IRObjectFile.cpp revision 309124
1//===- IRObjectFile.cpp - IR object file implementation ---------*- C++ -*-===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// Part of the IRObjectFile class implementation.
11//
12//===----------------------------------------------------------------------===//
13
14#include "llvm/Object/IRObjectFile.h"
15#include "RecordStreamer.h"
16#include "llvm/ADT/STLExtras.h"
17#include "llvm/Bitcode/ReaderWriter.h"
18#include "llvm/IR/GVMaterializer.h"
19#include "llvm/IR/LLVMContext.h"
20#include "llvm/IR/Mangler.h"
21#include "llvm/IR/Module.h"
22#include "llvm/MC/MCAsmInfo.h"
23#include "llvm/MC/MCContext.h"
24#include "llvm/MC/MCInstrInfo.h"
25#include "llvm/MC/MCObjectFileInfo.h"
26#include "llvm/MC/MCParser/MCAsmParser.h"
27#include "llvm/MC/MCParser/MCTargetAsmParser.h"
28#include "llvm/MC/MCRegisterInfo.h"
29#include "llvm/MC/MCSubtargetInfo.h"
30#include "llvm/Object/ObjectFile.h"
31#include "llvm/Support/MemoryBuffer.h"
32#include "llvm/Support/SourceMgr.h"
33#include "llvm/Support/TargetRegistry.h"
34#include "llvm/Support/raw_ostream.h"
35using namespace llvm;
36using namespace object;
37
38IRObjectFile::IRObjectFile(MemoryBufferRef Object, std::unique_ptr<Module> Mod)
39    : SymbolicFile(Binary::ID_IR, Object), M(std::move(Mod)) {
40  Mang.reset(new Mangler());
41  CollectAsmUndefinedRefs(Triple(M->getTargetTriple()), M->getModuleInlineAsm(),
42                          [this](StringRef Name, BasicSymbolRef::Flags Flags) {
43                            AsmSymbols.emplace_back(Name, std::move(Flags));
44                          });
45}
46
47// Parse inline ASM and collect the list of symbols that are not defined in
48// the current module. This is inspired from IRObjectFile.
49void IRObjectFile::CollectAsmUndefinedRefs(
50    const Triple &TT, StringRef InlineAsm,
51    function_ref<void(StringRef, BasicSymbolRef::Flags)> AsmUndefinedRefs) {
52  if (InlineAsm.empty())
53    return;
54
55  std::string Err;
56  const Target *T = TargetRegistry::lookupTarget(TT.str(), Err);
57  if (!T)
58    return;
59
60  std::unique_ptr<MCRegisterInfo> MRI(T->createMCRegInfo(TT.str()));
61  if (!MRI)
62    return;
63
64  std::unique_ptr<MCAsmInfo> MAI(T->createMCAsmInfo(*MRI, TT.str()));
65  if (!MAI)
66    return;
67
68  std::unique_ptr<MCSubtargetInfo> STI(
69      T->createMCSubtargetInfo(TT.str(), "", ""));
70  if (!STI)
71    return;
72
73  std::unique_ptr<MCInstrInfo> MCII(T->createMCInstrInfo());
74  if (!MCII)
75    return;
76
77  MCObjectFileInfo MOFI;
78  MCContext MCCtx(MAI.get(), MRI.get(), &MOFI);
79  MOFI.InitMCObjectFileInfo(TT, /*PIC*/ false, CodeModel::Default, MCCtx);
80  std::unique_ptr<RecordStreamer> Streamer(new RecordStreamer(MCCtx));
81  T->createNullTargetStreamer(*Streamer);
82
83  std::unique_ptr<MemoryBuffer> Buffer(MemoryBuffer::getMemBuffer(InlineAsm));
84  SourceMgr SrcMgr;
85  SrcMgr.AddNewSourceBuffer(std::move(Buffer), SMLoc());
86  std::unique_ptr<MCAsmParser> Parser(
87      createMCAsmParser(SrcMgr, MCCtx, *Streamer, *MAI));
88
89  MCTargetOptions MCOptions;
90  std::unique_ptr<MCTargetAsmParser> TAP(
91      T->createMCAsmParser(*STI, *Parser, *MCII, MCOptions));
92  if (!TAP)
93    return;
94
95  Parser->setTargetParser(*TAP);
96  if (Parser->Run(false))
97    return;
98
99  for (auto &KV : *Streamer) {
100    StringRef Key = KV.first();
101    RecordStreamer::State Value = KV.second;
102    uint32_t Res = BasicSymbolRef::SF_None;
103    switch (Value) {
104    case RecordStreamer::NeverSeen:
105      llvm_unreachable("foo");
106    case RecordStreamer::DefinedGlobal:
107      Res |= BasicSymbolRef::SF_Global;
108      break;
109    case RecordStreamer::Defined:
110      break;
111    case RecordStreamer::Global:
112    case RecordStreamer::Used:
113      Res |= BasicSymbolRef::SF_Undefined;
114      Res |= BasicSymbolRef::SF_Global;
115      break;
