1 //===--- ObjectFilePCHContainerOperations.cpp -----------------------------===//
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 #include "clang/CodeGen/ObjectFilePCHContainerOperations.h"
11 #include "CGDebugInfo.h"
12 #include "CodeGenModule.h"
13 #include "clang/AST/ASTContext.h"
14 #include "clang/AST/DeclObjC.h"
15 #include "clang/AST/Expr.h"
16 #include "clang/AST/RecursiveASTVisitor.h"
17 #include "clang/Basic/Diagnostic.h"
18 #include "clang/Basic/TargetInfo.h"
19 #include "clang/CodeGen/BackendUtil.h"
20 #include "clang/Frontend/CodeGenOptions.h"
21 #include "clang/Frontend/CompilerInstance.h"
22 #include "clang/Lex/HeaderSearch.h"
23 #include "clang/Lex/Preprocessor.h"
24 #include "clang/Serialization/ASTWriter.h"
25 #include "llvm/ADT/StringRef.h"
26 #include "llvm/Bitcode/BitstreamReader.h"
27 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
28 #include "llvm/IR/Constants.h"
29 #include "llvm/IR/DataLayout.h"
30 #include "llvm/IR/LLVMContext.h"
31 #include "llvm/IR/Module.h"
32 #include "llvm/Object/COFF.h"
33 #include "llvm/Object/ObjectFile.h"
34 #include "llvm/Support/Path.h"
35 #include "llvm/Support/TargetRegistry.h"
36 #include <memory>
37 #include <utility>
38 
39 using namespace clang;
40 
41 #define DEBUG_TYPE "pchcontainer"
42 
43 namespace {
44 class PCHContainerGenerator : public ASTConsumer {
45   DiagnosticsEngine &Diags;
46   const std::string MainFileName;
47   const std::string OutputFileName;
48   ASTContext *Ctx;
49   ModuleMap &MMap;
50   const HeaderSearchOptions &HeaderSearchOpts;
51   const PreprocessorOptions &PreprocessorOpts;
52   CodeGenOptions CodeGenOpts;
53   const TargetOptions TargetOpts;
54   const LangOptions LangOpts;
55   std::unique_ptr<llvm::LLVMContext> VMContext;
56   std::unique_ptr<llvm::Module> M;
57   std::unique_ptr<CodeGen::CodeGenModule> Builder;
58   std::unique_ptr<raw_pwrite_stream> OS;
59   std::shared_ptr<PCHBuffer> Buffer;
60 
61   /// Visit every type and emit debug info for it.
62   struct DebugTypeVisitor : public RecursiveASTVisitor<DebugTypeVisitor> {
63     clang::CodeGen::CGDebugInfo &DI;
64     ASTContext &Ctx;
65     DebugTypeVisitor(clang::CodeGen::CGDebugInfo &DI, ASTContext &Ctx)
66         : DI(DI), Ctx(Ctx) {}
67 
68     /// Determine whether this type can be represented in DWARF.
69     static bool CanRepresent(const Type *Ty) {
70       return !Ty->isDependentType() && !Ty->isUndeducedType();
71     }
72 
73     bool VisitImportDecl(ImportDecl *D) {
74       if (!D->getImportedOwningModule())
75         DI.EmitImportDecl(*D);
76       return true;
77     }
78 
79     bool VisitTypeDecl(TypeDecl *D) {
80       // TagDecls may be deferred until after all decls have been merged and we
81       // know the complete type. Pure forward declarations will be skipped, but
82       // they don't need to be emitted into the module anyway.
