1863cbfbeSPeter Collingbourne //===- ModuleSymbolTable.cpp - symbol table for in-memory IR ----*- C++ -*-===//
2863cbfbeSPeter Collingbourne //
3863cbfbeSPeter Collingbourne //                     The LLVM Compiler Infrastructure
4863cbfbeSPeter Collingbourne //
5863cbfbeSPeter Collingbourne // This file is distributed under the University of Illinois Open Source
6863cbfbeSPeter Collingbourne // License. See LICENSE.TXT for details.
7863cbfbeSPeter Collingbourne //
8863cbfbeSPeter Collingbourne //===----------------------------------------------------------------------===//
9863cbfbeSPeter Collingbourne //
10863cbfbeSPeter Collingbourne // This class represents a symbol table built from in-memory IR. It provides
11863cbfbeSPeter Collingbourne // access to GlobalValues and should only be used if such access is required
12863cbfbeSPeter Collingbourne // (e.g. in the LTO implementation).
13863cbfbeSPeter Collingbourne //
14863cbfbeSPeter Collingbourne //===----------------------------------------------------------------------===//
15863cbfbeSPeter Collingbourne 
16863cbfbeSPeter Collingbourne #include "llvm/Object/IRObjectFile.h"
17863cbfbeSPeter Collingbourne #include "RecordStreamer.h"
18863cbfbeSPeter Collingbourne #include "llvm/ADT/STLExtras.h"
19863cbfbeSPeter Collingbourne #include "llvm/Bitcode/BitcodeReader.h"
20863cbfbeSPeter Collingbourne #include "llvm/IR/GVMaterializer.h"
21863cbfbeSPeter Collingbourne #include "llvm/IR/LLVMContext.h"
22863cbfbeSPeter Collingbourne #include "llvm/IR/Mangler.h"
23863cbfbeSPeter Collingbourne #include "llvm/IR/Module.h"
24863cbfbeSPeter Collingbourne #include "llvm/MC/MCAsmInfo.h"
25863cbfbeSPeter Collingbourne #include "llvm/MC/MCContext.h"
26863cbfbeSPeter Collingbourne #include "llvm/MC/MCInstrInfo.h"
27863cbfbeSPeter Collingbourne #include "llvm/MC/MCObjectFileInfo.h"
28863cbfbeSPeter Collingbourne #include "llvm/MC/MCParser/MCAsmParser.h"
29863cbfbeSPeter Collingbourne #include "llvm/MC/MCParser/MCTargetAsmParser.h"
30863cbfbeSPeter Collingbourne #include "llvm/MC/MCRegisterInfo.h"
31863cbfbeSPeter Collingbourne #include "llvm/MC/MCSubtargetInfo.h"
32863cbfbeSPeter Collingbourne #include "llvm/Object/ObjectFile.h"
33863cbfbeSPeter Collingbourne #include "llvm/Support/MemoryBuffer.h"
34863cbfbeSPeter Collingbourne #include "llvm/Support/SourceMgr.h"
35863cbfbeSPeter Collingbourne #include "llvm/Support/TargetRegistry.h"
36863cbfbeSPeter Collingbourne #include "llvm/Support/raw_ostream.h"
37863cbfbeSPeter Collingbourne using namespace llvm;
38863cbfbeSPeter Collingbourne using namespace object;
39863cbfbeSPeter Collingbourne 
40863cbfbeSPeter Collingbourne void ModuleSymbolTable::addModule(Module *M) {
41863cbfbeSPeter Collingbourne   if (FirstMod)
42863cbfbeSPeter Collingbourne     assert(FirstMod->getTargetTriple() == M->getTargetTriple());
43863cbfbeSPeter Collingbourne   else
44863cbfbeSPeter Collingbourne     FirstMod = M;
45863cbfbeSPeter Collingbourne 
46863cbfbeSPeter Collingbourne   for (Function &F : *M)
47863cbfbeSPeter Collingbourne     SymTab.push_back(&F);
48863cbfbeSPeter Collingbourne   for (GlobalVariable &GV : M->globals())
49863cbfbeSPeter Collingbourne     SymTab.push_back(&GV);
50863cbfbeSPeter Collingbourne   for (GlobalAlias &GA : M->aliases())
51863cbfbeSPeter Collingbourne     SymTab.push_back(&GA);
52863cbfbeSPeter Collingbourne 
53*d8204472STeresa Johnson   CollectAsmSymbols(*M, [this](StringRef Name, BasicSymbolRef::Flags Flags) {
54*d8204472STeresa Johnson     SymTab.push_back(new (AsmSymbols.Allocate()) AsmSymbol(Name, Flags));
55863cbfbeSPeter Collingbourne   });
56863cbfbeSPeter Collingbourne }
57863cbfbeSPeter Collingbourne 
58*d8204472STeresa Johnson // Ensure ELF .symver aliases get the same binding as the defined symbol
59*d8204472STeresa Johnson // they alias with.
