10b57cec5SDimitry Andric //===- ModuleSymbolTable.cpp - symbol table for in-memory IR --------------===// 20b57cec5SDimitry Andric // 30b57cec5SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 40b57cec5SDimitry Andric // See https://llvm.org/LICENSE.txt for license information. 50b57cec5SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 60b57cec5SDimitry Andric // 70b57cec5SDimitry Andric //===----------------------------------------------------------------------===// 80b57cec5SDimitry Andric // 90b57cec5SDimitry Andric // This class represents a symbol table built from in-memory IR. It provides 100b57cec5SDimitry Andric // access to GlobalValues and should only be used if such access is required 110b57cec5SDimitry Andric // (e.g. in the LTO implementation). 120b57cec5SDimitry Andric // 130b57cec5SDimitry Andric //===----------------------------------------------------------------------===// 140b57cec5SDimitry Andric 150b57cec5SDimitry Andric #include "llvm/Object/ModuleSymbolTable.h" 160b57cec5SDimitry Andric #include "RecordStreamer.h" 170b57cec5SDimitry Andric #include "llvm/ADT/STLExtras.h" 180b57cec5SDimitry Andric #include "llvm/ADT/StringMap.h" 190b57cec5SDimitry Andric #include "llvm/ADT/StringRef.h" 200b57cec5SDimitry Andric #include "llvm/IR/Function.h" 210b57cec5SDimitry Andric #include "llvm/IR/GlobalAlias.h" 220b57cec5SDimitry Andric #include "llvm/IR/GlobalValue.h" 230b57cec5SDimitry Andric #include "llvm/IR/GlobalVariable.h" 245ffd83dbSDimitry Andric #include "llvm/IR/InlineAsm.h" 250b57cec5SDimitry Andric #include "llvm/IR/Module.h" 260b57cec5SDimitry Andric #include "llvm/MC/MCAsmInfo.h" 270b57cec5SDimitry Andric #include "llvm/MC/MCContext.h" 280b57cec5SDimitry Andric #include "llvm/MC/MCInstrInfo.h" 290b57cec5SDimitry Andric #include "llvm/MC/MCObjectFileInfo.h" 300b57cec5SDimitry Andric #include "llvm/MC/MCParser/MCAsmParser.h" 310b57cec5SDimitry Andric #include "llvm/MC/MCParser/MCTargetAsmParser.h" 320b57cec5SDimitry Andric #include "llvm/MC/MCRegisterInfo.h" 330b57cec5SDimitry Andric #include "llvm/MC/MCSubtargetInfo.h" 340b57cec5SDimitry Andric #include "llvm/MC/MCSymbol.h" 350b57cec5SDimitry Andric #include "llvm/MC/MCTargetOptions.h" 36349cc55cSDimitry Andric #include "llvm/MC/TargetRegistry.h" 370b57cec5SDimitry Andric #include "llvm/Object/SymbolicFile.h" 380b57cec5SDimitry Andric #include "llvm/Support/Casting.h" 390b57cec5SDimitry Andric #include "llvm/Support/ErrorHandling.h" 400b57cec5SDimitry Andric #include "llvm/Support/MemoryBuffer.h" 410b57cec5SDimitry Andric #include "llvm/Support/SMLoc.h" 420b57cec5SDimitry Andric #include "llvm/Support/SourceMgr.h" 430b57cec5SDimitry Andric #include "llvm/Support/raw_ostream.h" 44*fe013be4SDimitry Andric #include "llvm/TargetParser/Triple.h" 450b57cec5SDimitry Andric #include <algorithm> 460b57cec5SDimitry Andric #include <cassert> 470b57cec5SDimitry Andric #include <cstdint> 480b57cec5SDimitry Andric #include <memory> 490b57cec5SDimitry Andric #include <string> 500b57cec5SDimitry Andric 510b57cec5SDimitry Andric using namespace llvm; 520b57cec5SDimitry Andric using namespace object; 530b57cec5SDimitry Andric 540b57cec5SDimitry Andric void