1 //===- lib/MC/WasmObjectWriter.cpp - Wasm File Writer ---------------------===// 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 // This file implements Wasm object file writer information. 11 // 12 //===----------------------------------------------------------------------===// 13 14 #include "llvm/ADT/STLExtras.h" 15 #include "llvm/ADT/SmallPtrSet.h" 16 #include "llvm/BinaryFormat/Wasm.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 // For patching purposes, we need to remember where each section starts, both 42 // for patching up the section size field, and for patching up references to 43 // locations within the section. 44 struct SectionBookkeeping { 45 // Where the size of the section is written. 46 uint64_t SizeOffset; 47 // Where the contents of the section starts (after the header). 48 uint64_t ContentsOffset; 49 }; 50 51 // The signature of a wasm function, in a struct capable of being used as a 52 // DenseMap key. 53 struct WasmFunctionType { 54 // Support empty and tombstone instances, needed by DenseMap. 55 enum { Plain, Empty, Tombstone } State; 56 57 // The return types of the function. 58 SmallVector<wasm::ValType, 1> Returns; 59 60 // The parameter types of the function. 61 SmallVector<wasm::ValType, 4> Params; 62 63 WasmFunctionType() : State(Plain) {} 64 65 bool operator==(const WasmFunctionType &Other) const { 66 return State == Other.State && Returns == Other.Returns && 67 Params == Other.Params; 68 } 69 }; 70 71 // Traits for using WasmFunctionType in a DenseMap. 72 struct WasmFunctionTypeDenseMapInfo { 73 static WasmFunctionType getEmptyKey() { 74 WasmFunctionType FuncTy; 75 FuncTy.State = WasmFunctionType::Empty; 76 return FuncTy; 77 } 78 static WasmFunctionType getTombstoneKey() { 79 WasmFunctionType FuncTy; 80 FuncTy.State = WasmFunctionType::Tombstone; 81 return FuncTy; 82 } 83 static unsigned getHashValue(const WasmFunctionType &FuncTy) { 84 uintptr_t Value = FuncTy.State; 85 for (wasm::ValType Ret : FuncTy.Returns) 86 Value += DenseMapInfo<int32_t>::getHashValue(int32_t(Ret)); 87 for (wasm::ValType Param : FuncTy.Params) 88 Value += DenseMapInfo<int32_t>::getHashValue(int32_t(Param)); 89 return Value; 90 } 91 static bool isEqual(const WasmFunctionType &LHS, 92 const WasmFunctionType &RHS) { 93 return LHS == RHS; 94 } 95 }; 96 97 // A wasm data segment. A wasm binary contains only a single data section 98 // but that can contain many segments, each with their own virtual location 99 // in memory. Each MCSection data created by llvm is modeled as its own 100 // wasm data segment. 101 struct WasmDataSegment { 102 MCSectionWasm *Section; 103 StringRef Name; 104 uint32_t Offset; 105 uint32_t Alignment; 106 uint32_t Flags; 107 SmallVector<char, 4> Data; 108 }; 109 110 // A wasm import to be written into the import section. 111 struct WasmImport { 112 StringRef ModuleName; 113 StringRef FieldName; 114 unsigned Kind; 115 int32_t Type; 116 bool IsMutable; 117 }; 118 119 // A wasm function to be written into the function section. 120 struct WasmFunction { 121 int32_t Type; 122 const MCSymbolWasm *Sym; 123 }; 124 125 // A wasm export to be written into the export section. 126 struct WasmExport { 127 StringRef FieldName; 128 unsigned Kind; 129 uint32_t Index; 130 }; 131 132 // A wasm global to be written into the global section. 133 struct WasmGlobal { 134 wasm::ValType Type; 135 bool IsMutable; 136 bool HasImport; 137 uint64_t InitialValue; 138 uint32_t ImportIndex; 139 }; 140 141 // Information about a single relocation. 142 struct WasmRelocationEntry { 143 uint64_t Offset; // Where is the relocation. 144 const MCSymbolWasm *Symbol; // The symbol to relocate with. 145 int64_t Addend; // A value to add to the symbol. 146 unsigned Type; // The type of the relocation. 147 const MCSectionWasm *FixupSection;// The section the relocation is targeting. 148 149 WasmRelocationEntry(uint64_t Offset, const MCSymbolWasm *Symbol, 150 int64_t Addend, unsigned Type, 151 const MCSectionWasm *FixupSection) 152 : Offset(Offset), Symbol(Symbol), Addend(Addend), Type(Type), 153 FixupSection(FixupSection) {} 154 155 bool hasAddend() const { 156 switch (Type) { 157 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_LEB: 158 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_SLEB: 159 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_I32: 160 return true; 161 default: 162 return false; 163 } 164 } 165 166 void print(raw_ostream &Out) const { 167 Out << "Off=" << Offset << ", Sym=" << *Symbol << ", Addend=" << Addend 168 << ", Type=" << Type 169 << ", FixupSection=" << FixupSection->getSectionName(); 170 } 171 172 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) 173 LLVM_DUMP_METHOD void dump() const { print(dbgs()); } 174 #endif 175 }; 176 177 #if !defined(NDEBUG) 178 raw_ostream &operator<<(raw_ostream &OS, const WasmRelocationEntry &Rel) { 179 Rel.print(OS); 180 return OS; 181 } 182 #endif 183 184 class WasmObjectWriter : public MCObjectWriter { 185 /// Helper struct for containing some precomputed information on symbols. 186 struct WasmSymbolData { 187 const MCSymbolWasm *Symbol; 188 StringRef Name; 189 190 // Support lexicographic sorting. 191 bool operator<(const WasmSymbolData &RHS) const { return Name < RHS.Name; } 192 }; 193 194 /// The target specific Wasm writer instance. 195 std::unique_ptr<MCWasmObjectTargetWriter> TargetObjectWriter; 196 197 // Relocations for fixing up references in the code section. 198 std::vector<WasmRelocationEntry> CodeRelocations; 199 200 // Relocations for fixing up references in the data section. 201 std::vector<WasmRelocationEntry> DataRelocations; 202 203 // Index values to use for fixing up call_indirect type indices. 204 // Maps function symbols to the index of the type of the function 205 DenseMap<const MCSymbolWasm *, uint32_t> TypeIndices; 206 // Maps function symbols to the table element index space. Used 207 // for TABLE_INDEX relocation types (i.e. address taken functions). 208 DenseMap<const MCSymbolWasm *, uint32_t> IndirectSymbolIndices; 209 // Maps function/global symbols to the function/global index space. 