wasm::R_WASM_GLOBAL_INDEX_LEB:
 587   case wasm::R_WASM_EVENT_INDEX_LEB:
 588     // Provisional value is function/global/event Wasm index
 589     if (!WasmIndices.count(RelEntry.Symbol))
 590       report_fatal_error("symbol not found in wasm index space: " +
 591                          RelEntry.Symbol->getName());
 592     return WasmIndices[RelEntry.Symbol];
 593   case wasm::R_WASM_FUNCTION_OFFSET_I32:
 594   case wasm::R_WASM_SECTION_OFFSET_I32: {
 595     const auto &Section =
 596         static_cast<const MCSectionWasm &>(RelEntry.Symbol->getSection());
 597     return Section.getSectionOffset() + RelEntry.Addend;
 598   }
 599   case wasm::R_WASM_MEMORY_ADDR_LEB:
 600   case wasm::R_WASM_MEMORY_ADDR_I32:
 601   case wasm::R_WASM_MEMORY_ADDR_SLEB: {
 602     // Provisional value is address of the global
 603     const MCSymbolWasm *Sym = resolveSymbol(*RelEntry.Symbol);
 604     // For undefined symbols, use zero
 605     if (!Sym->isDefined())
 606       return 0;
 607     const wasm::WasmDataReference &Ref = DataLocations[Sym];
 608     const WasmDataSegment &Segment = DataSegments[Ref.Segment];
 609     // Ignore overflow. LLVM allows address arithmetic to silently wrap.
 610     return Segment.Offset + Ref.Offset + RelEntry.Addend;
 611   }
 612   default:
 613     llvm_unreachable("invalid relocation type");
 614   }
 615 }
 616
 617 static void addData(SmallVectorImpl<char> &DataBytes,
 618                     MCSectionWasm &DataSection) {
 619   LLVM_DEBUG(errs() << "addData: " << DataSection.getSectionName() << "\n");
 620
 621   DataBytes.resize(alignTo(DataBytes.size(), DataSection.getAlignment()));
 622
 623   for (const MCFragment &Frag : DataSection) {
 624     if (Frag.hasInstructions())
 625       report_fatal_error("only data supported in data sections");
 626
 627     if (auto *Align = dyn_cast<MCAlignFragment>(&Frag)) {
 628       if (Align->getValueSize() != 1)
 629         report_fatal_error("only byte values supported for alignment");
 630       // If nops are requested, use zeros, as this is the data section.
 631       uint8_t Value = Align->hasEmitNops() ? 0 : Align->getValue();
 632       uint64_t Size =
 633           std::min<uint64_t>(alignTo(DataBytes.size(), Align->getAlignment()),
 634                              DataBytes.size() + Align->getMaxBytesToEmit());
 635       DataBytes.resize(Size, Value);
 636     } else if (auto *Fill = dyn_cast<MCFillFragment>(&Frag)) {
 637       int64_t NumValues;
 638       if (!Fill->getNumValues().evaluateAsAbsolute(NumValues))
 639         llvm_unreachable("The fill should be an assembler constant");
 640       DataBytes.insert(DataBytes.end(), Fill->getValueSize() * NumValues,
 641                        Fill->getValue());
 642     } else if (auto *LEB = dyn_cast<MCLEBFragment>(&Frag)) {
 643       const SmallVectorImpl<char> &Contents = LEB->getContents();
 644       DataBytes.insert(DataBytes.end(), Contents.begin(), Contents.end());
 645     } else {
 646       const auto &DataFrag = cast<MCDataFragment>(Frag);
 647       const SmallVectorImpl<char> &Contents = DataFrag.getContents();
 648       DataBytes.insert(DataBytes.end(), Contents.begin(), Contents.end());
 649     }
 650   }
 651
 652   LLVM_DEBUG(dbgs() << "addData -> " << DataBytes.size() << "\n");
 653 }
 654
 655 uint32_t
 656 WasmObjectWriter::getRelocationIndexValue(const WasmRelocationEntry &RelEntry) {
 657   if (RelEntry.Type == wasm::R_WASM_TYPE_INDEX_LEB) {
 658     if (!TypeIndices.count(RelEntry.Symbol))
 659       report_fatal_error("symbol not found in type index space: " +
 660                          RelEntry.Symbol->getName());
 661     return TypeIndices[RelEntry.Symbol];
 662   }
 663
 664   return RelEntry.Symbol->getIndex();
 665 }
 666
 667 // Apply the portions of the relocation records that we can handle ourselves
 668 // directly.