ModuleSymbolTable::addModule(Module *M) { 550b57cec5SDimitry Andric if (FirstMod) 560b57cec5SDimitry Andric assert(FirstMod->getTargetTriple() == M->getTargetTriple()); 570b57cec5SDimitry Andric else 580b57cec5SDimitry Andric FirstMod = M; 590b57cec5SDimitry Andric 600b57cec5SDimitry Andric for (GlobalValue &GV : M->global_values()) 610b57cec5SDimitry Andric SymTab.push_back(&GV); 620b57cec5SDimitry Andric 630b57cec5SDimitry Andric CollectAsmSymbols(*M, [this](StringRef Name, BasicSymbolRef::Flags Flags) { 645ffd83dbSDimitry Andric SymTab.push_back(new (AsmSymbols.Allocate()) 655ffd83dbSDimitry Andric AsmSymbol(std::string(Name), Flags)); 660b57cec5SDimitry Andric }); 670b57cec5SDimitry Andric } 680b57cec5SDimitry Andric 690b57cec5SDimitry Andric static void 700b57cec5SDimitry Andric initializeRecordStreamer(const Module &M, 710b57cec5SDimitry Andric function_ref<void(RecordStreamer &)> Init) { 720b57cec5SDimitry Andric StringRef InlineAsm = M.getModuleInlineAsm(); 730b57cec5SDimitry Andric if (InlineAsm.empty()) 740b57cec5SDimitry Andric return; 750b57cec5SDimitry Andric 760b57cec5SDimitry Andric std::string Err; 770b57cec5SDimitry Andric const Triple TT(M.getTargetTriple()); 780b57cec5SDimitry Andric const Target *T = TargetRegistry::lookupTarget(TT.str(), Err); 790b57cec5SDimitry Andric assert(T && T->hasMCAsmParser()); 800b57cec5SDimitry Andric 810b57cec5SDimitry Andric std::unique_ptr<MCRegisterInfo> MRI(T->createMCRegInfo(TT.str())); 820b57cec5SDimitry Andric if (!MRI) 830b57cec5SDimitry Andric return; 840b57cec5SDimitry Andric 85480093f4SDimitry Andric MCTargetOptions MCOptions; 86480093f4SDimitry Andric std::unique_ptr<MCAsmInfo> MAI(T->createMCAsmInfo(*MRI, TT.str(), MCOptions)); 870b57cec5SDimitry Andric if (!MAI) 880b57cec5SDimitry Andric return; 890b57cec5SDimitry Andric 900b57cec5SDimitry Andric std::unique_ptr<MCSubtargetInfo> STI( 910b57cec5SDimitry Andric T->createMCSubtargetInfo(TT.str(), "", "")); 920b57cec5SDimitry Andric if (!STI) 930b57cec5SDimitry Andric return; 940b57cec5SDimitry Andric 950b57cec5SDimitry Andric std::unique_ptr<MCInstrInfo> MCII(T->createMCInstrInfo()); 960b57cec5SDimitry Andric if (!MCII) 970b57cec5SDimitry Andric return; 980b57cec5SDimitry Andric 990b57cec5SDimitry Andric std::unique_ptr<MemoryBuffer> Buffer(MemoryBuffer::getMemBuffer(InlineAsm)); 1000b57cec5SDimitry Andric SourceMgr SrcMgr; 1010b57cec5SDimitry Andric SrcMgr.AddNewSourceBuffer(std::move(Buffer), SMLoc()); 102fe6060f1SDimitry Andric 103fe6060f1SDimitry Andric MCContext MCCtx(TT, MAI.get(), MRI.get(), STI.get(), &SrcMgr); 104fe6060f1SDimitry Andric std::unique_ptr<MCObjectFileInfo> MOFI( 105fe6060f1SDimitry Andric T->createMCObjectFileInfo(MCCtx, /*PIC=*/false)); 106fe6060f1SDimitry Andric MOFI->setSDKVersion(M.getSDKVersion()); 107fe6060f1SDimitry Andric MCCtx.setObjectFileInfo(MOFI.get()); 108fe6060f1SDimitry Andric RecordStreamer Streamer(MCCtx, M); 109fe6060f1SDimitry Andric T->createNullTargetStreamer(Streamer); 110fe6060f1SDimitry Andric 1110b57cec5SDimitry Andric std::unique_ptr<MCAsmParser> Parser( 1120b57cec5SDimitry Andric createMCAsmParser(SrcMgr, MCCtx, Streamer, *MAI)); 1130b57cec5SDimitry Andric 1140b57cec5SDimitry Andric std::unique_ptr<MCTargetAsmParser> TAP( 1150b57cec5SDimitry Andric T->createMCAsmParser(*STI, *Parser, *MCII, MCOptions)); 1160b57cec5SDimitry Andric if (!TAP) 1170b57cec5SDimitry Andric return; 1180b57cec5SDimitry Andric 1195ffd83dbSDimitry Andric // Module-level inline asm is assumed to use At&t syntax (see 1205ffd83dbSDimitry Andric // AsmPrinter::doInitialization()). 