151690af2SDimitry Andric //===- ModuleSymbolTable.cpp - symbol table for in-memory IR --------------===//
2d88c1a5aSDimitry Andric //
3d88c1a5aSDimitry Andric //                     The LLVM Compiler Infrastructure
4d88c1a5aSDimitry Andric //
5d88c1a5aSDimitry Andric // This file is distributed under the University of Illinois Open Source
6d88c1a5aSDimitry Andric // License. See LICENSE.TXT for details.
7d88c1a5aSDimitry Andric //
8d88c1a5aSDimitry Andric //===----------------------------------------------------------------------===//
9d88c1a5aSDimitry Andric //
10d88c1a5aSDimitry Andric // This class represents a symbol table built from in-memory IR. It provides
11d88c1a5aSDimitry Andric // access to GlobalValues and should only be used if such access is required
12d88c1a5aSDimitry Andric // (e.g. in the LTO implementation).
13d88c1a5aSDimitry Andric //
14d88c1a5aSDimitry Andric //===----------------------------------------------------------------------===//
15d88c1a5aSDimitry Andric 
16db17bf38SDimitry Andric #include "llvm/Object/ModuleSymbolTable.h"
17d88c1a5aSDimitry Andric #include "RecordStreamer.h"
18d88c1a5aSDimitry Andric #include "llvm/ADT/STLExtras.h"
19db17bf38SDimitry Andric #include "llvm/ADT/SmallString.h"
2051690af2SDimitry Andric #include "llvm/ADT/StringMap.h"
2151690af2SDimitry Andric #include "llvm/ADT/StringRef.h"
2251690af2SDimitry Andric #include "llvm/ADT/Triple.h"
2351690af2SDimitry Andric #include "llvm/IR/Function.h"
2451690af2SDimitry Andric #include "llvm/IR/GlobalAlias.h"
2551690af2SDimitry Andric #include "llvm/IR/GlobalValue.h"
2651690af2SDimitry Andric #include "llvm/IR/GlobalVariable.h"
27d88c1a5aSDimitry Andric #include "llvm/IR/Mangler.h"
28d88c1a5aSDimitry Andric #include "llvm/IR/Module.h"
29d88c1a5aSDimitry Andric #include "llvm/MC/MCAsmInfo.h"
30d88c1a5aSDimitry Andric #include "llvm/MC/MCContext.h"
3151690af2SDimitry Andric #include "llvm/MC/MCDirectives.h"
32d88c1a5aSDimitry Andric #include "llvm/MC/MCInstrInfo.h"
33d88c1a5aSDimitry Andric #include "llvm/MC/MCObjectFileInfo.h"
34d88c1a5aSDimitry Andric #include "llvm/MC/MCParser/MCAsmParser.h"
35d88c1a5aSDimitry Andric #include "llvm/MC/MCParser/MCTargetAsmParser.h"
36d88c1a5aSDimitry Andric #include "llvm/MC/MCRegisterInfo.h"
37d88c1a5aSDimitry Andric #include "llvm/MC/MCSubtargetInfo.h"
3851690af2SDimitry Andric #include "llvm/MC/MCSymbol.h"
3951690af2SDimitry Andric #include "llvm/MC/MCTargetOptions.h"
4051690af2SDimitry Andric #include "llvm/Object/SymbolicFile.h"
4151690af2SDimitry Andric #include "llvm/Support/Casting.h"
4251690af2SDimitry Andric #include "llvm/Support/CodeGen.h"
4351690af2SDimitry Andric #include "llvm/Support/ErrorHandling.h"
44d88c1a5aSDimitry Andric #include "llvm/Support/MemoryBuffer.h"
4551690af2SDimitry Andric #include "llvm/Support/SMLoc.h"
46d88c1a5aSDimitry Andric #include "llvm/Support/SourceMgr.h"
47d88c1a5aSDimitry Andric #include "llvm/Support/TargetRegistry.h"
48db17bf38SDimitry Andric #include "llvm/Support/raw_ostream.h"
4951690af2SDimitry Andric #include <algorithm>
5051690af2SDimitry Andric #include <cassert>
5151690af2SDimitry Andric #include <cstdint>
