1*1df42facSEugene Zelenko //===- ModuleSymbolTable.cpp - symbol table for in-memory IR --------------===//
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 "RecordStreamer.h"
17*1df42facSEugene Zelenko #include "llvm/ADT/SmallString.h"
18863cbfbeSPeter Collingbourne #include "llvm/ADT/STLExtras.h"
19*1df42facSEugene Zelenko #include "llvm/ADT/StringMap.h"
20*1df42facSEugene Zelenko #include "llvm/ADT/StringRef.h"
21*1df42facSEugene Zelenko #include "llvm/ADT/Triple.h"
22*1df42facSEugene Zelenko #include "llvm/IR/Function.h"
23*1df42facSEugene Zelenko #include "llvm/IR/GlobalAlias.h"
24*1df42facSEugene Zelenko #include "llvm/IR/GlobalValue.h"
25*1df42facSEugene Zelenko #include "llvm/IR/GlobalVariable.h"
26863cbfbeSPeter Collingbourne #include "llvm/IR/Mangler.h"
27863cbfbeSPeter Collingbourne #include "llvm/IR/Module.h"
28863cbfbeSPeter Collingbourne #include "llvm/MC/MCAsmInfo.h"
29863cbfbeSPeter Collingbourne #include "llvm/MC/MCContext.h"
30*1df42facSEugene Zelenko #include "llvm/MC/MCDirectives.h"
31863cbfbeSPeter Collingbourne #include "llvm/MC/MCInstrInfo.h"
32863cbfbeSPeter Collingbourne #include "llvm/MC/MCObjectFileInfo.h"
33863cbfbeSPeter Collingbourne #include "llvm/MC/MCParser/MCAsmParser.h"
34863cbfbeSPeter Collingbourne #include "llvm/MC/MCParser/MCTargetAsmParser.h"
35863cbfbeSPeter Collingbourne #include "llvm/MC/MCRegisterInfo.h"
36863cbfbeSPeter Collingbourne #include "llvm/MC/MCSubtargetInfo.h"
37*1df42facSEugene Zelenko #include "llvm/MC/MCSymbol.h"
38*1df42facSEugene Zelenko #include "llvm/MC/MCTargetOptions.h"
39*1df42facSEugene Zelenko #include "llvm/Object/ModuleSymbolTable.h"
40*1df42facSEugene Zelenko #include "llvm/Object/SymbolicFile.h"
41*1df42facSEugene Zelenko #include "llvm/Support/Casting.h"
42*1df42facSEugene Zelenko #include "llvm/Support/CodeGen.h"
43*1df42facSEugene Zelenko #include "llvm/Support/ErrorHandling.h"
44863cbfbeSPeter Collingbourne #include "llvm/Support/MemoryBuffer.h"
45*1df42facSEugene Zelenko #include "llvm/Support/raw_ostream.h"
46*1df42facSEugene Zelenko #include "llvm/Support/SMLoc.h"
47863cbfbeSPeter Collingbourne #include "llvm/Support/SourceMgr.h"
48863cbfbeSPeter Collingbourne #include "llvm/Support/TargetRegistry.h"
49*1df42facSEugene Zelenko #include <algorithm>
50*1df42facSEugene Zelenko #include <cassert>
51*1df42facSEugene Zelenko #include <cstdint>
52*1df42facSEugene Zelenko #include <memory>
53*1df42facSEugene Zelenko #include <string>
54*1df42facSEugene Zelenko 
55863cbfbeSPeter Collingbourne using namespace llvm;
56863cbfbeSPeter Collingbourne using namespace object;
57863cbfbeSPeter Collingbourne 
58863cbfbeSPeter Collingbourne void ModuleSymbolTable::addModule(Module *M) {
59863cbfbeSPeter Collingbourne   if (FirstMod)
60863cbfbeSPeter Collingbourne     assert(FirstMod->getTargetTriple() == M->getTargetTriple());
61863cbfbeSPeter Collingbourne   else
62863cbfbeSPeter Collingbourne     FirstMod = M;
63863cbfbeSPeter Collingbourne 
64b26bc7fdSRafael Espindola   for (GlobalValue &GV : M->global_values())
65863cbfbeSPeter Collingbourne     SymTab.push_back(&GV);
66863cbfbeSPeter Collingbourne 
67d8204472STeresa Johnson   CollectAsmSymbols(*M, [this](StringRef Name, BasicSymbolRef::Flags Flags) {
68d8204472STeresa Johnson     SymTab.push_back(new (AsmSymbols.Allocate()) AsmSymbol(Name, Flags));
69863cbfbeSPeter Collingbourne   });
70863cbfbeSPeter Collingbourne }
71863cbfbeSPeter Collingbourne 
72d8204472STeresa Johnson // Ensure ELF .symver aliases get the same binding as the defined symbol
73d8204472STeresa Johnson // they alias with.