1215ffd83dbSDimitry Andric Parser->setAssemblerDialect(InlineAsm::AD_ATT); 1225ffd83dbSDimitry Andric 1230b57cec5SDimitry Andric Parser->setTargetParser(*TAP); 1240b57cec5SDimitry Andric if (Parser->Run(false)) 1250b57cec5SDimitry Andric return; 1260b57cec5SDimitry Andric 1270b57cec5SDimitry Andric Init(Streamer); 1280b57cec5SDimitry Andric } 1290b57cec5SDimitry Andric 1300b57cec5SDimitry Andric void ModuleSymbolTable::CollectAsmSymbols( 1310b57cec5SDimitry Andric const Module &M, 1320b57cec5SDimitry Andric function_ref<void(StringRef, BasicSymbolRef::Flags)> AsmSymbol) { 1330b57cec5SDimitry Andric initializeRecordStreamer(M, [&](RecordStreamer &Streamer) { 1340b57cec5SDimitry Andric Streamer.flushSymverDirectives(); 1350b57cec5SDimitry Andric 1360b57cec5SDimitry Andric for (auto &KV : Streamer) { 1370b57cec5SDimitry Andric StringRef Key = KV.first(); 1380b57cec5SDimitry Andric RecordStreamer::State Value = KV.second; 1390b57cec5SDimitry Andric // FIXME: For now we just assume that all asm symbols are executable. 1400b57cec5SDimitry Andric uint32_t Res = BasicSymbolRef::SF_Executable; 1410b57cec5SDimitry Andric switch (Value) { 1420b57cec5SDimitry Andric case RecordStreamer::NeverSeen: 1430b57cec5SDimitry Andric llvm_unreachable("NeverSeen should have been replaced earlier"); 1440b57cec5SDimitry Andric case RecordStreamer::DefinedGlobal: 1450b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Global; 1460b57cec5SDimitry Andric break; 1470b57cec5SDimitry Andric case RecordStreamer::Defined: 1480b57cec5SDimitry Andric break; 1490b57cec5SDimitry Andric case RecordStreamer::Global: 1500b57cec5SDimitry Andric case RecordStreamer::Used: 1510b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Undefined; 1520b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Global; 1530b57cec5SDimitry Andric break; 1540b57cec5SDimitry Andric case RecordStreamer::DefinedWeak: 1550b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Weak; 1560b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Global; 1570b57cec5SDimitry Andric break; 1580b57cec5SDimitry Andric case RecordStreamer::UndefinedWeak: 1590b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Weak; 1600b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Undefined; 1610b57cec5SDimitry Andric } 1620b57cec5SDimitry Andric AsmSymbol(Key, BasicSymbolRef::Flags(Res)); 1630b57cec5SDimitry Andric } 1640b57cec5SDimitry Andric }); 1650b57cec5SDimitry Andric } 1660b57cec5SDimitry Andric 1670b57cec5SDimitry Andric void ModuleSymbolTable::CollectAsmSymvers( 1680b57cec5SDimitry Andric const Module &M, function_ref<void(StringRef, StringRef)> AsmSymver) { 1690b57cec5SDimitry Andric initializeRecordStreamer(M, [&](RecordStreamer &Streamer) { 1700b57cec5SDimitry Andric for (auto &KV : Streamer.symverAliases()) 