5251690af2SDimitry Andric #include <memory>
5351690af2SDimitry Andric #include <string>
5451690af2SDimitry Andric 
55d88c1a5aSDimitry Andric using namespace llvm;
56d88c1a5aSDimitry Andric using namespace object;
57d88c1a5aSDimitry Andric 
58d88c1a5aSDimitry Andric void ModuleSymbolTable::addModule(Module *M) {
59d88c1a5aSDimitry Andric   if (FirstMod)
60d88c1a5aSDimitry Andric     assert(FirstMod->getTargetTriple() == M->getTargetTriple());
61d88c1a5aSDimitry Andric   else
62d88c1a5aSDimitry Andric     FirstMod = M;
63d88c1a5aSDimitry Andric 
647a7e6055SDimitry Andric   for (GlobalValue &GV : M->global_values())
65d88c1a5aSDimitry Andric     SymTab.push_back(&GV);
66d88c1a5aSDimitry Andric 
677a7e6055SDimitry Andric   CollectAsmSymbols(*M, [this](StringRef Name, BasicSymbolRef::Flags Flags) {
687a7e6055SDimitry Andric     SymTab.push_back(new (AsmSymbols.Allocate()) AsmSymbol(Name, Flags));
69d88c1a5aSDimitry Andric   });
70d88c1a5aSDimitry Andric }
71d88c1a5aSDimitry Andric 
727a7e6055SDimitry Andric // Ensure ELF .symver aliases get the same binding as the defined symbol
737a7e6055SDimitry Andric // they alias with.
747a7e6055SDimitry Andric static void handleSymverAliases(const Module &M, RecordStreamer &Streamer) {
757a7e6055SDimitry Andric   if (Streamer.symverAliases().empty())
767a7e6055SDimitry Andric     return;
777a7e6055SDimitry Andric 
787a7e6055SDimitry Andric   // The name in the assembler will be mangled, but the name in the IR
797a7e6055SDimitry Andric   // might not, so we first compute a mapping from mangled name to GV.
807a7e6055SDimitry Andric   Mangler Mang;
817a7e6055SDimitry Andric   SmallString<64> MangledName;
827a7e6055SDimitry Andric   StringMap<const GlobalValue *> MangledNameMap;
837a7e6055SDimitry Andric   auto GetMangledName = [&](const GlobalValue &GV) {
847a7e6055SDimitry Andric     if (!GV.hasName())
857a7e6055SDimitry Andric       return;
867a7e6055SDimitry Andric 
877a7e6055SDimitry Andric     MangledName.clear();
887a7e6055SDimitry Andric     MangledName.reserve(GV.getName().size() + 1);
897a7e6055SDimitry Andric     Mang.getNameWithPrefix(MangledName, &GV, /*CannotUsePrivateLabel=*/false);
907a7e6055SDimitry Andric     MangledNameMap[MangledName] = &GV;
917a7e6055SDimitry Andric   };
927a7e6055SDimitry Andric   for (const Function &F : M)
937a7e6055SDimitry Andric     GetMangledName(F);
947a7e6055SDimitry Andric   for (const GlobalVariable &GV : M.globals())
957a7e6055SDimitry Andric     GetMangledName(GV);
967a7e6055SDimitry Andric   for (const GlobalAlias &GA : M.aliases())
977a7e6055SDimitry Andric     GetMangledName(GA);
987a7e6055SDimitry Andric 
997a7e6055SDimitry Andric   // Walk all the recorded .symver aliases, and set up the binding
1007a7e6055SDimitry Andric   // for each alias.
1017a7e6055SDimitry Andric   for (auto &Symver : Streamer.symverAliases()) {
1027a7e6055SDimitry Andric     const MCSymbol *Aliasee = Symver.first;
1037a7e6055SDimitry Andric     MCSymbolAttr Attr = MCSA_Invalid;
1047a7e6055SDimitry Andric 
1057a7e6055SDimitry Andric     // First check if the aliasee binding was recorded in the asm.
