1*0b57cec5SDimitry Andric //===- ModuleSymbolTable.cpp - symbol table for in-memory IR --------------===//
2*0b57cec5SDimitry Andric //
3*0b57cec5SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4*0b57cec5SDimitry Andric // See https://llvm.org/LICENSE.txt for license information.
5*0b57cec5SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6*0b57cec5SDimitry Andric //
7*0b57cec5SDimitry Andric //===----------------------------------------------------------------------===//
8*0b57cec5SDimitry Andric //
9*0b57cec5SDimitry Andric // This class represents a symbol table built from in-memory IR. It provides
10*0b57cec5SDimitry Andric // access to GlobalValues and should only be used if such access is required
11*0b57cec5SDimitry Andric // (e.g. in the LTO implementation).
12*0b57cec5SDimitry Andric //
13*0b57cec5SDimitry Andric //===----------------------------------------------------------------------===//
14*0b57cec5SDimitry Andric 
15*0b57cec5SDimitry Andric #include "llvm/Object/ModuleSymbolTable.h"
16*0b57cec5SDimitry Andric #include "RecordStreamer.h"
17*0b57cec5SDimitry Andric #include "llvm/ADT/STLExtras.h"
18*0b57cec5SDimitry Andric #include "llvm/ADT/SmallString.h"
19*0b57cec5SDimitry Andric #include "llvm/ADT/StringMap.h"
20*0b57cec5SDimitry Andric #include "llvm/ADT/StringRef.h"
21*0b57cec5SDimitry Andric #include "llvm/ADT/Triple.h"
22*0b57cec5SDimitry Andric #include "llvm/IR/Function.h"
23*0b57cec5SDimitry Andric #include "llvm/IR/GlobalAlias.h"
24*0b57cec5SDimitry Andric #include "llvm/IR/GlobalValue.h"
25*0b57cec5SDimitry Andric #include "llvm/IR/GlobalVariable.h"
265ffd83dbSDimitry Andric #include "llvm/IR/InlineAsm.h"
27*0b57cec5SDimitry Andric #include "llvm/IR/Module.h"
28*0b57cec5SDimitry Andric #include "llvm/MC/MCAsmInfo.h"
29*0b57cec5SDimitry Andric #include "llvm/MC/MCContext.h"
30*0b57cec5SDimitry Andric #include "llvm/MC/MCDirectives.h"
31*0b57cec5SDimitry Andric #include "llvm/MC/MCInstrInfo.h"
32*0b57cec5SDimitry Andric #include "llvm/MC/MCObjectFileInfo.h"
33*0b57cec5SDimitry Andric #include "llvm/MC/MCParser/MCAsmParser.h"
34*0b57cec5SDimitry Andric #include "llvm/MC/MCParser/MCTargetAsmParser.h"
35*0b57cec5SDimitry Andric #include "llvm/MC/MCRegisterInfo.h"
36*0b57cec5SDimitry Andric #include "llvm/MC/MCSubtargetInfo.h"
37*0b57cec5SDimitry Andric #include "llvm/MC/MCSymbol.h"
38*0b57cec5SDimitry Andric #include "llvm/MC/MCTargetOptions.h"
39*0b57cec5SDimitry Andric #include "llvm/Object/SymbolicFile.h"
40*0b57cec5SDimitry Andric #include "llvm/Support/Casting.h"
41*0b57cec5SDimitry Andric #include "llvm/Support/CodeGen.h"
42*0b57cec5SDimitry Andric #include "llvm/Support/ErrorHandling.h"
43*0b57cec5SDimitry Andric #include "llvm/Support/MemoryBuffer.h"
44*0b57cec5SDimitry Andric #include "llvm/Support/SMLoc.h"
45*0b57cec5SDimitry Andric #include "llvm/Support/SourceMgr.h"
46*0b57cec5SDimitry Andric #include "llvm/Support/TargetRegistry.h"
47*0b57cec5SDimitry Andric #include "llvm/Support/raw_ostream.h"
48*0b57cec5SDimitry Andric #include <algorithm>
49*0b57cec5SDimitry Andric #include <cassert>
50*0b57cec5SDimitry Andric #include <cstdint>
51*0b57cec5SDimitry Andric #include <memory>
52*0b57cec5SDimitry Andric #include <string>
53*0b57cec5SDimitry Andric 
54*0b57cec5SDimitry Andric using namespace llvm;
55*0b57cec5SDimitry Andric using namespace object;
56*0b57cec5SDimitry Andric 
57*0b57cec5SDimitry Andric void ModuleSymbolTable::addModule(Module *M) {
58*0b57cec5SDimitry Andric   if (FirstMod)
59*0b57cec5SDimitry Andric     assert(FirstMod->getTargetTriple() == M->getTargetTriple());