 669 void WasmObjectWriter::applyRelocations(
 670     ArrayRef<WasmRelocationEntry> Relocations, uint64_t ContentsOffset) {
 671   auto &Stream = static_cast<raw_pwrite_stream &>(W.OS);
 672   for (const WasmRelocationEntry &RelEntry : Relocations) {
 673     uint64_t Offset = ContentsOffset +
 674                       RelEntry.FixupSection->getSectionOffset() +
 675                       RelEntry.Offset;
 676
 677     LLVM_DEBUG(dbgs() << "applyRelocation: " << RelEntry << "\n");
 678     uint32_t Value = getProvisionalValue(RelEntry);
 679
 680     switch (RelEntry.Type) {
 681     case wasm::R_WASM_FUNCTION_INDEX_LEB:
 682     case wasm::R_WASM_TYPE_INDEX_LEB:
 683     case wasm::R_WASM_GLOBAL_INDEX_LEB:
 684     case wasm::R_WASM_MEMORY_ADDR_LEB:
 685     case wasm::R_WASM_EVENT_INDEX_LEB:
 686       writePatchableLEB(Stream, Value, Offset);
 687       break;
 688     case wasm::R_WASM_TABLE_INDEX_I32:
 689     case wasm::R_WASM_MEMORY_ADDR_I32:
 690     case wasm::R_WASM_FUNCTION_OFFSET_I32:
 691     case wasm::R_WASM_SECTION_OFFSET_I32:
 692       writeI32(Stream, Value, Offset);
 693       break;
 694     case wasm::R_WASM_TABLE_INDEX_SLEB:
 695     case wasm::R_WASM_MEMORY_ADDR_SLEB:
 696       writePatchableSLEB(Stream, Value, Offset);
 697       break;
 698     default:
 699       llvm_unreachable("invalid relocation type");
 700     }
 701   }
 702 }
 703
 704 void WasmObjectWriter::writeTypeSection(ArrayRef<WasmSignature> Signatures) {
 705   if (Signatures.empty())
 706     return;
 707
 708   SectionBookkeeping Section;
 709   startSection(Section, wasm::WASM_SEC_TYPE);
 710
 711   encodeULEB128(Signatures.size(), W.OS);
 712
 713   for (const WasmSignature &Sig : Signatures) {
 714     W.OS << char(wasm::WASM_TYPE_FUNC);
 715     encodeULEB128(Sig.Params.size(), W.OS);
 716     for (wasm::ValType Ty : Sig.Params)
 717       writeValueType(Ty);
 718     encodeULEB128(Sig.Returns.size(), W.OS);
 719     for (wasm::ValType Ty : Sig.Returns)
 720       writeValueType(Ty);
 721   }
 722
 723   endSection(Section);
 724 }
 725
 726 void WasmObjectWriter::writeImportSection(ArrayRef<wasm::WasmImport> Imports,
 727                                           uint32_t DataSize,
 728                                           uint32_t NumElements) {
 729   if (Imports.empty())
 730     return;
 731
 732   uint32_t NumPages = (DataSize + wasm::WasmPageSize - 1) / wasm::WasmPageSize;
 733
 734   SectionBookkeeping Section;
 735   startSection(Section, wasm::WASM_SEC_IMPORT);
 736
 737   encodeULEB128(Imports.size(), W.OS);
 738   for (const wasm::WasmImport &Import : Imports) {
 739     writeString(Import.Module);
 740     writeString(Import.Field);
 741     W.OS << char(Import.Kind);
 742
 743     switch (Import.Kind) {
 744     case wasm::WASM_EXTERNAL_FUNCTION:
 745       encodeULEB128(Import.SigIndex, W.OS);
 746       break;
 747     case wasm::WASM_EXTERNAL_GLOBAL:
 748       W.OS << char(Import.Global.Type);
 749       W.OS << char(Import.Global.Mutable ? 1 : 0);
 750       break;
 751     case wasm::WASM_EXTERNAL_MEMORY:
 752       encodeULEB128(0, W.OS);        // flags
 753       encodeULEB128(NumPages, W.OS); // initial
 754       break;
 755     case wasm::WASM_EXTERNAL_TABLE:
 756       W.OS << char(Import.Table.ElemType);
 757       encodeULEB128(0, W.OS);           // flags
 758       encodeULEB128(NumElements, W.OS); // initial
 759       break;
 760     case wasm::WASM_EXTERNAL_EVENT:
 761       encodeULEB128(Import.Event.Attribute, W.OS);
 762       encodeULEB128(Import.Event.SigIndex, W.OS);
 763       break;
 764     default:
 765       llvm_unreachable("unsupported import kind");
 766     }
 767   }
 768
 769   endSection(Section);
 770 }
 771
 772 void WasmObjectWriter::writeFunctionSection(ArrayRef<WasmFunction> Functions) {
 773   if (Functions.empty())
 774     return;
 775
 776   SectionBookkeeping Section;
 777   startSection(Section, wasm::WASM_SEC_FUNCTION);
 778
 779   encodeULEB128(Functions.size(), W.OS);
 780   for (const WasmFunction &Func : Functions)
 781     encodeULEB128(Func.SigIndex, W.OS);
 782
 783   endSection(Section);
 784 }
 785
 786 void WasmObjectWriter::writeGlobalSection() {
 787   if (Globals.empty())
 788     return;
 789
 790   SectionBookkeeping Section;
 791   startSection(Section, wasm::WASM_SEC_GLOBAL);
 792
 793   encodeULEB128(Globals.size(), W.OS);
 794   for (const WasmGlobal &Global : Globals) {
 795     writeValueType(static_cast<wasm::ValType>(Global.Type.Type));
 796     W.OS << char(Global.Type.Mutable);
 797
 798     W.OS << char(wasm::WASM_OPCODE_I32_CONST);
 799     encodeSLEB128(Global.InitialValue, W.OS);
 800     W.OS << char(wasm::WASM_OPCODE_END);
 801   }
 802
 803   endSection(Section);
 804 }
 805
 806 void WasmObjectWriter::writeEventSection(ArrayRef<wasm::WasmEventType> Events) {
 807   if (Events.empty())
 808     return;
 809
 810   SectionBookkeeping Section;
 811   startSection(Section, wasm::WASM_SEC_EVENT);
 812
 813   encodeULEB128(Events.size(), W.OS);
 814   for (const wasm::WasmEventType &Event : Events) {
 815     encodeULEB128(Event.Attribute, W.OS);
 816     encodeULEB128(Event.SigIndex, W.OS);
 817   }
 818
 819   endSection(Section);
 820 }
 821