60*d8204472STeresa Johnson static void handleSymverAliases(const Module &M, RecordStreamer &Streamer) {
61*d8204472STeresa Johnson   if (Streamer.symverAliases().empty())
62*d8204472STeresa Johnson     return;
63*d8204472STeresa Johnson 
64*d8204472STeresa Johnson   // The name in the assembler will be mangled, but the name in the IR
65*d8204472STeresa Johnson   // might not, so we first compute a mapping from mangled name to GV.
66*d8204472STeresa Johnson   Mangler Mang;
67*d8204472STeresa Johnson   SmallString<64> MangledName;
68*d8204472STeresa Johnson   StringMap<const GlobalValue *> MangledNameMap;
69*d8204472STeresa Johnson   auto GetMangledName = [&](const GlobalValue &GV) {
70*d8204472STeresa Johnson     if (!GV.hasName())
71*d8204472STeresa Johnson       return;
72*d8204472STeresa Johnson 
73*d8204472STeresa Johnson     MangledName.clear();
74*d8204472STeresa Johnson     MangledName.reserve(GV.getName().size() + 1);
75*d8204472STeresa Johnson     Mang.getNameWithPrefix(MangledName, &GV, /*CannotUsePrivateLabel=*/false);
76*d8204472STeresa Johnson     MangledNameMap[MangledName] = &GV;
77*d8204472STeresa Johnson   };
78*d8204472STeresa Johnson   for (const Function &F : M)
79*d8204472STeresa Johnson     GetMangledName(F);
80*d8204472STeresa Johnson   for (const GlobalVariable &GV : M.globals())
81*d8204472STeresa Johnson     GetMangledName(GV);
82*d8204472STeresa Johnson   for (const GlobalAlias &GA : M.aliases())
83*d8204472STeresa Johnson     GetMangledName(GA);
84*d8204472STeresa Johnson 
85*d8204472STeresa Johnson   // Walk all the recorded .symver aliases, and set up the binding
86*d8204472STeresa Johnson   // for each alias.
87*d8204472STeresa Johnson   for (auto &Symver : Streamer.symverAliases()) {
88*d8204472STeresa Johnson     const MCSymbol *Aliasee = Symver.first;
89*d8204472STeresa Johnson     MCSymbolAttr Attr = MCSA_Invalid;
90*d8204472STeresa Johnson 
91*d8204472STeresa Johnson     // First check if the aliasee binding was recorded in the asm.
92*d8204472STeresa Johnson     RecordStreamer::State state = Streamer.getSymbolState(Aliasee);
93*d8204472STeresa Johnson     switch (state) {
94*d8204472STeresa Johnson     case RecordStreamer::Global:
95*d8204472STeresa Johnson     case RecordStreamer::DefinedGlobal:
96*d8204472STeresa Johnson       Attr = MCSA_Global;
97*d8204472STeresa Johnson       break;
98*d8204472STeresa Johnson     case RecordStreamer::UndefinedWeak:
99*d8204472STeresa Johnson     case RecordStreamer::DefinedWeak:
100*d8204472STeresa Johnson       Attr = MCSA_Weak;
101*d8204472STeresa Johnson       break;
102*d8204472STeresa Johnson     default:
103*d8204472STeresa Johnson       break;
104*d8204472STeresa Johnson     }
105*d8204472STeresa Johnson 
106*d8204472STeresa Johnson     // If we don't have a symbol attribute from assembly, then check if
107*d8204472STeresa Johnson     // the aliasee was defined in the IR.