60*0b57cec5SDimitry Andric   else
61*0b57cec5SDimitry Andric     FirstMod = M;
62*0b57cec5SDimitry Andric 
63*0b57cec5SDimitry Andric   for (GlobalValue &GV : M->global_values())
64*0b57cec5SDimitry Andric     SymTab.push_back(&GV);
65*0b57cec5SDimitry Andric 
66*0b57cec5SDimitry Andric   CollectAsmSymbols(*M, [this](StringRef Name, BasicSymbolRef::Flags Flags) {
675ffd83dbSDimitry Andric     SymTab.push_back(new (AsmSymbols.Allocate())
685ffd83dbSDimitry Andric                          AsmSymbol(std::string(Name), Flags));
69*0b57cec5SDimitry Andric   });
70*0b57cec5SDimitry Andric }
71*0b57cec5SDimitry Andric 
72*0b57cec5SDimitry Andric static void
73*0b57cec5SDimitry Andric initializeRecordStreamer(const Module &M,
74*0b57cec5SDimitry Andric                          function_ref<void(RecordStreamer &)> Init) {
75*0b57cec5SDimitry Andric   StringRef InlineAsm = M.getModuleInlineAsm();
76*0b57cec5SDimitry Andric   if (InlineAsm.empty())
77*0b57cec5SDimitry Andric     return;
78*0b57cec5SDimitry Andric 
79*0b57cec5SDimitry Andric   std::string Err;
80*0b57cec5SDimitry Andric   const Triple TT(M.getTargetTriple());
81*0b57cec5SDimitry Andric   const Target *T = TargetRegistry::lookupTarget(TT.str(), Err);
82*0b57cec5SDimitry Andric   assert(T && T->hasMCAsmParser());
83*0b57cec5SDimitry Andric 
84*0b57cec5SDimitry Andric   std::unique_ptr<MCRegisterInfo> MRI(T->createMCRegInfo(TT.str()));
85*0b57cec5SDimitry Andric   if (!MRI)
86*0b57cec5SDimitry Andric     return;
87*0b57cec5SDimitry Andric 
88480093f4SDimitry Andric   MCTargetOptions MCOptions;
89480093f4SDimitry Andric   std::unique_ptr<MCAsmInfo> MAI(T->createMCAsmInfo(*MRI, TT.str(), MCOptions));
90*0b57cec5SDimitry Andric   if (!MAI)
91*0b57cec5SDimitry Andric     return;
92*0b57cec5SDimitry Andric 
93*0b57cec5SDimitry Andric   std::unique_ptr<MCSubtargetInfo> STI(
94*0b57cec5SDimitry Andric       T->createMCSubtargetInfo(TT.str(), "", ""));
95*0b57cec5SDimitry Andric   if (!STI)
96*0b57cec5SDimitry Andric     return;
97*0b57cec5SDimitry Andric 
98*0b57cec5SDimitry Andric   std::unique_ptr<MCInstrInfo> MCII(T->createMCInstrInfo());
99*0b57cec5SDimitry Andric   if (!MCII)
100*0b57cec5SDimitry Andric     return;
101*0b57cec5SDimitry Andric 
102*0b57cec5SDimitry Andric   MCObjectFileInfo MOFI;
103*0b57cec5SDimitry Andric   MCContext MCCtx(MAI.get(), MRI.get(), &MOFI);
104*0b57cec5SDimitry Andric   MOFI.InitMCObjectFileInfo(TT, /*PIC*/ false, MCCtx);
105*0b57cec5SDimitry Andric   MOFI.setSDKVersion(M.getSDKVersion());
106*0b57cec5SDimitry Andric   RecordStreamer Streamer(MCCtx, M);
107*0b57cec5SDimitry Andric   T->createNullTargetStreamer(Streamer);
108*0b57cec5SDimitry Andric 
109*0b57cec5SDimitry Andric   std::unique_ptr<MemoryBuffer> Buffer(MemoryBuffer::getMemBuffer(InlineAsm));
110*0b57cec5SDimitry Andric   SourceMgr SrcMgr;
111*0b57cec5SDimitry Andric   SrcMgr.AddNewSourceBuffer(std::move(Buffer), SMLoc());
112*0b57cec5SDimitry Andric   std::unique_ptr<MCAsmParser> Parser(
113*0b57cec5SDimitry Andric       createMCAsmParser(SrcMgr, MCCtx, Streamer, *MAI));
114*0b57cec5SDimitry Andric 
115*0b57cec5SDimitry Andric   std::unique_ptr<MCTargetAsmParser> TAP(
116*0b57cec5SDimitry Andric       T->createMCAsmParser(*STI, *Parser, *MCII, MCOptions));
117*0b57cec5SDimitry Andric   if (!TAP)
118*0b57cec5SDimitry Andric     return;
119*0b57cec5SDimitry Andric 
1205ffd83dbSDimitry Andric   // Module-level inline asm is assumed to use At&t syntax (see
1215ffd83dbSDimitry Andric   // AsmPrinter::doInitialization()).