 822 void WasmObjectWriter::writeExportSection(ArrayRef<wasm::WasmExport> Exports) {
 823   if (Exports.empty())
 824     return;
 825
 826   SectionBookkeeping Section;
 827   startSection(Section, wasm::WASM_SEC_EXPORT);
 828
 829   encodeULEB128(Exports.size(), W.OS);
 830   for (const wasm::WasmExport &Export : Exports) {
 831     writeString(Export.Name);
 832     W.OS << char(Export.Kind);
 833     encodeULEB128(Export.Index, W.OS);
 834   }
 835
 836   endSection(Section);
 837 }
 838
 839 void WasmObjectWriter::writeElemSection(ArrayRef<uint32_t> TableElems) {
 840   if (TableElems.empty())
 841     return;
 842
 843   SectionBookkeeping Section;
 844   startSection(Section, wasm::WASM_SEC_ELEM);
 845
 846   encodeULEB128(1, W.OS); // number of "segments"
 847   encodeULEB128(0, W.OS); // the table index
 848
 849   // init expr for starting offset
 850   W.OS << char(wasm::WASM_OPCODE_I32_CONST);
 851   encodeSLEB128(InitialTableOffset, W.OS);
 852   W.OS << char(wasm::WASM_OPCODE_END);
 853
 854   encodeULEB128(TableElems.size(), W.OS);
 855   for (uint32_t Elem : TableElems)
 856     encodeULEB128(Elem, W.OS);
 857
 858   endSection(Section);
 859 }
 860
 861 void WasmObjectWriter::writeCodeSection(const MCAssembler &Asm,
 862                                         const MCAsmLayout &Layout,
 863                                         ArrayRef<WasmFunction> Functions) {
 864   if (Functions.empty())
 865     return;
 866
 867   SectionBookkeeping Section;
 868   startSection(Section, wasm::WASM_SEC_CODE);
 869   CodeSectionIndex = Section.Index;
 870
 871   encodeULEB128(Functions.size(), W.OS);
 872
 873   for (const WasmFunction &Func : Functions) {
 874     auto &FuncSection = static_cast<MCSectionWasm &>(Func.Sym->getSection());
 875
 876     int64_t Size = 0;
 877     if (!Func.Sym->getSize()->evaluateAsAbsolute(Size, Layout))
 878       report_fatal_error(".size expression must be evaluatable");
 879
 880     encodeULEB128(Size, W.OS);
 881     FuncSection.setSectionOffset(W.OS.tell() - Section.ContentsOffset);
 882     Asm.writeSectionData(W.OS, &FuncSection, Layout);
 883   }
 884
 885   // Apply fixups.
 886   applyRelocations(CodeRelocations, Section.ContentsOffset);
 887
 888   endSection(Section);
 889 }
 890
 891 void WasmObjectWriter::writeDataSection() {
 892   if (DataSegments.empty())
 893     return;
 894
 895   SectionBookkeeping Section;
 896   startSection(Section, wasm::WASM_SEC_DATA);
 897   DataSectionIndex = Section.Index;
 898
 899   encodeULEB128(DataSegments.size(), W.OS); // count
 900
 901   for (const WasmDataSegment &Segment : DataSegments) {
 902     encodeULEB128(0, W.OS); // memory index
 903     W.OS << char(wasm::WASM_OPCODE_I32_CONST);
 904     encodeSLEB128(Segment.Offset, W.OS); // offset
 905     W.OS << char(wasm::WASM_OPCODE_END);
 906     encodeULEB128(Segment.Data.size(), W.OS); // size
 907     Segment.Section->setSectionOffset(W.OS.tell() - Section.ContentsOffset);
 908     W.OS << Segment.Data; // data
 909   }
 910
 911   // Apply fixups.
 912   applyRelocations(DataRelocations, Section.ContentsOffset);
 913
 914   endSection(Section);
 915 }
 916
 917 void WasmObjectWriter::writeRelocSection(
 918     uint32_t SectionIndex, StringRef Name,
 919     std::vector<WasmRelocationEntry> &Relocs) {
 920   // See: https://github.com/WebAssembly/tool-conventions/blob/master/Linking.md
 921   // for descriptions of the reloc sections.
 922
 923   if (Relocs.empty())
 924     return;
 925
 926   // First, ensure the relocations are sorted in offset order.  In general they
 927   // should already be sorted since `recordRelocation` is called in offset
 928   // order, but for the code section we combine many MC sections into single
 929   // wasm section, and this order is determined by the order of Asm.Symbols()
 930   // not the sections order.
 931   std::stable_sort(
 932       Relocs.begin(), Relocs.end(),
 933       [](const WasmRelocationEntry &A, const WasmRelocationEntry &B) {
 934         return (A.Offset + A.FixupSection->getSectionOffset()) <
 935                (B.Offset + B.FixupSection->getSectionOffset());
 936       });
 937
 938   SectionBookkeeping Section;
 939   startCustomSection(Section, std::string("reloc.") + Name.str());
 940
 941   encodeULEB128(SectionIndex, W.OS);
 942   encodeULEB128(Relocs.size(), W.OS);
 943   for (const WasmRelocationEntry &RelEntry : Relocs) {
 944     uint64_t Offset =
 945         RelEntry.Offset + RelEntry.FixupSection->getSectionOffset();
 946     uint32_t Index = getRelocationIndexValue(RelEntry);
 947
 948     W.OS << char(RelEntry.Type);
 949     encodeULEB128(Offset, W.OS);
 950     encodeULEB128(Index, W.OS);
 951     if (RelEntry.hasAddend())
 952       encodeSLEB128(RelEntry.Addend, W.OS);
 953   }
 954
 955   endSection(Section);
 956 }
 957
 958 void WasmObjectWriter::writeCustomRelocSections() {
 959   for (const auto &Sec : CustomSections) {
 960     auto &Relocations = CustomSectionsRelocations[Sec.Section];
 961     writeRelocSection(Sec.OutputIndex, Sec.Name, Relocations);
 962   }
 963 }
 964
 965 void WasmObjectWriter::writeLinkingMetaDataSection(
 966     ArrayRef<wasm::WasmSymbolInfo> SymbolInfos,
 967     ArrayRef<std::pair<uint16_t, uint32_t>> InitFuncs,