108*d8204472STeresa Johnson     if (Attr == MCSA_Invalid) {
109*d8204472STeresa Johnson       const auto *GV = M.getNamedValue(Aliasee->getName());
110*d8204472STeresa Johnson       if (!GV) {
111*d8204472STeresa Johnson         auto MI = MangledNameMap.find(Aliasee->getName());
112*d8204472STeresa Johnson         if (MI != MangledNameMap.end())
113*d8204472STeresa Johnson           GV = MI->second;
114*d8204472STeresa Johnson         else
115*d8204472STeresa Johnson           continue;
116*d8204472STeresa Johnson       }
117*d8204472STeresa Johnson       if (GV->hasExternalLinkage())
118*d8204472STeresa Johnson         Attr = MCSA_Global;
119*d8204472STeresa Johnson       else if (GV->hasLocalLinkage())
120*d8204472STeresa Johnson         Attr = MCSA_Local;
121*d8204472STeresa Johnson       else if (GV->isWeakForLinker())
122*d8204472STeresa Johnson         Attr = MCSA_Weak;
123*d8204472STeresa Johnson     }
124*d8204472STeresa Johnson     if (Attr == MCSA_Invalid)
125*d8204472STeresa Johnson       continue;
126*d8204472STeresa Johnson 
127*d8204472STeresa Johnson     // Set the detected binding on each alias with this aliasee.
128*d8204472STeresa Johnson     for (auto &Alias : Symver.second)
129*d8204472STeresa Johnson       Streamer.EmitSymbolAttribute(Alias, Attr);
130*d8204472STeresa Johnson   }
131*d8204472STeresa Johnson }
132*d8204472STeresa Johnson 
133863cbfbeSPeter Collingbourne void ModuleSymbolTable::CollectAsmSymbols(
134*d8204472STeresa Johnson     const Module &M,
135863cbfbeSPeter Collingbourne     function_ref<void(StringRef, BasicSymbolRef::Flags)> AsmSymbol) {
136*d8204472STeresa Johnson   StringRef InlineAsm = M.getModuleInlineAsm();
137863cbfbeSPeter Collingbourne   if (InlineAsm.empty())
138863cbfbeSPeter Collingbourne     return;
139863cbfbeSPeter Collingbourne 
140863cbfbeSPeter Collingbourne   std::string Err;
141*d8204472STeresa Johnson   const Triple TT(M.getTargetTriple());
142863cbfbeSPeter Collingbourne   const Target *T = TargetRegistry::lookupTarget(TT.str(), Err);
143863cbfbeSPeter Collingbourne   assert(T && T->hasMCAsmParser());
144863cbfbeSPeter Collingbourne 
145863cbfbeSPeter Collingbourne   std::unique_ptr<MCRegisterInfo> MRI(T->createMCRegInfo(TT.str()));
146863cbfbeSPeter Collingbourne   if (!MRI)
147863cbfbeSPeter Collingbourne     return;
148863cbfbeSPeter Collingbourne 
149863cbfbeSPeter Collingbourne   std::unique_ptr<MCAsmInfo> MAI(T->createMCAsmInfo(*MRI, TT.str()));
150863cbfbeSPeter Collingbourne   if (!MAI)
151863cbfbeSPeter Collingbourne     return;
152863cbfbeSPeter Collingbourne 
153863cbfbeSPeter Collingbourne   std::unique_ptr<MCSubtargetInfo> STI(
154863cbfbeSPeter Collingbourne       T->createMCSubtargetInfo(TT.str(), "", ""));
155863cbfbeSPeter Collingbourne   if (!STI)
156863cbfbeSPeter Collingbourne     return;
157863cbfbeSPeter Collingbourne 
158863cbfbeSPeter Collingbourne   std::unique_ptr<MCInstrInfo> MCII(T->createMCInstrInfo());
159863cbfbeSPeter Collingbourne   if (!MCII)
160863cbfbeSPeter Collingbourne     return;
161863cbfbeSPeter Collingbourne 
162863cbfbeSPeter Collingbourne   MCObjectFileInfo MOFI;
163863cbfbeSPeter Collingbourne   MCContext MCCtx(MAI.get(), MRI.get(), &MOFI);