210 DenseMap<const MCSymbolWasm *, uint32_t> SymbolIndices; 211 212 DenseMap<WasmFunctionType, int32_t, WasmFunctionTypeDenseMapInfo> 213 FunctionTypeIndices; 214 SmallVector<WasmFunctionType, 4> FunctionTypes; 215 SmallVector<WasmGlobal, 4> Globals; 216 unsigned NumGlobalImports = 0; 217 218 // TargetObjectWriter wrappers. 219 bool is64Bit() const { return TargetObjectWriter->is64Bit(); } 220 unsigned getRelocType(const MCValue &Target, const MCFixup &Fixup) const { 221 return TargetObjectWriter->getRelocType(Target, Fixup); 222 } 223 224 void startSection(SectionBookkeeping &Section, unsigned SectionId, 225 const char *Name = nullptr); 226 void endSection(SectionBookkeeping &Section); 227 228 public: 229 WasmObjectWriter(std::unique_ptr<MCWasmObjectTargetWriter> MOTW, 230 raw_pwrite_stream &OS) 231 : MCObjectWriter(OS, /*IsLittleEndian=*/true), 232 TargetObjectWriter(std::move(MOTW)) {} 233 234 private: 235 ~WasmObjectWriter() override; 236 237 void reset() override { 238 CodeRelocations.clear(); 239 DataRelocations.clear(); 240 TypeIndices.clear(); 241 SymbolIndices.clear(); 242 IndirectSymbolIndices.clear(); 243 FunctionTypeIndices.clear(); 244 FunctionTypes.clear(); 245 Globals.clear(); 246 MCObjectWriter::reset(); 247 NumGlobalImports = 0; 248 } 249 250 void writeHeader(const MCAssembler &Asm); 251 252 void recordRelocation(MCAssembler &Asm, const MCAsmLayout &Layout, 253 const MCFragment *Fragment, const MCFixup &Fixup, 254 MCValue Target, uint64_t &FixedValue) override; 255 256 void executePostLayoutBinding(MCAssembler &Asm, 257 const MCAsmLayout &Layout) override; 258 259 void writeObject(MCAssembler &Asm, const MCAsmLayout &Layout) override; 260 261 void writeString(const StringRef Str) { 262 encodeULEB128(Str.size(), getStream()); 263 writeBytes(Str); 264 } 265 266 void writeValueType(wasm::ValType Ty) { 267 encodeSLEB128(int32_t(Ty), getStream()); 268 } 269 270 void writeTypeSection(ArrayRef<WasmFunctionType> FunctionTypes); 271 void writeImportSection(ArrayRef<WasmImport> Imports, uint32_t DataSize, 272 uint32_t NumElements); 273 void writeFunctionSection(ArrayRef<WasmFunction> Functions); 274 void writeGlobalSection(); 275 void writeExportSection(ArrayRef<WasmExport> Exports); 276 void writeElemSection(ArrayRef<uint32_t> TableElems); 277 void writeCodeSection(const MCAssembler &Asm, const MCAsmLayout &Layout, 278 ArrayRef<WasmFunction> Functions); 279 void writeDataSection(ArrayRef<WasmDataSegment> Segments); 280 void writeNameSection(ArrayRef<WasmFunction> Functions, 281 ArrayRef<WasmImport> Imports, 282 uint32_t NumFuncImports); 283 void writeCodeRelocSection(); 284 void writeDataRelocSection(); 285 void writeLinkingMetaDataSection( 286 ArrayRef<WasmDataSegment> Segments, uint32_t DataSize, 287 const SmallVector<std::pair<StringRef, uint32_t>, 4> &SymbolFlags, 288 const SmallVector<std::pair<uint16_t, uint32_t>, 2> &InitFuncs); 289 290 uint32_t getProvisionalValue(const WasmRelocationEntry &RelEntry); 291 void applyRelocations(ArrayRef<WasmRelocationEntry> Relocations, 292 uint64_t ContentsOffset); 293 294 void writeRelocations(ArrayRef<WasmRelocationEntry> Relocations); 295 uint32_t getRelocationIndexValue(const WasmRelocationEntry &RelEntry); 296 uint32_t getFunctionType(const MCSymbolWasm& Symbol); 297 uint32_t registerFunctionType(const MCSymbolWasm& Symbol); 298 }; 299 300 } // end anonymous namespace 301 302 WasmObjectWriter::~WasmObjectWriter() {} 303 304 // Write out a section header and a patchable section size field. 305 void WasmObjectWriter::startSection(SectionBookkeeping &Section, 306 unsigned SectionId, 307 const char *Name) { 308 assert((Name != nullptr) == (SectionId == wasm::WASM_SEC_CUSTOM) && 309 "Only custom sections can have names"); 310 311 DEBUG(dbgs() << "startSection " << SectionId << ": " << Name << "\n"); 312 encodeULEB128(SectionId, getStream()); 313 314 Section.SizeOffset = getStream().tell(); 315 316 // The section size. We don't know the size yet, so reserve enough space 317 // for any 32-bit value; we'll patch it later. 318 encodeULEB128(UINT32_MAX, getStream()); 319 320 // The position where the section starts, for measuring its size. 321 Section.ContentsOffset = getStream().tell(); 322 323 // Custom sections in wasm also have a string identifier. 324 if (SectionId == wasm::WASM_SEC_CUSTOM) { 325 assert(Name); 326 writeString(StringRef(Name)); 327 } 328 } 329 330 // Now that the section is complete and we know how big it is, patch up the 331 // section size field at the start of the section. 332 void WasmObjectWriter::endSection(SectionBookkeeping &Section) { 333 uint64_t Size = getStream().tell() - Section.ContentsOffset; 334 if (uint32_t(Size) != Size) 335 report_fatal_error("section size does not fit in a uint32_t"); 336 337 DEBUG(dbgs() << "endSection size=" << Size << "\n"); 338 339 // Write the final section size to the payload_len field, which follows 340 // the section id byte. 341 uint8_t Buffer[16]; 342 unsigned SizeLen = encodeULEB128(Size, Buffer, 5); 343 assert(SizeLen == 5); 344 getStream().pwrite((char *)Buffer, SizeLen, Section.SizeOffset); 345 } 346 347 // Emit the Wasm header. 348 void WasmObjectWriter::writeHeader(const MCAssembler &Asm) { 349 writeBytes(StringRef(wasm::WasmMagic, sizeof(wasm::WasmMagic))); 350 writeLE32(wasm::WasmVersion); 351 } 352 353 void WasmObjectWriter::executePostLayoutBinding(MCAssembler &Asm, 354 const MCAsmLayout &Layout) { 355 } 356 357 void WasmObjectWriter::recordRelocation(MCAssembler &Asm, 358 const MCAsmLayout &Layout, 359 const MCFragment *Fragment, 360 const MCFixup &Fixup, MCValue Target, 361 uint64_t &FixedValue) { 362 MCAsmBackend &Backend = Asm.getBackend(); 363 bool IsPCRel = Backend.getFixupKindInfo(Fixup.getKind()).Flags & 364 MCFixupKindInfo::FKF_IsPCRel; 365 const auto &FixupSection = cast<MCSectionWasm>(*Fragment->getParent()); 366 uint64_t C = Target.getConstant(); 367 uint64_t FixupOffset = Layout.getFragmentOffset(Fragment) + Fixup.getOffset(); 368 MCContext &Ctx = Asm.getContext(); 369 370 // The .init_array isn't translated as data, so don't do relocations in it. 371 if (FixupSection.getSectionName().startswith(".init_array")) 372 return; 373 374 if (const MCSymbolRefExpr *RefB = Target.getSymB()) { 375 assert(RefB->getKind() == MCSymbolRefExpr::VK_None && 376 "Should not have constructed this"); 377 378 // Let A, B and C being the components of Target and R be the location of 379 // the fixup. If the fixup is not pcrel, we want to compute (A - B + C). 