1067a7e6055SDimitry Andric     RecordStreamer::State state = Streamer.getSymbolState(Aliasee);
1077a7e6055SDimitry Andric     switch (state) {
1087a7e6055SDimitry Andric     case RecordStreamer::Global:
1097a7e6055SDimitry Andric     case RecordStreamer::DefinedGlobal:
1107a7e6055SDimitry Andric       Attr = MCSA_Global;
1117a7e6055SDimitry Andric       break;
1127a7e6055SDimitry Andric     case RecordStreamer::UndefinedWeak:
1137a7e6055SDimitry Andric     case RecordStreamer::DefinedWeak:
1147a7e6055SDimitry Andric       Attr = MCSA_Weak;
1157a7e6055SDimitry Andric       break;
1167a7e6055SDimitry Andric     default:
1177a7e6055SDimitry Andric       break;
1187a7e6055SDimitry Andric     }
1197a7e6055SDimitry Andric 
1207a7e6055SDimitry Andric     // If we don't have a symbol attribute from assembly, then check if
1217a7e6055SDimitry Andric     // the aliasee was defined in the IR.
1227a7e6055SDimitry Andric     if (Attr == MCSA_Invalid) {
1237a7e6055SDimitry Andric       const auto *GV = M.getNamedValue(Aliasee->getName());
1247a7e6055SDimitry Andric       if (!GV) {
1257a7e6055SDimitry Andric         auto MI = MangledNameMap.find(Aliasee->getName());
1267a7e6055SDimitry Andric         if (MI != MangledNameMap.end())
1277a7e6055SDimitry Andric           GV = MI->second;
1287a7e6055SDimitry Andric         else
1297a7e6055SDimitry Andric           continue;
1307a7e6055SDimitry Andric       }
1317a7e6055SDimitry Andric       if (GV->hasExternalLinkage())
1327a7e6055SDimitry Andric         Attr = MCSA_Global;
1337a7e6055SDimitry Andric       else if (GV->hasLocalLinkage())
1347a7e6055SDimitry Andric         Attr = MCSA_Local;
1357a7e6055SDimitry Andric       else if (GV->isWeakForLinker())
1367a7e6055SDimitry Andric         Attr = MCSA_Weak;
1377a7e6055SDimitry Andric     }
1387a7e6055SDimitry Andric     if (Attr == MCSA_Invalid)
1397a7e6055SDimitry Andric       continue;
1407a7e6055SDimitry Andric 
1417a7e6055SDimitry Andric     // Set the detected binding on each alias with this aliasee.
1427a7e6055SDimitry Andric     for (auto &Alias : Symver.second)
1437a7e6055SDimitry Andric       Streamer.EmitSymbolAttribute(Alias, Attr);
1447a7e6055SDimitry Andric   }
1457a7e6055SDimitry Andric }
1467a7e6055SDimitry Andric 
147d88c1a5aSDimitry Andric void ModuleSymbolTable::CollectAsmSymbols(
1487a7e6055SDimitry Andric     const Module &M,
149d88c1a5aSDimitry Andric     function_ref<void(StringRef, BasicSymbolRef::Flags)> AsmSymbol) {
1507a7e6055SDimitry Andric   StringRef InlineAsm = M.getModuleInlineAsm();
151d88c1a5aSDimitry Andric   if (InlineAsm.empty())
152d88c1a5aSDimitry Andric     return;
153d88c1a5aSDimitry Andric 
154d88c1a5aSDimitry Andric   std::string Err;