74d8204472STeresa Johnson static void handleSymverAliases(const Module &M, RecordStreamer &Streamer) {
75d8204472STeresa Johnson   if (Streamer.symverAliases().empty())
76d8204472STeresa Johnson     return;
77d8204472STeresa Johnson 
78d8204472STeresa Johnson   // The name in the assembler will be mangled, but the name in the IR
79d8204472STeresa Johnson   // might not, so we first compute a mapping from mangled name to GV.
80d8204472STeresa Johnson   Mangler Mang;
81d8204472STeresa Johnson   SmallString<64> MangledName;
82d8204472STeresa Johnson   StringMap<const GlobalValue *> MangledNameMap;
83d8204472STeresa Johnson   auto GetMangledName = [&](const GlobalValue &GV) {
84d8204472STeresa Johnson     if (!GV.hasName())
85d8204472STeresa Johnson       return;
86d8204472STeresa Johnson 
87d8204472STeresa Johnson     MangledName.clear();
88d8204472STeresa Johnson     MangledName.reserve(GV.getName().size() + 1);
89d8204472STeresa Johnson     Mang.getNameWithPrefix(MangledName, &GV, /*CannotUsePrivateLabel=*/false);
90d8204472STeresa Johnson     MangledNameMap[MangledName] = &GV;
91d8204472STeresa Johnson   };
92d8204472STeresa Johnson   for (const Function &F : M)
93d8204472STeresa Johnson     GetMangledName(F);
94d8204472STeresa Johnson   for (const GlobalVariable &GV : M.globals())
95d8204472STeresa Johnson     GetMangledName(GV);
96d8204472STeresa Johnson   for (const GlobalAlias &GA : M.aliases())
97d8204472STeresa Johnson     GetMangledName(GA);
98d8204472STeresa Johnson 
99d8204472STeresa Johnson   // Walk all the recorded .symver aliases, and set up the binding
100d8204472STeresa Johnson   // for each alias.
101d8204472STeresa Johnson   for (auto &Symver : Streamer.symverAliases()) {
102d8204472STeresa Johnson     const MCSymbol *Aliasee = Symver.first;
103d8204472STeresa Johnson     MCSymbolAttr Attr = MCSA_Invalid;
104d8204472STeresa Johnson 
105d8204472STeresa Johnson     // First check if the aliasee binding was recorded in the asm.
106d8204472STeresa Johnson     RecordStreamer::State state = Streamer.getSymbolState(Aliasee);
107d8204472STeresa Johnson     switch (state) {
108d8204472STeresa Johnson     case RecordStreamer::Global:
109d8204472STeresa Johnson     case RecordStreamer::DefinedGlobal:
110d8204472STeresa Johnson       Attr = MCSA_Global;
111d8204472STeresa Johnson       break;
112d8204472STeresa Johnson     case RecordStreamer::UndefinedWeak:
113d8204472STeresa Johnson     case RecordStreamer::DefinedWeak:
114d8204472STeresa Johnson       Attr = MCSA_Weak;
115d8204472STeresa Johnson       break;
116d8204472STeresa Johnson     default:
117d8204472STeresa Johnson       break;
118d8204472STeresa Johnson     }
119d8204472STeresa Johnson 
120d8204472STeresa Johnson     // If we don't have a symbol attribute from assembly, then check if
121d8204472STeresa Johnson     // the aliasee was defined in the IR.
