X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=tools%2Fgold%2Fgold-plugin.cpp;h=e563376ca9abda504ee14452246d4608a52c8034;hb=019c78a994327bb1d41b2a1f6f8305993058d7a7;hp=3c2da94af532c2bca2d257da930790126d50ae4f;hpb=8483dacfd804c71adc73764cee7f1304d1c8977c;p=oota-llvm.git diff --git a/tools/gold/gold-plugin.cpp b/tools/gold/gold-plugin.cpp index 3c2da94af53..e563376ca9a 100644 --- a/tools/gold/gold-plugin.cpp +++ b/tools/gold/gold-plugin.cpp @@ -13,34 +13,42 @@ //===----------------------------------------------------------------------===// #include "llvm/Config/config.h" // plugin-api.h requires HAVE_STDINT_H -#include "llvm-c/lto.h" +#include "llvm/ADT/DenseSet.h" #include "llvm/ADT/StringSet.h" +#include "llvm/Analysis/TargetLibraryInfo.h" +#include "llvm/Analysis/TargetTransformInfo.h" +#include "llvm/Bitcode/ReaderWriter.h" +#include "llvm/CodeGen/Analysis.h" #include "llvm/CodeGen/CommandFlags.h" -#include "llvm/LTO/LTOCodeGenerator.h" -#include "llvm/LTO/LTOModule.h" -#include "llvm/Support/Errno.h" -#include "llvm/Support/FileSystem.h" +#include "llvm/CodeGen/ParallelCG.h" +#include "llvm/IR/AutoUpgrade.h" +#include "llvm/IR/Constants.h" +#include "llvm/IR/DiagnosticInfo.h" +#include "llvm/IR/DiagnosticPrinter.h" +#include "llvm/IR/LLVMContext.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/Module.h" +#include "llvm/IR/Verifier.h" +#include "llvm/Linker/Linker.h" +#include "llvm/MC/SubtargetFeature.h" +#include "llvm/Object/IRObjectFile.h" +#include "llvm/Object/FunctionIndexObjectFile.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Support/Host.h" +#include "llvm/Support/ManagedStatic.h" #include "llvm/Support/MemoryBuffer.h" -#include "llvm/Support/Path.h" -#include "llvm/Support/Program.h" +#include "llvm/Support/TargetRegistry.h" #include "llvm/Support/TargetSelect.h" -#include "llvm/Support/ToolOutputFile.h" -#include -#include -#include -#include +#include "llvm/Transforms/IPO.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/GlobalStatus.h" +#include "llvm/Transforms/Utils/ModuleUtils.h" +#include "llvm/Transforms/Utils/ValueMapper.h" #include #include #include #include -// Support Windows/MinGW crazyness. -#ifdef _WIN32 -# include -# define lseek _lseek -# define read _read -#endif - #ifndef LDPO_PIE // FIXME: remove this declaration when we stop maintaining Ubuntu Quantal and // Precise and Debian Wheezy (binutils 2.23 is required) @@ -62,39 +70,54 @@ static ld_plugin_status discard_message(int level, const char *format, ...) { abort(); } -static ld_plugin_add_symbols add_symbols = NULL; -static ld_plugin_get_symbols get_symbols = NULL; -static ld_plugin_add_input_file add_input_file = NULL; -static ld_plugin_set_extra_library_path set_extra_library_path = NULL; -static ld_plugin_get_view get_view = NULL; +static ld_plugin_get_input_file get_input_file = nullptr; +static ld_plugin_release_input_file release_input_file = nullptr; +static ld_plugin_add_symbols add_symbols = nullptr; +static ld_plugin_get_symbols get_symbols = nullptr; +static ld_plugin_add_input_file add_input_file = nullptr; +static ld_plugin_set_extra_library_path set_extra_library_path = nullptr; +static ld_plugin_get_view get_view = nullptr; static ld_plugin_message message = discard_message; -static lto_codegen_model output_type = LTO_CODEGEN_PIC_MODEL_STATIC; +static Reloc::Model RelocationModel = Reloc::Default; static std::string output_name = ""; static std::list Modules; static std::vector Cleanup; -static LTOCodeGenerator *CodeGen = nullptr; -static StringSet<> CannotBeHidden; static llvm::TargetOptions TargetOpts; namespace options { - enum generate_bc { BC_NO, BC_ALSO, BC_ONLY }; + enum OutputType { + OT_NORMAL, + OT_DISABLE, + OT_BC_ONLY, + OT_SAVE_TEMPS + }; static bool generate_api_file = false; - static generate_bc generate_bc_file = BC_NO; - static std::string bc_path; + static OutputType TheOutputType = OT_NORMAL; + static unsigned OptLevel = 2; + static unsigned Parallelism = 1; +#ifdef NDEBUG + static bool DisableVerify = true; +#else + static bool DisableVerify = false; +#endif static std::string obj_path; static std::string extra_library_path; static std::string triple; static std::string mcpu; + // When the thinlto plugin option is specified, only read the function + // the information from intermediate files and write a combined + // global index for the ThinLTO backends. + static bool thinlto = false; // Additional options to pass into the code generator. // Note: This array will contain all plugin options which are not claimed // as plugin exclusive to pass to the code generator. // For example, "generate-api-file" and "as"options are for the plugin // use only and will not be passed. - static std::vector extra; + static std::vector extra; - static void process_plugin_option(const char* opt_) + static void process_plugin_option(const char *opt_) { - if (opt_ == NULL) + if (opt_ == nullptr) return; llvm::StringRef opt = opt_; @@ -109,21 +132,30 @@ namespace options { } else if (opt.startswith("obj-path=")) { obj_path = opt.substr(strlen("obj-path=")); } else if (opt == "emit-llvm") { - generate_bc_file = BC_ONLY; - } else if (opt == "also-emit-llvm") { - generate_bc_file = BC_ALSO; - } else if (opt.startswith("also-emit-llvm=")) { - llvm::StringRef path = opt.substr(strlen("also-emit-llvm=")); - generate_bc_file = BC_ALSO; - if (!bc_path.empty()) { - (*message)(LDPL_WARNING, "Path to the output IL file specified twice. " - "Discarding %s", opt_); - } else { - bc_path = path; - } + TheOutputType = OT_BC_ONLY; + } else if (opt == "save-temps") { + TheOutputType = OT_SAVE_TEMPS; + } else if (opt == "disable-output") { + TheOutputType = OT_DISABLE; + } else if (opt == "thinlto") { + thinlto = true; + } else if (opt.size() == 2 && opt[0] == 'O') { + if (opt[1] < '0' || opt[1] > '3') + message(LDPL_FATAL, "Optimization level must be between 0 and 3"); + OptLevel = opt[1] - '0'; + } else if (opt.startswith("jobs=")) { + if (StringRef(opt_ + 5).getAsInteger(10, Parallelism)) + message(LDPL_FATAL, "Invalid parallelism level: %s", opt_ + 5); + } else if (opt == "disable-verify") { + DisableVerify = true; } else { // Save this option to pass to the code generator. - extra.push_back(opt); + // ParseCommandLineOptions() expects argv[0] to be program name. Lazily + // add that. + if (extra.empty()) + extra.push_back("LLVMgold"); + + extra.push_back(opt_); } } } @@ -135,6 +167,12 @@ static ld_plugin_status cleanup_hook(void); extern "C" ld_plugin_status onload(ld_plugin_tv *tv); ld_plugin_status onload(ld_plugin_tv *tv) { + InitializeAllTargetInfos(); + InitializeAllTargets(); + InitializeAllTargetMCs(); + InitializeAllAsmParsers(); + InitializeAllAsmPrinters(); + // We're given a pointer to the first transfer vector. We read through them // until we find one where tv_tag == LDPT_NULL. The REGISTER_* tagged values // contain pointers to functions that we need to call to register our own @@ -154,14 +192,13 @@ ld_plugin_status onload(ld_plugin_tv *tv) { case LDPO_REL: // .o case LDPO_DYN: // .so case LDPO_PIE: // position independent executable - output_type = LTO_CODEGEN_PIC_MODEL_DYNAMIC; + RelocationModel = Reloc::PIC_; break; case LDPO_EXEC: // .exe - output_type = LTO_CODEGEN_PIC_MODEL_STATIC; + RelocationModel = Reloc::Static; break; default: - (*message)(LDPL_ERROR, "Unknown output file type %d", - tv->tv_u.tv_val); + message(LDPL_ERROR, "Unknown output file type %d", tv->tv_u.tv_val); return LDPS_ERR; } break; @@ -172,7 +209,7 @@ ld_plugin_status onload(ld_plugin_tv *tv) { ld_plugin_register_claim_file callback; callback = tv->tv_u.tv_register_claim_file; - if ((*callback)(claim_file_hook) != LDPS_OK) + if (callback(claim_file_hook) != LDPS_OK) return LDPS_ERR; registeredClaimFile = true; @@ -181,7 +218,7 @@ ld_plugin_status onload(ld_plugin_tv *tv) { ld_plugin_register_all_symbols_read callback; callback = tv->tv_u.tv_register_all_symbols_read; - if ((*callback)(all_symbols_read_hook) != LDPS_OK) + if (callback(all_symbols_read_hook) != LDPS_OK) return LDPS_ERR; RegisteredAllSymbolsRead = true; @@ -190,9 +227,15 @@ ld_plugin_status onload(ld_plugin_tv *tv) { ld_plugin_register_cleanup callback; callback = tv->tv_u.tv_register_cleanup; - if ((*callback)(cleanup_hook) != LDPS_OK) + if (callback(cleanup_hook) != LDPS_OK) return LDPS_ERR; } break; + case LDPT_GET_INPUT_FILE: + get_input_file = tv->tv_u.tv_get_input_file; + break; + case LDPT_RELEASE_INPUT_FILE: + release_input_file = tv->tv_u.tv_release_input_file; + break; case LDPT_ADD_SYMBOLS: add_symbols = tv->tv_u.tv_add_symbols; break; @@ -217,61 +260,94 @@ ld_plugin_status onload(ld_plugin_tv *tv) { } if (!registeredClaimFile) { - (*message)(LDPL_ERROR, "register_claim_file not passed to LLVMgold."); + message(LDPL_ERROR, "register_claim_file not passed to LLVMgold."); return LDPS_ERR; } if (!add_symbols) { - (*message)(LDPL_ERROR, "add_symbols not passed to LLVMgold."); + message(LDPL_ERROR, "add_symbols not passed to LLVMgold."); return LDPS_ERR; } if (!RegisteredAllSymbolsRead) return LDPS_OK; - InitializeAllTargetInfos(); - InitializeAllTargets(); - InitializeAllTargetMCs(); - InitializeAllAsmParsers(); - InitializeAllAsmPrinters(); - CodeGen = new LTOCodeGenerator(); - if (MAttrs.size()) { - std::string