Index: wasm/CMakeLists.txt =================================================================== --- wasm/CMakeLists.txt +++ wasm/CMakeLists.txt @@ -13,8 +13,10 @@ LTO.cpp MarkLive.cpp OutputSections.cpp + Relocations.cpp SymbolTable.cpp Symbols.cpp + SyntheticSections.cpp Writer.cpp WriterUtils.cpp @@ -35,4 +37,4 @@ DEPENDS WasmOptionsTableGen ${tablegen_deps} - ) \ No newline at end of file + ) Index: wasm/InputChunks.h =================================================================== --- wasm/InputChunks.h +++ wasm/InputChunks.h @@ -31,6 +31,7 @@ class ObjFile; class OutputSegment; +class OutputSection; class InputChunk { public: @@ -207,6 +208,8 @@ StringRef getDebugName() const override { return StringRef(); } uint32_t getComdat() const override { return UINT32_MAX; } + OutputSection *OutputSec = nullptr; + protected: ArrayRef data() const override { return Section.Content; } Index: wasm/InputFiles.cpp =================================================================== --- wasm/InputFiles.cpp +++ wasm/InputFiles.cpp @@ -75,7 +75,10 @@ assert(TypeIsUsed[Reloc.Index]); return TypeMap[Reloc.Index]; } - return Symbols[Reloc.Index]->getOutputSymbolIndex(); + const Symbol *Sym = Symbols[Reloc.Index]; + if (auto *SS = dyn_cast(Sym)) + Sym = SS->getOutputSectionSymbol(); + return Sym->getOutputSymbolIndex(); } // Relocations can contain addend for combined sections. This function takes a @@ -395,7 +398,7 @@ case WASM_SYMBOL_TYPE_SECTION: { InputSection *Section = CustomSectionsByIndex[Sym.Info.ElementIndex]; assert(Sym.isBindingLocal()); - return make(Name, Flags, Section, this); + return make(Flags, Section, this); } case WASM_SYMBOL_TYPE_EVENT: { InputEvent *Event = Index: wasm/OutputSections.h =================================================================== --- wasm/OutputSections.h +++ wasm/OutputSections.h @@ -38,6 +38,7 @@ Offset = NewOffset; } void createHeader(size_t BodySize); + virtual bool isNeeded() const { return true; } virtual size_t getSize() const = 0; virtual void writeTo(uint8_t *Buf) = 0; virtual void finalizeContents() = 0; @@ -46,54 +47,24 @@ std::string Header; uint32_t Type; + uint32_t SectionIndex = UINT32_MAX; std::string Name; + OutputSectionSymbol *SectionSym = nullptr; protected: size_t Offset = 0; }; -class SyntheticSection : public OutputSection { -public: - SyntheticSection(uint32_t Type, std::string Name = "") - : OutputSection(Type, Name), BodyOutputStream(Body) { - if (!Name.empty()) - writeStr(BodyOutputStream, Name, "section name"); - } - - void writeTo(uint8_t *Buf) override { - assert(Offset); - log("writing " + toString(*this)); - memcpy(Buf + Offset, Header.data(), Header.size()); - memcpy(Buf + Offset + Header.size(), Body.data(), Body.size()); - } - - size_t getSize() const override { return Header.size() + Body.size(); } - - virtual void writeBody() {} - - void finalizeContents() override { - writeBody(); - BodyOutputStream.flush(); - createHeader(Body.size()); - } - - raw_ostream &getStream() { return BodyOutputStream; } - - std::string Body; - -protected: - llvm::raw_string_ostream BodyOutputStream; -}; - class CodeSection : public OutputSection { public: explicit CodeSection(ArrayRef Functions) : OutputSection(llvm::wasm::WASM_SEC_CODE), Functions(Functions) {} - size_t getSize() const override { assert(BodySize); return Header.size() + BodySize; } + size_t getSize() const override { return Header.size() + BodySize; } void writeTo(uint8_t *Buf) override; uint32_t numRelocations() const override; void writeRelocations(raw_ostream &OS) const override; + bool isNeeded() const override { return Functions.size() > 0; } void finalizeContents() override; protected: @@ -111,6 +82,7 @@ void writeTo(uint8_t *Buf) override; uint32_t numRelocations() const override; void writeRelocations(raw_ostream &OS) const override; + bool isNeeded() const override { return Segments.size() > 0; } void finalizeContents() override; protected: @@ -145,19 +117,6 @@ std::string NameData; }; -class RelocSection : public SyntheticSection { -public: - RelocSection(StringRef Name, OutputSection *Sec, uint32_t SectionIndex) - : SyntheticSection(llvm::wasm::WASM_SEC_CUSTOM, Name), Sec(Sec), - SectionIndex(SectionIndex) {} - void writeBody() override; - -protected: - OutputSection* Sec; - uint32_t SectionIndex; -}; - - } // namespace wasm } // namespace lld Index: wasm/OutputSections.cpp =================================================================== --- wasm/OutputSections.cpp +++ wasm/OutputSections.cpp @@ -206,6 +206,7 @@ for (InputSection *Section : InputSections) { Section->OutputOffset = PayloadSize; + Section->OutputSec = this; PayloadSize += Section->getSize(); } @@ -241,9 +242,3 @@ for (const InputSection *S : InputSections) S->writeRelocations(OS); } - -void RelocSection::writeBody() { - writeUleb128(BodyOutputStream, SectionIndex, "reloc section"); - writeUleb128(BodyOutputStream, Sec->numRelocations(), "reloc count"); - Sec->writeRelocations(BodyOutputStream); -} Index: wasm/Relocations.h =================================================================== --- wasm/Relocations.h +++ wasm/Relocations.h @@ -0,0 +1,22 @@ +//===- Relocations.h -------------------------------------------*- C++ -*-===// +// +// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. +// See https://llvm.org/LICENSE.txt for license information. +// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception +// +//===----------------------------------------------------------------------===// + +#ifndef LLD_WASM_RELOCATIONS_H +#define LLD_WASM_RELOCATIONS_H + +namespace lld { +namespace wasm { + +class InputChunk; + +void scanRelocations(InputChunk *Chunk); + +} // namespace wasm +} // namespace lld + +#endif Index: wasm/Relocations.cpp =================================================================== --- wasm/Relocations.cpp +++ wasm/Relocations.cpp @@ -0,0 +1,86 @@ +//===- Relocations.cpp ----------------------------------------------------===// +// +// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. +// See https://llvm.org/LICENSE.txt for license information. +// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception +// +//===----------------------------------------------------------------------===// + +#include "Relocations.h" + +#include "InputChunks.h" +#include "SyntheticSections.h" + +using namespace llvm; +using namespace llvm::wasm; + +using namespace lld; +using namespace lld::wasm; + +static bool requiresGOTAccess(const Symbol *Sym) { + return Config->Pic && !Sym->isHidden() && !Sym->isLocal(); +} + +void lld::wasm::scanRelocations(InputChunk *Chunk) { + if (!Chunk->Live) + return; + ObjFile *File = Chunk->File; + ArrayRef Types = File->getWasmObj()->types(); + for (const WasmRelocation &Reloc : Chunk->getRelocations()) { + if (Reloc.Type == R_WASM_TYPE_INDEX_LEB) { + // Mark target type as live + File->TypeMap[Reloc.Index] = + Out.TypeSec->registerType(Types[Reloc.Index]); + File->TypeIsUsed[Reloc.Index] = true; + continue; + } + + // Other relocation types all have a corresponding symbol + Symbol *Sym = File->getSymbols()[Reloc.Index]; + + switch (Reloc.Type) { + case R_WASM_TABLE_INDEX_I32: + case R_WASM_TABLE_INDEX_SLEB: + case R_WASM_TABLE_INDEX_REL_SLEB: + if (requiresGOTAccess(Sym)) + break; + Out.ElemSec->addEntry(cast(Sym)); + break; + case R_WASM_GLOBAL_INDEX_LEB: + if (!isa(Sym)) + Out.ImportSec->addGOTEntry(Sym); + break; + case R_WASM_MEMORY_ADDR_SLEB: + case R_WASM_MEMORY_ADDR_LEB: + case R_WASM_MEMORY_ADDR_REL_SLEB: + if (!Config->Relocatable && Sym->isUndefined() && !Sym->isWeak()) { + error(toString(File) + ": cannot resolve relocation of type " + + relocTypeToString(Reloc.Type) + + " against undefined (non-weak) data symbol: " + toString(*Sym)); + } + break; + } + + if (Config->Pic) { + switch (Reloc.Type) { + case R_WASM_TABLE_INDEX_SLEB: + case R_WASM_MEMORY_ADDR_SLEB: + case R_WASM_MEMORY_ADDR_LEB: + // Certain relocation types can't be used when building PIC output, + // since they would require absolute symbol addresses at link time. + error(toString(File) + ": relocation " + relocTypeToString(Reloc.Type) + + " cannot be used againt symbol " + toString(*Sym) + + "; recompile with -fPIC"); + break; + case R_WASM_TABLE_INDEX_I32: + case R_WASM_MEMORY_ADDR_I32: + // These relocation types are only present in the data section and + // will be converted into code by `generateRelocationCode`. This code + // requires the symbols to have GOT entires. + if (requiresGOTAccess(Sym)) + Out.ImportSec->addGOTEntry(Sym); + break; + } + } + } +} Index: wasm/Symbols.h =================================================================== --- wasm/Symbols.h +++ wasm/Symbols.h @@ -17,6 +17,14 @@ namespace lld { namespace wasm { +// Shared string constants + +// The default module name to use for symbol imports. +extern const char *DefaultModule; + +// The name under which to import or export the wasm table. +extern const char *FunctionTableName; + using llvm::wasm::WasmSymbolType; class InputFile; @@ -26,6 +34,7 @@ class InputGlobal; class InputEvent; class InputSection; +class OutputSection; #define INVALID_INDEX UINT32_MAX @@ -38,6 +47,7 @@ DefinedGlobalKind, DefinedEventKind, SectionKind, + OutputSectionKind, UndefinedFunctionKind, UndefinedDataKind, UndefinedGlobalKind, @@ -46,11 +56,7 @@ Kind kind() const { return SymbolKind; } - bool isDefined() const { - return SymbolKind == DefinedFunctionKind || SymbolKind == DefinedDataKind || - SymbolKind == DefinedGlobalKind || SymbolKind == DefinedEventKind || - SymbolKind == SectionKind; - } + bool isDefined() const { return !isLazy() && !isUndefined(); } bool isUndefined() const { return SymbolKind == UndefinedFunctionKind || @@ -195,21 +201,33 @@ StringRef ImportModule; }; +// Section symbols for output sections are different from those for input +// section. These