164863cbfbeSPeter Collingbourne   MOFI.InitMCObjectFileInfo(TT, /*PIC*/ false, CodeModel::Default, MCCtx);
165863cbfbeSPeter Collingbourne   RecordStreamer Streamer(MCCtx);
166863cbfbeSPeter Collingbourne   T->createNullTargetStreamer(Streamer);
167863cbfbeSPeter Collingbourne 
168863cbfbeSPeter Collingbourne   std::unique_ptr<MemoryBuffer> Buffer(MemoryBuffer::getMemBuffer(InlineAsm));
169863cbfbeSPeter Collingbourne   SourceMgr SrcMgr;
170863cbfbeSPeter Collingbourne   SrcMgr.AddNewSourceBuffer(std::move(Buffer), SMLoc());
171863cbfbeSPeter Collingbourne   std::unique_ptr<MCAsmParser> Parser(
172863cbfbeSPeter Collingbourne       createMCAsmParser(SrcMgr, MCCtx, Streamer, *MAI));
173863cbfbeSPeter Collingbourne 
174863cbfbeSPeter Collingbourne   MCTargetOptions MCOptions;
175863cbfbeSPeter Collingbourne   std::unique_ptr<MCTargetAsmParser> TAP(
176863cbfbeSPeter Collingbourne       T->createMCAsmParser(*STI, *Parser, *MCII, MCOptions));
177863cbfbeSPeter Collingbourne   if (!TAP)
178863cbfbeSPeter Collingbourne     return;
179863cbfbeSPeter Collingbourne 
180863cbfbeSPeter Collingbourne   Parser->setTargetParser(*TAP);
181863cbfbeSPeter Collingbourne   if (Parser->Run(false))
182863cbfbeSPeter Collingbourne     return;
183863cbfbeSPeter Collingbourne 
184*d8204472STeresa Johnson   handleSymverAliases(M, Streamer);
185*d8204472STeresa Johnson 
186863cbfbeSPeter Collingbourne   for (auto &KV : Streamer) {
187863cbfbeSPeter Collingbourne     StringRef Key = KV.first();
188863cbfbeSPeter Collingbourne     RecordStreamer::State Value = KV.second;
189e2f1b4a6SPeter Collingbourne     // FIXME: For now we just assume that all asm symbols are executable.
190e2f1b4a6SPeter Collingbourne     uint32_t Res = BasicSymbolRef::SF_Executable;
191863cbfbeSPeter Collingbourne     switch (Value) {
192863cbfbeSPeter Collingbourne     case RecordStreamer::NeverSeen:
193863cbfbeSPeter Collingbourne       llvm_unreachable("NeverSeen should have been replaced earlier");
194863cbfbeSPeter Collingbourne     case RecordStreamer::DefinedGlobal:
195863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Global;
196863cbfbeSPeter Collingbourne       break;
197863cbfbeSPeter Collingbourne     case RecordStreamer::Defined:
198863cbfbeSPeter Collingbourne       break;
199863cbfbeSPeter Collingbourne     case RecordStreamer::Global:
200863cbfbeSPeter Collingbourne     case RecordStreamer::Used:
201863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Undefined;
202863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Global;
203863cbfbeSPeter Collingbourne       break;
204863cbfbeSPeter Collingbourne     case RecordStreamer::DefinedWeak:
205863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Weak;
206863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Global;
207863cbfbeSPeter Collingbourne       break;
208863cbfbeSPeter Collingbourne     case RecordStreamer::UndefinedWeak:
209863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Weak;
210863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Undefined;
211863cbfbeSPeter Collingbourne     }
212863cbfbeSPeter Collingbourne     AsmSymbol(Key, BasicSymbolRef::Flags(Res));
213863cbfbeSPeter Collingbourne   }
214863cbfbeSPeter Collingbourne }