1710b57cec5SDimitry Andric for (auto &Alias : KV.second) 1720b57cec5SDimitry Andric AsmSymver(KV.first->getName(), Alias); 1730b57cec5SDimitry Andric }); 1740b57cec5SDimitry Andric } 1750b57cec5SDimitry Andric 1760b57cec5SDimitry Andric void ModuleSymbolTable::printSymbolName(raw_ostream &OS, Symbol S) const { 177*fe013be4SDimitry Andric if (isa<AsmSymbol *>(S)) { 178*fe013be4SDimitry Andric OS << cast<AsmSymbol *>(S)->first; 1790b57cec5SDimitry Andric return; 1800b57cec5SDimitry Andric } 1810b57cec5SDimitry Andric 182*fe013be4SDimitry Andric auto *GV = cast<GlobalValue *>(S); 1830b57cec5SDimitry Andric if (GV->hasDLLImportStorageClass()) 1840b57cec5SDimitry Andric OS << "__imp_"; 1850b57cec5SDimitry Andric 1860b57cec5SDimitry Andric Mang.getNameWithPrefix(OS, GV, false); 1870b57cec5SDimitry Andric } 1880b57cec5SDimitry Andric 1890b57cec5SDimitry Andric uint32_t ModuleSymbolTable::getSymbolFlags(Symbol S) const { 190*fe013be4SDimitry Andric if (isa<AsmSymbol *>(S)) 191*fe013be4SDimitry Andric return cast<AsmSymbol *>(S)->second; 1920b57cec5SDimitry Andric 193*fe013be4SDimitry Andric auto *GV = cast<GlobalValue *>(S); 1940b57cec5SDimitry Andric 1950b57cec5SDimitry Andric uint32_t Res = BasicSymbolRef::SF_None; 1960b57cec5SDimitry Andric if (GV->isDeclarationForLinker()) 1970b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Undefined; 1980b57cec5SDimitry Andric else if (GV->hasHiddenVisibility() && !GV->hasLocalLinkage()) 1990b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Hidden; 2000b57cec5SDimitry Andric if (const GlobalVariable *GVar = dyn_cast<GlobalVariable>(GV)) { 2010b57cec5SDimitry Andric if (GVar->isConstant()) 2020b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Const; 2030b57cec5SDimitry Andric } 204349cc55cSDimitry Andric if (const GlobalObject *GO = GV->getAliaseeObject()) 205349cc55cSDimitry Andric if (isa<Function>(GO) || isa<GlobalIFunc>(GO)) 2060b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Executable; 2070b57cec5SDimitry Andric if (isa<GlobalAlias>(GV)) 2080b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Indirect; 2090b57cec5SDimitry Andric if (GV->hasPrivateLinkage()) 2100b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_FormatSpecific; 2110b57cec5SDimitry Andric if (!GV->hasLocalLinkage()) 2120b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Global; 2130b57cec5SDimitry Andric if (GV->hasCommonLinkage()) 2140b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Common; 2150b57cec5SDimitry Andric if (GV->hasLinkOnceLinkage() || GV->hasWeakLinkage() || 2160b57cec5SDimitry Andric GV->hasExternalWeakLinkage()) 2170b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_Weak; 2180b57cec5SDimitry Andric 2190b57cec5SDimitry Andric if (GV->getName().startswith("llvm.")) 2200b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_FormatSpecific; 2210b57cec5SDimitry Andric else if (auto *Var = dyn_cast<GlobalVariable>(GV)) { 2220b57cec5SDimitry Andric if (Var->getSection() == "llvm.metadata") 2230b57cec5SDimitry Andric Res |= BasicSymbolRef::SF_FormatSpecific; 2240b57cec5SDimitry Andric } 2250b57cec5SDimitry Andric 2260b57cec5SDimitry Andric return Res; 2270b57cec5SDimitry Andric } 228