380 // If it is pcrel, we want to compute (A - B + C - R). 381 382 // In general, Wasm has no relocations for -B. It can only represent (A + C) 383 // or (A + C - R). If B = R + K and the relocation is not pcrel, we can 384 // replace B to implement it: (A - R - K + C) 385 if (IsPCRel) { 386 Ctx.reportError( 387 Fixup.getLoc(), 388 "No relocation available to represent this relative expression"); 389 return; 390 } 391 392 const auto &SymB = cast<MCSymbolWasm>(RefB->getSymbol()); 393 394 if (SymB.isUndefined()) { 395 Ctx.reportError(Fixup.getLoc(), 396 Twine("symbol '") + SymB.getName() + 397 "' can not be undefined in a subtraction expression"); 398 return; 399 } 400 401 assert(!SymB.isAbsolute() && "Should have been folded"); 402 const MCSection &SecB = SymB.getSection(); 403 if (&SecB != &FixupSection) { 404 Ctx.reportError(Fixup.getLoc(), 405 "Cannot represent a difference across sections"); 406 return; 407 } 408 409 uint64_t SymBOffset = Layout.getSymbolOffset(SymB); 410 uint64_t K = SymBOffset - FixupOffset; 411 IsPCRel = true; 412 C -= K; 413 } 414 415 // We either rejected the fixup or folded B into C at this point. 416 const MCSymbolRefExpr *RefA = Target.getSymA(); 417 const auto *SymA = RefA ? cast<MCSymbolWasm>(&RefA->getSymbol()) : nullptr; 418 419 if (SymA && SymA->isVariable()) { 420 const MCExpr *Expr = SymA->getVariableValue(); 421 const auto *Inner = cast<MCSymbolRefExpr>(Expr); 422 if (Inner->getKind() == MCSymbolRefExpr::VK_WEAKREF) 423 llvm_unreachable("weakref used in reloc not yet implemented"); 424 } 425 426 // Put any constant offset in an addend. Offsets can be negative, and 427 // LLVM expects wrapping, in contrast to wasm's immediates which can't 428 // be negative and don't wrap. 429 FixedValue = 0; 430 431 if (SymA) 432 SymA->setUsedInReloc(); 433 434 assert(!IsPCRel); 435 assert(SymA); 436 437 unsigned Type = getRelocType(Target, Fixup); 438 439 WasmRelocationEntry Rec(FixupOffset, SymA, C, Type, &FixupSection); 440 DEBUG(dbgs() << "WasmReloc: " << Rec << "\n"); 441 442 if (FixupSection.isWasmData()) 443 DataRelocations.push_back(Rec); 444 else if (FixupSection.getKind().isText()) 445 CodeRelocations.push_back(Rec); 446 else if (!FixupSection.getKind().isMetadata()) 447 // TODO(sbc): Add support for debug sections. 448 llvm_unreachable("unexpected section type"); 449 } 450 451 // Write X as an (unsigned) LEB value at offset Offset in Stream, padded 452 // to allow patching. 453 static void 454 WritePatchableLEB(raw_pwrite_stream &Stream, uint32_t X, uint64_t Offset) { 455 uint8_t Buffer[5]; 456 unsigned SizeLen = encodeULEB128(X, Buffer, 5); 457 assert(SizeLen == 5); 458 Stream.pwrite((char *)Buffer, SizeLen, Offset); 459 } 460 461 // Write X as an signed LEB value at offset Offset in Stream, padded 462 // to allow patching. 463 static void 464 WritePatchableSLEB(raw_pwrite_stream &Stream, int32_t X, uint64_t Offset) { 465 uint8_t Buffer[5]; 466 unsigned SizeLen = encodeSLEB128(X, Buffer, 5); 467 assert(SizeLen == 5); 468 Stream.pwrite((char *)Buffer, SizeLen, Offset); 469 } 470 471 // Write X as a plain integer value at offset Offset in Stream. 472 static void WriteI32(raw_pwrite_stream &Stream, uint32_t X, uint64_t Offset) { 473 uint8_t Buffer[4]; 474 support::endian::write32le(Buffer, X); 475 Stream.pwrite((char *)Buffer, sizeof(Buffer), Offset); 476 } 477 478 static const MCSymbolWasm* ResolveSymbol(const MCSymbolWasm& Symbol) { 479 if (Symbol.isVariable()) { 480 const MCExpr *Expr = Symbol.getVariableValue(); 481 auto *Inner = cast<MCSymbolRefExpr>(Expr); 482 return cast<MCSymbolWasm>(&Inner->getSymbol()); 483 } 484 return &Symbol; 485 } 486 487 // Compute a value to write into the code at the location covered 488 // by RelEntry. This value isn't used by the static linker, since 489 // we have addends; it just serves to make the code more readable 490 // and to make standalone wasm modules directly usable. 491 uint32_t 492 WasmObjectWriter::getProvisionalValue(const WasmRelocationEntry &RelEntry) { 493 const MCSymbolWasm *Sym = ResolveSymbol(*RelEntry.Symbol); 494 495 // For undefined symbols, use a hopefully invalid value. 496 if (!Sym->isDefined(/*SetUsed=*/false)) 497 return UINT32_MAX; 498 499 uint32_t GlobalIndex = SymbolIndices[Sym]; 500 const WasmGlobal& Global = Globals[GlobalIndex - NumGlobalImports]; 501 uint64_t Address = Global.InitialValue + RelEntry.Addend; 502 503 // Ignore overflow. LLVM allows address arithmetic to silently wrap. 504 uint32_t Value = Address; 505 506 return Value; 507 } 508 509 static void addData(SmallVectorImpl<char> &DataBytes, 510 MCSectionWasm &DataSection) { 511 DEBUG(errs() << "addData: " << DataSection.getSectionName() << "\n"); 512 513 DataBytes.resize(alignTo(DataBytes.size(), DataSection.getAlignment())); 514 515 size_t LastFragmentSize = 0; 516 for (const MCFragment &Frag : DataSection) { 517 if (Frag.hasInstructions()) 518 report_fatal_error("only data supported in data sections"); 519 520 if (auto *Align = dyn_cast<MCAlignFragment>(&Frag)) { 521 if (Align->getValueSize() != 1) 522 report_fatal_error("only byte values supported for alignment"); 523 // If nops are requested, use zeros, as this is the data section. 524 uint8_t Value = Align->hasEmitNops() ? 