122d8204472STeresa Johnson     if (Attr == MCSA_Invalid) {
123d8204472STeresa Johnson       const auto *GV = M.getNamedValue(Aliasee->getName());
124d8204472STeresa Johnson       if (!GV) {
125d8204472STeresa Johnson         auto MI = MangledNameMap.find(Aliasee->getName());
126d8204472STeresa Johnson         if (MI != MangledNameMap.end())
127d8204472STeresa Johnson           GV = MI->second;
128d8204472STeresa Johnson         else
129d8204472STeresa Johnson           continue;
130d8204472STeresa Johnson       }
131d8204472STeresa Johnson       if (GV->hasExternalLinkage())
132d8204472STeresa Johnson         Attr = MCSA_Global;
133d8204472STeresa Johnson       else if (GV->hasLocalLinkage())
134d8204472STeresa Johnson         Attr = MCSA_Local;
135d8204472STeresa Johnson       else if (GV->isWeakForLinker())
136d8204472STeresa Johnson         Attr = MCSA_Weak;
137d8204472STeresa Johnson     }
138d8204472STeresa Johnson     if (Attr == MCSA_Invalid)
139d8204472STeresa Johnson       continue;
140d8204472STeresa Johnson 
141d8204472STeresa Johnson     // Set the detected binding on each alias with this aliasee.
142d8204472STeresa Johnson     for (auto &Alias : Symver.second)
143d8204472STeresa Johnson       Streamer.EmitSymbolAttribute(Alias, Attr);
144d8204472STeresa Johnson   }
145d8204472STeresa Johnson }
146d8204472STeresa Johnson 
147863cbfbeSPeter Collingbourne void ModuleSymbolTable::CollectAsmSymbols(
148d8204472STeresa Johnson     const Module &M,
149863cbfbeSPeter Collingbourne     function_ref<void(StringRef, BasicSymbolRef::Flags)> AsmSymbol) {
150d8204472STeresa Johnson   StringRef InlineAsm = M.getModuleInlineAsm();
151863cbfbeSPeter Collingbourne   if (InlineAsm.empty())
152863cbfbeSPeter Collingbourne     return;
153863cbfbeSPeter Collingbourne 
154863cbfbeSPeter Collingbourne   std::string Err;
155d8204472STeresa Johnson   const Triple TT(M.getTargetTriple());
156863cbfbeSPeter Collingbourne   const Target *T = TargetRegistry::lookupTarget(TT.str(), Err);
157863cbfbeSPeter Collingbourne   assert(T && T->hasMCAsmParser());
158863cbfbeSPeter Collingbourne 
159863cbfbeSPeter Collingbourne   std::unique_ptr<MCRegisterInfo> MRI(T->createMCRegInfo(TT.str()));
160863cbfbeSPeter Collingbourne   if (!MRI)
161863cbfbeSPeter Collingbourne     return;
162863cbfbeSPeter Collingbourne 
163863cbfbeSPeter Collingbourne   std::unique_ptr<MCAsmInfo> MAI(T->createMCAsmInfo(*MRI, TT.str()));
164863cbfbeSPeter Collingbourne   if (!MAI)
165863cbfbeSPeter Collingbourne     return;
166863cbfbeSPeter Collingbourne 
167863cbfbeSPeter Collingbourne   std::unique_ptr<MCSubtargetInfo> STI(
168863cbfbeSPeter Collingbourne       T->createMCSubtargetInfo(TT.str(), "", ""));
169863cbfbeSPeter Collingbourne   if (!STI)
170863cbfbeSPeter Collingbourne     return;
171863cbfbeSPeter Collingbourne 
172863cbfbeSPeter Collingbourne   std::unique_ptr<MCInstrInfo> MCII(T->createMCInstrInfo());
173863cbfbeSPeter Collingbourne   if (!MCII)
174863cbfbeSPeter Collingbourne     return;
175863cbfbeSPeter Collingbourne 
176863cbfbeSPeter Collingbourne   MCObjectFileInfo MOFI;
177863cbfbeSPeter Collingbourne   MCContext MCCtx(MAI.get(), MRI.get(), &MOFI);