1557a7e6055SDimitry Andric   const Triple TT(M.getTargetTriple());
156d88c1a5aSDimitry Andric   const Target *T = TargetRegistry::lookupTarget(TT.str(), Err);
157d88c1a5aSDimitry Andric   assert(T && T->hasMCAsmParser());
158d88c1a5aSDimitry Andric 
159d88c1a5aSDimitry Andric   std::unique_ptr<MCRegisterInfo> MRI(T->createMCRegInfo(TT.str()));
160d88c1a5aSDimitry Andric   if (!MRI)
161d88c1a5aSDimitry Andric     return;
162d88c1a5aSDimitry Andric 
163d88c1a5aSDimitry Andric   std::unique_ptr<MCAsmInfo> MAI(T->createMCAsmInfo(*MRI, TT.str()));
164d88c1a5aSDimitry Andric   if (!MAI)
165d88c1a5aSDimitry Andric     return;
166d88c1a5aSDimitry Andric 
167d88c1a5aSDimitry Andric   std::unique_ptr<MCSubtargetInfo> STI(
168d88c1a5aSDimitry Andric       T->createMCSubtargetInfo(TT.str(), "", ""));
169d88c1a5aSDimitry Andric   if (!STI)
170d88c1a5aSDimitry Andric     return;
171d88c1a5aSDimitry Andric 
172d88c1a5aSDimitry Andric   std::unique_ptr<MCInstrInfo> MCII(T->createMCInstrInfo());
173d88c1a5aSDimitry Andric   if (!MCII)
174d88c1a5aSDimitry Andric     return;
175d88c1a5aSDimitry Andric 
176d88c1a5aSDimitry Andric   MCObjectFileInfo MOFI;
177d88c1a5aSDimitry Andric   MCContext MCCtx(MAI.get(), MRI.get(), &MOFI);
1782cab237bSDimitry Andric   MOFI.InitMCObjectFileInfo(TT, /*PIC*/ false, MCCtx);
179d88c1a5aSDimitry Andric   RecordStreamer Streamer(MCCtx);
180d88c1a5aSDimitry Andric   T->createNullTargetStreamer(Streamer);
181d88c1a5aSDimitry Andric 
182d88c1a5aSDimitry Andric   std::unique_ptr<MemoryBuffer> Buffer(MemoryBuffer::getMemBuffer(InlineAsm));
183d88c1a5aSDimitry Andric   SourceMgr SrcMgr;
184d88c1a5aSDimitry Andric   SrcMgr.AddNewSourceBuffer(std::move(Buffer), SMLoc());
185d88c1a5aSDimitry Andric   std::unique_ptr<MCAsmParser> Parser(
186d88c1a5aSDimitry Andric       createMCAsmParser(SrcMgr, MCCtx, Streamer, *MAI));
187d88c1a5aSDimitry Andric 
188d88c1a5aSDimitry Andric   MCTargetOptions MCOptions;
189d88c1a5aSDimitry Andric   std::unique_ptr<MCTargetAsmParser> TAP(
190d88c1a5aSDimitry Andric       T->createMCAsmParser(*STI, *Parser, *MCII, MCOptions));
191d88c1a5aSDimitry Andric   if (!TAP)
192d88c1a5aSDimitry Andric     return;
193d88c1a5aSDimitry Andric 
194d88c1a5aSDimitry Andric   Parser->setTargetParser(*TAP);
195d88c1a5aSDimitry Andric   if (Parser->Run(false))
196d88c1a5aSDimitry Andric     return;
197d88c1a5aSDimitry Andric 
1987a7e6055SDimitry Andric   handleSymverAliases(M, Streamer);
1997a7e6055SDimitry Andric 
200d88c1a5aSDimitry Andric   for (auto &KV : Streamer) {
201d88c1a5aSDimitry Andric     StringRef Key = KV.first();
202d88c1a5aSDimitry Andric     RecordStreamer::State Value = KV.second;
203d88c1a5aSDimitry Andric     // FIXME: For now we just assume that all asm symbols are executable.