215863cbfbeSPeter Collingbourne 
216863cbfbeSPeter Collingbourne void ModuleSymbolTable::printSymbolName(raw_ostream &OS, Symbol S) const {
217863cbfbeSPeter Collingbourne   if (S.is<AsmSymbol *>()) {
218863cbfbeSPeter Collingbourne     OS << S.get<AsmSymbol *>()->first;
219863cbfbeSPeter Collingbourne     return;
220863cbfbeSPeter Collingbourne   }
221863cbfbeSPeter Collingbourne 
222863cbfbeSPeter Collingbourne   auto *GV = S.get<GlobalValue *>();
223863cbfbeSPeter Collingbourne   if (GV->hasDLLImportStorageClass())
224863cbfbeSPeter Collingbourne     OS << "__imp_";
225863cbfbeSPeter Collingbourne 
226863cbfbeSPeter Collingbourne   Mang.getNameWithPrefix(OS, GV, false);
227863cbfbeSPeter Collingbourne }
228863cbfbeSPeter Collingbourne 
229863cbfbeSPeter Collingbourne uint32_t ModuleSymbolTable::getSymbolFlags(Symbol S) const {
230863cbfbeSPeter Collingbourne   if (S.is<AsmSymbol *>())
231863cbfbeSPeter Collingbourne     return S.get<AsmSymbol *>()->second;
232863cbfbeSPeter Collingbourne 
233863cbfbeSPeter Collingbourne   auto *GV = S.get<GlobalValue *>();
234863cbfbeSPeter Collingbourne 
235863cbfbeSPeter Collingbourne   uint32_t Res = BasicSymbolRef::SF_None;
236863cbfbeSPeter Collingbourne   if (GV->isDeclarationForLinker())
237863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Undefined;
238863cbfbeSPeter Collingbourne   else if (GV->hasHiddenVisibility() && !GV->hasLocalLinkage())
239863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Hidden;
240863cbfbeSPeter Collingbourne   if (const GlobalVariable *GVar = dyn_cast<GlobalVariable>(GV)) {
241863cbfbeSPeter Collingbourne     if (GVar->isConstant())
242863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Const;
243863cbfbeSPeter Collingbourne   }
244e2f1b4a6SPeter Collingbourne   if (dyn_cast_or_null<Function>(GV->getBaseObject()))
245e2f1b4a6SPeter Collingbourne     Res |= BasicSymbolRef::SF_Executable;
246d64ecf26SPeter Collingbourne   if (isa<GlobalAlias>(GV))
247d64ecf26SPeter Collingbourne     Res |= BasicSymbolRef::SF_Indirect;
248863cbfbeSPeter Collingbourne   if (GV->hasPrivateLinkage())
249863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_FormatSpecific;
250863cbfbeSPeter Collingbourne   if (!GV->hasLocalLinkage())
251863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Global;
252863cbfbeSPeter Collingbourne   if (GV->hasCommonLinkage())
253863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Common;
254863cbfbeSPeter Collingbourne   if (GV->hasLinkOnceLinkage() || GV->hasWeakLinkage() ||
255863cbfbeSPeter Collingbourne       GV->hasExternalWeakLinkage())
256863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Weak;
257863cbfbeSPeter Collingbourne 
258863cbfbeSPeter Collingbourne   if (GV->getName().startswith("llvm."))
259863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_FormatSpecific;
260863cbfbeSPeter Collingbourne   else if (auto *Var = dyn_cast<GlobalVariable>(GV)) {
261863cbfbeSPeter Collingbourne     if (Var->getSection() == "llvm.metadata")
262863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_FormatSpecific;
263863cbfbeSPeter Collingbourne   }
264863cbfbeSPeter Collingbourne 
265863cbfbeSPeter Collingbourne   return Res;
266863cbfbeSPeter Collingbourne }
267