116    case RecordStreamer::GlobalWeak:
117      Res |= BasicSymbolRef::SF_Weak;
118      Res |= BasicSymbolRef::SF_Global;
119      break;
120    }
121    AsmUndefinedRefs(Key, BasicSymbolRef::Flags(Res));
122  }
123}
124
125IRObjectFile::~IRObjectFile() {
126 }
127
128static GlobalValue *getGV(DataRefImpl &Symb) {
129  if ((Symb.p & 3) == 3)
130    return nullptr;
131
132  return reinterpret_cast<GlobalValue*>(Symb.p & ~uintptr_t(3));
133}
134
135static uintptr_t skipEmpty(Module::const_alias_iterator I, const Module &M) {
136  if (I == M.alias_end())
137    return 3;
138  const GlobalValue *GV = &*I;
139  return reinterpret_cast<uintptr_t>(GV) | 2;
140}
141
142static uintptr_t skipEmpty(Module::const_global_iterator I, const Module &M) {
143  if (I == M.global_end())
144    return skipEmpty(M.alias_begin(), M);
145  const GlobalValue *GV = &*I;
146  return reinterpret_cast<uintptr_t>(GV) | 1;
147}
148
149static uintptr_t skipEmpty(Module::const_iterator I, const Module &M) {
150  if (I == M.end())
151    return skipEmpty(M.global_begin(), M);
152  const GlobalValue *GV = &*I;
153  return reinterpret_cast<uintptr_t>(GV) | 0;
154}
155
156static unsigned getAsmSymIndex(DataRefImpl Symb) {
157  assert((Symb.p & uintptr_t(3)) == 3);
158  uintptr_t Index = Symb.p & ~uintptr_t(3);
159  Index >>= 2;
160  return Index;
161}
162
163void IRObjectFile::moveSymbolNext(DataRefImpl &Symb) const {
164  const GlobalValue *GV = getGV(Symb);
165  uintptr_t Res;
166
167  switch (Symb.p & 3) {
168  case 0: {
169    Module::const_iterator Iter(static_cast<const Function*>(GV));
170    ++Iter;
171    Res = skipEmpty(Iter, *M);
172    break;
173  }
174  case 1: {
175    Module::const_global_iterator Iter(static_cast<const GlobalVariable*>(GV));
176    ++Iter;
177    Res = skipEmpty(Iter, *M);
178    break;
179  }
180  case 2: {
181    Module::const_alias_iterator Iter(static_cast<const GlobalAlias*>(GV));
182    ++Iter;
183    Res = skipEmpty(Iter, *M);
184    break;
185  }
186  case 3: {
187    unsigned Index = getAsmSymIndex(Symb);
188    assert(Index < AsmSymbols.size());
189    ++Index;
190    Res = (Index << 2) | 3;
191    break;
192  }
193  default:
194    llvm_unreachable("unreachable case");
195  }
196
197  Symb.p = Res;
198}
199
200std::error_code IRObjectFile::printSymbolName(raw_ostream &OS,
201                                              DataRefImpl Symb) const {
202  const GlobalValue *GV = getGV(Symb);
203  if (!GV) {
204    unsigned Index = getAsmSymIndex(Symb);
205    assert(Index <= AsmSymbols.size());
206    OS << AsmSymbols[Index].first;
207    return std::error_code();
208  }
209
210  if (GV->hasDLLImportStorageClass())
211    OS << "__imp_";
212
213  if (Mang)
214    Mang->getNameWithPrefix(OS, GV, false);
215  else
216    OS << GV->getName();
217
218  return std::error_code();
219}
220
221uint32_t IRObjectFile::getSymbolFlags(DataRefImpl Symb) const {
222  const GlobalValue *GV = getGV(Symb);
223
224  if (!GV) {
225    unsigned Index = getAsmSymIndex(Symb);
226    assert(Index <= AsmSymbols.size());
227    return AsmSymbols[Index].second;
228  }
229
230  uint32_t Res = BasicSymbolRef::SF_None;
231  if (GV->isDeclarationForLinker())
232    Res |= BasicSymbolRef::SF_Undefined;
233  else if (GV->hasHiddenVisibility() && !GV->hasLocalLinkage())
234    Res |= BasicSymbolRef::SF_Hidden;
235  if (const GlobalVariable *GVar = dyn_cast<GlobalVariable>(GV)) {
236    if (GVar->isConstant())
237      Res |= BasicSymbolRef::SF_Const;
238  }
239  if (GV->hasPrivateLinkage())
240    Res |= BasicSymbolRef::SF_FormatSpecific;
241  if (!GV->hasLocalLinkage())
242    Res |= BasicSymbolRef::SF_Global;
243  if (GV->hasCommonLinkage())
244    Res |= BasicSymbolRef::SF_Common;
245  if (GV->hasLinkOnceLinkage() || GV->hasWeakLinkage() ||
246      GV->hasExternalWeakLinkage())
247    Res |= BasicSymbolRef::SF_Weak;
248
249  if (GV->getName().startswith("llvm."))