 968     const std::map<StringRef, std::vector<WasmComdatEntry>> &Comdats) {
 969   SectionBookkeeping Section;
 970   startCustomSection(Section, "linking");
 971   encodeULEB128(wasm::WasmMetadataVersion, W.OS);
 972
 973   SectionBookkeeping SubSection;
 974   if (SymbolInfos.size() != 0) {
 975     startSection(SubSection, wasm::WASM_SYMBOL_TABLE);
 976     encodeULEB128(SymbolInfos.size(), W.OS);
 977     for (const wasm::WasmSymbolInfo &Sym : SymbolInfos) {
 978       encodeULEB128(Sym.Kind, W.OS);
 979       encodeULEB128(Sym.Flags, W.OS);
 980       switch (Sym.Kind) {
 981       case wasm::WASM_SYMBOL_TYPE_FUNCTION:
 982       case wasm::WASM_SYMBOL_TYPE_GLOBAL:
 983       case wasm::WASM_SYMBOL_TYPE_EVENT:
 984         encodeULEB128(Sym.ElementIndex, W.OS);
 985         if ((Sym.Flags & wasm::WASM_SYMBOL_UNDEFINED) == 0 ||
 986             (Sym.Flags & wasm::WASM_SYMBOL_EXPLICIT_NAME) != 0)
 987           writeString(Sym.Name);
 988         break;
 989       case wasm::WASM_SYMBOL_TYPE_DATA:
 990         writeString(Sym.Name);
 991         if ((Sym.Flags & wasm::WASM_SYMBOL_UNDEFINED) == 0) {
 992           encodeULEB128(Sym.DataRef.Segment, W.OS);
 993           encodeULEB128(Sym.DataRef.Offset, W.OS);
 994           encodeULEB128(Sym.DataRef.Size, W.OS);
 995         }
 996         break;
 997       case wasm::WASM_SYMBOL_TYPE_SECTION: {
 998         const uint32_t SectionIndex =
 999             CustomSections[Sym.ElementIndex].OutputIndex;
1000         encodeULEB128(SectionIndex, W.OS);
1001         break;
1002       }
1003       default:
1004         llvm_unreachable("unexpected kind");
1005       }
1006     }
1007     endSection(SubSection);
1008   }
1009
1010   if (DataSegments.size()) {
1011     startSection(SubSection, wasm::WASM_SEGMENT_INFO);
1012     encodeULEB128(DataSegments.size(), W.OS);
1013     for (const WasmDataSegment &Segment : DataSegments) {
1014       writeString(Segment.Name);
1015       encodeULEB128(Segment.Alignment, W.OS);
1016       encodeULEB128(Segment.Flags, W.OS);
1017     }
1018     endSection(SubSection);
1019   }
1020
1021   if (!InitFuncs.empty()) {
1022     startSection(SubSection, wasm::WASM_INIT_FUNCS);
1023     encodeULEB128(InitFuncs.size(), W.OS);
1024     for (auto &StartFunc : InitFuncs) {
1025       encodeULEB128(StartFunc.first, W.OS);  // priority
1026       encodeULEB128(StartFunc.second, W.OS); // function index
1027     }
1028     endSection(SubSection);
1029   }
1030
1031   if (Comdats.size()) {
1032     startSection(SubSection, wasm::WASM_COMDAT_INFO);
1033     encodeULEB128(Comdats.size(), W.OS);
1034     for (const auto &C : Comdats) {
1035       writeString(C.first);
1036       encodeULEB128(0, W.OS); // flags for future use
1037       encodeULEB128(C.second.size(), W.OS);
1038       for (const WasmComdatEntry &Entry : C.second) {
1039         encodeULEB128(Entry.Kind, W.OS);
1040         encodeULEB128(Entry.Index, W.OS);
1041       }
1042     }
1043     endSection(SubSection);
1044   }
1045
1046   endSection(Section);
1047 }
1048
1049 void WasmObjectWriter::writeCustomSection(WasmCustomSection &CustomSection,
1050                                           const MCAssembler &Asm,
1051                                           const MCAsmLayout &Layout) {
1052   SectionBookkeeping Section;
1053   auto *Sec = CustomSection.Section;
1054   startCustomSection(Section, CustomSection.Name);
1055
1056   Sec->setSectionOffset(W.OS.tell() - Section.ContentsOffset);
1057   Asm.writeSectionData(W.OS, Sec, Layout);
1058
1059   CustomSection.OutputContentsOffset = Section.ContentsOffset;
1060   CustomSection.OutputIndex = Section.Index;
1061
1062   endSection(Section);
1063
1064   // Apply fixups.
1065   auto &Relocations = CustomSectionsRelocations[CustomSection.Section];
1066   applyRelocations(Relocations, CustomSection.OutputContentsOffset);
1067 }
1068
1069 uint32_t WasmObjectWriter::getFunctionType(const MCSymbolWasm &Symbol) {
1070   assert(Symbol.isFunction());
1071   assert(TypeIndices.count(&Symbol));
1072   return TypeIndices[&Symbol];
1073 }
1074
1075 uint32_t WasmObjectWriter::getEventType(const MCSymbolWasm &Symbol) {
1076   assert(Symbol.isEvent());
1077   assert(TypeIndices.count(&Symbol));
1078   return TypeIndices[&Symbol];
1079 }
1080
1081 void WasmObjectWriter::registerFunctionType(const MCSymbolWasm &Symbol) {
1082   assert(Symbol.isFunction());
1083
1084   WasmSignature S;
1085   const MCSymbolWasm *ResolvedSym = resolveSymbol(Symbol);
1086   if (auto *Sig = ResolvedSym->getSignature()) {
1087     S.Returns = Sig->Returns;
1088     S.Params = Sig->Params;
1089   }
1090
1091   auto Pair = SignatureIndices.insert(std::make_pair(S, Signatures.size()));
1092   if (Pair.second)
1093     Signatures.push_back(S);
1094   TypeIndices[&Symbol] = Pair.first->second;
1095
1096   LLVM_DEBUG(dbgs() << "registerFunctionType: " << Symbol
1097                     << " new:" << Pair.second << "\n");
1098   LLVM_DEBUG(dbgs() << "  -> type index: " << Pair.first->second << "\n");
1099 }
1100
1101 void WasmObjectWriter::registerEventType(const MCSymbolWasm &Symbol) {
1102   assert(Symbol.isEvent());
1103
1104   // TODO Currently we don't generate imported exceptions, but if we do, we
1105   // should have a way of infering types of imported exceptions.