1225ffd83dbSDimitry Andric   Parser->setAssemblerDialect(InlineAsm::AD_ATT);
1235ffd83dbSDimitry Andric 
124*0b57cec5SDimitry Andric   Parser->setTargetParser(*TAP);
125*0b57cec5SDimitry Andric   if (Parser->Run(false))
126*0b57cec5SDimitry Andric     return;
127*0b57cec5SDimitry Andric 
128*0b57cec5SDimitry Andric   Init(Streamer);
129*0b57cec5SDimitry Andric }
130*0b57cec5SDimitry Andric 
131*0b57cec5SDimitry Andric void ModuleSymbolTable::CollectAsmSymbols(
132*0b57cec5SDimitry Andric     const Module &M,
133*0b57cec5SDimitry Andric     function_ref<void(StringRef, BasicSymbolRef::Flags)> AsmSymbol) {
134*0b57cec5SDimitry Andric   initializeRecordStreamer(M, [&](RecordStreamer &Streamer) {
135*0b57cec5SDimitry Andric     Streamer.flushSymverDirectives();
136*0b57cec5SDimitry Andric 
137*0b57cec5SDimitry Andric     for (auto &KV : Streamer) {
138*0b57cec5SDimitry Andric       StringRef Key = KV.first();
139*0b57cec5SDimitry Andric       RecordStreamer::State Value = KV.second;
140*0b57cec5SDimitry Andric       // FIXME: For now we just assume that all asm symbols are executable.
141*0b57cec5SDimitry Andric       uint32_t Res = BasicSymbolRef::SF_Executable;
142*0b57cec5SDimitry Andric       switch (Value) {
143*0b57cec5SDimitry Andric       case RecordStreamer::NeverSeen:
144*0b57cec5SDimitry Andric         llvm_unreachable("NeverSeen should have been replaced earlier");
145*0b57cec5SDimitry Andric       case RecordStreamer::DefinedGlobal:
146*0b57cec5SDimitry Andric         Res |= BasicSymbolRef::SF_Global;
147*0b57cec5SDimitry Andric         break;
148*0b57cec5SDimitry Andric       case RecordStreamer::Defined:
149*0b57cec5SDimitry Andric         break;
150*0b57cec5SDimitry Andric       case RecordStreamer::Global:
151*0b57cec5SDimitry Andric       case RecordStreamer::Used:
152*0b57cec5SDimitry Andric         Res |= BasicSymbolRef::SF_Undefined;
153*0b57cec5SDimitry Andric         Res |= BasicSymbolRef::SF_Global;
154*0b57cec5SDimitry Andric         break;
155*0b57cec5SDimitry Andric       case RecordStreamer::DefinedWeak:
156*0b57cec5SDimitry Andric         Res |= BasicSymbolRef::SF_Weak;
157*0b57cec5SDimitry Andric         Res |= BasicSymbolRef::SF_Global;
158*0b57cec5SDimitry Andric         break;
159*0b57cec5SDimitry Andric       case RecordStreamer::UndefinedWeak:
160*0b57cec5SDimitry Andric         Res |= BasicSymbolRef::SF_Weak;
161*0b57cec5SDimitry Andric         Res |= BasicSymbolRef::SF_Undefined;
162*0b57cec5SDimitry Andric       }
163*0b57cec5SDimitry Andric       AsmSymbol(Key, BasicSymbolRef::Flags(Res));
164*0b57cec5SDimitry Andric     }
165*0b57cec5SDimitry Andric   });
166*0b57cec5SDimitry Andric }
167*0b57cec5SDimitry Andric 
168*0b57cec5SDimitry Andric void ModuleSymbolTable::CollectAsmSymvers(
169*0b57cec5SDimitry Andric     const Module &M, function_ref<void(StringRef, StringRef)> AsmSymver) {
170*0b57cec5SDimitry Andric   initializeRecordStreamer(M, [&](RecordStreamer &Streamer) {
171*0b57cec5SDimitry Andric     for (auto &KV : Streamer.symverAliases())
172*0b57cec5SDimitry Andric       for (auto &Alias : KV.second)
173*0b57cec5SDimitry Andric         AsmSymver(KV.first->getName(), Alias);