are generated by the linker and point the OutputSection +// rather than an InputSection. +class OutputSectionSymbol : public Symbol { +public: + OutputSectionSymbol(const OutputSection *S) + : Symbol("", OutputSectionKind, llvm::wasm::WASM_SYMBOL_BINDING_LOCAL, + nullptr), + Section(S) {} + + static bool classof(const Symbol *S) { + return S->kind() == OutputSectionKind; + } + + const OutputSection *Section; +}; + class SectionSymbol : public Symbol { public: + SectionSymbol(uint32_t Flags, const InputSection *S, InputFile *F = nullptr) + : Symbol("", SectionKind, Flags, F), Section(S) {} + static bool classof(const Symbol *S) { return S->kind() == SectionKind; } - SectionSymbol(StringRef Name, uint32_t Flags, const InputSection *S, - InputFile *F = nullptr) - : Symbol(Name, SectionKind, Flags, F), Section(S) {} + const OutputSectionSymbol *getOutputSectionSymbol() const; const InputSection *Section; - - uint32_t getOutputSectionIndex() const; - void setOutputSectionIndex(uint32_t Index); - -protected: - uint32_t OutputSectionIndex = INVALID_INDEX; }; class DataSymbol : public Symbol { Index: wasm/Symbols.cpp =================================================================== --- wasm/Symbols.cpp +++ wasm/Symbols.cpp @@ -12,6 +12,7 @@ #include "InputEvent.h" #include "InputFiles.h" #include "InputGlobal.h" +#include "OutputSections.h" #include "OutputSegment.h" #include "lld/Common/ErrorHandler.h" #include "lld/Common/Strings.h" @@ -41,7 +42,7 @@ return WASM_SYMBOL_TYPE_GLOBAL; if (isa(this)) return WASM_SYMBOL_TYPE_EVENT; - if (isa(this)) + if (isa(this) || isa(this)) return WASM_SYMBOL_TYPE_SECTION; llvm_unreachable("invalid symbol kind"); } @@ -262,17 +263,9 @@ Event ? &Event->Signature : nullptr), Event(Event) {} -uint32_t SectionSymbol::getOutputSectionIndex() const { - LLVM_DEBUG(dbgs() << "getOutputSectionIndex: " << getName() << "\n"); - assert(OutputSectionIndex != INVALID_INDEX); - return OutputSectionIndex; -} - -void SectionSymbol::setOutputSectionIndex(uint32_t Index) { - LLVM_DEBUG(dbgs() << "setOutputSectionIndex: " << getName() << " -> " << Index - << "\n"); - assert(Index != INVALID_INDEX); - OutputSectionIndex = Index; +const OutputSectionSymbol *SectionSymbol::getOutputSectionSymbol() const { + assert(Section->OutputSec && Section->OutputSec->SectionSym); + return Section->OutputSec->SectionSym; } void LazySymbol::fetch() { cast(File)->addMember(&ArchiveSymbol); } @@ -308,6 +301,8 @@ return "LazyKind"; case wasm::Symbol::SectionKind: return "SectionKind"; + case wasm::Symbol::OutputSectionKind: + return "OutputSectionKind"; } llvm_unreachable("invalid symbol kind"); } @@ -324,3 +319,6 @@ message(toString(Sym->getFile()) + S + Sym->getName()); } + +const char *lld::wasm::DefaultModule = "env"; +const char *lld::wasm::FunctionTableName = "__indirect_function_table"; Index: wasm/SyntheticSections.h =================================================================== --- wasm/SyntheticSections.h +++ wasm/SyntheticSections.h @@ -0,0 +1,323 @@ +//===- SyntheticSection.h ---------------------------------------*- C++ -*-===// +// +// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. +// See https://llvm.org/LICENSE.txt for license information. +// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception +// +//===----------------------------------------------------------------------===// +// +// Synthetic sections represent chunks of linker-created data. If you +// need to create a chunk of data that to be included in some section +// in the result, you probably want to create that as a synthetic section. +// +//===----------------------------------------------------------------------===// + +#ifndef LLD_WASM_SYNTHETIC_SECTIONS_H +#define LLD_WASM_SYNTHETIC_SECTIONS_H + +#include "OutputSections.h" + +#include "llvm/ADT/SmallSet.h" +#include "llvm/ADT/StringMap.h" +#include "llvm/Object/WasmTraits.h" + +#define DEBUG_TYPE "lld" + +namespace lld { +namespace wasm { + +// An init entry to be written to either the synthetic init func or the +// linking metadata. +struct WasmInitEntry { + const FunctionSymbol *Sym; + uint32_t Priority; +}; + +class SyntheticSection : public OutputSection { +public: + SyntheticSection(uint32_t Type, std::string Name = "") + : OutputSection(Type, Name), BodyOutputStream(Body) { + if (!Name.empty()) + writeStr(BodyOutputStream, Name, "section name"); + } + + void writeTo(uint8_t *Buf) override { + assert(Offset); + log("writing " + toString(*this)); + memcpy(Buf + Offset, Header.data(), Header.size()); + memcpy(Buf + Offset + Header.size(), Body.data(), Body.size()); + } + + size_t getSize() const override { return Header.size() + Body.size(); } + + virtual void writeBody() {} + + void finalizeContents() override { + writeBody(); + BodyOutputStream.flush(); + createHeader(Body.size()); + } + + raw_ostream &getStream() { return BodyOutputStream; } + + std::string Body; + +protected: + llvm::raw_string_ostream BodyOutputStream; +}; + +// Create the custom "dylink" section containing information for the dynamic +// linker. +// See +// https://github.com/WebAssembly/tool-conventions/blob/master/DynamicLinking.md +class DylinkSection : public SyntheticSection { +public: + DylinkSection() : SyntheticSection(llvm::wasm::WASM_SEC_CUSTOM, "dylink") {} + bool isNeeded() const override { return Config->Pic; } + void writeBody() override; + + uint32_t MemAlign = 0; + uint32_t MemSize = 0; +}; + +class TypeSection : public SyntheticSection { +public: + TypeSection() : SyntheticSection(llvm::wasm::WASM_SEC_TYPE) {} + + bool isNeeded() const override { return Types.size() > 0; }; + void writeBody() override; + uint32_t registerType(const WasmSignature &Sig); + uint32_t lookupType(const WasmSignature &Sig); + +protected: + std::vector Types; + llvm::DenseMap TypeIndices; +}; + +class ImportSection : public SyntheticSection { +public: + ImportSection() : SyntheticSection(llvm::wasm::WASM_SEC_IMPORT) {} + bool isNeeded() const override { return numImports() > 0; } + void writeBody() override; + void addImport(Symbol *Sym); + void addGOTEntry(Symbol *Sym); + uint32_t numImports() const; + + unsigned NumImportedGlobals = 0; + unsigned NumImportedFunctions = 0; + unsigned NumImportedEvents = 0; + std::vector ImportedSymbols; + +protected: + std::vector GOTSymbols; +}; + +class FunctionSection : public SyntheticSection { +public: + FunctionSection() : SyntheticSection(llvm::wasm::WASM_SEC_FUNCTION) {} + + bool isNeeded() const override { return InputFunctions.size() > 0; }; + void writeBody() override; + void addFunction(InputFunction *Func); + + std::vector InputFunctions; + +protected: +}; + +class MemorySection : public SyntheticSection { +public: + MemorySection() : SyntheticSection(llvm::wasm::WASM_SEC_MEMORY) {} + + bool isNeeded() const override { return !Config->ImportMemory; } + void writeBody() override; + + uint32_t NumMemoryPages = 0; + uint32_t MaxMemoryPages = 0; +}; + +class TableSection : public SyntheticSection { +public: + TableSection() : SyntheticSection(llvm::wasm::WASM_SEC_TABLE) {} + + bool isNeeded() const override { + // Always output a table section (or table import), even if there are no + // indirect calls. There are two reasons for this: + // 1. For executables it is useful to have an empty table slot at 0 + // which can be filled with a null function call handler. + // 2. If we don't do this, any program that contains a call_indirect but + // no address-taken function will fail at validation time since it is + // a validation error to include a call_indirect instruction if there + // is not table. + return !Config->ImportTable; + } + + void writeBody() override; +}; + +class GlobalSection : public SyntheticSection { +public: + GlobalSection() : SyntheticSection(llvm::wasm::WASM_SEC_GLOBAL) {} + uint32_t numGlobals() const { + return InputGlobals.size() + DefinedFakeGlobals.size(); + } + bool isNeeded() const override { return numGlobals() > 0; } + void writeBody() override; + void addGlobal(InputGlobal *Global); + + std::vector DefinedFakeGlobals; + std::vector InputGlobals; +}; + +// The event section contains a list of declared wasm events associated with the +// module. Currently the only supported event kind is exceptions. A single event +// entry represents a single event with an event tag. All C++ exceptions are +// represented by a single event. An event entry in this section contains +// information on what kind of event it is (e.g. exception) and the type of +// values contained in a single event object. (In wasm, an event can contain +// multiple values of primitive types. But for C++ exceptions, we just throw a +// pointer which is an i32 value (for wasm32 architecture), so the signature of +// C++ exception is (i32)->(void), because all event types are assumed to have +// void return type to share WasmSignature with functions.) +class EventSection : public SyntheticSection { +public: + EventSection() : SyntheticSection(llvm::wasm::WASM_SEC_EVENT) {} + void writeBody() override; + bool isNeeded() const override { return InputEvents.size() > 0; } + void addEvent(InputEvent *Event); + + std::vector InputEvents; +}; + +class ExportSection : public SyntheticSection { +public: + ExportSection() : SyntheticSection(llvm::wasm::WASM_SEC_EXPORT) {} + bool isNeeded() const override { return Exports.size() > 0; } + void writeBody() override; + + std::vector Exports; +}; + +class ElemSection : public SyntheticSection { +public: + ElemSection(uint32_t Offset) + : SyntheticSection(llvm::wasm::WASM_SEC_ELEM), ElemOffset(Offset) {} + bool isNeeded() const override { return IndirectFunctions.size() > 0; }; + void writeBody() override; + void addEntry(FunctionSymbol *Sym); + uint32_t numEntries() const { return IndirectFunctions.size(); } + uint32_t ElemOffset; + +protected: + std::vector