1106   WasmSignature S;
1107   if (auto *Sig = Symbol.getSignature()) {
1108     S.Returns = Sig->Returns;
1109     S.Params = Sig->Params;
1110   }
1111
1112   auto Pair = SignatureIndices.insert(std::make_pair(S, Signatures.size()));
1113   if (Pair.second)
1114     Signatures.push_back(S);
1115   TypeIndices[&Symbol] = Pair.first->second;
1116
1117   LLVM_DEBUG(dbgs() << "registerEventType: " << Symbol << " new:" << Pair.second
1118                     << "\n");
1119   LLVM_DEBUG(dbgs() << "  -> type index: " << Pair.first->second << "\n");
1120 }
1121
1122 static bool isInSymtab(const MCSymbolWasm &Sym) {
1123   if (Sym.isUsedInReloc())
1124     return true;
1125
1126   if (Sym.isComdat() && !Sym.isDefined())
1127     return false;
1128
1129   if (Sym.isTemporary() && Sym.getName().empty())
1130     return false;
1131
1132   if (Sym.isTemporary() && Sym.isData() && !Sym.getSize())
1133     return false;
1134
1135   if (Sym.isSection())
1136     return false;
1137
1138   return true;
1139 }
1140
1141 uint64_t WasmObjectWriter::writeObject(MCAssembler &Asm,
1142                                        const MCAsmLayout &Layout) {
1143   uint64_t StartOffset = W.OS.tell();
1144
1145   LLVM_DEBUG(dbgs() << "WasmObjectWriter::writeObject\n");
1146   MCContext &Ctx = Asm.getContext();
1147
1148   // Collect information from the available symbols.
1149   SmallVector<WasmFunction, 4> Functions;
1150   SmallVector<uint32_t, 4> TableElems;
1151   SmallVector<wasm::WasmImport, 4> Imports;
1152   SmallVector<wasm::WasmExport, 4> Exports;
1153   SmallVector<wasm::WasmEventType, 1> Events;
1154   SmallVector<wasm::WasmSymbolInfo, 4> SymbolInfos;
1155   SmallVector<std::pair<uint16_t, uint32_t>, 2> InitFuncs;
1156   std::map<StringRef, std::vector<WasmComdatEntry>> Comdats;
1157   uint32_t DataSize = 0;
1158
1159   // For now, always emit the memory import, since loads and stores are not
1160   // valid without it. In the future, we could perhaps be more clever and omit
1161   // it if there are no loads or stores.
1162   auto *MemorySym =
1163       cast<MCSymbolWasm>(Ctx.getOrCreateSymbol("__linear_memory"));
1164   wasm::WasmImport MemImport;
1165   MemImport.Module = MemorySym->getImportModule();
1166   MemImport.Field = MemorySym->getImportName();
1167   MemImport.Kind = wasm::WASM_EXTERNAL_MEMORY;
1168   Imports.push_back(MemImport);
1169
1170   // For now, always emit the table section, since indirect calls are not
1171   // valid without it. In the future, we could perhaps be more clever and omit
1172   // it if there are no indirect calls.
1173   auto *TableSym =
1174       cast<MCSymbolWasm>(Ctx.getOrCreateSymbol("__indirect_function_table"));
1175   wasm::WasmImport TableImport;
1176   TableImport.Module = TableSym->getImportModule();
1177   TableImport.Field = TableSym->getImportName();
1178   TableImport.Kind = wasm::WASM_EXTERNAL_TABLE;
1179   TableImport.Table.ElemType = wasm::WASM_TYPE_FUNCREF;
1180   Imports.push_back(TableImport);
1181
1182   // Populate SignatureIndices, and Imports and WasmIndices for undefined
1183   // symbols.  This must be done before populating WasmIndices for defined
1184   // symbols.
1185   for (const MCSymbol &S : Asm.symbols()) {
1186     const auto &WS = static_cast<const MCSymbolWasm &>(S);
1187
1188     // Register types for all functions, including those with private linkage
1189     // (because wasm always needs a type signature).
1190     if (WS.isFunction())
1191       registerFunctionType(WS);
1192
1193     if (WS.isEvent())
1194       registerEventType(WS);
1195
1196     if (WS.isTemporary())
1197       continue;
1198
1199     // If the symbol is not defined in this translation unit, import it.
1200     if (!WS.isDefined() && !WS.isComdat()) {
1201       if (WS.isFunction()) {
1202         wasm::WasmImport Import;
1203         Import.Module = WS.getImportModule();
1204         Import.Field = WS.getImportName();
1205         Import.Kind = wasm::WASM_EXTERNAL_FUNCTION;
1206         Import.SigIndex = getFunctionType(WS);
1207         Imports.push_back(Import);
1208         WasmIndices[&WS] = NumFunctionImports++;
1209       } else if (WS.isGlobal()) {
1210         if (WS.isWeak())
1211           report_fatal_error("undefined global symbol cannot be weak");
1212
1213         wasm::WasmImport Import;
1214         Import.Module = WS.getImportModule();
1215         Import.Field = WS.getImportName();
1216         Import.Kind = wasm::WASM_EXTERNAL_GLOBAL;
1217         Import.Global = WS.getGlobalType();
1218         Imports.push_back(Import);
1219         WasmIndices[&WS] = NumGlobalImports++;
1220       } else if (WS.isEvent()) {
1221         if (WS.isWeak())
1222           report_fatal_error("undefined event symbol cannot be weak");
1223
1224         wasm::WasmImport Import;
1225         Import.Module = WS.getImportModule();
1226         Import.Field = WS.getImportName();
1227         Import.Kind = wasm::WASM_EXTERNAL_EVENT;
1228         Import.Event.Attribute = wasm::WASM_EVENT_ATTRIBUTE_EXCEPTION;
1229         Import.Event.SigIndex = getEventType(WS);
1230         Imports.push_back(Import);
1231         WasmIndices[&WS] = NumEventImports++;
1232       }
1233     }
1234   }
1235
1236   // Populate DataSegments and CustomSections, which must be done before
1237   // populating DataLocations.