Attrs; - for (unsigned I = 0; I < MAttrs.size(); ++I) { - if (I > 0) - Attrs.append(","); - Attrs.append(MAttrs[I]); - } - CodeGen->setAttr(Attrs.c_str()); + if (!get_input_file) { + message(LDPL_ERROR, "get_input_file not passed to LLVMgold."); + return LDPS_ERR; + } + if (!release_input_file) { + message(LDPL_ERROR, "relesase_input_file not passed to LLVMgold."); + return LDPS_ERR; } - // Pass through extra options to the code generator. - if (!options::extra.empty()) { - for (std::vector::iterator it = options::extra.begin(); - it != options::extra.end(); ++it) { - CodeGen->setCodeGenDebugOptions((*it).c_str()); - } + return LDPS_OK; +} + +static const GlobalObject *getBaseObject(const GlobalValue &GV) { + if (auto *GA = dyn_cast(&GV)) + return GA->getBaseObject(); + return cast(&GV); +} + +static bool shouldSkip(uint32_t Symflags) { + if (!(Symflags & object::BasicSymbolRef::SF_Global)) + return true; + if (Symflags & object::BasicSymbolRef::SF_FormatSpecific) + return true; + return false; +} + +static void diagnosticHandler(const DiagnosticInfo &DI) { + if (const auto *BDI = dyn_cast(&DI)) { + std::error_code EC = BDI->getError(); + if (EC == BitcodeError::InvalidBitcodeSignature) + return; } - CodeGen->parseCodeGenDebugOptions(); - TargetOpts = InitTargetOptionsFromCodeGenFlags(); - CodeGen->setTargetOptions(TargetOpts); + std::string ErrStorage; + { + raw_string_ostream OS(ErrStorage); + DiagnosticPrinterRawOStream DP(OS); + DI.print(DP); + } + ld_plugin_level Level; + switch (DI.getSeverity()) { + case DS_Error: + message(LDPL_FATAL, "LLVM gold plugin has failed to create LTO module: %s", + ErrStorage.c_str()); + llvm_unreachable("Fatal doesn't return."); + case DS_Warning: + Level = LDPL_WARNING; + break; + case DS_Note: + case DS_Remark: + Level = LDPL_INFO; + break; + } + message(Level, "LLVM gold plugin: %s", ErrStorage.c_str()); +} - return LDPS_OK; +static void diagnosticHandlerForContext(const DiagnosticInfo &DI, + void *Context) { + diagnosticHandler(DI); } -/// claim_file_hook - called by gold to see whether this file is one that -/// our plugin can handle. We'll try to open it and register all the symbols -/// with add_symbol if possible. +/// Called by gold to see whether this file is one that our plugin can handle. +/// We'll try to open it and register all the symbols with add_symbol if +/// possible. static ld_plugin_status claim_file_hook(const ld_plugin_input_file *file, int *claimed) { - LTOModule *M; - const void *view; - std::unique_ptr buffer; + LLVMContext Context; + MemoryBufferRef BufferRef; + std::unique_ptr Buffer; if (get_view) { + const void *view; if (get_view(file->handle, &view) != LDPS_OK) { - (*message)(LDPL_ERROR, "Failed to get a view of %s", file->name); + message(LDPL_ERROR, "Failed to get a view of %s", file->name); return LDPS_ERR; } + BufferRef = + MemoryBufferRef(StringRef((const char *)view, file->filesize), ""); } else { int64_t offset = 0; // Gold has found what might be IR part-way inside of a file, such as @@ -279,226 +355,684 @@ static ld_plugin_status claim_file_hook(const ld_plugin_input_file *file, if (file->offset) { offset = file->offset; } - if (std::error_code ec = MemoryBuffer::getOpenFileSlice( - file->fd, file->name, buffer, file->filesize, offset)) { - (*message)(LDPL_ERROR, ec.message().c_str()); + ErrorOr> BufferOrErr = + MemoryBuffer::getOpenFileSlice(file->fd, file->name, file->filesize, + offset); + if (std::error_code EC = BufferOrErr.getError()) { + message(LDPL_ERROR, EC.message().c_str()); return LDPS_ERR; } - view = buffer->getBufferStart(); + Buffer = std::move(BufferOrErr.get()); + BufferRef = Buffer->getMemBufferRef(); } - if (!LTOModule::isBitcodeFile(view, file->filesize)) + Context.setDiagnosticHandler(diagnosticHandlerForContext); + ErrorOr> ObjOrErr = + object::IRObjectFile::create(BufferRef, Context); + std::error_code EC = ObjOrErr.getError(); + if (EC == object::object_error::invalid_file_type || + EC == object::object_error::bitcode_section_not_found) return LDPS_OK; - std::string Error; - M = LTOModule::makeLTOModule(view, file->filesize, TargetOpts, Error); - if (!M) { - (*message)(LDPL_ERROR, - "LLVM gold plugin has failed to create LTO module: %s", - Error.c_str()); - return LDPS_OK; + *claimed = 1; + + if (EC) { + message(LDPL_ERROR, "LLVM gold plugin has failed to create LTO module: %s", + EC.message().c_str()); + return LDPS_ERR; } + std::unique_ptr Obj = std::move(*ObjOrErr); - *claimed = 1; Modules.resize(Modules.size() + 1); claimed_file &cf = Modules.back(); - if (!options::triple.empty()) - M->setTargetTriple(options::triple.c_str()); - cf.handle = file->handle; - unsigned sym_count = M->getSymbolCount(); - cf.syms.reserve(sym_count); - for (unsigned i = 0; i != sym_count; ++i) { - lto_symbol_attributes attrs = M->getSymbolAttributes(i); - if ((attrs & LTO_SYMBOL_SCOPE_MASK) == LTO_SYMBOL_SCOPE_INTERNAL) + // If we are doing ThinLTO compilation, don't need to process the symbols. + // Later we simply build a combined index file after all files are claimed. + if (options::thinlto) + return LDPS_OK; + + for (auto &Sym : Obj->symbols()) { + uint32_t Symflags = Sym.getFlags(); + if (shouldSkip(Symflags)) continue; cf.syms.push_back(ld_plugin_symbol()); ld_plugin_symbol &sym = cf.syms.back(); - sym.name = strdup(M->getSymbolName(i)); - sym.version = NULL; - - int scope = attrs & LTO_SYMBOL_SCOPE_MASK; - bool CanBeHidden = scope == LTO_SYMBOL_SCOPE_DEFAULT_CAN_BE_HIDDEN; - if (!CanBeHidden) - CannotBeHidden.insert(sym.name); - switch (scope) { - case LTO_SYMBOL_SCOPE_HIDDEN: + sym.version = nullptr; + + SmallString<64> Name; + { + raw_svector_ostream OS(Name); + Sym.printName(OS); + } + sym.name = strdup(Name.c_str()); + + const GlobalValue *GV = Obj->getSymbolGV(Sym.getRawDataRefImpl()); + + sym.visibility = LDPV_DEFAULT; + if (GV) { + switch (GV->getVisibility()) { + case GlobalValue::DefaultVisibility: + sym.visibility = LDPV_DEFAULT; + break; + case GlobalValue::HiddenVisibility: sym.visibility = LDPV_HIDDEN; break; - case LTO_SYMBOL_SCOPE_PROTECTED: + case GlobalValue::ProtectedVisibility: sym.visibility = LDPV_PROTECTED; break; - case 0: // extern - case LTO_SYMBOL_SCOPE_DEFAULT: - case LTO_SYMBOL_SCOPE_DEFAULT_CAN_BE_HIDDEN: - sym.visibility = LDPV_DEFAULT; - break; - default: - (*message)(LDPL_ERROR, "Unknown scope attribute: %d", scope); - return LDPS_ERR; + } } - int definition = attrs & LTO_SYMBOL_DEFINITION_MASK; - sym.comdat_key = NULL; - switch (definition) { - case LTO_SYMBOL_DEFINITION_REGULAR: - sym.def = LDPK_DEF; - break; - case LTO_SYMBOL_DEFINITION_UNDEFINED: - sym.def = LDPK_UNDEF; - break; - case LTO_SYMBOL_DEFINITION_TENTATIVE: - sym.def = LDPK_COMMON; - break; - case LTO_SYMBOL_DEFINITION_WEAK: - sym.comdat_key = sym.name; - sym.def = LDPK_WEAKDEF; - break; - case LTO_SYMBOL_DEFINITION_WEAKUNDEF: + if (Symflags & object::BasicSymbolRef::SF_Undefined) { + sym.def = LDPK_UNDEF; + if (GV && GV->hasExternalWeakLinkage()) sym.def = LDPK_WEAKUNDEF; - break; - default: - (*message)(LDPL_ERROR, "Unknown definition attribute: %d", definition); - return LDPS_ERR; + } else { + sym.def = LDPK_DEF; + if (GV) { + assert(!GV->hasExternalWeakLinkage() && + !GV->hasAvailableExternallyLinkage() && "Not a declaration!"); + if (GV->hasCommonLinkage()) + sym.def = LDPK_COMMON; + else if (GV->isWeakForLinker()) + sym.def = LDPK_WEAKDEF; + } } sym.size = 0; + sym.comdat_key = nullptr; + if (GV) { + const GlobalObject *Base = getBaseObject(*GV); + if (!Base) + message(LDPL_FATAL, "Unable to determine comdat of alias!"); + const Comdat *C = Base->getComdat(); + if (C) + sym.comdat_key = strdup(C->getName().str().c_str()); + } sym.resolution = LDPR_UNKNOWN; } - cf.syms.reserve(cf.syms.size()); - if (!cf.syms.empty()) { - if ((*add_symbols)(cf.handle, cf.syms.size(), &cf.syms[0]) != LDPS_OK) { - (*message)(LDPL_ERROR, "Unable to add symbols!"); + if (add_symbols(cf.handle, cf.syms.size(), cf.syms.data()) != LDPS_OK) { + message(LDPL_ERROR, "Unable to add symbols!"); return LDPS_ERR; } } - if (CodeGen) { - std::string Error; - if (!CodeGen->addModule(M, Error)) { - (*message)(LDPL_ERROR, "Error linking module: %s", Error.c_str()); - return LDPS_ERR; - } + return LDPS_OK; +} + +static void keepGlobalValue(GlobalValue &GV, + std::vector &KeptAliases) { + assert(!GV.hasLocalLinkage()); + + if (auto *GA = dyn_cast(&GV)) + KeptAliases.push_back(GA); + + switch (GV.getLinkage()) { + default: + break; + case GlobalValue::LinkOnceAnyLinkage: + GV.setLinkage(GlobalValue::WeakAnyLinkage); + break; + case GlobalValue::LinkOnceODRLinkage: + GV.setLinkage(GlobalValue::WeakODRLinkage); + break; } - delete M; + assert(!GV.isDiscardableIfUnused()); +} - return LDPS_OK; +static void internalize(GlobalValue &GV) { + if (GV.isDeclarationForLinker()) + return; // We get here if there is a matching asm definition. + if (!GV.hasLocalLinkage()) + GV.setLinkage(GlobalValue::InternalLinkage); } -static bool mustPreserve(const claimed_file &F, int i) { - if (F.syms[i].resolution == LDPR_PREVAILING_DEF) - return true; - if (F.syms[i].resolution == LDPR_PREVAILING_DEF_IRONLY_EXP) - return CannotBeHidden.count(F.syms[i].name); - return false; +static void drop(GlobalValue &GV) { + if (auto *F = dyn_cast(&GV)) { + F->deleteBody(); + F->setComdat(nullptr); // Should deleteBody do this? + return; + } + + if (auto *Var = dyn_cast(&GV)) { + Var->setInitializer(nullptr); + Var->setLinkage( + GlobalValue::ExternalLinkage); // Should setInitializer do this? + Var->setComdat(nullptr); // and this? + return; + } + + auto &Alias = cast(GV); + Module &M = *Alias.getParent(); + PointerType &Ty = *cast(Alias.getType()); + GlobalValue::LinkageTypes L = Alias.getLinkage(); + auto *Var = + new GlobalVariable(M, Ty.getElementType(), /*isConstant*/ false, L, + /*Initializer*/ nullptr); + Var->takeName(&Alias); + Alias.replaceAllUsesWith(Var); + Alias.eraseFromParent(); } -/// all_symbols_read_hook - gold informs us that all symbols have been read. -/// At this point, we use get_symbols to see if any of our definitions have -/// been overridden by a native object file. Then, perform optimization and -/// codegen. -static ld_plugin_status all_symbols_read_hook(void) { - std::ofstream api_file; - assert(CodeGen); - - if (options::generate_api_file) { - api_file.open("apifile.txt", std::ofstream::out | std::ofstream::trunc); - if (!api_file.is_open()) { - (*message)(LDPL_FATAL, "Unable to open apifile.txt for writing."); - abort(); - } +static const char *getResolutionName(ld_plugin_symbol_resolution R) { + switch (R) { + case LDPR_UNKNOWN: + return "UNKNOWN"; + case LDPR_UNDEF: + return "UNDEF"; + case LDPR_PREVAILING_DEF: + return "PREVAILING_DEF"; + case LDPR_PREVAILING_DEF_IRONLY: + return "PREVAILING_DEF_IRONLY"; + case LDPR_PREEMPTED_REG: + return "PREEMPTED_REG"; + case LDPR_PREEMPTED_IR: + return "PREEMPTED_IR"; + case LDPR_RESOLVED_IR: + return "RESOLVED_IR"; + case LDPR_RESOLVED_EXEC: + return "RESOLVED_EXEC"; + case LDPR_RESOLVED_DYN: + return "RESOLVED_DYN"; + case LDPR_PREVAILING_DEF_IRONLY_EXP: + return "PREVAILING_DEF_IRONLY_EXP"; + } + llvm_unreachable("Unknown resolution"); +} + +namespace { +class LocalValueMaterializer final : public ValueMaterializer { + DenseSet &Dropped; + DenseMap LocalVersions; + +public: + LocalValueMaterializer(DenseSet &Dropped) : Dropped(Dropped) {} + Value *materializeValueFor(Value *V) override; +}; +} + +Value *LocalValueMaterializer::materializeValueFor(Value *V) { + auto *GO = dyn_cast(V); + if (!GO) + return nullptr; + + auto I = LocalVersions.find(GO); + if (I != LocalVersions.end()) + return I->second; + + if (!Dropped.count(GO)) + return nullptr; + + Module &M = *GO->getParent(); + GlobalValue::LinkageTypes L = GO->getLinkage(); + GlobalObject *Declaration; + if (auto *F = dyn_cast(GO)) { + Declaration = Function::Create(F->getFunctionType(), L, "", &M); + } else { + auto *Var = cast(GO); + Declaration = new GlobalVariable(M, Var->getType()->getElementType(), + Var->isConstant(), L, + /*Initializer*/ nullptr); } + Declaration->takeName(GO); + Declaration->copyAttributesFrom(GO); + + GO->setLinkage(GlobalValue::InternalLinkage); + GO->setName(Declaration->getName()); + Dropped.erase(GO); + GO->replaceAllUsesWith(Declaration); + + LocalVersions[Declaration] = GO; + + return GO; +} + +static Constant *mapConstantToLocalCopy(Constant *C, ValueToValueMapTy &VM, + LocalValueMaterializer *Materializer) { + return MapValue(C, VM, RF_IgnoreMissingEntries, nullptr, Materializer); +} + +static void freeSymName(ld_plugin_symbol &Sym) { + free(Sym.name); + free(Sym.comdat_key); + Sym.name = nullptr; + Sym.comdat_key = nullptr; +} - for (std::list::iterator I = Modules.begin(), - E = Modules.end(); I != E; ++I) { - if (I->syms.empty()) +static std::unique_ptr +getFunctionIndexForFile(claimed_file &F, ld_plugin_input_file &Info) { + + if (get_symbols(F.handle, F.syms.size(), &F.syms[0]) != LDPS_OK) + message(LDPL_FATAL, "Failed to get symbol information"); + + const void *View; + if (get_view(F.handle, &View) != LDPS_OK) + message(LDPL_FATAL, "Failed to get a view of file"); + + MemoryBufferRef BufferRef(StringRef((const char *)View, Info.filesize), + Info.name); + + // Don't bother trying to build an index if there is no summary information + // in this bitcode file. + if (!object::FunctionIndexObjectFile::hasFunctionSummaryInMemBuffer( + BufferRef, diagnosticHandler)) + return std::unique_ptr(nullptr); + + ErrorOr> ObjOrErr = + object::FunctionIndexObjectFile::create(BufferRef, diagnosticHandler); + + if (std::error_code EC = ObjOrErr.getError()) + message(LDPL_FATAL, "Could not read function index bitcode from file : %s", + EC.message().c_str()); + + object::FunctionIndexObjectFile &Obj = **ObjOrErr; + + return Obj.takeIndex(); +} + +static std::unique_ptr +getModuleForFile(LLVMContext &Context, claimed_file &F, + ld_plugin_input_file &Info, raw_fd_ostream *ApiFile, + StringSet<> &Internalize, StringSet<> &Maybe) { + + if (get_symbols(F.handle, F.syms.size(), F.syms.data()) != LDPS_OK) + message(LDPL_FATAL, "Failed to get symbol information"); + + const void *View; + if (get_view(F.handle, &View) != LDPS_OK) + message(LDPL_FATAL, "Failed to get a view of file"); + + MemoryBufferRef BufferRef(StringRef((const char *)View, Info.filesize), + Info.name); + ErrorOr> ObjOrErr = + object::IRObjectFile::create(BufferRef, Context); + + if (std::error_code EC = ObjOrErr.getError()) + message(LDPL_FATAL, "Could not read bitcode from file : %s", + EC.message().c_str()); + + object::IRObjectFile &Obj = **ObjOrErr; + + Module &M = Obj.getModule(); + + M.materializeMetadata(); + UpgradeDebugInfo(M); + + SmallPtrSet Used; + collectUsedGlobalVariables(M, Used, /*CompilerUsed*/ false); + + DenseSet Drop; + std::vector KeptAliases; + + unsigned SymNum = 0; + for (auto &ObjSym : Obj.symbols()) { + if (shouldSkip(ObjSym.getFlags())) continue; - (*get_symbols)(I->handle, I->syms.size(), &I->syms[0]); - for (unsigned i = 0, e = I->syms.size(); i != e; i++) { - if (mustPreserve(*I, i)) { - CodeGen->addMustPreserveSymbol(I->syms[i].name); + ld_plugin_symbol &Sym = F.syms[SymNum]; + ++SymNum; + + ld_plugin_symbol_resolution Resolution = + (ld_plugin_symbol_resolution)Sym.resolution; + + if (options::generate_api_file) + *ApiFile << Sym.name << ' ' << getResolutionName(Resolution) << '\n'; + + GlobalValue *GV = Obj.getSymbolGV(ObjSym.getRawDataRefImpl()); + if (!GV) { + freeSymName(Sym); + continue; // Asm symbol. + } + + if (Resolution != LDPR_PREVAILING_DEF_IRONLY && GV->hasCommonLinkage()) { + // Common linkage is special. There is no single symbol that wins the + // resolution. Instead we have to collect the maximum alignment and size. + // The IR linker does that for us if we just pass it every common GV. + // We still have to keep track of LDPR_PREVAILING_DEF_IRONLY so we + // internalize once the IR linker has done its job. + freeSymName(Sym); + continue; + } - if (options::generate_api_file) - api_file << I->syms[i].name << "\n"; + switch (Resolution) { + case LDPR_UNKNOWN: + llvm_unreachable("Unexpected resolution"); + + case LDPR_RESOLVED_IR: + case LDPR_RESOLVED_EXEC: + case LDPR_RESOLVED_DYN: + assert(GV->isDeclarationForLinker()); + break; + + case LDPR_UNDEF: + if (!GV->isDeclarationForLinker()) { + assert(GV->hasComdat()); + Drop.insert(GV); + } + break; + + case LDPR_PREVAILING_DEF_IRONLY: { + keepGlobalValue(*GV, KeptAliases); + if (!Used.count(GV)) { + // Since we use the regular lib/Linker, we cannot just internalize GV + // now or it will not be copied to the merged module. Instead we force + // it to be copied and then internalize it. + Internalize.insert(GV->getName()); } + break; } - } - if (options::generate_api_file) - api_file.close(); + case LDPR_PREVAILING_DEF: + keepGlobalValue(*GV, KeptAliases); + break; - CodeGen->setCodePICModel(output_type); - CodeGen->setDebugInfo(LTO_DEBUG_MODEL_DWARF); - if (!options::mcpu.empty()) - CodeGen->setCpu(options::mcpu.c_str()); + case LDPR_PREEMPTED_IR: + // Gold might have selected a linkonce_odr and preempted a weak_odr. + // In that case we have to make sure we don't end up internalizing it. + if (!GV->isDiscardableIfUnused()) + Maybe.erase(GV->getName()); - if (options::generate_bc_file != options::BC_NO) { - std::string path; - if (options::generate_bc_file == options::BC_ONLY) - path = output_name; - else if (!options::bc_path.empty()) - path = options::bc_path; - else - path = output_name + ".bc"; - std::string Error; - if (!CodeGen->writeMergedModules(path.c_str(), Error)) - (*message)(LDPL_FATAL, "Failed to write the output file."); - if (options::generate_bc_file == options::BC_ONLY) { - delete CodeGen; - exit(0); + // fall-through + case LDPR_PREEMPTED_REG: + Drop.insert(GV); + break; + + case LDPR_PREVAILING_DEF_IRONLY_EXP: { + // We can only check for address uses after we merge the modules. The + // reason is that this GV might have a copy in another module + // and in that module the address might be significant, but that + // copy will be LDPR_PREEMPTED_IR. + if (GV->hasLinkOnceODRLinkage()) + Maybe.insert(GV->getName()); + keepGlobalValue(*GV, KeptAliases); + break; } + } + + freeSymName(Sym); + } + + ValueToValueMapTy VM; + LocalValueMaterializer Materializer(Drop); + for (GlobalAlias *GA : KeptAliases) { + // Gold told us to keep GA. It is possible that a GV usied in the aliasee + // expression is being dropped. If that is the case, that GV must be copied. + Constant *Aliasee = GA->getAliasee(); + Constant *Replacement = mapConstantToLocalCopy(Aliasee, VM, &Materializer); + GA->setAliasee(Replacement); } - std::string ObjPath; + for (auto *GV : Drop) + drop(*GV); + + return Obj.takeModule(); +} + +static void runLTOPasses(Module &M, TargetMachine &TM) { + M.setDataLayout(TM.createDataLayout()); + + legacy::PassManager passes; + passes.add(createTargetTransformInfoWrapperPass(TM.getTargetIRAnalysis())); + + PassManagerBuilder PMB; + PMB.LibraryInfo = new TargetLibraryInfoImpl(Triple(TM.getTargetTriple())); + PMB.Inliner = createFunctionInliningPass(); + // Unconditionally verify input since it is not verified before this + // point and has unknown origin. + PMB.VerifyInput = true; + PMB.VerifyOutput = !options::DisableVerify; + PMB.LoopVectorize = true; + PMB.SLPVectorize = true; + PMB.OptLevel = options::OptLevel; + PMB.populateLTOPassManager(passes); + passes.run(M); +} + +static void saveBCFile(StringRef Path, Module &M) { + std::error_code EC; + raw_fd_ostream OS(Path, EC, sys::fs::OpenFlags::F_None); + if (EC) + message(LDPL_FATAL, "Failed to write the output file."); + WriteBitcodeToFile(&M, OS, /* ShouldPreserveUseListOrder */ true); +} + +static void codegen(std::unique_ptr M) { + const std::string &TripleStr = M->getTargetTriple(); + Triple TheTriple(TripleStr); + + std::string ErrMsg; + const Target *TheTarget = TargetRegistry::lookupTarget(TripleStr, ErrMsg); + if (!TheTarget) + message(LDPL_FATAL, "Target not found: %s", ErrMsg.c_str()); + + if (unsigned NumOpts = options::extra.size()) + cl::ParseCommandLineOptions(NumOpts, &options::extra[0]); + + SubtargetFeatures Features; + Features.getDefaultSubtargetFeatures(TheTriple); + for (const std::string &A : MAttrs) + Features.AddFeature(A); + + TargetOptions Options = InitTargetOptionsFromCodeGenFlags(); + CodeGenOpt::Level CGOptLevel; + switch (options::OptLevel) { + case 0: + CGOptLevel = CodeGenOpt::None; + break; + case 1: + CGOptLevel = CodeGenOpt::Less; + break; + case 2: + CGOptLevel = CodeGenOpt::Default; + break; + case 3: + CGOptLevel = CodeGenOpt::Aggressive; + break; + } + std::unique_ptr TM(TheTarget->createTargetMachine( + TripleStr, options::mcpu, Features.getString(), Options, RelocationModel, + CodeModel::Default, CGOptLevel)); + + runLTOPasses(*M, *TM); + + if (options::TheOutputType == options::OT_SAVE_TEMPS) + saveBCFile(output_name + ".opt.bc", *M); + + SmallString<128> Filename; + if (!options::obj_path.empty()) + Filename = options::obj_path; + else if (options::TheOutputType == options::OT_SAVE_TEMPS) + Filename = output_name + ".o"; + + std::vector> Filenames(options::Parallelism); + bool TempOutFile = Filename.empty(); { - const char *Temp; - std::string Error; - if (!CodeGen->compile_to_file(&Temp, /*DisableOpt*/ false, /*DisableInline*/ - false, /*DisableGVNLoadPRE*/ false, Error)) - (*message)(LDPL_ERROR, "Could not produce a combined object file\n"); - ObjPath = Temp; + // Open a file descriptor for each backend thread. This is done in a block + // so that the output file descriptors are closed before gold opens them. + std::list OSs; + std::vector OSPtrs(options::Parallelism); + for (unsigned I = 0; I != options::Parallelism; ++I) { + int FD; + if (TempOutFile) { + std::error_code EC = + sys::fs::createTemporaryFile("lto-llvm", "o", FD, Filenames[I]); + if (EC) + message(LDPL_FATAL, "Could not create temporary file: %s", + EC.message().c_str()); + } else { + Filenames[I] = Filename; + if (options::Parallelism != 1) + Filenames[I] += utostr(I); + std::error_code EC = + sys::fs::openFileForWrite(Filenames[I], FD, sys::fs::F_None); + if (EC) + message(LDPL_FATAL, "Could not open file: %s", EC.message().c_str()); + } + OSs.emplace_back(FD, true); + OSPtrs[I] = &OSs.back(); + } + + // Run backend threads. + splitCodeGen(std::move(M), OSPtrs, options::mcpu, Features.getString(), + Options, RelocationModel, CodeModel::Default, CGOptLevel); } - delete CodeGen; - for (std::list::iterator I = Modules.begin(), - E = Modules.end(); I != E; ++I) { - for (unsigned i = 0; i != I->syms.size(); ++i) { - ld_plugin_symbol &sym = I->syms[i]; - free(sym.name); + for (auto &Filename : Filenames) { + if (add_input_file(Filename.c_str()) != LDPS_OK) + message(LDPL_FATAL, + "Unable to add .o file to the link. File left behind in: %s", + Filename.c_str()); + if (TempOutFile) + Cleanup.push_back(Filename.c_str()); + } +} + +/// gold informs us that all symbols have been read. At this point, we use +/// get_symbols to see if any of our definitions have been overridden by a +/// native object file. Then, perform optimization and codegen. +static