IndirectFunctions; +}; + +class DataCountSection : public SyntheticSection { +public: + DataCountSection(uint32_t NumSegments) + : SyntheticSection(llvm::wasm::WASM_SEC_DATACOUNT), + NumSegments(NumSegments) {} + bool isNeeded() const override; + void writeBody() override; + +protected: + uint32_t NumSegments; +}; + +// Create the custom "linking" section containing linker metadata. +// This is only created when relocatable output is requested. +class LinkingSection : public SyntheticSection { +public: + LinkingSection(const std::vector &InitFunctions, + const std::vector &DataSegments) + : SyntheticSection(llvm::wasm::WASM_SEC_CUSTOM, "linking"), + InitFunctions(InitFunctions), DataSegments(DataSegments) {} + bool isNeeded() const override { return Config->Relocatable; } + void writeBody() override; + void addToSymtab(Symbol *Sym); + +protected: + std::vector SymtabEntries; + llvm::StringMap SectionSymbolIndices; + const std::vector &InitFunctions; + const std::vector &DataSegments; +}; + +// Create the custom "name" section containing debug symbol names. +class NameSection : public SyntheticSection { +public: + NameSection() : SyntheticSection(llvm::wasm::WASM_SEC_CUSTOM, "name") {} + bool isNeeded() const override { + return !Config->StripDebug && !Config->StripAll && numNames() > 0; + } + void writeBody() override; + unsigned numNames() const; +}; + +class ProducersSection : public SyntheticSection { +public: + ProducersSection() + : SyntheticSection(llvm::wasm::WASM_SEC_CUSTOM, "producers") {} + bool isNeeded() const override { + return !Config->StripAll && fieldCount() > 0; + } + void writeBody() override; + void addInfo(const llvm::wasm::WasmProducerInfo &Info); + +protected: + int fieldCount() const { + return int(!Languages.empty()) + int(!Tools.empty()) + int(!SDKs.empty()); + } + SmallVector, 8> Languages; + SmallVector, 8> Tools; + SmallVector, 8> SDKs; +}; + +class TargetFeaturesSection : public SyntheticSection { +public: + TargetFeaturesSection() + : SyntheticSection(llvm::wasm::WASM_SEC_CUSTOM, "target_features") {} + bool isNeeded() const override { + return !Config->StripAll && Features.size() > 0; + } + void writeBody() override; + + llvm::SmallSet Features; +}; + +class RelocSection : public SyntheticSection { +public: + RelocSection(StringRef Name, OutputSection *Sec) + : SyntheticSection(llvm::wasm::WASM_SEC_CUSTOM, Name), Sec(Sec) {} + void writeBody() override; + bool isNeeded() const override { return Sec->numRelocations() > 0; }; + +protected: + OutputSection *Sec; +}; + +// Linker generated output sections +struct OutStruct { + DylinkSection *DylinkSec; + TypeSection *TypeSec; + FunctionSection *FunctionSec; + ImportSection *ImportSec; + TableSection *TableSec; + MemorySection *MemorySec; + GlobalSection *GlobalSec; + EventSection *EventSec; + ExportSection *ExportSec; + ElemSection *ElemSec; + DataCountSection *DataCountSec; + LinkingSection *LinkingSec; + NameSection *NameSec; + ProducersSection *ProducersSec; + TargetFeaturesSection *TargetFeaturesSec; +}; + +extern OutStruct Out; + +} // namespace wasm +} // namespace lld + +#endif Index: wasm/SyntheticSections.cpp =================================================================== --- wasm/SyntheticSections.cpp +++ wasm/SyntheticSections.cpp @@ -0,0 +1,543 @@ +//===- SyntheticSections.cpp ----------------------------------------------===// +// +// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. +// See https://llvm.org/LICENSE.txt for license information. +// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception +// +//===----------------------------------------------------------------------===// +// +// This file contains linker-synthesized sections. +// +//===----------------------------------------------------------------------===// + +#include "SyntheticSections.h" + +#include "InputChunks.h" +#include "InputEvent.h" +#include "InputGlobal.h" +#include "OutputSegment.h" +#include "SymbolTable.h" +#include "llvm/Support/Path.h" + +using namespace llvm; +using namespace llvm::wasm; + +using namespace lld; +using namespace lld::wasm; + +OutStruct lld::wasm::Out; + +namespace { + +// Some synthetic sections (e.g. "name" and "linking") have subsections. +// Just like the synthetic sections themselves these need to be created before +// they can be written out (since they are preceded by their length). This +// class is used to create subsections and then write them into the stream +// of the parent section. +class SubSection { +public: + explicit SubSection(uint32_t Type) : Type(Type) {} + + void writeTo(raw_ostream &To) { + OS.flush(); + writeUleb128(To, Type, "subsection type"); + writeUleb128(To, Body.size(), "subsection size"); + To.write(Body.data(), Body.size()); + } + +private: + uint32_t Type; + std::string Body; + +public: + raw_string_ostream OS{Body}; +}; + +} // namespace + +void DylinkSection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + writeUleb128(OS, MemSize, "MemSize"); + writeUleb128(OS, MemAlign, "MemAlign"); + writeUleb128(OS, Out.ElemSec->numEntries(), "TableSize"); + writeUleb128(OS, 0, "TableAlign"); + writeUleb128(OS, Symtab->SharedFiles.size(), "Needed"); + for (auto *SO : Symtab->SharedFiles) + writeStr(OS, llvm::sys::path::filename(SO->getName()), "so name"); +} + +uint32_t TypeSection::registerType(const WasmSignature &Sig) { + auto Pair = TypeIndices.insert(std::make_pair(Sig, Types.size())); + if (Pair.second) { + LLVM_DEBUG(llvm::dbgs() << "type " << toString(Sig) << "\n"); + Types.push_back(&Sig); + } + return Pair.first->second; +} + +uint32_t TypeSection::lookupType(const WasmSignature &Sig) { + auto It = TypeIndices.find(Sig); + if (It == TypeIndices.end()) { + error("type not found: " + toString(Sig)); + return 0; + } + return It->second; +} + +void TypeSection::writeBody() { + writeUleb128(BodyOutputStream, Types.size(), "type count"); + for (const WasmSignature *Sig : Types) + writeSig(BodyOutputStream, *Sig); +} + +uint32_t ImportSection::numImports() const { + uint32_t NumImports = ImportedSymbols.size() + GOTSymbols.size(); + if (Config->ImportMemory) + ++NumImports; + if (Config->ImportTable) + ++NumImports; + return NumImports; +} + +void ImportSection::addGOTEntry(Symbol *Sym) { + if (Sym->hasGOTIndex()) + return; + Sym->setGOTIndex(NumImportedGlobals++); + GOTSymbols.push_back(Sym); +} + +void ImportSection::addImport(Symbol *Sym) { + ImportedSymbols.emplace_back(Sym); + if (auto *F = dyn_cast(Sym)) + F->setFunctionIndex(NumImportedFunctions++); + else if (auto *G = dyn_cast(Sym)) + G->setGlobalIndex(NumImportedGlobals++); + else + cast(Sym)->setEventIndex(NumImportedEvents++); +} + +void ImportSection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + writeUleb128(OS, numImports(), "import count"); + + if (Config->ImportMemory) { + WasmImport Import; + Import.Module = DefaultModule; + Import.Field = "memory"; + Import.Kind = WASM_EXTERNAL_MEMORY; + Import.Memory.Flags = 0; + Import.Memory.Initial = Out.MemorySec->NumMemoryPages; + if (Out.MemorySec->MaxMemoryPages != 0 || Config->SharedMemory) { + Import.Memory.Flags |= WASM_LIMITS_FLAG_HAS_MAX; + Import.Memory.Maximum = Out.MemorySec->MaxMemoryPages; + } + if (Config->SharedMemory) + Import.Memory.Flags |= WASM_LIMITS_FLAG_IS_SHARED; + writeImport(OS, Import); + } + + if (Config->ImportTable) { + uint32_t TableSize = Out.ElemSec->ElemOffset + Out.ElemSec->numEntries(); + WasmImport Import; + Import.Module = DefaultModule; + Import.Field = FunctionTableName; + Import.Kind = WASM_EXTERNAL_TABLE; + Import.Table.ElemType = WASM_TYPE_FUNCREF; + Import.Table.Limits = {0, TableSize, 0}; + writeImport(OS, Import); + } + + for (const Symbol *Sym : ImportedSymbols) { + WasmImport Import; + if (auto *F = dyn_cast(Sym)) { + Import.Field = F->ImportName; + Import.Module = F->ImportModule; + } else if (auto *G = dyn_cast(Sym)) { + Import.Field = G->ImportName; + Import.Module = G->ImportModule; + } else { + Import.Field = Sym->getName(); + Import.Module = DefaultModule; + } + + if (auto *FunctionSym = dyn_cast(Sym)) { + Import.Kind = WASM_EXTERNAL_FUNCTION; + Import.SigIndex = Out.TypeSec->lookupType(*FunctionSym->Signature); + } else if (auto *GlobalSym = dyn_cast(Sym)) { + Import.Kind = WASM_EXTERNAL_GLOBAL; + Import.Global = *GlobalSym->getGlobalType(); + } else { + auto *EventSym = cast(Sym); + Import.Kind = WASM_EXTERNAL_EVENT; + Import.Event.Attribute = EventSym->getEventType()->Attribute; + Import.Event.SigIndex = Out.TypeSec->lookupType(*EventSym->Signature); + } + writeImport(OS, Import); + } + + for (const Symbol *Sym : GOTSymbols) { + WasmImport Import; + Import.Kind = WASM_EXTERNAL_GLOBAL; + Import.Global = {WASM_TYPE_I32, true}; + if (isa(Sym)) + Import.Module = "GOT.mem"; + else + Import.Module = "GOT.func"; + Import.Field = Sym->getName(); + writeImport(OS, Import); + } +} + +void FunctionSection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + writeUleb128(OS, InputFunctions.size(), "function count"); + for (const InputFunction *Func : InputFunctions) + writeUleb128(OS, Out.TypeSec->lookupType(Func->Signature), "sig index"); +} + +void FunctionSection::addFunction(InputFunction *Func) { + if (!Func->Live) + return; + uint32_t FunctionIndex = + Out.ImportSec->NumImportedFunctions + InputFunctions.size(); + InputFunctions.emplace_back(Func); + Func->setFunctionIndex(FunctionIndex); +} + +void TableSection::writeBody() { + uint32_t TableSize = Out.ElemSec->ElemOffset + Out.ElemSec->numEntries(); + + raw_ostream &OS = BodyOutputStream; + writeUleb128(OS, 1, "table count"); + WasmLimits Limits = {WASM_LIMITS_FLAG_HAS_MAX, TableSize, TableSize}; + writeTableType(OS, WasmTable{WASM_TYPE_FUNCREF, Limits}); +} + +void MemorySection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + bool HasMax = MaxMemoryPages != 0 || Config->SharedMemory; + writeUleb128(OS, 