1238   for (MCSection &Sec : Asm) {
1239     auto &Section = static_cast<MCSectionWasm &>(Sec);
1240     StringRef SectionName = Section.getSectionName();
1241
1242     // .init_array sections are handled specially elsewhere.
1243     if (SectionName.startswith(".init_array"))
1244       continue;
1245
1246     // Code is handled separately
1247     if (Section.getKind().isText())
1248       continue;
1249
1250     if (Section.isWasmData()) {
1251       uint32_t SegmentIndex = DataSegments.size();
1252       DataSize = alignTo(DataSize, Section.getAlignment());
1253       DataSegments.emplace_back();
1254       WasmDataSegment &Segment = DataSegments.back();
1255       Segment.Name = SectionName;
1256       Segment.Offset = DataSize;
1257       Segment.Section = &Section;
1258       addData(Segment.Data, Section);
1259       Segment.Alignment = Log2_32(Section.getAlignment());
1260       Segment.Flags = 0;
1261       DataSize += Segment.Data.size();
1262       Section.setSegmentIndex(SegmentIndex);
1263
1264       if (const MCSymbolWasm *C = Section.getGroup()) {
1265         Comdats[C->getName()].emplace_back(
1266             WasmComdatEntry{wasm::WASM_COMDAT_DATA, SegmentIndex});
1267       }
1268     } else {
1269       // Create custom sections
1270       assert(Sec.getKind().isMetadata());
1271
1272       StringRef Name = SectionName;
1273
1274       // For user-defined custom sections, strip the prefix
1275       if (Name.startswith(".custom_section."))
1276         Name = Name.substr(strlen(".custom_section."));
1277
1278       MCSymbol *Begin = Sec.getBeginSymbol();
1279       if (Begin) {
1280         WasmIndices[cast<MCSymbolWasm>(Begin)] = CustomSections.size();
1281         if (SectionName != Begin->getName())
1282           report_fatal_error("section name and begin symbol should match: " +
1283                              Twine(SectionName));
1284       }
1285
1286       // Separate out the producers section
1287       if (Name == "producers") {
1288         ProducersSection = llvm::make_unique<WasmCustomSection>(Name, &Section);
1289         continue;
1290       }
1291
1292       CustomSections.emplace_back(Name, &Section);
1293     }
1294   }
1295
1296   // Populate WasmIndices and DataLocations for defined symbols.
1297   for (const MCSymbol &S : Asm.symbols()) {
1298     // Ignore unnamed temporary symbols, which aren't ever exported, imported,
1299     // or used in relocations.
1300     if (S.isTemporary() && S.getName().empty())
1301       continue;
1302
1303     const auto &WS = static_cast<const MCSymbolWasm &>(S);
1304     LLVM_DEBUG(
1305         dbgs() << "MCSymbol: " << toString(WS.getType()) << " '" << S << "'"
1306                << " isDefined=" << S.isDefined() << " isExternal="
1307                << S.isExternal() << " isTemporary=" << S.isTemporary()
1308                << " isWeak=" << WS.isWeak() << " isHidden=" << WS.isHidden()
1309                << " isVariable=" << WS.isVariable() << "\n");
1310
1311     if (WS.isVariable())
1312       continue;
1313     if (WS.isComdat() && !WS.isDefined())
1314       continue;
1315
1316     if (WS.isFunction()) {
1317       unsigned Index;
1318       if (WS.isDefined()) {
1319         if (WS.getOffset() != 0)
1320           report_fatal_error(
1321               "function sections must contain one function each");
1322
1323         if (WS.getSize() == nullptr)
1324           report_fatal_error(
1325               "function symbols must have a size set with .size");
1326
1327         // A definition. Write out the function body.
1328         Index = NumFunctionImports + Functions.size();
1329         WasmFunction Func;
1330         Func.SigIndex = getFunctionType(WS);
1331         Func.Sym = &WS;
1332         WasmIndices[&WS] = Index;
1333         Functions.push_back(Func);
1334
1335         auto &Section = static_cast<MCSectionWasm &>(WS.getSection());
1336         if (const MCSymbolWasm *C = Section.getGroup()) {
1337           Comdats[C->getName()].emplace_back(
1338               WasmComdatEntry{wasm::WASM_COMDAT_FUNCTION, Index});
1339         }
1340       } else {
1341         // An import; the index was assigned above.
1342         Index = WasmIndices.find(&WS)->second;
1343       }
1344
1345       LLVM_DEBUG(dbgs() << "  -> function index: " << Index << "\n");
1346
1347     } else if (WS.isData()) {
1348       if (WS.isTemporary() && !WS.getSize())
1349         continue;
1350
1351       if (!WS.isDefined()) {
1352         LLVM_DEBUG(dbgs() << "  -> segment index: -1"
1353                           << "\n");
1354         continue;
1355       }
1356
1357       if (!WS.getSize())
1358         report_fatal_error("data symbols must have a size set with .size: " +
1359                            WS.getName());
1360
1361       int64_t Size = 0;
1362       if (!WS.getSize()->evaluateAsAbsolute(Size, Layout))
1363         report_fatal_error(".size expression must be evaluatable");
1364
1365       auto &DataSection = static_cast<MCSectionWasm &>(WS.getSection());
1366       assert(DataSection.isWasmData());
1367
1368       // For each data symbol, export it in the symtab as a reference to the
1369       // corresponding Wasm data segment.