83       if (auto *TD = dyn_cast<TagDecl>(D))
84         if (!TD->isCompleteDefinition())
85           return true;
86 
87       QualType QualTy = Ctx.getTypeDeclType(D);
88       if (!QualTy.isNull() && CanRepresent(QualTy.getTypePtr()))
89         DI.getOrCreateStandaloneType(QualTy, D->getLocation());
90       return true;
91     }
92 
93     bool VisitObjCInterfaceDecl(ObjCInterfaceDecl *D) {
94       QualType QualTy(D->getTypeForDecl(), 0);
95       if (!QualTy.isNull() && CanRepresent(QualTy.getTypePtr()))
96         DI.getOrCreateStandaloneType(QualTy, D->getLocation());
97       return true;
98     }
99 
100     bool VisitFunctionDecl(FunctionDecl *D) {
101       if (isa<CXXMethodDecl>(D))
102         // This is not yet supported. Constructing the `this' argument
103         // mandates a CodeGenFunction.
104         return true;
105 
106       SmallVector<QualType, 16> ArgTypes;
107       for (auto i : D->parameters())
108         ArgTypes.push_back(i->getType());
109       QualType RetTy = D->getReturnType();
110       QualType FnTy = Ctx.getFunctionType(RetTy, ArgTypes,
111                                           FunctionProtoType::ExtProtoInfo());
112       if (CanRepresent(FnTy.getTypePtr()))
113         DI.EmitFunctionDecl(D, D->getLocation(), FnTy);
114       return true;
115     }
116 
117     bool VisitObjCMethodDecl(ObjCMethodDecl *D) {
118       if (!D->getClassInterface())
119         return true;
120 
121       bool selfIsPseudoStrong, selfIsConsumed;
122       SmallVector<QualType, 16> ArgTypes;
123       ArgTypes.push_back(D->getSelfType(Ctx, D->getClassInterface(),
124                                         selfIsPseudoStrong, selfIsConsumed));
125       ArgTypes.push_back(Ctx.getObjCSelType());
126       for (auto i : D->parameters())
127         ArgTypes.push_back(i->getType());
128       QualType RetTy = D->getReturnType();
129       QualType FnTy = Ctx.getFunctionType(RetTy, ArgTypes,
130                                           FunctionProtoType::ExtProtoInfo());
131       if (CanRepresent(FnTy.getTypePtr()))
132         DI.EmitFunctionDecl(D, D->getLocation(), FnTy);
133       return true;
134     }
135   };
136 
137 public:
138   PCHContainerGenerator(CompilerInstance &CI, const std::string &MainFileName,
139                         const std::string &OutputFileName,
140                         std::unique_ptr<raw_pwrite_stream> OS,
141                         std::shared_ptr<PCHBuffer> Buffer)
142       : Diags(CI.getDiagnostics()), MainFileName(MainFileName),
143         OutputFileName(OutputFileName), Ctx(nullptr),
144         MMap(CI.getPreprocessor().getHeaderSearchInfo().getModuleMap()),
145         HeaderSearchOpts(CI.getHeaderSearchOpts()),
146         PreprocessorOpts(CI.getPreprocessorOpts()),
147         TargetOpts(CI.getTargetOpts()), LangOpts(CI.getLangOpts()),
148         OS(std::move(OS)), Buffer(std::move(Buffer)) {
149     // The debug info output isn't affected by CodeModel and
150     // ThreadModel, but the backend expects them to be nonempty.
151     CodeGenOpts.CodeModel = "default";
152     CodeGenOpts.ThreadModel = "single";
153     CodeGenOpts.DebugTypeExtRefs = true;
154     // When building a module MainFileName is the name of the modulemap file.
155     CodeGenOpts.MainFileName =
156         LangOpts.CurrentModule.empty() ? MainFileName : LangOpts.CurrentModule;
157     CodeGenOpts.setDebugInfo(codegenoptions::FullDebugInfo);
158     CodeGenOpts.setDebuggerTuning(CI.getCodeGenOpts().getDebuggerTuning());
159   }
160 
161   ~PCHContainerGenerator() override = default;
162 
163   void Initialize(ASTContext &Context) override {
164     assert(!Ctx && "initialized multiple times");
165 
166     Ctx = &Context;
167     VMContext.reset(new llvm::LLVMContext());
168     M.reset(new llvm::Module(MainFileName, *VMContext));
169     M->setDataLayout(Ctx->getTargetInfo().getDataLayout());
170     Builder.reset(new CodeGen::CodeGenModule(
171         *Ctx, HeaderSearchOpts, PreprocessorOpts, CodeGenOpts, *M, Diags));
172 
173     // Prepare CGDebugInfo to emit debug info for a clang module.