1, "memory count"); + unsigned Flags = 0; + if (HasMax) + Flags |= WASM_LIMITS_FLAG_HAS_MAX; + if (Config->SharedMemory) + Flags |= WASM_LIMITS_FLAG_IS_SHARED; + writeUleb128(OS, Flags, "memory limits flags"); + writeUleb128(OS, NumMemoryPages, "initial pages"); + if (HasMax) + writeUleb128(OS, MaxMemoryPages, "max pages"); +} + +void GlobalSection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + writeUleb128(OS, numGlobals(), "global count"); + for (const InputGlobal *G : InputGlobals) + writeGlobal(OS, G->Global); + for (const DefinedData *Sym : DefinedFakeGlobals) { + WasmGlobal Global; + Global.Type = {WASM_TYPE_I32, false}; + Global.InitExpr.Opcode = WASM_OPCODE_I32_CONST; + Global.InitExpr.Value.Int32 = Sym->getVirtualAddress(); + writeGlobal(OS, Global); + } +} + +void GlobalSection::addGlobal(InputGlobal *Global) { + if (!Global->Live) + return; + uint32_t GlobalIndex = + Out.ImportSec->NumImportedGlobals + InputGlobals.size(); + LLVM_DEBUG(dbgs() << "addGlobal: " << GlobalIndex << "\n"); + Global->setGlobalIndex(GlobalIndex); + Out.GlobalSec->InputGlobals.push_back(Global); +} + +void EventSection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + writeUleb128(OS, InputEvents.size(), "event count"); + for (InputEvent *E : InputEvents) { + E->Event.Type.SigIndex = Out.TypeSec->lookupType(E->Signature); + writeEvent(OS, E->Event); + } +} + +void EventSection::addEvent(InputEvent *Event) { + if (!Event->Live) + return; + uint32_t EventIndex = Out.ImportSec->NumImportedEvents + InputEvents.size(); + LLVM_DEBUG(dbgs() << "addEvent: " << EventIndex << "\n"); + Event->setEventIndex(EventIndex); + InputEvents.push_back(Event); +} + +void ExportSection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + writeUleb128(OS, Exports.size(), "export count"); + for (const WasmExport &Export : Exports) + writeExport(OS, Export); +} + +void ElemSection::addEntry(FunctionSymbol *Sym) { + if (Sym->hasTableIndex()) + return; + Sym->setTableIndex(ElemOffset + IndirectFunctions.size()); + IndirectFunctions.emplace_back(Sym); +} + +void ElemSection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + writeUleb128(OS, 1, "segment count"); + writeUleb128(OS, 0, "table index"); + WasmInitExpr InitExpr; + if (Config->Pic) { + InitExpr.Opcode = WASM_OPCODE_GLOBAL_GET; + InitExpr.Value.Global = WasmSym::TableBase->getGlobalIndex(); + } else { + InitExpr.Opcode = WASM_OPCODE_I32_CONST; + InitExpr.Value.Int32 = ElemOffset; + } + writeInitExpr(OS, InitExpr); + writeUleb128(OS, IndirectFunctions.size(), "elem count"); + + uint32_t TableIndex = ElemOffset; + for (const FunctionSymbol *Sym : IndirectFunctions) { + assert(Sym->getTableIndex() == TableIndex); + writeUleb128(OS, Sym->getFunctionIndex(), "function index"); + ++TableIndex; + } +} + +void DataCountSection::writeBody() { + writeUleb128(BodyOutputStream, NumSegments, "data count"); +} + +bool DataCountSection::isNeeded() const { + return NumSegments && Out.TargetFeaturesSec->Features.count("bulk-memory"); +} + +static uint32_t getWasmFlags(const Symbol *Sym) { + uint32_t Flags = 0; + if (Sym->isLocal()) + Flags |= WASM_SYMBOL_BINDING_LOCAL; + if (Sym->isWeak()) + Flags |= WASM_SYMBOL_BINDING_WEAK; + if (Sym->isHidden()) + Flags |= WASM_SYMBOL_VISIBILITY_HIDDEN; + if (Sym->isUndefined()) + Flags |= WASM_SYMBOL_UNDEFINED; + if (auto *F = dyn_cast(Sym)) { + if (F->getName() != F->ImportName) + Flags |= WASM_SYMBOL_EXPLICIT_NAME; + } else if (auto *G = dyn_cast(Sym)) { + if (G->getName() != G->ImportName) + Flags |= WASM_SYMBOL_EXPLICIT_NAME; + } + return Flags; +} + +void LinkingSection::writeBody() { + raw_ostream &OS = BodyOutputStream; + + writeUleb128(OS, WasmMetadataVersion, "Version"); + + if (!SymtabEntries.empty()) { + SubSection Sub(WASM_SYMBOL_TABLE); + writeUleb128(Sub.OS, SymtabEntries.size(), "num symbols"); + + for (const Symbol *Sym : SymtabEntries) { + assert(Sym->isDefined() || Sym->isUndefined()); + WasmSymbolType Kind = Sym->getWasmType(); + uint32_t Flags = getWasmFlags(Sym); + + writeU8(Sub.OS, Kind, "sym kind"); + writeUleb128(Sub.OS, Flags, "sym flags"); + + if (auto *F = dyn_cast(Sym)) { + writeUleb128(Sub.OS, F->getFunctionIndex(), "index"); + if (Sym->isDefined() || (Flags & WASM_SYMBOL_EXPLICIT_NAME) != 0) + writeStr(Sub.OS, Sym->getName(), "sym name"); + } else if (auto *G = dyn_cast(Sym)) { + writeUleb128(Sub.OS, G->getGlobalIndex(), "index"); + if (Sym->isDefined() || (Flags & WASM_SYMBOL_EXPLICIT_NAME) != 0) + writeStr(Sub.OS, Sym->getName(), "sym name"); + } else if (auto *E = dyn_cast(Sym)) { + writeUleb128(Sub.OS, E->getEventIndex(), "index"); + if (Sym->isDefined() || (Flags & WASM_SYMBOL_EXPLICIT_NAME) != 0) + writeStr(Sub.OS, Sym->getName(), "sym name"); + } else if (isa(Sym)) { + writeStr(Sub.OS, Sym->getName(), "sym name"); + if (auto *DataSym = dyn_cast(Sym)) { + writeUleb128(Sub.OS, DataSym->getOutputSegmentIndex(), "index"); + writeUleb128(Sub.OS, DataSym->getOutputSegmentOffset(), + "data offset"); + writeUleb128(Sub.OS, DataSym->getSize(), "data size"); + } + } else { + auto *S = cast(Sym); + writeUleb128(Sub.OS, S->Section->SectionIndex, "sym section index"); + } + } + + Sub.writeTo(OS); + } + + if (DataSegments.size()) { + SubSection Sub(WASM_SEGMENT_INFO); + writeUleb128(Sub.OS, DataSegments.size(), "num data segments"); + for (const OutputSegment *S : DataSegments) { + writeStr(Sub.OS, S->Name, "segment name"); + writeUleb128(Sub.OS, S->Alignment, "alignment"); + writeUleb128(Sub.OS, 0, "flags"); + } + Sub.writeTo(OS); + } + + if (!InitFunctions.empty()) { + SubSection Sub(WASM_INIT_FUNCS); + writeUleb128(Sub.OS, InitFunctions.size(), "num init functions"); + for (const WasmInitEntry &F : InitFunctions) { + writeUleb128(Sub.OS, F.Priority, "priority"); + writeUleb128(Sub.OS, F.Sym->getOutputSymbolIndex(), "function index"); + } + Sub.writeTo(OS); + } + + struct ComdatEntry { + unsigned Kind; + uint32_t Index; + }; + std::map> Comdats; + + for (const InputFunction *F : Out.FunctionSec->InputFunctions) { + StringRef Comdat = F->getComdatName(); + if (!Comdat.empty()) + Comdats[Comdat].emplace_back( + ComdatEntry{WASM_COMDAT_FUNCTION, F->getFunctionIndex()}); + } + for (uint32_t I = 0; I < DataSegments.size(); ++I) { + const auto &InputSegments = DataSegments[I]->InputSegments; + if (InputSegments.empty()) + continue; + StringRef Comdat = InputSegments[0]->getComdatName(); +#ifndef NDEBUG + for (const InputSegment *IS : InputSegments) + assert(IS->getComdatName() == Comdat); +#endif + if (!Comdat.empty()) + Comdats[Comdat].emplace_back(ComdatEntry{WASM_COMDAT_DATA, I}); + } + + if (!Comdats.empty()) { + SubSection Sub(WASM_COMDAT_INFO); + writeUleb128(Sub.OS, Comdats.size(), "num comdats"); + for (const auto &C : Comdats) { + writeStr(Sub.OS, C.first, "comdat name"); + writeUleb128(Sub.OS, 0, "comdat flags"); // flags for future use + writeUleb128(Sub.OS, C.second.size(), "num entries"); + for (const ComdatEntry &Entry : C.second) { + writeU8(Sub.OS, Entry.Kind, "entry kind"); + writeUleb128(Sub.OS, Entry.Index, "entry index"); + } + } + Sub.writeTo(OS); + } +} + +void LinkingSection::addToSymtab(Symbol *Sym) { + Sym->setOutputSymbolIndex(SymtabEntries.size()); + SymtabEntries.emplace_back(Sym); +} + +unsigned NameSection::numNames() const { + unsigned NumNames = Out.ImportSec->NumImportedFunctions; + for (const InputFunction *F : Out.FunctionSec->InputFunctions) + if (!F->getName().empty() || !F->getDebugName().empty()) + ++NumNames; + + return NumNames; +} + +// Create the custom "name" section containing debug symbol names. +void NameSection::writeBody() { + SubSection Sub(WASM_NAMES_FUNCTION); + writeUleb128(Sub.OS, numNames(), "name count"); + + // Names must appear in function index order. As it happens ImportedSymbols + // and InputFunctions are numbered in order with imported functions coming + // first. + for (const Symbol *S : Out.ImportSec->ImportedSymbols) { + if (auto *F = dyn_cast(S)) { + writeUleb128(Sub.OS, F->getFunctionIndex(), "func index"); + writeStr(Sub.OS, toString(*S), "symbol name"); + } + } + for (const InputFunction *F : Out.FunctionSec->InputFunctions) { + if (!F->getName().empty()) { + writeUleb128(Sub.OS, F->getFunctionIndex(), "func index"); + if (!F->getDebugName().empty()) { + writeStr(Sub.OS, F->getDebugName(), "symbol name"); + } else { + writeStr(Sub.OS, maybeDemangleSymbol(F->getName()), "symbol name"); + } + } + } + + Sub.writeTo(BodyOutputStream); +} + +void ProducersSection::addInfo(const WasmProducerInfo &Info) { + for (auto &Producers : + {std::make_pair(&Info.Languages, &Languages), + std::make_pair(&Info.Tools, &Tools), std::make_pair(&Info.SDKs, &SDKs)}) + for (auto &Producer : *Producers.first) + if (Producers.second->end() == + llvm::find_if(*Producers.second, + [&](std::pair Seen) { + return Seen.first == Producer.first; + })) + Producers.second->push_back(Producer); +} + +void ProducersSection::writeBody() { + auto &OS = BodyOutputStream; + writeUleb128(OS, fieldCount(), "field count"); + for (auto &Field : + {std::make_pair("language", Languages), + std::make_pair("processed-by", Tools), std::make_pair("sdk", SDKs)}) { + if (Field.second.empty()) + continue; + writeStr(OS, Field.first, "field name"); + writeUleb128(OS, Field.second.size(), "number of entries"); + for (auto &Entry : Field.second) { + writeStr(OS, Entry.first, "producer name"); + writeStr(OS, Entry.second, "producer version"); + } + } +} + +void TargetFeaturesSection::writeBody() { + SmallVector Emitted(Features.begin(), Features.end()); + llvm::sort(Emitted); + auto &OS = BodyOutputStream; + writeUleb128(OS, Emitted.size(), "feature count"); + for (auto &Feature : Emitted) { + writeU8(OS, WASM_FEATURE_PREFIX_USED, "feature used prefix"); + writeStr(OS, Feature, "feature name"); + } +} + +void RelocSection::writeBody() { + uint32_t Count = Sec->numRelocations(); + assert(Sec->SectionIndex != UINT32_MAX); + writeUleb128(BodyOutputStream, Sec->SectionIndex, "reloc section"); + writeUleb128(BodyOutputStream, Count, "reloc count"); + Sec->writeRelocations(BodyOutputStream); +} Index: wasm/Writer.h =================================================================== --- wasm/Writer.h +++ wasm/Writer.h @@ -14,8 +14,6 @@ void writeResult(); -extern const char *DefaultModule; - } // namespace wasm } // namespace lld Index: wasm/Writer.cpp =================================================================== --- wasm/Writer.cpp +++ wasm/Writer.cpp @@ -13,7 +13,9 @@ #include "InputGlobal.h" #include "OutputSections.h" #include "OutputSegment.h" +#include "Relocations.h" #include "SymbolTable.h" +#include "SyntheticSections.h" #include "WriterUtils.h" #include "lld/Common/ErrorHandler.h" #include "lld/Common/Memory.h" @@ -29,7 +31,6 @@ #include "llvm/Support/Format.h" #include "llvm/Support/FormatVariadic.h" #include "llvm/Support/LEB128.h" -#include "llvm/Support/Path.h" #include #include @@ -42,18 +43,9 @@ using namespace lld::wasm; static constexpr int StackAlignment = 16; -static constexpr const char *FunctionTableName = "__indirect_function_table"; -const char *lld::wasm::DefaultModule = "env"; namespace { -// An init entry to be written to either the synthetic init func or the -// linking metadata. -struct WasmInitEntry { - const FunctionSymbol *Sym; - uint32_t Priority; -}; - // The writer writes a SymbolTable result to a file. class Writer { public: @@ -62,81 +54,40 @@ private: void openFile(); - uint32_t lookupType(const WasmSignature &Sig); - uint32_t registerType(const WasmSignature &Sig); - void createApplyRelocationsFunction(); void createCallCtorsFunction(); - void calculateInitFunctions(); - void processRelocations(InputChunk *Chunk); void assignIndexes(); - void calculateTargetFeatures(); + void populateSymtab(); + void populateProducers(); + void populateTargetFeatures(); + void calculateInitFunctions(); void calculateImports(); void calculateExports(); void calculateCustomSections(); - void assignSymtab(); void calculateTypes(); void createOutputSegments(); void layoutMemory(); void createHeader(); - void createSections(); - SyntheticSection *createSyntheticSection(uint32_t Type, StringRef Name = ""); - // Builtin sections - void createTypeSection(); - void createFunctionSection(); - void createTableSection(); - void createGlobalSection(); - void createEventSection(); - void createExportSection(); - void createImportSection(); - void createMemorySection(); - void createElemSection(); - void createDataCountSection(); - void createCodeSection(); - void createDataSection(); + void addSection(OutputSection *Sec); + + void addSections(); void createCustomSections(); + void createSyntheticSections(); + void finalizeSections(); // Custom sections - void createDylinkSection(); void createRelocSections(); - void createLinkingSection(); - void createNameSection(); - void createProducersSection(); - void createTargetFeaturesSection(); void writeHeader(); void writeSections(); uint64_t FileSize = 0; uint32_t TableBase = 0; - uint32_t NumMemoryPages = 0; - uint32_t MaxMemoryPages = 0; - // Memory size and aligment. Written to the "dylink" section - // when build with -shared or -pie. - uint32_t MemAlign = 0; - uint32_t MemSize = 0; - - std::vector Types; - DenseMap TypeIndices; - std::vector ImportedSymbols; - std::vector GOTSymbols; - unsigned NumImportedFunctions = 0; - unsigned NumImportedGlobals = 0; - unsigned NumImportedEvents = 0; - std::vector Exports; - std::vector DefinedFakeGlobals; - std::vector InputGlobals; - std::vector InputFunctions; - std::vector InputEvents; - std::vector IndirectFunctions; - std::vector SymtabEntries; - std::vector InitFunctions; + std::vector InitFunctions; llvm::StringMap> CustomSectionMapping; - llvm::StringMap CustomSectionSymbols; - llvm::SmallSet TargetFeatures; // Elements that are used to construct the final output std::string Header; @@ -150,210 +101,6 @@ } // anonymous namespace -void Writer::createImportSection() { - uint32_t NumImports = ImportedSymbols.size() + GOTSymbols.size(); - if (Config->ImportMemory) - ++NumImports; - if (Config->ImportTable) - ++NumImports; - - if (NumImports == 0) - return; - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_IMPORT); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, NumImports, "import count"); - - if (Config->ImportMemory) { - WasmImport Import; - Import.Module = DefaultModule; - Import.Field = "memory"; - Import.Kind = WASM_EXTERNAL_MEMORY; - Import.Memory.Flags = 0; - Import.Memory.Initial = NumMemoryPages; - if (MaxMemoryPages != 0 || Config->SharedMemory) { - Import.Memory.Flags |= WASM_LIMITS_FLAG_HAS_MAX; - Import.Memory.Maximum = MaxMemoryPages; - } - if (Config->SharedMemory) - Import.Memory.Flags |= WASM_LIMITS_FLAG_IS_SHARED; - writeImport(OS, Import); - } - - if (Config->ImportTable) { - uint32_t TableSize = TableBase + IndirectFunctions.size(); - WasmImport Import; - Import.Module = DefaultModule; - Import.Field = FunctionTableName; - Import.Kind = WASM_EXTERNAL_TABLE; - Import.Table.ElemType = WASM_TYPE_FUNCREF; - Import.Table.Limits = {0, TableSize, 0}; - writeImport(OS, Import); - } - - for (const Symbol *Sym : ImportedSymbols) { - WasmImport Import; - if (auto *F = dyn_cast(Sym)) { - Import.Field = F->ImportName; - Import.Module = F->ImportModule; - } else if (auto *G = dyn_cast(Sym)) { - Import.Field = G->ImportName; - Import.Module = G->ImportModule; - } else { - Import.Field = Sym->getName(); - Import.Module = DefaultModule; - } - - if (auto *FunctionSym = dyn_cast(Sym)) { - Import.Kind = WASM_EXTERNAL_FUNCTION; - Import.SigIndex = lookupType(*FunctionSym->Signature); - } else if (auto *GlobalSym = dyn_cast(Sym)) { - Import.Kind = WASM_EXTERNAL_GLOBAL; - Import.Global = *GlobalSym->getGlobalType(); - } else { - auto *EventSym = cast(Sym); - Import.Kind = WASM_EXTERNAL_EVENT; - Import.Event.Attribute = EventSym->getEventType()->Attribute; - Import.Event.SigIndex = lookupType(*EventSym->Signature); - } - writeImport(OS, Import); - } - - for (const Symbol *Sym : GOTSymbols) { - WasmImport Import; - Import.Kind = WASM_EXTERNAL_GLOBAL; - Import.Global = {WASM_TYPE_I32, true}; - if (isa(Sym)) - Import.Module = "GOT.mem"; - else - Import.Module = "GOT.func"; - Import.Field = Sym->getName(); - writeImport(OS, Import); - } -} - -void Writer::createTypeSection() { - if (!Types.size()) - return; - SyntheticSection *Section = createSyntheticSection(WASM_SEC_TYPE); - raw_ostream &OS = Section->getStream(); - writeUleb128(OS, Types.size(), "type count"); - for (const WasmSignature *Sig : Types) - writeSig(OS, *Sig); -} - -void Writer::createFunctionSection() { - if (InputFunctions.empty()) - return; - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_FUNCTION); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, InputFunctions.size(), "function count"); - for (const InputFunction *Func : InputFunctions) - writeUleb128(OS, lookupType(Func->Signature), "sig index"); -} - -void Writer::createMemorySection() { - if (Config->ImportMemory) - return; - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_MEMORY); - raw_ostream &OS = Section->getStream(); - - bool HasMax = MaxMemoryPages != 0 || Config->SharedMemory; - writeUleb128(OS, 1, "memory count"); - unsigned Flags = 0; - if (HasMax) - Flags |= WASM_LIMITS_FLAG_HAS_MAX; - if (Config->SharedMemory) - Flags |= WASM_LIMITS_FLAG_IS_SHARED; - writeUleb128(OS, Flags, "memory limits flags"); - writeUleb128(OS, NumMemoryPages, "initial pages"); - if (HasMax) - writeUleb128(OS, MaxMemoryPages, "max pages"); -} - -void Writer::createGlobalSection() { - unsigned NumGlobals = InputGlobals.size() + DefinedFakeGlobals.size(); - if (NumGlobals == 0) - return; - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_GLOBAL); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, NumGlobals, "global count"); - for (const InputGlobal *G : InputGlobals) - writeGlobal(OS, G->Global); - for (const DefinedData *Sym : DefinedFakeGlobals) { - WasmGlobal Global; - Global.Type = {WASM_TYPE_I32, false}; - Global.InitExpr.Opcode = WASM_OPCODE_I32_CONST; - Global.InitExpr.Value.Int32 = Sym->getVirtualAddress(); - writeGlobal(OS, Global); - } -} - -// The event section contains a list of declared wasm events associated with the -// module. Currently the only supported event kind is exceptions. A single event -// entry represents a single event with an event tag. All C++ exceptions are -// represented by a single event. An event entry in this section contains -// information on what kind of event it is (e.g. exception) and the type of -// values contained in a single event object. (In wasm, an event can contain -// multiple values of primitive types. But for C++ exceptions, we just throw a -// pointer which is an i32 value (for wasm32 architecture), so the signature of -// C++ exception is (i32)->(void), because all event types are assumed to have -// void return type to share WasmSignature with functions.) -void Writer::createEventSection() { - unsigned NumEvents = InputEvents.size(); - if (NumEvents == 0) - return; - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_EVENT); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, NumEvents, "event count"); - for (InputEvent *E : InputEvents) { - E->Event.Type.SigIndex = lookupType(E->Signature); - writeEvent(OS, E->Event); - } -} - -void Writer::createTableSection() { - if (Config->ImportTable) - return; - - // Always output a table section (or table import), even if there are no - // indirect calls. There are two reasons for this: - // 1. For executables it is useful to have an empty table slot at 0 - // which can be filled with a null function call handler. - // 2. If we don't do this, any program that