0 : Align->getValue(); 525 uint64_t Size = std::min<uint64_t>(alignTo(DataBytes.size(), 526 Align->getAlignment()), 527 DataBytes.size() + 528 Align->getMaxBytesToEmit()); 529 DataBytes.resize(Size, Value); 530 } else if (auto *Fill = dyn_cast<MCFillFragment>(&Frag)) { 531 DataBytes.insert(DataBytes.end(), Fill->getSize(), Fill->getValue()); 532 } else { 533 const auto &DataFrag = cast<MCDataFragment>(Frag); 534 const SmallVectorImpl<char> &Contents = DataFrag.getContents(); 535 536 DataBytes.insert(DataBytes.end(), Contents.begin(), Contents.end()); 537 LastFragmentSize = Contents.size(); 538 } 539 } 540 541 // Don't allow empty segments, or segments that end with zero-sized 542 // fragment, otherwise the linker cannot map symbols to a unique 543 // data segment. This can be triggered by zero-sized structs 544 // See: test/MC/WebAssembly/bss.ll 545 if (LastFragmentSize == 0) 546 DataBytes.resize(DataBytes.size() + 1); 547 DEBUG(dbgs() << "addData -> " << DataBytes.size() << "\n"); 548 } 549 550 uint32_t WasmObjectWriter::getRelocationIndexValue( 551 const WasmRelocationEntry &RelEntry) { 552 switch (RelEntry.Type) { 553 case wasm::R_WEBASSEMBLY_TABLE_INDEX_SLEB: 554 case wasm::R_WEBASSEMBLY_TABLE_INDEX_I32: 555 if (!IndirectSymbolIndices.count(RelEntry.Symbol)) 556 report_fatal_error("symbol not found table index space: " + 557 RelEntry.Symbol->getName()); 558 return IndirectSymbolIndices[RelEntry.Symbol]; 559 case wasm::R_WEBASSEMBLY_FUNCTION_INDEX_LEB: 560 case wasm::R_WEBASSEMBLY_GLOBAL_INDEX_LEB: 561 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_LEB: 562 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_SLEB: 563 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_I32: 564 if (!SymbolIndices.count(RelEntry.Symbol)) 565 report_fatal_error("symbol not found function/global index space: " + 566 RelEntry.Symbol->getName()); 567 return SymbolIndices[RelEntry.Symbol]; 568 case wasm::R_WEBASSEMBLY_TYPE_INDEX_LEB: 569 if (!TypeIndices.count(RelEntry.Symbol)) 570 report_fatal_error("symbol not found in type index space: " + 571 RelEntry.Symbol->getName()); 572 return TypeIndices[RelEntry.Symbol]; 573 default: 574 llvm_unreachable("invalid relocation type"); 575 } 576 } 577 578 // Apply the portions of the relocation records that we can handle ourselves 579 // directly. 580 void WasmObjectWriter::applyRelocations( 581 ArrayRef<WasmRelocationEntry> Relocations, uint64_t ContentsOffset) { 582 raw_pwrite_stream &Stream = getStream(); 583 for (const WasmRelocationEntry &RelEntry : Relocations) { 584 uint64_t Offset = ContentsOffset + 585 RelEntry.FixupSection->getSectionOffset() + 586 RelEntry.Offset; 587 588 DEBUG(dbgs() << "applyRelocation: " << RelEntry << "\n"); 589 switch (RelEntry.Type) { 590 case wasm::R_WEBASSEMBLY_TABLE_INDEX_SLEB: 591 case wasm::R_WEBASSEMBLY_FUNCTION_INDEX_LEB: 592 case wasm::R_WEBASSEMBLY_TYPE_INDEX_LEB: 593 case wasm::R_WEBASSEMBLY_GLOBAL_INDEX_LEB: { 594 uint32_t Index = getRelocationIndexValue(RelEntry); 595 WritePatchableSLEB(Stream, Index, Offset); 596 break; 597 } 598 case wasm::R_WEBASSEMBLY_TABLE_INDEX_I32: { 599 uint32_t Index = getRelocationIndexValue(RelEntry); 600 WriteI32(Stream, Index, Offset); 601 break; 602 } 603 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_SLEB: { 604 uint32_t Value = getProvisionalValue(RelEntry); 605 WritePatchableSLEB(Stream, Value, Offset); 606 break; 607 } 608 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_LEB: { 609 uint32_t Value = getProvisionalValue(RelEntry); 610 WritePatchableLEB(Stream, Value, Offset); 611 break; 612 } 613 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_I32: { 614 uint32_t Value = getProvisionalValue(RelEntry); 615 WriteI32(Stream, Value, Offset); 616 break; 617 } 618 default: 619 llvm_unreachable("invalid relocation type"); 620 } 621 } 622 } 623 624 // Write out the portions of the relocation records that the linker will 625 // need to handle. 626 void WasmObjectWriter::writeRelocations( 627 ArrayRef<WasmRelocationEntry> Relocations) { 628 raw_pwrite_stream &Stream = getStream(); 629 for (const WasmRelocationEntry& RelEntry : Relocations) { 630 631 uint64_t Offset = RelEntry.Offset + 632 RelEntry.FixupSection->getSectionOffset(); 633 uint32_t Index = getRelocationIndexValue(RelEntry); 634 635 encodeULEB128(RelEntry.Type, Stream); 636 encodeULEB128(Offset, Stream); 637 encodeULEB128(Index, Stream); 638 if (RelEntry.hasAddend()) 639 encodeSLEB128(RelEntry.Addend, Stream); 640 } 641 } 642 643 void WasmObjectWriter::writeTypeSection( 644 ArrayRef<WasmFunctionType> FunctionTypes) { 645 if (FunctionTypes.empty()) 646 return; 647 648 SectionBookkeeping Section; 649 startSection(Section, wasm::WASM_SEC_TYPE); 650 651 encodeULEB128(FunctionTypes.size(), getStream()); 652 653 for (const WasmFunctionType &FuncTy : FunctionTypes) { 654 encodeSLEB128(wasm::WASM_TYPE_FUNC, getStream()); 655 encodeULEB128(FuncTy.Params.size(), getStream()); 656 for (wasm::ValType Ty : FuncTy.Params) 657 writeValueType(Ty); 658 encodeULEB128(FuncTy.Returns.size(), getStream()); 659 for (wasm::ValType Ty : FuncTy.Returns) 660 writeValueType(Ty); 661 } 662 663 endSection(Section); 664 } 665 666 void WasmObjectWriter::writeImportSection(ArrayRef<WasmImport> Imports, 667 uint32_t DataSize, 668 uint32_t NumElements) { 669 if (Imports.empty()) 670 return; 671 672 uint32_t NumPages = (DataSize + wasm::WasmPageSize - 1) / wasm::WasmPageSize; 673 674 SectionBookkeeping Section; 675 startSection(Section, wasm::WASM_SEC_IMPORT); 676 677 encodeULEB128(Imports.size(), getStream()); 678 for (const WasmImport &Import : Imports) { 679 writeString(Import.ModuleName); 680 writeString(Import.FieldName); 681 682 encodeULEB128(Import.Kind, getStream()); 683 684 switch (Import.Kind) { 685 case wasm::WASM_EXTERNAL_FUNCTION: 686 encodeULEB128(Import.Type, getStream()); 687 break; 688 case wasm::WASM_EXTERNAL_GLOBAL: 689 encodeSLEB128(int32_t(Import.Type), getStream()); 690 encodeULEB128(int32_t(Import.IsMutable), getStream()); 691 break; 692 case wasm::WASM_EXTERNAL_MEMORY: 693 encodeULEB128(0, getStream()); // flags 694 encodeULEB128(NumPages, getStream()); // initial 695 break; 696 case wasm::WASM_EXTERNAL_TABLE: 697 encodeSLEB128(int32_t(Import.Type), getStream()); 698 encodeULEB128(0, getStream()); // flags 699 encodeULEB128(NumElements, getStream()); // initial 700 break; 701 default: 702 llvm_unreachable("unsupported