178863cbfbeSPeter Collingbourne   MOFI.InitMCObjectFileInfo(TT, /*PIC*/ false, CodeModel::Default, MCCtx);
179863cbfbeSPeter Collingbourne   RecordStreamer Streamer(MCCtx);
180863cbfbeSPeter Collingbourne   T->createNullTargetStreamer(Streamer);
181863cbfbeSPeter Collingbourne 
182863cbfbeSPeter Collingbourne   std::unique_ptr<MemoryBuffer> Buffer(MemoryBuffer::getMemBuffer(InlineAsm));
183863cbfbeSPeter Collingbourne   SourceMgr SrcMgr;
184863cbfbeSPeter Collingbourne   SrcMgr.AddNewSourceBuffer(std::move(Buffer), SMLoc());
185863cbfbeSPeter Collingbourne   std::unique_ptr<MCAsmParser> Parser(
186863cbfbeSPeter Collingbourne       createMCAsmParser(SrcMgr, MCCtx, Streamer, *MAI));
187863cbfbeSPeter Collingbourne 
188863cbfbeSPeter Collingbourne   MCTargetOptions MCOptions;
189863cbfbeSPeter Collingbourne   std::unique_ptr<MCTargetAsmParser> TAP(
190863cbfbeSPeter Collingbourne       T->createMCAsmParser(*STI, *Parser, *MCII, MCOptions));
191863cbfbeSPeter Collingbourne   if (!TAP)
192863cbfbeSPeter Collingbourne     return;
193863cbfbeSPeter Collingbourne 
194863cbfbeSPeter Collingbourne   Parser->setTargetParser(*TAP);
195863cbfbeSPeter Collingbourne   if (Parser->Run(false))
196863cbfbeSPeter Collingbourne     return;
197863cbfbeSPeter Collingbourne 
198d8204472STeresa Johnson   handleSymverAliases(M, Streamer);
199d8204472STeresa Johnson 
200863cbfbeSPeter Collingbourne   for (auto &KV : Streamer) {
201863cbfbeSPeter Collingbourne     StringRef Key = KV.first();
202863cbfbeSPeter Collingbourne     RecordStreamer::State Value = KV.second;
203e2f1b4a6SPeter Collingbourne     // FIXME: For now we just assume that all asm symbols are executable.
204e2f1b4a6SPeter Collingbourne     uint32_t Res = BasicSymbolRef::SF_Executable;
205863cbfbeSPeter Collingbourne     switch (Value) {
206863cbfbeSPeter Collingbourne     case RecordStreamer::NeverSeen:
207863cbfbeSPeter Collingbourne       llvm_unreachable("NeverSeen should have been replaced earlier");
208863cbfbeSPeter Collingbourne     case RecordStreamer::DefinedGlobal:
209863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Global;
210863cbfbeSPeter Collingbourne       break;
211863cbfbeSPeter Collingbourne     case RecordStreamer::Defined:
212863cbfbeSPeter Collingbourne       break;
213863cbfbeSPeter Collingbourne     case RecordStreamer::Global:
214863cbfbeSPeter Collingbourne     case RecordStreamer::Used:
215863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Undefined;
216863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Global;
217863cbfbeSPeter Collingbourne       break;
218863cbfbeSPeter Collingbourne     case RecordStreamer::DefinedWeak:
219863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Weak;
220863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Global;
221863cbfbeSPeter Collingbourne       break;
222863cbfbeSPeter Collingbourne     case RecordStreamer::UndefinedWeak:
223863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Weak;
224863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Undefined;
225863cbfbeSPeter Collingbourne     }
226863cbfbeSPeter Collingbourne     AsmSymbol(Key, BasicSymbolRef::Flags(Res));
227863cbfbeSPeter Collingbourne   }
228863cbfbeSPeter Collingbourne }
229863cbfbeSPeter Collingbourne 