ld_plugin_status allSymbolsReadHook(raw_fd_ostream *ApiFile) { + if (Modules.empty()) + return LDPS_OK; + + LLVMContext Context; + Context.setDiagnosticHandler(diagnosticHandlerForContext, nullptr, true); + + // If we are doing ThinLTO compilation, simply build the combined + // function index/summary and emit it. We don't need to parse the modules + // and link them in this case. + if (options::thinlto) { + FunctionInfoIndex CombinedIndex; + uint64_t NextModuleId = 0; + for (claimed_file &F : Modules) { + ld_plugin_input_file File; + if (get_input_file(F.handle, &File) != LDPS_OK) + message(LDPL_FATAL, "Failed to get file information"); + + std::unique_ptr Index = + getFunctionIndexForFile(F, File); + + // Skip files without a function summary. + if (!Index) + continue; + + CombinedIndex.mergeFrom(std::move(Index), ++NextModuleId); } + + std::error_code EC; + raw_fd_ostream OS(output_name + ".thinlto.bc", EC, + sys::fs::OpenFlags::F_None); + if (EC) + message(LDPL_FATAL, "Unable to open %s.thinlto.bc for writing: %s", + output_name.data(), EC.message().c_str()); + WriteFunctionSummaryToFile(CombinedIndex, OS); + OS.close(); + + cleanup_hook(); + exit(0); } - if ((*add_input_file)(ObjPath.c_str()) != LDPS_OK) { - (*message)(LDPL_ERROR, "Unable to add .o file to the link."); - (*message)(LDPL_ERROR, "File left behind in: %s", ObjPath.c_str()); - return LDPS_ERR; + std::unique_ptr Combined(new Module("ld-temp.o", Context)); + Linker L(Combined.get()); + + std::string DefaultTriple = sys::getDefaultTargetTriple(); + + StringSet<> Internalize; + StringSet<> Maybe; + for (claimed_file &F : Modules) { + ld_plugin_input_file File; + if (get_input_file(F.handle, &File) != LDPS_OK) + message(LDPL_FATAL, "Failed to get file information"); + std::unique_ptr M = + getModuleForFile(Context, F, File, ApiFile, Internalize, Maybe); + if (!options::triple.empty()) + M->setTargetTriple(options::triple.c_str()); + else if (M->getTargetTriple().empty()) { + M->setTargetTriple(DefaultTriple); + } + + if (L.linkInModule(M.get())) + message(LDPL_FATAL, "Failed to link module"); + if (release_input_file(F.handle) != LDPS_OK) + message(LDPL_FATAL, "Failed to release file information"); } - if (!options::extra_library_path.empty() && - set_extra_library_path(options::extra_library_path.c_str()) != LDPS_OK) { - (*message)(LDPL_ERROR, "Unable to set the extra library path."); - return LDPS_ERR; + for (const auto &Name : Internalize) { + GlobalValue *GV = Combined->getNamedValue(Name.first()); + if (GV) + internalize(*GV); + } + + for (const auto &Name : Maybe) { + GlobalValue *GV = Combined->getNamedValue(Name.first()); + if (!GV) + continue; + GV->setLinkage(GlobalValue::LinkOnceODRLinkage); + if (canBeOmittedFromSymbolTable(GV)) + internalize(*GV); + } + + if (options::TheOutputType == options::OT_DISABLE) + return LDPS_OK; + + if (options::TheOutputType != options::OT_NORMAL) { + std::string path; + if (options::TheOutputType == options::OT_BC_ONLY) + path = output_name; + else + path = output_name + ".bc"; + saveBCFile(path, *L.getModule()); + if (options::TheOutputType == options::OT_BC_ONLY) + return LDPS_OK; } - if (options::obj_path.empty()) - Cleanup.push_back(ObjPath); + codegen(std::move(Combined)); + + if (!options::extra_library_path.empty() && + set_extra_library_path(options::extra_library_path.c_str()) != LDPS_OK) + message(LDPL_FATAL, "Unable to set the extra library path."); return LDPS_OK; } +static ld_plugin_status all_symbols_read_hook(void) { + ld_plugin_status Ret; + if (!options::generate_api_file) { + Ret = allSymbolsReadHook(nullptr); + } else { + std::error_code EC; + raw_fd_ostream ApiFile("apifile.txt", EC, sys::fs::F_None); + if (EC) + message(LDPL_FATAL, "Unable to open apifile.txt for writing: %s", + EC.message().c_str()); + Ret = allSymbolsReadHook(&ApiFile); + } + + llvm_shutdown(); + + if (options::TheOutputType == options::OT_BC_ONLY || + options::TheOutputType == options::OT_DISABLE) { + if (options::TheOutputType == options::OT_DISABLE) + // Remove the output file here since ld.bfd creates the output file + // early. + sys::fs::remove(output_name); + exit(0); + } + + return Ret; +} + static ld_plugin_status cleanup_hook(void) { - for (int i = 0, e = Cleanup.size(); i != e; ++i) { - std::error_code EC = sys::fs::remove(Cleanup[i]); + for (std::string &Name : Cleanup) { + std::error_code EC = sys::fs::remove(Name); if (EC) - (*message)(LDPL_ERROR, "Failed to delete '%s': %s", Cleanup[i].c_str(), - EC.message().c_str()); + message(LDPL_ERROR, "Failed to delete '%s': %s", Name.c_str(), + EC.message().c_str()); } return LDPS_OK;