import kind"); 703 } 704 } 705 706 endSection(Section); 707 } 708 709 void WasmObjectWriter::writeFunctionSection(ArrayRef<WasmFunction> Functions) { 710 if (Functions.empty()) 711 return; 712 713 SectionBookkeeping Section; 714 startSection(Section, wasm::WASM_SEC_FUNCTION); 715 716 encodeULEB128(Functions.size(), getStream()); 717 for (const WasmFunction &Func : Functions) 718 encodeULEB128(Func.Type, getStream()); 719 720 endSection(Section); 721 } 722 723 void WasmObjectWriter::writeGlobalSection() { 724 if (Globals.empty()) 725 return; 726 727 SectionBookkeeping Section; 728 startSection(Section, wasm::WASM_SEC_GLOBAL); 729 730 encodeULEB128(Globals.size(), getStream()); 731 for (const WasmGlobal &Global : Globals) { 732 writeValueType(Global.Type); 733 write8(Global.IsMutable); 734 735 if (Global.HasImport) { 736 assert(Global.InitialValue == 0); 737 write8(wasm::WASM_OPCODE_GET_GLOBAL); 738 encodeULEB128(Global.ImportIndex, getStream()); 739 } else { 740 assert(Global.ImportIndex == 0); 741 write8(wasm::WASM_OPCODE_I32_CONST); 742 encodeSLEB128(Global.InitialValue, getStream()); // offset 743 } 744 write8(wasm::WASM_OPCODE_END); 745 } 746 747 endSection(Section); 748 } 749 750 void WasmObjectWriter::writeExportSection(ArrayRef<WasmExport> Exports) { 751 if (Exports.empty()) 752 return; 753 754 SectionBookkeeping Section; 755 startSection(Section, wasm::WASM_SEC_EXPORT); 756 757 encodeULEB128(Exports.size(), getStream()); 758 for (const WasmExport &Export : Exports) { 759 writeString(Export.FieldName); 760 encodeSLEB128(Export.Kind, getStream()); 761 encodeULEB128(Export.Index, getStream()); 762 } 763 764 endSection(Section); 765 } 766 767 void WasmObjectWriter::writeElemSection(ArrayRef<uint32_t> TableElems) { 768 if (TableElems.empty()) 769 return; 770 771 SectionBookkeeping Section; 772 startSection(Section, wasm::WASM_SEC_ELEM); 773 774 encodeULEB128(1, getStream()); // number of "segments" 775 encodeULEB128(0, getStream()); // the table index 776 777 // init expr for starting offset 778 write8(wasm::WASM_OPCODE_I32_CONST); 779 encodeSLEB128(0, getStream()); 780 write8(wasm::WASM_OPCODE_END); 781 782 encodeULEB128(TableElems.size(), getStream()); 783 for (uint32_t Elem : TableElems) 784 encodeULEB128(Elem, getStream()); 785 786 endSection(Section); 787 } 788 789 void WasmObjectWriter::writeCodeSection(const MCAssembler &Asm, 790 const MCAsmLayout &Layout, 791 ArrayRef<WasmFunction> Functions) { 792 if (Functions.empty()) 793 return; 794 795 SectionBookkeeping Section; 796 startSection(Section, wasm::WASM_SEC_CODE); 797 798 encodeULEB128(Functions.size(), getStream()); 799 800 for (const WasmFunction &Func : Functions) { 801 auto &FuncSection = static_cast<MCSectionWasm &>(Func.Sym->getSection()); 802 803 int64_t Size = 0; 804 if (!Func.Sym->getSize()->evaluateAsAbsolute(Size, Layout)) 805 report_fatal_error(".size expression must be evaluatable"); 806 807 encodeULEB128(Size, getStream()); 808 FuncSection.setSectionOffset(getStream().tell() - Section.ContentsOffset); 809 Asm.writeSectionData(&FuncSection, Layout); 810 } 811 812 // Apply fixups. 813 applyRelocations(CodeRelocations, Section.ContentsOffset); 814 815 endSection(Section); 816 } 817 818 void WasmObjectWriter::writeDataSection(ArrayRef<WasmDataSegment> Segments) { 819 if (Segments.empty()) 820 return; 821 822 SectionBookkeeping Section; 823 startSection(Section, wasm::WASM_SEC_DATA); 824 825 encodeULEB128(Segments.size(), getStream()); // count 826 827 for (const WasmDataSegment & Segment : Segments) { 828 encodeULEB128(0, getStream()); // memory index 829 write8(wasm::WASM_OPCODE_I32_CONST); 830 encodeSLEB128(Segment.Offset, getStream()); // offset 831 write8(wasm::WASM_OPCODE_END); 832 encodeULEB128(Segment.Data.size(), getStream()); // size 833 Segment.Section->setSectionOffset(getStream().tell() - Section.ContentsOffset); 834 writeBytes(Segment.Data); // data 835 } 836 837 // Apply fixups. 838 applyRelocations(DataRelocations, Section.ContentsOffset); 839 840 endSection(Section); 841 } 842 843 void WasmObjectWriter::writeNameSection( 844 ArrayRef<WasmFunction> Functions, 845 ArrayRef<WasmImport> Imports, 846 unsigned NumFuncImports) { 847 uint32_t TotalFunctions = NumFuncImports + Functions.size(); 848 if (TotalFunctions == 0) 849 return; 850 851 SectionBookkeeping Section; 852 startSection(Section, wasm::WASM_SEC_CUSTOM, "name"); 853 SectionBookkeeping SubSection; 854 startSection(SubSection, wasm::WASM_NAMES_FUNCTION); 855 856 encodeULEB128(TotalFunctions, getStream()); 857 uint32_t Index = 0; 858 for (const WasmImport &Import : Imports) { 859 if (Import.Kind == wasm::WASM_EXTERNAL_FUNCTION) { 860 encodeULEB128(Index, getStream()); 861 writeString(Import.FieldName); 862 ++Index; 863 } 864 } 865 for (const WasmFunction &Func : Functions) { 866 encodeULEB128(Index, getStream()); 867 writeString(Func.Sym->getName()); 868 ++Index; 869 } 870 871 endSection(SubSection); 872 endSection(Section); 873 } 874 875 void WasmObjectWriter::writeCodeRelocSection() { 876 // See: https://github.com/WebAssembly/tool-conventions/blob/master/Linking.md 877 // for descriptions of the reloc sections. 878 879 if (CodeRelocations.empty()) 880 return; 881 882 SectionBookkeeping Section; 883 startSection(Section, wasm::WASM_SEC_CUSTOM, "reloc.CODE"); 884 885 encodeULEB128(wasm::WASM_SEC_CODE, getStream()); 886 encodeULEB128(CodeRelocations.size(), getStream()); 887 888 writeRelocations(CodeRelocations); 889 890 endSection(Section); 891 } 892 893 void WasmObjectWriter::writeDataRelocSection() { 894 // See: https://github.com/WebAssembly/tool-conventions/blob/master/Linking.md 895 // for descriptions of the reloc sections. 