250    Res |= BasicSymbolRef::SF_FormatSpecific;
251  else if (auto *Var = dyn_cast<GlobalVariable>(GV)) {
252    if (Var->getSection() == "llvm.metadata")
253      Res |= BasicSymbolRef::SF_FormatSpecific;
254  }
255
256  return Res;
257}
258
259GlobalValue *IRObjectFile::getSymbolGV(DataRefImpl Symb) { return getGV(Symb); }
260
261std::unique_ptr<Module> IRObjectFile::takeModule() { return std::move(M); }
262
263basic_symbol_iterator IRObjectFile::symbol_begin_impl() const {
264  Module::const_iterator I = M->begin();
265  DataRefImpl Ret;
266  Ret.p = skipEmpty(I, *M);
267  return basic_symbol_iterator(BasicSymbolRef(Ret, this));
268}
269
270basic_symbol_iterator IRObjectFile::symbol_end_impl() const {
271  DataRefImpl Ret;
272  uint64_t NumAsm = AsmSymbols.size();
273  NumAsm <<= 2;
274  Ret.p = 3 | NumAsm;
275  return basic_symbol_iterator(BasicSymbolRef(Ret, this));
276}
277
278ErrorOr<MemoryBufferRef> IRObjectFile::findBitcodeInObject(const ObjectFile &Obj) {
279  for (const SectionRef &Sec : Obj.sections()) {
280    if (Sec.isBitcode()) {
281      StringRef SecContents;
282      if (std::error_code EC = Sec.getContents(SecContents))
283        return EC;
284      return MemoryBufferRef(SecContents, Obj.getFileName());
285    }
286  }
287
288  return object_error::bitcode_section_not_found;
289}
290
291ErrorOr<MemoryBufferRef> IRObjectFile::findBitcodeInMemBuffer(MemoryBufferRef Object) {
292  sys::fs::file_magic Type = sys::fs::identify_magic(Object.getBuffer());
293  switch (Type) {
294  case sys::fs::file_magic::bitcode:
295    return Object;
296  case sys::fs::file_magic::elf_relocatable:
297  case sys::fs::file_magic::macho_object:
298  case sys::fs::file_magic::coff_object: {
299    Expected<std::unique_ptr<ObjectFile>> ObjFile =
300        ObjectFile::createObjectFile(Object, Type);
301    if (!ObjFile)
302      return errorToErrorCode(ObjFile.takeError());
303    return findBitcodeInObject(*ObjFile->get());
304  }
305  default:
306    return object_error::invalid_file_type;
307  }
308}
309
310ErrorOr<std::unique_ptr<IRObjectFile>>
311llvm::object::IRObjectFile::create(MemoryBufferRef Object,
312                                   LLVMContext &Context) {
313  ErrorOr<MemoryBufferRef> BCOrErr = findBitcodeInMemBuffer(Object);
314  if (!BCOrErr)
315    return BCOrErr.getError();
316
317  std::unique_ptr<MemoryBuffer> Buff =
318      MemoryBuffer::getMemBuffer(BCOrErr.get(), false);
319
320  ErrorOr<std::unique_ptr<Module>> MOrErr =
321      getLazyBitcodeModule(std::move(Buff), Context,
322                           /*ShouldLazyLoadMetadata*/ true);
323  if (std::error_code EC = MOrErr.getError())
324    return EC;
325
326  std::unique_ptr<Module> &M = MOrErr.get();
327  return llvm::make_unique<IRObjectFile>(Object, std::move(M));
328}
329