1370       wasm::WasmDataReference Ref = wasm::WasmDataReference{
1371           DataSection.getSegmentIndex(),
1372           static_cast<uint32_t>(Layout.getSymbolOffset(WS)),
1373           static_cast<uint32_t>(Size)};
1374       DataLocations[&WS] = Ref;
1375       LLVM_DEBUG(dbgs() << "  -> segment index: " << Ref.Segment << "\n");
1376
1377     } else if (WS.isGlobal()) {
1378       // A "true" Wasm global (currently just __stack_pointer)
1379       if (WS.isDefined())
1380         report_fatal_error("don't yet support defined globals");
1381
1382       // An import; the index was assigned above
1383       LLVM_DEBUG(dbgs() << "  -> global index: "
1384                         << WasmIndices.find(&WS)->second << "\n");
1385
1386     } else if (WS.isEvent()) {
1387       // C++ exception symbol (__cpp_exception)
1388       unsigned Index;
1389       if (WS.isDefined()) {
1390         Index = NumEventImports + Events.size();
1391         wasm::WasmEventType Event;
1392         Event.SigIndex = getEventType(WS);
1393         Event.Attribute = wasm::WASM_EVENT_ATTRIBUTE_EXCEPTION;
1394         WasmIndices[&WS] = Index;
1395         Events.push_back(Event);
1396       } else {
1397         // An import; the index was assigned above.
1398         Index = WasmIndices.find(&WS)->second;
1399       }
1400       LLVM_DEBUG(dbgs() << "  -> event index: " << WasmIndices.find(&WS)->second
1401                         << "\n");
1402
1403     } else {
1404       assert(WS.isSection());
1405     }
1406   }
1407
1408   // Populate WasmIndices and DataLocations for aliased symbols.  We need to
1409   // process these in a separate pass because we need to have processed the
1410   // target of the alias before the alias itself and the symbols are not
1411   // necessarily ordered in this way.
1412   for (const MCSymbol &S : Asm.symbols()) {
1413     if (!S.isVariable())
1414       continue;
1415
1416     assert(S.isDefined());
1417
1418     // Find the target symbol of this weak alias and export that index
1419     const auto &WS = static_cast<const MCSymbolWasm &>(S);
1420     const MCSymbolWasm *ResolvedSym = resolveSymbol(WS);
1421     LLVM_DEBUG(dbgs() << WS.getName() << ": weak alias of '" << *ResolvedSym
1422                       << "'\n");
1423
1424     if (WS.isFunction()) {
1425       assert(WasmIndices.count(ResolvedSym) > 0);
1426       uint32_t WasmIndex = WasmIndices.find(ResolvedSym)->second;
1427       WasmIndices[&WS] = WasmIndex;
1428       LLVM_DEBUG(dbgs() << "  -> index:" << WasmIndex << "\n");
1429     } else if (WS.isData()) {
1430       assert(DataLocations.count(ResolvedSym) > 0);
1431       const wasm::WasmDataReference &Ref =
1432           DataLocations.find(ResolvedSym)->second;
1433       DataLocations[&WS] = Ref;
1434       LLVM_DEBUG(dbgs() << "  -> index:" << Ref.Segment << "\n");
1435     } else {
1436       report_fatal_error("don't yet support global/event aliases");
1437     }
1438   }
1439
1440   // Finally, populate the symbol table itself, in its "natural" order.
1441   for (const MCSymbol &S : Asm.symbols()) {
1442     const auto &WS = static_cast<const MCSymbolWasm &>(S);
1443     if (!isInSymtab(WS)) {
1444       WS.setIndex(InvalidIndex);
1445       continue;
1446     }
1447     LLVM_DEBUG(dbgs() << "adding to symtab: " << WS << "\n");
1448
1449     uint32_t Flags = 0;
1450     if (WS.isWeak())
1451       Flags |= wasm::WASM_SYMBOL_BINDING_WEAK;
1452     if (WS.isHidden())
1453       Flags |= wasm::WASM_SYMBOL_VISIBILITY_HIDDEN;
1454     if (!WS.isExternal() && WS.isDefined())
1455       Flags |= wasm::WASM_SYMBOL_BINDING_LOCAL;
1456     if (WS.isUndefined())
1457       Flags |= wasm::WASM_SYMBOL_UNDEFINED;
1458     if (WS.isExported())
1459       Flags |= wasm::WASM_SYMBOL_EXPORTED;
1460     if (WS.getName() != WS.getImportName())
1461       Flags |= wasm::WASM_SYMBOL_EXPLICIT_NAME;
1462
1463     wasm::WasmSymbolInfo Info;
1464     Info.Name = WS.getName();
1465     Info.Kind = WS.getType();
1466     Info.Flags = Flags;
1467     if (!WS.isData()) {
1468       assert(WasmIndices.count(&WS) > 0);
1469       Info.ElementIndex = WasmIndices.find(&WS)->second;
1470     } else if (WS.isDefined()) {
1471       assert(DataLocations.count(&WS) > 0);
1472       Info.DataRef = DataLocations.find(&WS)->second;
1473     }
1474     WS.setIndex(SymbolInfos.size());
1475     SymbolInfos.emplace_back(Info);
1476   }
1477
1478   {
1479     auto HandleReloc = [&](const WasmRelocationEntry &Rel) {
1480       // Functions referenced by a relocation need to put in the table.  This is
1481       // purely to make the object file's provisional values readable, and is
1482       // ignored by the linker, which re-calculates the relocations itself.