896 897 if (DataRelocations.empty()) 898 return; 899 900 SectionBookkeeping Section; 901 startSection(Section, wasm::WASM_SEC_CUSTOM, "reloc.DATA"); 902 903 encodeULEB128(wasm::WASM_SEC_DATA, getStream()); 904 encodeULEB128(DataRelocations.size(), getStream()); 905 906 writeRelocations(DataRelocations); 907 908 endSection(Section); 909 } 910 911 void WasmObjectWriter::writeLinkingMetaDataSection( 912 ArrayRef<WasmDataSegment> Segments, uint32_t DataSize, 913 const SmallVector<std::pair<StringRef, uint32_t>, 4> &SymbolFlags, 914 const SmallVector<std::pair<uint16_t, uint32_t>, 2> &InitFuncs) { 915 SectionBookkeeping Section; 916 startSection(Section, wasm::WASM_SEC_CUSTOM, "linking"); 917 SectionBookkeeping SubSection; 918 919 if (SymbolFlags.size() != 0) { 920 startSection(SubSection, wasm::WASM_SYMBOL_INFO); 921 encodeULEB128(SymbolFlags.size(), getStream()); 922 for (auto Pair: SymbolFlags) { 923 writeString(Pair.first); 924 encodeULEB128(Pair.second, getStream()); 925 } 926 endSection(SubSection); 927 } 928 929 if (DataSize > 0) { 930 startSection(SubSection, wasm::WASM_DATA_SIZE); 931 encodeULEB128(DataSize, getStream()); 932 endSection(SubSection); 933 } 934 935 if (Segments.size()) { 936 startSection(SubSection, wasm::WASM_SEGMENT_INFO); 937 encodeULEB128(Segments.size(), getStream()); 938 for (const WasmDataSegment &Segment : Segments) { 939 writeString(Segment.Name); 940 encodeULEB128(Segment.Alignment, getStream()); 941 encodeULEB128(Segment.Flags, getStream()); 942 } 943 endSection(SubSection); 944 } 945 946 if (!InitFuncs.empty()) { 947 startSection(SubSection, wasm::WASM_INIT_FUNCS); 948 encodeULEB128(InitFuncs.size(), getStream()); 949 for (auto &StartFunc : InitFuncs) { 950 encodeULEB128(StartFunc.first, getStream()); // priority 951 encodeULEB128(StartFunc.second, getStream()); // function index 952 } 953 endSection(SubSection); 954 } 955 956 endSection(Section); 957 } 958 959 uint32_t WasmObjectWriter::getFunctionType(const MCSymbolWasm& Symbol) { 960 assert(Symbol.isFunction()); 961 assert(TypeIndices.count(&Symbol)); 962 return TypeIndices[&Symbol]; 963 } 964 965 uint32_t WasmObjectWriter::registerFunctionType(const MCSymbolWasm& Symbol) { 966 assert(Symbol.isFunction()); 967 968 WasmFunctionType F; 969 const MCSymbolWasm* ResolvedSym = ResolveSymbol(Symbol); 970 F.Returns = ResolvedSym->getReturns(); 971 F.Params = ResolvedSym->getParams(); 972 973 auto Pair = 974 FunctionTypeIndices.insert(std::make_pair(F, FunctionTypes.size())); 975 if (Pair.second) 976 FunctionTypes.push_back(F); 977 TypeIndices[&Symbol] = Pair.first->second; 978 979 DEBUG(dbgs() << "registerFunctionType: " << Symbol << " new:" << Pair.second << "\n"); 980 DEBUG(dbgs() << " -> type index: " << Pair.first->second << "\n"); 981 return Pair.first->second; 982 } 983 984 void WasmObjectWriter::writeObject(MCAssembler &Asm, 985 const MCAsmLayout &Layout) { 986 DEBUG(dbgs() << "WasmObjectWriter::writeObject\n"); 987 MCContext &Ctx = Asm.getContext(); 988 wasm::ValType PtrType = is64Bit() ? wasm::ValType::I64 : wasm::ValType::I32; 989 990 // Collect information from the available symbols. 991 SmallVector<WasmFunction, 4> Functions; 992 SmallVector<uint32_t, 4> TableElems; 993 SmallVector<WasmImport, 4> Imports; 994 SmallVector<WasmExport, 4> Exports; 995 SmallVector<std::pair<StringRef, uint32_t>, 4> SymbolFlags; 996 SmallVector<std::pair<uint16_t, uint32_t>, 2> InitFuncs; 997 SmallPtrSet<const MCSymbolWasm *, 4> IsAddressTaken; 998 unsigned NumFuncImports = 0; 999 SmallVector<WasmDataSegment, 4> DataSegments; 1000 uint32_t DataSize = 0; 1001 1002 // Populate the IsAddressTaken set. 1003 for (const WasmRelocationEntry &RelEntry : CodeRelocations) { 1004 switch (RelEntry.Type) { 1005 case wasm::R_WEBASSEMBLY_TABLE_INDEX_SLEB: 1006 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_SLEB: 1007 IsAddressTaken.insert(RelEntry.Symbol); 1008 break; 1009 default: 1010 break; 1011 } 1012 } 1013 for (const WasmRelocationEntry &RelEntry : DataRelocations) { 1014 switch (RelEntry.Type) { 1015 case wasm::R_WEBASSEMBLY_TABLE_INDEX_I32: 1016 case wasm::R_WEBASSEMBLY_MEMORY_ADDR_I32: 1017 IsAddressTaken.insert(RelEntry.Symbol); 1018 break; 1019 default: 1020 break; 1021 } 1022 } 1023 1024 // In the special .global_variables section, we've encoded global 1025 // variables used by the function. Translate them into the Globals 1026 // list. 1027 MCSectionWasm *GlobalVars = 1028 Ctx.getWasmSection(".global_variables", SectionKind::getMetadata()); 1029 if (!GlobalVars->getFragmentList().empty()) { 1030 if (GlobalVars->getFragmentList().size() != 1) 1031 report_fatal_error("only one .global_variables fragment supported"); 1032 const MCFragment &Frag = *GlobalVars->begin(); 1033 if (Frag.hasInstructions() || Frag.getKind() != MCFragment::FT_Data) 1034 report_fatal_error("only data supported in .global_variables"); 1035 const auto &DataFrag = cast<MCDataFragment>(Frag); 1036 if (!DataFrag.getFixups().empty()) 1037 report_fatal_error("fixups not supported in .global_variables"); 1038 const SmallVectorImpl<char> &Contents = DataFrag.getContents(); 1039 for (const uint8_t *p = (const uint8_t *)Contents.data(), 1040 *end = (const uint8_t *)Contents.data() + Contents.size(); 1041 p != end; ) { 1042 WasmGlobal G; 1043 if (end - p < 3) 1044 report_fatal_error("truncated global variable encoding"); 1045 G.Type = wasm::ValType(int8_t(*p++)); 1046 G.IsMutable = bool(*p++); 1047 G.HasImport = bool(*p++); 1048 if (G.HasImport) { 1049 G.InitialValue = 0; 1050 1051 WasmImport Import; 1052 Import.ModuleName = (const char *)p; 1053 const uint8_t *nul = (const uint8_t *)memchr(p, '\0', end - p); 1054 if (!nul) 1055 report_fatal_error("global module name must be nul-terminated"); 1056 p = nul + 1; 1057 nul = (const uint8_t *)memchr(p, '\0', end - p); 1058 if (!nul) 1059 report_fatal_error("global base name must be nul-terminated"); 1060 Import.FieldName = (const char *)p; 1061 p = nul + 1; 1062 1063 Import.Kind = wasm::WASM_EXTERNAL_GLOBAL; 1064 Import.Type = int32_t(G.Type); 1065 1066 G.ImportIndex = NumGlobalImports; 1067 ++NumGlobalImports; 1068 1069 Imports.push_back(Import); 1070 } else { 1071 unsigned n; 1072 G.InitialValue = decodeSLEB128(p, &n); 1073 G.ImportIndex = 0; 1074 if ((ptrdiff_t)n > end - p) 1075 report_fatal_error("global initial value must be valid SLEB128"); 1076 p += n; 1077 } 1078 Globals.push_back(G); 1079 } 1080 } 1081 1082 // For now, always emit the memory import, since loads and stores are not 1083 // valid without it. In the future, we could perhaps be more clever and omit 1084 // it if there are no loads or stores. 