174     auto *DI = Builder->getModuleDebugInfo();
175     StringRef ModuleName = llvm::sys::path::filename(MainFileName);
176     DI->setPCHDescriptor({ModuleName, "", OutputFileName,
177                           ASTFileSignature{{{~0U, ~0U, ~0U, ~0U, ~1U}}}});
178     DI->setModuleMap(MMap);
179   }
180 
181   bool HandleTopLevelDecl(DeclGroupRef D) override {
182     if (Diags.hasErrorOccurred())
183       return true;
184 
185     // Collect debug info for all decls in this group.
186     for (auto *I : D)
187       if (!I->isFromASTFile()) {
188         DebugTypeVisitor DTV(*Builder->getModuleDebugInfo(), *Ctx);
189         DTV.TraverseDecl(I);
190       }
191     return true;
192   }
193 
194   void HandleTopLevelDeclInObjCContainer(DeclGroupRef D) override {
195     HandleTopLevelDecl(D);
196   }
197 
198   void HandleTagDeclDefinition(TagDecl *D) override {
199     if (Diags.hasErrorOccurred())
200       return;
201 
202     if (D->isFromASTFile())
203       return;
204 
205     // Anonymous tag decls are deferred until we are building their declcontext.
206     if (D->getName().empty())
207       return;
208 
209     // Defer tag decls until their declcontext is complete.
210     auto *DeclCtx = D->getDeclContext();
211     while (DeclCtx) {
212       if (auto *D = dyn_cast<TagDecl>(DeclCtx))
213         if (!D->isCompleteDefinition())
214           return;
215       DeclCtx = DeclCtx->getParent();
216     }
217 
218     DebugTypeVisitor DTV(*Builder->getModuleDebugInfo(), *Ctx);
219     DTV.TraverseDecl(D);
220     Builder->UpdateCompletedType(D);
221   }
222 
223   void HandleTagDeclRequiredDefinition(const TagDecl *D) override {
224     if (Diags.hasErrorOccurred())
225       return;
226 
227     if (const RecordDecl *RD = dyn_cast<RecordDecl>(D))
228       Builder->getModuleDebugInfo()->completeRequiredType(RD);
229   }
230 
231   /// Emit a container holding the serialized AST.
232   void HandleTranslationUnit(ASTContext &Ctx) override {
233     assert(M && VMContext && Builder);
234     // Delete these on function exit.
235     std::unique_ptr<llvm::LLVMContext> VMContext = std::move(this->VMContext);
236     std::unique_ptr<llvm::Module> M = std::move(this->M);
237     std::unique_ptr<CodeGen::CodeGenModule> Builder = std::move(this->Builder);
238 
239     if (Diags.hasErrorOccurred())
240       return;
241 
242     M->setTargetTriple(Ctx.getTargetInfo().getTriple().getTriple());
243     M->setDataLayout(Ctx.getTargetInfo().getDataLayout());
244 
245     // PCH files don't have a signature field in the control block,
246     // but LLVM detects DWO CUs by looking for a non-zero DWO id.
247     // We use the lower 64 bits for debug info.
248     uint64_t Signature =
249         Buffer->Signature
250             ? (uint64_t)Buffer->Signature[1] << 32 | Buffer->Signature[0]
251             : ~1ULL;
252     Builder->getModuleDebugInfo()->setDwoId(Signature);
253 
254     // Finalize the Builder.
255     if (Builder)
256       Builder->Release();
257 
258     // Ensure the target exists.
259     std::string Error;
260     auto Triple = Ctx.getTargetInfo().getTriple();
261     if (!llvm::TargetRegistry::lookupTarget(Triple.getTriple(), Error))
262       llvm::report_fatal_error(Error);
263 
264     // Emit the serialized Clang AST into its own section.