174*0b57cec5SDimitry Andric   });
175*0b57cec5SDimitry Andric }
176*0b57cec5SDimitry Andric 
177*0b57cec5SDimitry Andric void ModuleSymbolTable::printSymbolName(raw_ostream &OS, Symbol S) const {
178*0b57cec5SDimitry Andric   if (S.is<AsmSymbol *>()) {
179*0b57cec5SDimitry Andric     OS << S.get<AsmSymbol *>()->first;
180*0b57cec5SDimitry Andric     return;
181*0b57cec5SDimitry Andric   }
182*0b57cec5SDimitry Andric 
183*0b57cec5SDimitry Andric   auto *GV = S.get<GlobalValue *>();
184*0b57cec5SDimitry Andric   if (GV->hasDLLImportStorageClass())
185*0b57cec5SDimitry Andric     OS << "__imp_";
186*0b57cec5SDimitry Andric 
187*0b57cec5SDimitry Andric   Mang.getNameWithPrefix(OS, GV, false);
188*0b57cec5SDimitry Andric }
189*0b57cec5SDimitry Andric 
190*0b57cec5SDimitry Andric uint32_t ModuleSymbolTable::getSymbolFlags(Symbol S) const {
191*0b57cec5SDimitry Andric   if (S.is<AsmSymbol *>())
192*0b57cec5SDimitry Andric     return S.get<AsmSymbol *>()->second;
193*0b57cec5SDimitry Andric 
194*0b57cec5SDimitry Andric   auto *GV = S.get<GlobalValue *>();
195*0b57cec5SDimitry Andric 
196*0b57cec5SDimitry Andric   uint32_t Res = BasicSymbolRef::SF_None;
197*0b57cec5SDimitry Andric   if (GV->isDeclarationForLinker())
198*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_Undefined;
199*0b57cec5SDimitry Andric   else if (GV->hasHiddenVisibility() && !GV->hasLocalLinkage())
200*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_Hidden;
201*0b57cec5SDimitry Andric   if (const GlobalVariable *GVar = dyn_cast<GlobalVariable>(GV)) {
202*0b57cec5SDimitry Andric     if (GVar->isConstant())
203*0b57cec5SDimitry Andric       Res |= BasicSymbolRef::SF_Const;
204*0b57cec5SDimitry Andric   }
205*0b57cec5SDimitry Andric   if (dyn_cast_or_null<Function>(GV->getBaseObject()))
206*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_Executable;
207*0b57cec5SDimitry Andric   if (isa<GlobalAlias>(GV))
208*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_Indirect;
209*0b57cec5SDimitry Andric   if (GV->hasPrivateLinkage())
210*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_FormatSpecific;
211*0b57cec5SDimitry Andric   if (!GV->hasLocalLinkage())
212*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_Global;
213*0b57cec5SDimitry Andric   if (GV->hasCommonLinkage())
214*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_Common;
215*0b57cec5SDimitry Andric   if (GV->hasLinkOnceLinkage() || GV->hasWeakLinkage() ||
216*0b57cec5SDimitry Andric       GV->hasExternalWeakLinkage())
217*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_Weak;
218*0b57cec5SDimitry Andric 
219*0b57cec5SDimitry Andric   if (GV->getName().startswith("llvm."))
220*0b57cec5SDimitry Andric     Res |= BasicSymbolRef::SF_FormatSpecific;
221*0b57cec5SDimitry Andric   else if (auto *Var = dyn_cast<GlobalVariable>(GV)) {
222*0b57cec5SDimitry Andric     if (Var->getSection() == "llvm.metadata")
223*0b57cec5SDimitry Andric       Res |= BasicSymbolRef::SF_FormatSpecific;
224*0b57cec5SDimitry Andric   }
225*0b57cec5SDimitry Andric 
226*0b57cec5SDimitry Andric   return Res;
227*0b57cec5SDimitry Andric }
228