1085 MCSymbolWasm *MemorySym = 1086 cast<MCSymbolWasm>(Ctx.getOrCreateSymbol("__linear_memory")); 1087 WasmImport MemImport; 1088 MemImport.ModuleName = MemorySym->getModuleName(); 1089 MemImport.FieldName = MemorySym->getName(); 1090 MemImport.Kind = wasm::WASM_EXTERNAL_MEMORY; 1091 Imports.push_back(MemImport); 1092 1093 // For now, always emit the table section, since indirect calls are not 1094 // valid without it. In the future, we could perhaps be more clever and omit 1095 // it if there are no indirect calls. 1096 MCSymbolWasm *TableSym = 1097 cast<MCSymbolWasm>(Ctx.getOrCreateSymbol("__indirect_function_table")); 1098 WasmImport TableImport; 1099 TableImport.ModuleName = TableSym->getModuleName(); 1100 TableImport.FieldName = TableSym->getName(); 1101 TableImport.Kind = wasm::WASM_EXTERNAL_TABLE; 1102 TableImport.Type = wasm::WASM_TYPE_ANYFUNC; 1103 Imports.push_back(TableImport); 1104 1105 // Populate FunctionTypeIndices and Imports. 1106 for (const MCSymbol &S : Asm.symbols()) { 1107 const auto &WS = static_cast<const MCSymbolWasm &>(S); 1108 1109 // Register types for all functions, including those with private linkage 1110 // (making them 1111 // because wasm always needs a type signature. 1112 if (WS.isFunction()) 1113 registerFunctionType(WS); 1114 1115 if (WS.isTemporary()) 1116 continue; 1117 1118 // If the symbol is not defined in this translation unit, import it. 1119 if (!WS.isDefined(/*SetUsed=*/false)) { 1120 WasmImport Import; 1121 Import.ModuleName = WS.getModuleName(); 1122 Import.FieldName = WS.getName(); 1123 1124 if (WS.isFunction()) { 1125 Import.Kind = wasm::WASM_EXTERNAL_FUNCTION; 1126 Import.Type = getFunctionType(WS); 1127 SymbolIndices[&WS] = NumFuncImports; 1128 ++NumFuncImports; 1129 } else { 1130 Import.Kind = wasm::WASM_EXTERNAL_GLOBAL; 1131 Import.Type = int32_t(PtrType); 1132 Import.IsMutable = false; 1133 SymbolIndices[&WS] = NumGlobalImports; 1134 1135 // If this global is the stack pointer, make it mutable and remember it 1136 // so that we can emit metadata for it. 1137 if (WS.getName() == "__stack_pointer") 1138 Import.IsMutable = true; 1139 1140 ++NumGlobalImports; 1141 } 1142 1143 Imports.push_back(Import); 1144 } 1145 } 1146 1147 for (MCSection &Sec : Asm) { 1148 auto &Section = static_cast<MCSectionWasm &>(Sec); 1149 if (!Section.isWasmData()) 1150 continue; 1151 1152 // .init_array sections are handled specially elsewhere. 1153 if (cast<MCSectionWasm>(Sec).getSectionName().startswith(".init_array")) 1154 continue; 1155 1156 DataSize = alignTo(DataSize, Section.getAlignment()); 1157 DataSegments.emplace_back(); 1158 WasmDataSegment &Segment = DataSegments.back(); 1159 Segment.Name = Section.getSectionName(); 1160 Segment.Offset = DataSize; 1161 Segment.Section = &Section; 1162 addData(Segment.Data, Section); 1163 Segment.Alignment = Section.getAlignment(); 1164 Segment.Flags = 0; 1165 DataSize += Segment.Data.size(); 1166 Section.setMemoryOffset(Segment.Offset); 1167 } 1168 1169 // Handle regular defined and undefined symbols. 1170 for (const MCSymbol &S : Asm.symbols()) { 1171 // Ignore unnamed temporary symbols, which aren't ever exported, imported, 1172 // or used in relocations. 1173 if (S.isTemporary() && S.getName().empty()) 1174 continue; 1175 1176 const auto &WS = static_cast<const MCSymbolWasm &>(S); 1177 DEBUG(dbgs() << "MCSymbol: '" << S << "'" 1178 << " isDefined=" << S.isDefined() << " isExternal=" 1179 << S.isExternal() << " isTemporary=" << S.isTemporary() 1180 << " isFunction=" << WS.isFunction() 1181 << " isWeak=" << WS.isWeak() 1182 << " isHidden=" << WS.isHidden() 1183 << " isVariable=" << WS.isVariable() << "\n"); 1184 1185 if (WS.isWeak() || WS.isHidden()) { 1186 uint32_t Flags = (WS.isWeak() ? wasm::WASM_SYMBOL_BINDING_WEAK : 0) | 1187 (WS.isHidden() ? wasm::WASM_SYMBOL_VISIBILITY_HIDDEN : 0); 1188 SymbolFlags.emplace_back(WS.getName(), Flags); 1189 } 1190 1191 if (WS.isVariable()) 1192 continue; 1193 1194 unsigned Index; 1195 1196 if (WS.isFunction()) { 1197 if (WS.isDefined(/*SetUsed=*/false)) { 1198 if (WS.getOffset() != 0) 1199 report_fatal_error( 1200 "function sections must contain one function each"); 1201 1202 if (WS.getSize() == 0) 1203 report_fatal_error( 1204 "function symbols must have a size set with .size"); 1205 1206 // A definition. Take the next available index. 1207 Index = NumFuncImports + Functions.size(); 1208 1209 // Prepare the function. 1210 WasmFunction Func; 1211 Func.Type = getFunctionType(WS); 1212 Func.Sym = &WS; 1213 SymbolIndices[&WS] = Index; 1214 Functions.push_back(Func); 1215 } else { 1216 // An import; the index was assigned above. 1217 Index = SymbolIndices.find(&WS)->second; 1218 } 1219 1220 DEBUG(dbgs() << " -> function index: " << Index << "\n"); 1221 1222 // If needed, prepare the function to be called indirectly. 1223 if (IsAddressTaken.count(&WS) != 0) { 1224 IndirectSymbolIndices[&WS] = TableElems.size(); 1225 DEBUG(dbgs() << " -> adding to table: " << TableElems.size() << "\n"); 1226 TableElems.push_back(Index); 1227 } 1228 } else { 1229 if (WS.isTemporary() && !WS.getSize()) 1230 continue; 1231 1232 if (!WS.isDefined(/*SetUsed=*/false)) 1233 continue; 1234 1235 if (!WS.getSize()) 1236 report_fatal_error("data symbols must have a size set with .size: " + 1237 WS.getName()); 1238 1239 int64_t Size = 0; 1240 if (!WS.getSize()->evaluateAsAbsolute(Size, Layout)) 1241 report_fatal_error(".size expression must be evaluatable"); 1242 1243 // For each global, prepare a corresponding wasm global holding its 1244 // address. For externals these will also be named exports. 