204d88c1a5aSDimitry Andric     uint32_t Res = BasicSymbolRef::SF_Executable;
205d88c1a5aSDimitry Andric     switch (Value) {
206d88c1a5aSDimitry Andric     case RecordStreamer::NeverSeen:
207d88c1a5aSDimitry Andric       llvm_unreachable("NeverSeen should have been replaced earlier");
208d88c1a5aSDimitry Andric     case RecordStreamer::DefinedGlobal:
209d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_Global;
210d88c1a5aSDimitry Andric       break;
211d88c1a5aSDimitry Andric     case RecordStreamer::Defined:
212d88c1a5aSDimitry Andric       break;
213d88c1a5aSDimitry Andric     case RecordStreamer::Global:
214d88c1a5aSDimitry Andric     case RecordStreamer::Used:
215d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_Undefined;
216d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_Global;
217d88c1a5aSDimitry Andric       break;
218d88c1a5aSDimitry Andric     case RecordStreamer::DefinedWeak:
219d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_Weak;
220d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_Global;
221d88c1a5aSDimitry Andric       break;
222d88c1a5aSDimitry Andric     case RecordStreamer::UndefinedWeak:
223d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_Weak;
224d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_Undefined;
225d88c1a5aSDimitry Andric     }
226d88c1a5aSDimitry Andric     AsmSymbol(Key, BasicSymbolRef::Flags(Res));
227d88c1a5aSDimitry Andric   }
228d88c1a5aSDimitry Andric }
229d88c1a5aSDimitry Andric 
230d88c1a5aSDimitry Andric void ModuleSymbolTable::printSymbolName(raw_ostream &OS, Symbol S) const {
231d88c1a5aSDimitry Andric   if (S.is<AsmSymbol *>()) {
232d88c1a5aSDimitry Andric     OS << S.get<AsmSymbol *>()->first;
233d88c1a5aSDimitry Andric     return;
234d88c1a5aSDimitry Andric   }
235d88c1a5aSDimitry Andric 
236d88c1a5aSDimitry Andric   auto *GV = S.get<GlobalValue *>();
237d88c1a5aSDimitry Andric   if (GV->hasDLLImportStorageClass())
238d88c1a5aSDimitry Andric     OS << "__imp_";
239d88c1a5aSDimitry Andric 
240d88c1a5aSDimitry Andric   Mang.getNameWithPrefix(OS, GV, false);
241d88c1a5aSDimitry Andric }
242d88c1a5aSDimitry Andric 
243d88c1a5aSDimitry Andric uint32_t ModuleSymbolTable::getSymbolFlags(Symbol S) const {
244d88c1a5aSDimitry Andric   if (S.is<AsmSymbol *>())
245d88c1a5aSDimitry Andric     return S.get<AsmSymbol *>()->second;
246d88c1a5aSDimitry Andric 
247d88c1a5aSDimitry Andric   auto *GV = S.get<GlobalValue *>();
248d88c1a5aSDimitry Andric 
249d88c1a5aSDimitry Andric   uint32_t Res = BasicSymbolRef::SF_None;
250d88c1a5aSDimitry Andric   if (GV->isDeclarationForLinker())
251d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_Undefined;
252d88c1a5aSDimitry Andric   else if (GV->hasHiddenVisibility() && !GV->hasLocalLinkage())
253d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_Hidden;
254d88c1a5aSDimitry Andric   if (const GlobalVariable *GVar = dyn_cast<GlobalVariable>(GV)) {
255d88c1a5aSDimitry Andric     if (GVar->isConstant())
256d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_Const;
257d88c1a5aSDimitry Andric   }
258d88c1a5aSDimitry Andric   if (dyn_cast_or_null<Function>(GV->getBaseObject()))
259d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_Executable;
260d88c1a5aSDimitry Andric   if (isa<GlobalAlias>(GV))
261d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_Indirect;
262d88c1a5aSDimitry Andric   if (GV->hasPrivateLinkage())
263d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_FormatSpecific;
264d88c1a5aSDimitry Andric   if (!GV->hasLocalLinkage())
265d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_Global;
266d88c1a5aSDimitry Andric   if (GV->hasCommonLinkage())
267d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_Common;
268d88c1a5aSDimitry Andric   if (GV->hasLinkOnceLinkage() || GV->hasWeakLinkage() ||
269d88c1a5aSDimitry Andric       GV->hasExternalWeakLinkage())
270d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_Weak;
271d88c1a5aSDimitry Andric 
272d88c1a5aSDimitry Andric   if (GV->getName().startswith("llvm."))
273d88c1a5aSDimitry Andric     Res |= BasicSymbolRef::SF_FormatSpecific;
274d88c1a5aSDimitry Andric   else if (auto *Var = dyn_cast<GlobalVariable>(GV)) {
275d88c1a5aSDimitry Andric     if (Var->getSection() == "llvm.metadata")
276d88c1a5aSDimitry Andric       Res |= BasicSymbolRef::SF_FormatSpecific;
277d88c1a5aSDimitry Andric   }
278d88c1a5aSDimitry Andric 
279d88c1a5aSDimitry Andric   return Res;
280d88c1a5aSDimitry Andric }
281