contains a call_indirect but - // no address-taken function will fail at validation time since it is - // a validation error to include a call_indirect instruction if there - // is not table. - uint32_t TableSize = TableBase + IndirectFunctions.size(); - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_TABLE); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, 1, "table count"); - WasmLimits Limits = {WASM_LIMITS_FLAG_HAS_MAX, TableSize, TableSize}; - writeTableType(OS, WasmTable{WASM_TYPE_FUNCREF, Limits}); -} - -void Writer::createExportSection() { - if (!Exports.size()) - return; - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_EXPORT); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, Exports.size(), "export count"); - for (const WasmExport &Export : Exports) - writeExport(OS, Export); -} - void Writer::calculateCustomSections() { log("calculateCustomSections"); bool StripDebug = Config->StripDebug || Config->StripAll; @@ -377,75 +124,18 @@ log("createCustomSections"); for (auto &Pair : CustomSectionMapping) { StringRef Name = Pair.first(); - - auto P = CustomSectionSymbols.find(Name); - if (P != CustomSectionSymbols.end()) { - uint32_t SectionIndex = OutputSections.size(); - P->second->setOutputSectionIndex(SectionIndex); - } - LLVM_DEBUG(dbgs() << "createCustomSection: " << Name << "\n"); - OutputSections.push_back(make(Name, Pair.second)); - } -} -void Writer::createElemSection() { - if (IndirectFunctions.empty()) - return; - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_ELEM); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, 1, "segment count"); - writeUleb128(OS, 0, "table index"); - WasmInitExpr InitExpr; - if (Config->Pic) { - InitExpr.Opcode = WASM_OPCODE_GLOBAL_GET; - InitExpr.Value.Global = WasmSym::TableBase->getGlobalIndex(); - } else { - InitExpr.Opcode = WASM_OPCODE_I32_CONST; - InitExpr.Value.Int32 = TableBase; - } - writeInitExpr(OS, InitExpr); - writeUleb128(OS, IndirectFunctions.size(), "elem count"); - - uint32_t TableIndex = TableBase; - for (const FunctionSymbol *Sym : IndirectFunctions) { - assert(Sym->getTableIndex() == TableIndex); - writeUleb128(OS, Sym->getFunctionIndex(), "function index"); - ++TableIndex; + OutputSection *Sec = make(Name, Pair.second); + if (Config->Relocatable) { + auto *Sym = make(Sec); + Out.LinkingSec->addToSymtab(Sym); + Sec->SectionSym = Sym; + } + addSection(Sec); } } -void Writer::createDataCountSection() { - if (!Segments.size() || !TargetFeatures.count("bulk-memory")) - return; - - log("createDataCountSection"); - SyntheticSection *Section = createSyntheticSection(WASM_SEC_DATACOUNT); - raw_ostream &OS = Section->getStream(); - writeUleb128(OS, Segments.size(), "data count"); -} - -void Writer::createCodeSection() { - if (InputFunctions.empty()) - return; - - log("createCodeSection"); - - auto Section = make(InputFunctions); - OutputSections.push_back(Section); -} - -void Writer::createDataSection() { - if (!Segments.size()) - return; - - log("createDataSection"); - auto Section = make(Segments); - OutputSections.push_back(Section); -} - // Create relocations sections in the final output. // These are only created when relocatable output is requested. void Writer::createRelocSections() { @@ -453,293 +143,33 @@ // Don't use iterator here since we are adding to OutputSection size_t OrigSize = OutputSections.size(); for (size_t I = 0; I < OrigSize; I++) { - OutputSection *OSec = OutputSections[I]; - uint32_t Count = OSec->numRelocations(); + LLVM_DEBUG(dbgs() << "check section " << I << "\n"); + OutputSection *Sec = OutputSections[I]; + + // Count the number of needed sections. + uint32_t Count = Sec->numRelocations(); if (!Count) continue; StringRef Name; - if (OSec->Type == WASM_SEC_DATA) + if (Sec->Type == WASM_SEC_DATA) Name = "reloc.DATA"; - else if (OSec->Type == WASM_SEC_CODE) + else if (Sec->Type == WASM_SEC_CODE) Name = "reloc.CODE"; - else if (OSec->Type == WASM_SEC_CUSTOM) - Name = Saver.save("reloc." + OSec->Name); + else if (Sec->Type == WASM_SEC_CUSTOM) + Name = Saver.save("reloc." + Sec->Name); else llvm_unreachable( "relocations only supported for code, data, or custom sections"); - OutputSections.push_back(make(Name, OSec, I)); - } -} - -static uint32_t getWasmFlags(const Symbol *Sym) { - uint32_t Flags = 0; - if (Sym->isLocal()) - Flags |= WASM_SYMBOL_BINDING_LOCAL; - if (Sym->isWeak()) - Flags |= WASM_SYMBOL_BINDING_WEAK; - if (Sym->isHidden()) - Flags |= WASM_SYMBOL_VISIBILITY_HIDDEN; - if (Sym->isUndefined()) - Flags |= WASM_SYMBOL_UNDEFINED; - if (auto *F = dyn_cast(Sym)) { - if (F->getName() != F->ImportName) - Flags |= WASM_SYMBOL_EXPLICIT_NAME; - } else if (auto *G = dyn_cast(Sym)) { - if (G->getName() != G->ImportName) - Flags |= WASM_SYMBOL_EXPLICIT_NAME; - } - return Flags; -} - -// Some synthetic sections (e.g. "name" and "linking") have subsections. -// Just like the synthetic sections themselves these need to be created before -// they can be written out (since they are preceded by their length). This -// class is used to create subsections and then write them into the stream -// of the parent section. -class SubSection { -public: - explicit SubSection(uint32_t Type) : Type(Type) {} - - void writeTo(raw_ostream &To) { - OS.flush(); - writeUleb128(To, Type, "subsection type"); - writeUleb128(To, Body.size(), "subsection size"); - To.write(Body.data(), Body.size()); - } - -private: - uint32_t Type; - std::string Body; - -public: - raw_string_ostream OS{Body}; -}; - -// Create the custom "dylink" section containing information for the dynamic -// linker. -// See -// https://github.com/WebAssembly/tool-conventions/blob/master/DynamicLinking.md -void Writer::createDylinkSection() { - SyntheticSection *Section = createSyntheticSection(WASM_SEC_CUSTOM, "dylink"); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, MemSize, "MemSize"); - writeUleb128(OS, MemAlign, "MemAlign"); - writeUleb128(OS, IndirectFunctions.size(), "TableSize"); - writeUleb128(OS, 0, "TableAlign"); - writeUleb128(OS, Symtab->SharedFiles.size(), "Needed"); - for (auto *SO : Symtab->SharedFiles) - writeStr(OS, llvm::sys::path::filename(SO->getName()), "so name"); -} - -// Create the custom "linking" section containing linker metadata. -// This is only created when relocatable output is requested. -void Writer::createLinkingSection() { - SyntheticSection *Section = - createSyntheticSection(WASM_SEC_CUSTOM, "linking"); - raw_ostream &OS = Section->getStream(); - - writeUleb128(OS, WasmMetadataVersion, "Version"); - - if (!SymtabEntries.empty()) { - SubSection Sub(WASM_SYMBOL_TABLE); - writeUleb128(Sub.OS, SymtabEntries.size(), "num symbols"); - - for (const Symbol *Sym : SymtabEntries) { - assert(Sym->isDefined() || Sym->isUndefined()); - WasmSymbolType Kind = Sym->getWasmType(); - uint32_t Flags = getWasmFlags(Sym); - - writeU8(Sub.OS, Kind, "sym kind"); - writeUleb128(Sub.OS, Flags, "sym flags"); - - if (auto *F = dyn_cast(Sym)) { - writeUleb128(Sub.OS, F->getFunctionIndex(), "index"); - if (Sym->isDefined() || - (Flags & WASM_SYMBOL_EXPLICIT_NAME) != 0) - writeStr(Sub.OS, Sym->getName(), "sym name"); - } else if (auto *G = dyn_cast(Sym)) { - writeUleb128(Sub.OS, G->getGlobalIndex(), "index"); - if (Sym->isDefined() || - (Flags & WASM_SYMBOL_EXPLICIT_NAME) != 0) - writeStr(Sub.OS, Sym->getName(), "sym name"); - } else if (auto *E = dyn_cast(Sym)) { - writeUleb128(Sub.OS, E->getEventIndex(), "index"); - if (Sym->isDefined() || - (Flags & WASM_SYMBOL_EXPLICIT_NAME) != 0) - writeStr(Sub.OS, Sym->getName(), "sym name"); - } else if (isa(Sym)) { - writeStr(Sub.OS, Sym->getName(), "sym name"); - if (auto *DataSym = dyn_cast(Sym)) { - writeUleb128(Sub.OS, DataSym->getOutputSegmentIndex(), "index"); - writeUleb128(Sub.OS, DataSym->getOutputSegmentOffset(), - "data offset"); - writeUleb128(Sub.OS, DataSym->getSize(), "data size"); - } - } else { - auto *S = cast(Sym); - writeUleb128(Sub.OS, S->getOutputSectionIndex(), "sym section index"); - } - } - - Sub.writeTo(OS); + addSection(make(Name, Sec)); } - - if (Segments.size()) { - SubSection Sub(WASM_SEGMENT_INFO); - writeUleb128(Sub.OS, Segments.size(), "num data segments"); - for (const OutputSegment *S : Segments) { - writeStr(Sub.OS, S->Name, "segment name"); - writeUleb128(Sub.OS, S->Alignment, "alignment"); - writeUleb128(Sub.OS, 0, "flags"); - } - Sub.writeTo(OS); - } - - if (!InitFunctions.empty()) { - SubSection Sub(WASM_INIT_FUNCS); - writeUleb128(Sub.OS, InitFunctions.size(), "num init functions"); - for (const WasmInitEntry &F : InitFunctions) { - writeUleb128(Sub.OS, F.Priority, "priority"); - writeUleb128(Sub.OS, F.Sym->getOutputSymbolIndex(), "function index"); - } - Sub.writeTo(OS); - } - - struct ComdatEntry { - unsigned Kind; - uint32_t Index; - }; - std::map> Comdats; - - for (const InputFunction *F : InputFunctions) { - StringRef Comdat = F->getComdatName(); - if (!Comdat.empty()) - Comdats[Comdat].emplace_back( - ComdatEntry{WASM_COMDAT_FUNCTION, F->getFunctionIndex()}); - } - for (uint32_t I = 0; I < Segments.size(); ++I) { - const auto &InputSegments = Segments[I]->InputSegments; - if (InputSegments.empty()) - continue; - StringRef Comdat = InputSegments[0]->getComdatName(); -#ifndef NDEBUG - for (const InputSegment *IS : InputSegments) - assert(IS->getComdatName() == Comdat); -#endif - if (!Comdat.empty()) - Comdats[Comdat].emplace_back(ComdatEntry{WASM_COMDAT_DATA, I}); - } - - if (!Comdats.empty()) { - SubSection Sub(WASM_COMDAT_INFO); - writeUleb128(Sub.OS, Comdats.size(), "num comdats"); - for (const auto &C : Comdats) { - writeStr(Sub.OS, C.first, "comdat name"); - writeUleb128(Sub.OS, 0, "comdat flags"); // flags for future use - writeUleb128(Sub.OS, C.second.size(), "num entries"); - for (const ComdatEntry &Entry : C.second) { - writeU8(Sub.OS, Entry.Kind, "entry kind"); - writeUleb128(Sub.OS, Entry.Index, "entry