1245 Index = NumGlobalImports + Globals.size(); 1246 auto &DataSection = static_cast<MCSectionWasm &>(WS.getSection()); 1247 1248 WasmGlobal Global; 1249 Global.Type = PtrType; 1250 Global.IsMutable = false; 1251 Global.HasImport = false; 1252 Global.InitialValue = DataSection.getMemoryOffset() + Layout.getSymbolOffset(WS); 1253 Global.ImportIndex = 0; 1254 SymbolIndices[&WS] = Index; 1255 DEBUG(dbgs() << " -> global index: " << Index << "\n"); 1256 Globals.push_back(Global); 1257 } 1258 1259 // If the symbol is visible outside this translation unit, export it. 1260 if (WS.isDefined(/*SetUsed=*/false)) { 1261 WasmExport Export; 1262 Export.FieldName = WS.getName(); 1263 Export.Index = Index; 1264 if (WS.isFunction()) 1265 Export.Kind = wasm::WASM_EXTERNAL_FUNCTION; 1266 else 1267 Export.Kind = wasm::WASM_EXTERNAL_GLOBAL; 1268 DEBUG(dbgs() << " -> export " << Exports.size() << "\n"); 1269 Exports.push_back(Export); 1270 if (!WS.isExternal()) 1271 SymbolFlags.emplace_back(WS.getName(), wasm::WASM_SYMBOL_BINDING_LOCAL); 1272 } 1273 } 1274 1275 // Handle weak aliases. We need to process these in a separate pass because 1276 // we need to have processed the target of the alias before the alias itself 1277 // and the symbols are not necessarily ordered in this way. 1278 for (const MCSymbol &S : Asm.symbols()) { 1279 if (!S.isVariable()) 1280 continue; 1281 1282 assert(S.isDefined(/*SetUsed=*/false)); 1283 1284 // Find the target symbol of this weak alias and export that index 1285 const auto &WS = static_cast<const MCSymbolWasm &>(S); 1286 const MCSymbolWasm *ResolvedSym = ResolveSymbol(WS); 1287 DEBUG(dbgs() << WS.getName() << ": weak alias of '" << *ResolvedSym << "'\n"); 1288 assert(SymbolIndices.count(ResolvedSym) > 0); 1289 uint32_t Index = SymbolIndices.find(ResolvedSym)->second; 1290 DEBUG(dbgs() << " -> index:" << Index << "\n"); 1291 1292 SymbolIndices[&WS] = Index; 1293 WasmExport Export; 1294 Export.FieldName = WS.getName(); 1295 Export.Index = Index; 1296 if (WS.isFunction()) 1297 Export.Kind = wasm::WASM_EXTERNAL_FUNCTION; 1298 else 1299 Export.Kind = wasm::WASM_EXTERNAL_GLOBAL; 1300 DEBUG(dbgs() << " -> export " << Exports.size() << "\n"); 1301 Exports.push_back(Export); 1302 1303 if (!WS.isExternal()) 1304 SymbolFlags.emplace_back(WS.getName(), wasm::WASM_SYMBOL_BINDING_LOCAL); 1305 } 1306 1307 // Add types for indirect function calls. 1308 for (const WasmRelocationEntry &Fixup : CodeRelocations) { 1309 if (Fixup.Type != wasm::R_WEBASSEMBLY_TYPE_INDEX_LEB) 1310 continue; 1311 1312 registerFunctionType(*Fixup.Symbol); 1313 } 1314 1315 // Translate .init_array section contents into start functions. 1316 for (const MCSection &S : Asm) { 1317 const auto &WS = static_cast<const MCSectionWasm &>(S); 1318 if (WS.getSectionName().startswith(".fini_array")) 1319 report_fatal_error(".fini_array sections are unsupported"); 1320 if (!WS.getSectionName().startswith(".init_array")) 1321 continue; 1322 if (WS.getFragmentList().empty()) 1323 continue; 1324 if (WS.getFragmentList().size() != 2) 1325 report_fatal_error("only one .init_array section fragment supported"); 1326 const MCFragment &AlignFrag = *WS.begin(); 1327 if (AlignFrag.getKind() != MCFragment::FT_Align) 1328 report_fatal_error(".init_array section should be aligned"); 1329 if (cast<MCAlignFragment>(AlignFrag).getAlignment() != (is64Bit() ? 8 : 4)) 1330 report_fatal_error(".init_array section should be aligned for pointers"); 1331 const MCFragment &Frag = *std::next(WS.begin()); 1332 if (Frag.hasInstructions() || Frag.getKind() != MCFragment::FT_Data) 1333 report_fatal_error("only data supported in .init_array section"); 1334 uint16_t Priority = UINT16_MAX; 1335 if (WS.getSectionName().size() != 11) { 1336 if (WS.getSectionName()[11] != '.') 1337 report_fatal_error(".init_array section priority should start with '.'"); 1338 if (WS.getSectionName().substr(12).getAsInteger(10, Priority)) 1339 report_fatal_error("invalid .init_array section priority"); 1340 } 1341 const auto &DataFrag = cast<MCDataFragment>(Frag); 1342 const SmallVectorImpl<char> &Contents = DataFrag.getContents(); 1343 for (const uint8_t *p = (const uint8_t *)Contents.data(), 1344 *end = (const uint8_t *)Contents.data() + Contents.size(); 1345 p != end; ++p) { 1346 if (*p != 0) 1347 report_fatal_error("non-symbolic data in .init_array section"); 1348 } 1349 for (const MCFixup &Fixup : DataFrag.getFixups()) { 1350 assert(Fixup.getKind() == MCFixup::getKindForSize(is64Bit() ? 8 : 4, false)); 1351 const MCExpr *Expr = Fixup.getValue(); 1352 auto *Sym = dyn_cast<MCSymbolRefExpr>(Expr); 1353 if (!Sym) 1354 report_fatal_error("fixups in .init_array should be symbol references"); 1355 if (Sym->getKind() != MCSymbolRefExpr::VK_WebAssembly_FUNCTION) 1356 report_fatal_error("symbols in .init_array should be for functions"); 1357 auto I = SymbolIndices.find(cast<MCSymbolWasm>(&Sym->getSymbol())); 1358 if (I == SymbolIndices.end()) 1359 report_fatal_error("symbols in .init_array should be defined"); 1360 uint32_t Index = I->second; 1361 InitFuncs.push_back(std::make_pair(Priority, Index)); 1362 } 1363 } 1364 1365 // Write out the Wasm header. 1366 writeHeader(Asm); 1367 1368 writeTypeSection(FunctionTypes); 1369 writeImportSection(Imports, DataSize, TableElems.size()); 1370 writeFunctionSection(Functions); 1371 // Skip the "table" section; we import the table instead. 1372 // Skip the "memory" section; we import the memory instead. 1373 writeGlobalSection(); 1374 writeExportSection(Exports); 1375 writeElemSection(TableElems); 1376 writeCodeSection(Asm, Layout, Functions); 1377 writeDataSection(DataSegments); 1378 writeNameSection(Functions, Imports, NumFuncImports); 1379 writeCodeRelocSection(); 1380 writeDataRelocSection(); 1381 writeLinkingMetaDataSection(DataSegments, DataSize, SymbolFlags, 1382 InitFuncs); 1383 1384 // TODO: Translate the .comment section to the output. 1385 // TODO: Translate debug sections to the output. 1386 } 1387 1388 std::unique_ptr<MCObjectWriter> 1389 llvm::createWasmObjectWriter(std::unique_ptr<MCWasmObjectTargetWriter> MOTW, 1390 raw_pwrite_stream &OS) { 1391 // FIXME: Can't use make_unique<WasmObjectWriter>(...) as WasmObjectWriter's 1392 // destructor is private. Is that necessary? 1393 return std::unique_ptr<MCObjectWriter>( 1394 new WasmObjectWriter(std::move(MOTW), OS)); 1395 } 1396