230863cbfbeSPeter Collingbourne void ModuleSymbolTable::printSymbolName(raw_ostream &OS, Symbol S) const {
231863cbfbeSPeter Collingbourne   if (S.is<AsmSymbol *>()) {
232863cbfbeSPeter Collingbourne     OS << S.get<AsmSymbol *>()->first;
233863cbfbeSPeter Collingbourne     return;
234863cbfbeSPeter Collingbourne   }
235863cbfbeSPeter Collingbourne 
236863cbfbeSPeter Collingbourne   auto *GV = S.get<GlobalValue *>();
237863cbfbeSPeter Collingbourne   if (GV->hasDLLImportStorageClass())
238863cbfbeSPeter Collingbourne     OS << "__imp_";
239863cbfbeSPeter Collingbourne 
240863cbfbeSPeter Collingbourne   Mang.getNameWithPrefix(OS, GV, false);
241863cbfbeSPeter Collingbourne }
242863cbfbeSPeter Collingbourne 
243863cbfbeSPeter Collingbourne uint32_t ModuleSymbolTable::getSymbolFlags(Symbol S) const {
244863cbfbeSPeter Collingbourne   if (S.is<AsmSymbol *>())
245863cbfbeSPeter Collingbourne     return S.get<AsmSymbol *>()->second;
246863cbfbeSPeter Collingbourne 
247863cbfbeSPeter Collingbourne   auto *GV = S.get<GlobalValue *>();
248863cbfbeSPeter Collingbourne 
249863cbfbeSPeter Collingbourne   uint32_t Res = BasicSymbolRef::SF_None;
250863cbfbeSPeter Collingbourne   if (GV->isDeclarationForLinker())
251863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Undefined;
252863cbfbeSPeter Collingbourne   else if (GV->hasHiddenVisibility() && !GV->hasLocalLinkage())
253863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Hidden;
254863cbfbeSPeter Collingbourne   if (const GlobalVariable *GVar = dyn_cast<GlobalVariable>(GV)) {
255863cbfbeSPeter Collingbourne     if (GVar->isConstant())
256863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_Const;
257863cbfbeSPeter Collingbourne   }
258e2f1b4a6SPeter Collingbourne   if (dyn_cast_or_null<Function>(GV->getBaseObject()))
259e2f1b4a6SPeter Collingbourne     Res |= BasicSymbolRef::SF_Executable;
260d64ecf26SPeter Collingbourne   if (isa<GlobalAlias>(GV))
261d64ecf26SPeter Collingbourne     Res |= BasicSymbolRef::SF_Indirect;
262863cbfbeSPeter Collingbourne   if (GV->hasPrivateLinkage())
263863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_FormatSpecific;
264863cbfbeSPeter Collingbourne   if (!GV->hasLocalLinkage())
265863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Global;
266863cbfbeSPeter Collingbourne   if (GV->hasCommonLinkage())
267863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Common;
268863cbfbeSPeter Collingbourne   if (GV->hasLinkOnceLinkage() || GV->hasWeakLinkage() ||
269863cbfbeSPeter Collingbourne       GV->hasExternalWeakLinkage())
270863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_Weak;
271863cbfbeSPeter Collingbourne 
272863cbfbeSPeter Collingbourne   if (GV->getName().startswith("llvm."))
273863cbfbeSPeter Collingbourne     Res |= BasicSymbolRef::SF_FormatSpecific;
274863cbfbeSPeter Collingbourne   else if (auto *Var = dyn_cast<GlobalVariable>(GV)) {
275863cbfbeSPeter Collingbourne     if (Var->getSection() == "llvm.metadata")
276863cbfbeSPeter Collingbourne       Res |= BasicSymbolRef::SF_FormatSpecific;
277863cbfbeSPeter Collingbourne   }
278863cbfbeSPeter Collingbourne 
279863cbfbeSPeter Collingbourne   return Res;
280863cbfbeSPeter Collingbourne }
281