index"); - } - } - Sub.writeTo(OS); - } -} - -// Create the custom "name" section containing debug symbol names. -void Writer::createNameSection() { - unsigned NumNames = NumImportedFunctions; - for (const InputFunction *F : InputFunctions) - if (!F->getName().empty() || !F->getDebugName().empty()) - ++NumNames; - - if (NumNames == 0) - return; - - SyntheticSection *Section = createSyntheticSection(WASM_SEC_CUSTOM, "name"); - - SubSection Sub(WASM_NAMES_FUNCTION); - writeUleb128(Sub.OS, NumNames, "name count"); - - // Names must appear in function index order. As it happens ImportedSymbols - // and InputFunctions are numbered in order with imported functions coming - // first. - for (const Symbol *S : ImportedSymbols) { - if (auto *F = dyn_cast(S)) { - writeUleb128(Sub.OS, F->getFunctionIndex(), "func index"); - writeStr(Sub.OS, toString(*S), "symbol name"); - } - } - for (const InputFunction *F : InputFunctions) { - if (!F->getName().empty()) { - writeUleb128(Sub.OS, F->getFunctionIndex(), "func index"); - if (!F->getDebugName().empty()) { - writeStr(Sub.OS, F->getDebugName(), "symbol name"); - } else { - writeStr(Sub.OS, maybeDemangleSymbol(F->getName()), "symbol name"); - } - } - } - - Sub.writeTo(Section->getStream()); } -void Writer::createProducersSection() { - SmallVector, 8> Languages; - SmallVector, 8> Tools; - SmallVector, 8> SDKs; +void Writer::populateProducers() { for (ObjFile *File : Symtab->ObjectFiles) { const WasmProducerInfo &Info = File->getWasmObj()->getProducerInfo(); - for (auto &Producers : {std::make_pair(&Info.Languages, &Languages), - std::make_pair(&Info.Tools, &Tools), - std::make_pair(&Info.SDKs, &SDKs)}) - for (auto &Producer : *Producers.first) - if (Producers.second->end() == - llvm::find_if(*Producers.second, - [&](std::pair Seen) { - return Seen.first == Producer.first; - })) - Producers.second->push_back(Producer); - } - int FieldCount = - int(!Languages.empty()) + int(!Tools.empty()) + int(!SDKs.empty()); - if (FieldCount == 0) - return; - SyntheticSection *Section = - createSyntheticSection(WASM_SEC_CUSTOM, "producers"); - auto &OS = Section->getStream(); - writeUleb128(OS, FieldCount, "field count"); - for (auto &Field : - {std::make_pair("language", Languages), - std::make_pair("processed-by", Tools), std::make_pair("sdk", SDKs)}) { - if (Field.second.empty()) - continue; - writeStr(OS, Field.first, "field name"); - writeUleb128(OS, Field.second.size(), "number of entries"); - for (auto &Entry : Field.second) { - writeStr(OS, Entry.first, "producer name"); - writeStr(OS, Entry.second, "producer version"); - } - } -} - -void Writer::createTargetFeaturesSection() { - if (TargetFeatures.empty()) - return; - - SmallVector Emitted(TargetFeatures.begin(), - TargetFeatures.end()); - llvm::sort(Emitted); - SyntheticSection *Section = - createSyntheticSection(WASM_SEC_CUSTOM, "target_features"); - auto &OS = Section->getStream(); - writeUleb128(OS, Emitted.size(), "feature count"); - for (auto &Feature : Emitted) { - writeU8(OS, WASM_FEATURE_PREFIX_USED, "feature used prefix"); - writeStr(OS, Feature, "feature name"); + Out.ProducersSec->addInfo(Info); } } @@ -749,7 +179,10 @@ void Writer::writeSections() { uint8_t *Buf = Buffer->getBufferStart(); - parallelForEach(OutputSections, [Buf](OutputSection *S) { S->writeTo(Buf); }); + parallelForEach(OutputSections, [Buf](OutputSection *S) { + assert(S->isNeeded()); + S->writeTo(Buf); + }); } // Fix the memory layout of the output binary. This assigns memory offsets @@ -766,8 +199,6 @@ // rather than overwriting global data, but also increases code size since all // static data loads and stores requires larger offsets. void Writer::layoutMemory() { - createOutputSegments(); - uint32_t MemoryPtr = 0; auto PlaceStack = [&]() { @@ -798,9 +229,9 @@ if (WasmSym::DsoHandle) WasmSym::DsoHandle->setVirtualAddress(DataStart); - MemAlign = 0; + Out.DylinkSec->MemAlign = 0; for (OutputSegment *Seg : Segments) { - MemAlign = std::max(MemAlign, Seg->Alignment); + Out.DylinkSec->MemAlign = std::max(Out.DylinkSec->MemAlign, Seg->Alignment); MemoryPtr = alignTo(MemoryPtr, 1ULL << Seg->Alignment); Seg->StartVA = MemoryPtr; log(formatv("mem: {0,-15} offset={1,-8} size={2,-8} align={3}", Seg->Name, @@ -815,7 +246,7 @@ log("mem: static data = " + Twine(MemoryPtr - DataStart)); if (Config->Shared) { - MemSize = MemoryPtr; + Out.DylinkSec->MemSize = MemoryPtr; return; } @@ -838,9 +269,10 @@ else MemoryPtr = Config->InitialMemory; } - MemSize = MemoryPtr; - NumMemoryPages = alignTo(MemoryPtr, WasmPageSize) / WasmPageSize; - log("mem: total pages = " + Twine(NumMemoryPages)); + Out.DylinkSec->MemSize = MemoryPtr; + Out.MemorySec->NumMemoryPages = + alignTo(MemoryPtr, WasmPageSize) / WasmPageSize; + log("mem: total pages = " + Twine(Out.MemorySec->NumMemoryPages)); // Check max if explicitly supplied or required by shared memory if (Config->MaxMemory != 0 || Config->SharedMemory) { @@ -848,51 +280,48 @@ error("maximum memory must be " + Twine(WasmPageSize) + "-byte aligned"); if (MemoryPtr > Config->MaxMemory) error("maximum memory too small, " + Twine(MemoryPtr) + " bytes needed"); - MaxMemoryPages = Config->MaxMemory / WasmPageSize; - log("mem: max pages = " + Twine(MaxMemoryPages)); + Out.MemorySec->MaxMemoryPages = Config->MaxMemory / WasmPageSize; + log("mem: max pages = " + Twine(Out.MemorySec->MaxMemoryPages)); } } -SyntheticSection *Writer::createSyntheticSection(uint32_t Type, - StringRef Name) { - auto Sec = make(Type, Name); - log("createSection: " + toString(*Sec)); +void Writer::addSection(OutputSection *Sec) { + if (!Sec->isNeeded()) + return; + log("addSection: " + toString(*Sec)); + Sec->SectionIndex = OutputSections.size(); OutputSections.push_back(Sec); - return Sec; } -void Writer::createSections() { - // Known sections - if (Config->Pic) - createDylinkSection(); - createTypeSection(); - createImportSection(); - createFunctionSection(); - createTableSection(); - createMemorySection(); - createGlobalSection(); - createEventSection(); - createExportSection(); - createElemSection(); - createDataCountSection(); - createCodeSection(); - createDataSection(); +void Writer::addSections() { + addSection(Out.DylinkSec); + addSection(Out.TypeSec); + addSection(Out.ImportSec); + addSection(Out.FunctionSec); + addSection(Out.TableSec); + addSection(Out.MemorySec); + addSection(Out.GlobalSec); + addSection(Out.EventSec); + addSection(Out.ExportSec); + addSection(Out.ElemSec); + addSection(Out.DataCountSec); + + addSection(make(Out.FunctionSec->InputFunctions)); + addSection(make(Segments)); + createCustomSections(); - // Custom sections + addSection(Out.LinkingSec); if (Config->Relocatable) { - createLinkingSection(); createRelocSections(); } - if (!Config->StripDebug && !Config->StripAll) - createNameSection(); - - if (!Config->StripAll) { - createProducersSection(); - createTargetFeaturesSection(); - } + addSection(Out.NameSec); + addSection(Out.ProducersSec); + addSection(Out.TargetFeaturesSec); +} +void Writer::finalizeSections() { for (OutputSection *S : OutputSections) { S->setOffset(FileSize); S->finalizeContents(); @@ -900,7 +329,7 @@ } } -void Writer::calculateTargetFeatures() { +void Writer::populateTargetFeatures() { SmallSet Used; SmallSet Required; SmallSet Disallowed; @@ -910,7 +339,7 @@ if (!InferFeatures) { for (auto &Feature : Config->Features.getValue()) - TargetFeatures.insert(Feature); + Out.TargetFeaturesSec->Features.insert(Feature); // No need to read or check features if (!Config->CheckFeatures) return; @@ -938,9 +367,9 @@ } if (InferFeatures) - TargetFeatures.insert(Used.begin(), Used.end()); + Out.TargetFeaturesSec->Features.insert(Used.begin(), Used.end()); - if (TargetFeatures.count("atomics") && !Config->SharedMemory) + if (Out.TargetFeaturesSec->Features.count("atomics") && !Config->SharedMemory) error("'atomics' feature is used, so --shared-memory must be used"); if (!Config->CheckFeatures) @@ -953,7 +382,7 @@ // Validate that used features are allowed in output if (!InferFeatures) { for (auto &Feature : Used) { - if (!TargetFeatures.count(Feature)) + if (!Out.TargetFeaturesSec->Features.count(Feature)) error(Twine("Target feature '") + Feature + "' is not allowed."); } } @@ -993,13 +422,7 @@ continue; LLVM_DEBUG(dbgs() << "import: " << Sym->getName() << "\n"); - ImportedSymbols.emplace_back(Sym); - if (auto *F = dyn_cast(Sym)) - F->setFunctionIndex(NumImportedFunctions++); - else if (auto *G = dyn_cast(Sym)) - G->setGlobalIndex(NumImportedGlobals++); - else - cast(Sym)->setEventIndex(NumImportedEvents++); + Out.ImportSec->addImport(Sym); } } @@ -1008,12 +431,15 @@ return; if (!Config->Relocatable && !Config->ImportMemory) - Exports.push_back(WasmExport{"memory", WASM_EXTERNAL_MEMORY, 0}); + Out.ExportSec->Exports.push_back( + WasmExport{"memory", WASM_EXTERNAL_MEMORY, 0}); if (!Config->Relocatable && Config->ExportTable) - Exports.push_back(WasmExport{FunctionTableName, WASM_EXTERNAL_TABLE, 0}); + Out.ExportSec->Exports.push_back( + WasmExport{FunctionTableName, WASM_EXTERNAL_TABLE, 0}); - unsigned FakeGlobalIndex = NumImportedGlobals + InputGlobals.size(); + unsigned FakeGlobalIndex = + Out.ImportSec->NumImportedGlobals + Out.GlobalSec->InputGlobals.size(); for (Symbol *Sym : Symtab->getSymbols()) { if (!Sym->isExported()) @@ -1039,78 +465,31 @@ Export = {Name, WASM_EXTERNAL_EVENT, E->getEventIndex()}; } else { auto *D = cast(Sym); - DefinedFakeGlobals.emplace_back(D); + Out.GlobalSec->DefinedFakeGlobals.emplace_back(D); Export = {Name, WASM_EXTERNAL_GLOBAL, FakeGlobalIndex++}; } LLVM_DEBUG(dbgs() << "Export: " << Name << "\n"); - Exports.push_back(Export); + Out.ExportSec->Exports.push_back(Export); } } -void Writer::assignSymtab() { +void Writer::populateSymtab() { if (!Config->Relocatable) return; - StringMap SectionSymbolIndices; - - unsigned