265     assert(Buffer->IsComplete && "serialization did not complete");
266     auto &SerializedAST = Buffer->Data;
267     auto Size = SerializedAST.size();
268     auto Int8Ty = llvm::Type::getInt8Ty(*VMContext);
269     auto *Ty = llvm::ArrayType::get(Int8Ty, Size);
270     auto *Data = llvm::ConstantDataArray::getString(
271         *VMContext, StringRef(SerializedAST.data(), Size),
272         /*AddNull=*/false);
273     auto *ASTSym = new llvm::GlobalVariable(
274         *M, Ty, /*constant*/ true, llvm::GlobalVariable::InternalLinkage, Data,
275         "__clang_ast");
276     // The on-disk hashtable needs to be aligned.
277     ASTSym->setAlignment(8);
278 
279     // Mach-O also needs a segment name.
280     if (Triple.isOSBinFormatMachO())
281       ASTSym->setSection("__CLANG,__clangast");
282     // COFF has an eight character length limit.
283     else if (Triple.isOSBinFormatCOFF())
284       ASTSym->setSection("clangast");
285     else
286       ASTSym->setSection("__clangast");
287 
288     DEBUG({
289       // Print the IR for the PCH container to the debug output.
290       llvm::SmallString<0> Buffer;
291       clang::EmitBackendOutput(
292           Diags, HeaderSearchOpts, CodeGenOpts, TargetOpts, LangOpts,
293           Ctx.getTargetInfo().getDataLayout(), M.get(),
294           BackendAction::Backend_EmitLL,
295           llvm::make_unique<llvm::raw_svector_ostream>(Buffer));
296       llvm::dbgs() << Buffer;
297     });
298 
299     // Use the LLVM backend to emit the pch container.
300     clang::EmitBackendOutput(Diags, HeaderSearchOpts, CodeGenOpts, TargetOpts,
301                              LangOpts, Ctx.getTargetInfo().getDataLayout(),
302                              M.get(), BackendAction::Backend_EmitObj,
303                              std::move(OS));
304 
305     // Free the memory for the temporary buffer.
306     llvm::SmallVector<char, 0> Empty;
307     SerializedAST = std::move(Empty);
308   }
309 };
310 
311 } // anonymous namespace
312 
313 std::unique_ptr<ASTConsumer>
314 ObjectFilePCHContainerWriter::CreatePCHContainerGenerator(
315     CompilerInstance &CI, const std::string &MainFileName,
316     const std::string &OutputFileName,
317     std::unique_ptr<llvm::raw_pwrite_stream> OS,
318     std::shared_ptr<PCHBuffer> Buffer) const {
319   return llvm::make_unique<PCHContainerGenerator>(
320       CI, MainFileName, OutputFileName, std::move(OS), Buffer);
321 }
322 
323 StringRef
324 ObjectFilePCHContainerReader::ExtractPCH(llvm::MemoryBufferRef Buffer) const {
325   StringRef PCH;
326   auto OFOrErr = llvm::object::ObjectFile::createObjectFile(Buffer);
327   if (OFOrErr) {
328     auto &OF = OFOrErr.get();
329     bool IsCOFF = isa<llvm::object::COFFObjectFile>(*OF);
330     // Find the clang AST section in the container.
331     for (auto &Section : OF->sections()) {
332       StringRef Name;
333       Section.getName(Name);
334       if ((!IsCOFF && Name == "__clangast") || (IsCOFF && Name == "clangast")) {
335         Section.getContents(PCH);
336         return PCH;
337       }
338     }
339   }
340   handleAllErrors(OFOrErr.takeError(), [&](const llvm::ErrorInfoBase &EIB) {
341     if (EIB.convertToErrorCode() ==
342         llvm::object::object_error::invalid_file_type)
343       // As a fallback, treat the buffer as a raw AST.
344       PCH = Buffer.getBuffer();
345     else
346       EIB.log(llvm::errs());
347   });
348   return PCH;
349 }
350