1483       if (Rel.Type != wasm::R_WASM_TABLE_INDEX_I32 &&
1484           Rel.Type != wasm::R_WASM_TABLE_INDEX_SLEB)
1485         return;
1486       assert(Rel.Symbol->isFunction());
1487       const MCSymbolWasm &WS = *resolveSymbol(*Rel.Symbol);
1488       uint32_t FunctionIndex = WasmIndices.find(&WS)->second;
1489       uint32_t TableIndex = TableElems.size() + InitialTableOffset;
1490       if (TableIndices.try_emplace(&WS, TableIndex).second) {
1491         LLVM_DEBUG(dbgs() << "  -> adding " << WS.getName()
1492                           << " to table: " << TableIndex << "\n");
1493         TableElems.push_back(FunctionIndex);
1494         registerFunctionType(WS);
1495       }
1496     };
1497
1498     for (const WasmRelocationEntry &RelEntry : CodeRelocations)
1499       HandleReloc(RelEntry);
1500     for (const WasmRelocationEntry &RelEntry : DataRelocations)
1501       HandleReloc(RelEntry);
1502   }
1503
1504   // Translate .init_array section contents into start functions.
1505   for (const MCSection &S : Asm) {
1506     const auto &WS = static_cast<const MCSectionWasm &>(S);
1507     if (WS.getSectionName().startswith(".fini_array"))
1508       report_fatal_error(".fini_array sections are unsupported");
1509     if (!WS.getSectionName().startswith(".init_array"))
1510       continue;
1511     if (WS.getFragmentList().empty())
1512       continue;
1513
1514     // init_array is expected to contain a single non-empty data fragment
1515     if (WS.getFragmentList().size() != 3)
1516       report_fatal_error("only one .init_array section fragment supported");
1517
1518     auto IT = WS.begin();
1519     const MCFragment &EmptyFrag = *IT;
1520     if (EmptyFrag.getKind() != MCFragment::FT_Data)
1521       report_fatal_error(".init_array section should be aligned");
1522
1523     IT = std::next(IT);
1524     const MCFragment &AlignFrag = *IT;
1525     if (AlignFrag.getKind() != MCFragment::FT_Align)
1526       report_fatal_error(".init_array section should be aligned");
1527     if (cast<MCAlignFragment>(AlignFrag).getAlignment() != (is64Bit() ? 8 : 4))
1528       report_fatal_error(".init_array section should be aligned for pointers");
1529
1530     const MCFragment &Frag = *std::next(IT);
1531     if (Frag.hasInstructions() || Frag.getKind() != MCFragment::FT_Data)
1532       report_fatal_error("only data supported in .init_array section");
1533
1534     uint16_t Priority = UINT16_MAX;
1535     unsigned PrefixLength = strlen(".init_array");
1536     if (WS.getSectionName().size() > PrefixLength) {
1537       if (WS.getSectionName()[PrefixLength] != '.')
1538         report_fatal_error(
1539             ".init_array section priority should start with '.'");
1540       if (WS.getSectionName()
1541               .substr(PrefixLength + 1)
1542               .getAsInteger(10, Priority))
1543         report_fatal_error("invalid .init_array section priority");
1544     }
1545     const auto &DataFrag = cast<MCDataFragment>(Frag);
1546     const SmallVectorImpl<char> &Contents = DataFrag.getContents();
1547     for (const uint8_t *
1548              P = (const uint8_t *)Contents.data(),
1549             *End = (const uint8_t *)Contents.data() + Contents.size();
1550          P != End; ++P) {
1551       if (*P != 0)
1552         report_fatal_error("non-symbolic data in .init_array section");
1553     }
1554     for (const MCFixup &Fixup : DataFrag.getFixups()) {
1555       assert(Fixup.getKind() ==
1556              MCFixup::getKindForSize(is64Bit() ? 8 : 4, false));
1557       const MCExpr *Expr = Fixup.getValue();
1558       auto *Sym = dyn_cast<MCSymbolRefExpr>(Expr);
1559       if (!Sym)
1560         report_fatal_error("fixups in .init_array should be symbol references");
1561       if (Sym->getKind() != MCSymbolRefExpr::VK_WebAssembly_FUNCTION)
1562         report_fatal_error("symbols in .init_array should be for functions");
1563       if (Sym->getSymbol().getIndex() == InvalidIndex)
1564         report_fatal_error("symbols in .init_array should exist in symbtab");
1565       InitFuncs.push_back(
1566           std::make_pair(Priority, Sym->getSymbol().getIndex()));
1567     }
1568   }
1569
1570   // Write out the Wasm header.
1571   writeHeader(Asm);
1572
1573   writeTypeSection(Signatures);
1574   writeImportSection(Imports, DataSize, TableElems.size());
1575   writeFunctionSection(Functions);
1576   // Skip the "table" section; we import the table instead.
1577   // Skip the "memory" section; we import the memory instead.
1578   writeGlobalSection();
1579   writeEventSection(Events);
1580   writeExportSection(Exports);
1581   writeElemSection(TableElems);
1582   writeCodeSection(Asm, Layout, Functions);
1583   writeDataSection();
1584   for (auto &CustomSection : CustomSections)
1585     writeCustomSection(CustomSection, Asm, Layout);
1586   writeLinkingMetaDataSection(SymbolInfos, InitFuncs, Comdats);
1587   writeRelocSection(CodeSectionIndex, "CODE", CodeRelocations);
1588   writeRelocSection(DataSectionIndex, "DATA", DataRelocations);
1589   writeCustomRelocSections();
1590   if (ProducersSection)
1591     writeCustomSection(*ProducersSection, Asm, Layout);
1592
1593   // TODO: Translate the .comment section to the output.
1594   return W.OS.tell() - StartOffset;
1595 }
1596
1597 std::unique_ptr<MCObjectWriter>
1598 llvm::createWasmObjectWriter(std::unique_ptr<MCWasmObjectTargetWriter> MOTW,
1599                              raw_pwrite_stream &OS) {
1600   return llvm::make_unique<WasmObjectWriter>(std::move(MOTW), OS);
1601 }