SymbolIndex = SymtabEntries.size(); - - auto AddSymbol = [&](Symbol *Sym) { - if (auto *S = dyn_cast(Sym)) { - StringRef Name = S->getName(); - if (CustomSectionMapping.count(Name) == 0) - return; - - auto SSI = SectionSymbolIndices.find(Name); - if (SSI != SectionSymbolIndices.end()) { - Sym->setOutputSymbolIndex(SSI->second); - return; - } - - SectionSymbolIndices[Name] = SymbolIndex; - CustomSectionSymbols[Name] = cast(Sym); - - Sym->markLive(); - } - - // (Since this is relocatable output, GC is not performed so symbols must - // be live.) - assert(Sym->isLive()); - Sym->setOutputSymbolIndex(SymbolIndex++); - SymtabEntries.emplace_back(Sym); - }; - for (Symbol *Sym : Symtab->getSymbols()) if (Sym->IsUsedInRegularObj) - AddSymbol(Sym); + Out.LinkingSec->addToSymtab(Sym); for (ObjFile *File : Symtab->ObjectFiles) { LLVM_DEBUG(dbgs() << "Local symtab entries: " << File->getName() << "\n"); for (Symbol *Sym : File->getSymbols()) - if (Sym->isLocal()) - AddSymbol(Sym); + if (Sym->isLocal() && !isa(Sym)) + Out.LinkingSec->addToSymtab(Sym); } } -uint32_t Writer::lookupType(const WasmSignature &Sig) { - auto It = TypeIndices.find(Sig); - if (It == TypeIndices.end()) { - error("type not found: " + toString(Sig)); - return 0; - } - return It->second; -} - -uint32_t Writer::registerType(const WasmSignature &Sig) { - auto Pair = TypeIndices.insert(std::make_pair(Sig, Types.size())); - if (Pair.second) { - LLVM_DEBUG(dbgs() << "type " << toString(Sig) << "\n"); - Types.push_back(&Sig); - } - return Pair.first->second; -} - void Writer::calculateTypes() { // The output type section is the union of the following sets: // 1. Any signature used in the TYPE relocation @@ -1123,162 +502,62 @@ ArrayRef Types = File->getWasmObj()->types(); for (uint32_t I = 0; I < Types.size(); I++) if (File->TypeIsUsed[I]) - File->TypeMap[I] = registerType(Types[I]); + File->TypeMap[I] = Out.TypeSec->registerType(Types[I]); } - for (const Symbol *Sym : ImportedSymbols) { + for (const Symbol *Sym : Out.ImportSec->ImportedSymbols) { if (auto *F = dyn_cast(Sym)) - registerType(*F->Signature); + Out.TypeSec->registerType(*F->Signature); else if (auto *E = dyn_cast(Sym)) - registerType(*E->Signature); + Out.TypeSec->registerType(*E->Signature); } - for (const InputFunction *F : InputFunctions) - registerType(F->Signature); + for (const InputFunction *F : Out.FunctionSec->InputFunctions) + Out.TypeSec->registerType(F->Signature); - for (const InputEvent *E : InputEvents) - registerType(E->Signature); + for (const InputEvent *E : Out.EventSec->InputEvents) + Out.TypeSec->registerType(E->Signature); } -static bool requiresGOTAccess(const Symbol* Sym) { - return Config->Pic && !Sym->isHidden() && !Sym->isLocal(); -} - -void Writer::processRelocations(InputChunk *Chunk) { - if (!Chunk->Live) - return; - ObjFile *File = Chunk->File; - ArrayRef Types = File->getWasmObj()->types(); - for (const WasmRelocation &Reloc : Chunk->getRelocations()) { - if (Reloc.Type == R_WASM_TYPE_INDEX_LEB) { - // Mark target type as live - File->TypeMap[Reloc.Index] = registerType(Types[Reloc.Index]); - File->TypeIsUsed[Reloc.Index] = true; - continue; - } - - // Other relocation types all have a corresponding symbol - auto *Sym = File->getSymbols()[Reloc.Index]; - switch (Reloc.Type) { - case R_WASM_TABLE_INDEX_I32: - case R_WASM_TABLE_INDEX_SLEB: - case R_WASM_TABLE_INDEX_REL_SLEB: { - auto *F = cast(Sym); - if (F->hasTableIndex() || !F->hasFunctionIndex() || requiresGOTAccess(F)) - break; - F->setTableIndex(TableBase + IndirectFunctions.size()); - IndirectFunctions.emplace_back(F); - break; - } - case R_WASM_TYPE_INDEX_LEB: - break; - case R_WASM_GLOBAL_INDEX_LEB: - if (!isa(Sym) && !Sym->isInGOT()) { - Sym->setGOTIndex(NumImportedGlobals++); - GOTSymbols.push_back(Sym); - } - break; - case R_WASM_MEMORY_ADDR_SLEB: - case R_WASM_MEMORY_ADDR_LEB: - case R_WASM_MEMORY_ADDR_REL_SLEB: - if (!Config->Relocatable) { - if (Sym->isUndefined() && !Sym->isWeak()) { - error(toString(File) + ": cannot resolve relocation of type " + - relocTypeToString(Reloc.Type) + - " against undefined (non-weak) data symbol: " + toString(*Sym)); - } - } - break; - } - - if (Config->Pic) { - switch (Reloc.Type) { - case R_WASM_TABLE_INDEX_SLEB: - case R_WASM_MEMORY_ADDR_SLEB: - case R_WASM_MEMORY_ADDR_LEB: - // Certain relocation types can't be used when building PIC output, since - // they would require absolute symbol addresses at link time. - error(toString(File) + ": relocation " + - relocTypeToString(Reloc.Type) + " cannot be used againt symbol " + - toString(*Sym) + "; recompile with -fPIC"); - break; - case R_WASM_TABLE_INDEX_I32: - case R_WASM_MEMORY_ADDR_I32: - // These relocation types are only present in the data section and - // will be converted into code by `generateRelocationCode`. This code - // requires the symbols to have GOT entires. - if (requiresGOTAccess(Sym) && !Sym->isInGOT()) { - Sym->setGOTIndex(NumImportedGlobals++); - GOTSymbols.push_back(Sym); - } - break; - } - } +static void scanRelocations() { + for (ObjFile *File : Symtab->ObjectFiles) { + LLVM_DEBUG(dbgs() << "scanRelocations: " << File->getName() << "\n"); + for (InputChunk *Chunk : File->Functions) + scanRelocations(Chunk); + for (InputChunk *Chunk : File->Segments) + scanRelocations(Chunk); + for (auto &P : File->CustomSections) + scanRelocations(P); } } void Writer::assignIndexes() { - assert(InputFunctions.empty()); - uint32_t FunctionIndex = NumImportedFunctions; - auto AddDefinedFunction = [&](InputFunction *Func) { - if (!Func->Live) - return; - InputFunctions.emplace_back(Func); - Func->setFunctionIndex(FunctionIndex++); - }; + assert(Out.FunctionSec->InputFunctions.empty()); for (InputFunction *Func : Symtab->SyntheticFunctions) - AddDefinedFunction(Func); + Out.FunctionSec->addFunction(Func); for (ObjFile *File : Symtab->ObjectFiles) { LLVM_DEBUG(dbgs() << "Functions: " << File->getName() << "\n"); for (InputFunction *Func : File->Functions) - AddDefinedFunction(Func); - } - - for (ObjFile *File : Symtab->ObjectFiles) { - LLVM_DEBUG(dbgs() << "Handle relocs: " << File->getName() << "\n"); - for (InputChunk *Chunk : File->Functions) - processRelocations(Chunk); - for (InputChunk *Chunk : File->Segments) - processRelocations(Chunk); - for (auto &P : File->CustomSections) - processRelocations(P); + Out.FunctionSec->addFunction(Func); } - assert(InputGlobals.empty()); - uint32_t GlobalIndex = NumImportedGlobals; - auto AddDefinedGlobal = [&](InputGlobal *Global) { - if (Global->Live) { - LLVM_DEBUG(dbgs() << "AddDefinedGlobal: " << GlobalIndex << "\n"); - Global->setGlobalIndex(GlobalIndex++); - InputGlobals.push_back(Global); - } - }; + scanRelocations(); for (InputGlobal *Global : Symtab->SyntheticGlobals) - AddDefinedGlobal(Global); + Out.GlobalSec->addGlobal(Global); for (ObjFile *File : Symtab->ObjectFiles) { LLVM_DEBUG(dbgs() << "Globals: " << File->getName() << "\n"); for (InputGlobal *Global : File->Globals) - AddDefinedGlobal(Global); + Out.GlobalSec->addGlobal(Global); } - assert(InputEvents.empty()); - uint32_t EventIndex = NumImportedEvents; - auto AddDefinedEvent = [&](InputEvent *Event) { - if (Event->Live) { - LLVM_DEBUG(dbgs() << "AddDefinedEvent: " << EventIndex << "\n"); - Event->setEventIndex(EventIndex++); - InputEvents.push_back(Event); - } - }; - for (ObjFile *File : Symtab->ObjectFiles) { LLVM_DEBUG(dbgs() << "Events: " << File->getName() << "\n"); for (InputEvent *Event : File->Events) - AddDefinedEvent(Event); + Out.EventSec->addEvent(Event); } } @@ -1408,6 +687,24 @@ }); } +void Writer::createSyntheticSections() { + Out.DylinkSec = make(); + Out.TypeSec = make(); + Out.ImportSec = make(); + Out.FunctionSec = make(); + Out.TableSec = make(); + Out.MemorySec = make(); + Out.GlobalSec = make(); + Out.EventSec = make(); + Out.ExportSec = make(); + Out.ElemSec = make(TableBase); + Out.DataCountSec = make(Segments.size()); + Out.LinkingSec = make(InitFunctions, Segments); + Out.NameSec = make(); + Out.ProducersSec = make(); + Out.TargetFeaturesSec = make(); +} + void Writer::run() { if (Config->Relocatable || Config->Pic) Config->GlobalBase = 0; @@ -1417,8 +714,14 @@ if (!Config->Pic) TableBase = 1; - log("-- calculateTargetFeatures"); - calculateTargetFeatures(); + log("-- createOutputSegments"); + createOutputSegments(); + log("-- createSyntheticSections"); + createSyntheticSections(); + log("-- populateProducers"); + populateProducers(); + log("-- populateTargetFeatures"); + populateTargetFeatures(); log("-- calculateImports"); calculateImports(); log("-- assignIndexes"); @@ -1438,23 +741,25 @@ calculateExports(); log("-- calculateCustomSections"); calculateCustomSections(); - log("-- assignSymtab"); - assignSymtab(); + log("-- populateSymtab"); + populateSymtab(); + log("-- addSections"); + addSections(); if (errorHandler().Verbose) { - log("Defined Functions: " + Twine(InputFunctions.size())); - log("Defined Globals : " + Twine(InputGlobals.size())); - log("Defined Events : " + Twine(InputEvents.size())); - log("Function Imports : " + Twine(NumImportedFunctions)); - log("Global Imports : " + Twine(NumImportedGlobals)); - log("Event Imports : " + Twine(NumImportedEvents)); + log("Defined Functions: " + Twine(Out.FunctionSec->InputFunctions.size())); + log("Defined Globals : " + Twine(Out.GlobalSec->InputGlobals.size())); + log("Defined Events : " + Twine(Out.EventSec->InputEvents.size())); + log("Function Imports : " + Twine(Out.ImportSec->NumImportedFunctions)); + log("Global Imports : " + Twine(Out.ImportSec->NumImportedGlobals)); + log("Event Imports : " + Twine(Out.ImportSec->NumImportedEvents)); for (ObjFile *File : Symtab->ObjectFiles) File->dumpInfo(); } createHeader(); - log("-- createSections"); - createSections(); + log("-- finalizeSections"); + finalizeSections(); log("-- openFile"); openFile();