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       auto *Import = cast<ImportDecl>(D);
75       if (!Import->getImportedOwningModule())
76         DI.EmitImportDecl(*Import);
77       return true;
78     }
79 
80     bool VisitTypeDecl(TypeDecl *D) {
81       // TagDecls may be deferred until after all decls have been merged and we
82       // know the complete type. Pure forward declarations will be skipped, but
83       // they don't need to be emitted into the module anyway.
84       if (auto *TD = dyn_cast<TagDecl>(D))
85         if (!TD->isCompleteDefinition())
86           return true;
87 
88       QualType QualTy = Ctx.getTypeDeclType(D);
89       if (!QualTy.isNull() && CanRepresent(QualTy.getTypePtr()))
90         DI.getOrCreateStandaloneType(QualTy, D->getLocation());
91       return true;
92     }
93 
94     bool VisitObjCInterfaceDecl(ObjCInterfaceDecl *D) {
95       QualType QualTy(D->getTypeForDecl(), 0);
96       if (!QualTy.isNull() && CanRepresent(QualTy.getTypePtr()))
97         DI.getOrCreateStandaloneType(QualTy, D->getLocation());
98       return true;
99     }
100 
101     bool VisitFunctionDecl(FunctionDecl *D) {
102       if (isa<CXXMethodDecl>(D))
103         // This is not yet supported. Constructing the `this' argument
104         // mandates a CodeGenFunction.
105         return true;
106 
107       SmallVector<QualType, 16> ArgTypes;
108       for (auto i : D->parameters())
109         ArgTypes.push_back(i->getType());
110       QualType RetTy = D->getReturnType();
111       QualType FnTy = Ctx.getFunctionType(RetTy, ArgTypes,
112                                           FunctionProtoType::ExtProtoInfo());
113       if (CanRepresent(FnTy.getTypePtr()))
114         DI.EmitFunctionDecl(D, D->getLocation(), FnTy);
115       return true;
116     }
117 
118     bool VisitObjCMethodDecl(ObjCMethodDecl *D) {
119       if (!D->getClassInterface())
120         return true;
121 
122       bool selfIsPseudoStrong, selfIsConsumed;
123       SmallVector<QualType, 16> ArgTypes;
124       ArgTypes.push_back(D->getSelfType(Ctx, D->getClassInterface(),
125                                         selfIsPseudoStrong, selfIsConsumed));
126       ArgTypes.push_back(Ctx.getObjCSelType());
127       for (auto i : D->parameters())
128         ArgTypes.push_back(i->getType());
129       QualType RetTy = D->getReturnType();
130       QualType FnTy = Ctx.getFunctionType(RetTy, ArgTypes,
131                                           FunctionProtoType::ExtProtoInfo());
132       if (CanRepresent(FnTy.getTypePtr()))
133         DI.EmitFunctionDecl(D, D->getLocation(), FnTy);
134       return true;
135     }
136   };
137 
138 public:
139   PCHContainerGenerator(CompilerInstance &CI, const std::string &MainFileName,
140                         const std::string &OutputFileName,
141                         std::unique_ptr<raw_pwrite_stream> OS,
142                         std::shared_ptr<PCHBuffer> Buffer)
143       : Diags(CI.getDiagnostics()), MainFileName(MainFileName),
144         OutputFileName(OutputFileName), Ctx(nullptr),
145         MMap(CI.getPreprocessor().getHeaderSearchInfo().getModuleMap()),
146         HeaderSearchOpts(CI.getHeaderSearchOpts()),
147         PreprocessorOpts(CI.getPreprocessorOpts()),
148         TargetOpts(CI.getTargetOpts()), LangOpts(CI.getLangOpts()),
149         OS(std::move(OS)), Buffer(std::move(Buffer)) {
150     // The debug info output isn't affected by CodeModel and
151     // ThreadModel, but the backend expects them to be nonempty.
152     CodeGenOpts.CodeModel = "default";
153     CodeGenOpts.ThreadModel = "single";
154     CodeGenOpts.DebugTypeExtRefs = true;
155     // When building a module MainFileName is the name of the modulemap file.
156     CodeGenOpts.MainFileName =
157         LangOpts.CurrentModule.empty() ? MainFileName : LangOpts.CurrentModule;
158     CodeGenOpts.setDebugInfo(codegenoptions::FullDebugInfo);
159     CodeGenOpts.setDebuggerTuning(CI.getCodeGenOpts().getDebuggerTuning());
160   }
161 
162   ~PCHContainerGenerator() override = default;
163 
164   void Initialize(ASTContext &Context) override {
165     assert(!Ctx && "initialized multiple times");
166 
167     Ctx = &Context;
168     VMContext.reset(new llvm::LLVMContext());
169     M.reset(new llvm::Module(MainFileName, *VMContext));
170     M->setDataLayout(Ctx->getTargetInfo().getDataLayout());
171     Builder.reset(new CodeGen::CodeGenModule(
172         *Ctx, HeaderSearchOpts, PreprocessorOpts, CodeGenOpts, *M, Diags));
173 
174     // Prepare CGDebugInfo to emit debug info for a clang module.
175     auto *DI = Builder->getModuleDebugInfo();
176     StringRef ModuleName = llvm::sys::path::filename(MainFileName);
177     DI->setPCHDescriptor({ModuleName, "", OutputFileName,
178                           ASTFileSignature{{{~0U, ~0U, ~0U, ~0U, ~1U}}}});
179     DI->setModuleMap(MMap);
180   }
181 
182   bool HandleTopLevelDecl(DeclGroupRef D) override {
183     if (Diags.hasErrorOccurred())
184       return true;
185 
186     // Collect debug info for all decls in this group.
187     for (auto *I : D)
188       if (!I->isFromASTFile()) {
189         DebugTypeVisitor DTV(*Builder->getModuleDebugInfo(), *Ctx);
190         DTV.TraverseDecl(I);
191       }
192     return true;
193   }
194 
195   void HandleTopLevelDeclInObjCContainer(DeclGroupRef D) override {
196     HandleTopLevelDecl(D);
197   }
198 
199   void HandleTagDeclDefinition(TagDecl *D) override {
200     if (Diags.hasErrorOccurred())
201       return;
202 
203     if (D->isFromASTFile())
204       return;
205 
206     // Anonymous tag decls are deferred until we are building their declcontext.
207     if (D->getName().empty())
208       return;
209 
210     // Defer tag decls until their declcontext is complete.
211     auto *DeclCtx = D->getDeclContext();
212     while (DeclCtx) {
213       if (auto *D = dyn_cast<TagDecl>(DeclCtx))
214         if (!D->isCompleteDefinition())
215           return;
216       DeclCtx = DeclCtx->getParent();
217     }
218 
219     DebugTypeVisitor DTV(*Builder->getModuleDebugInfo(), *Ctx);
220     DTV.TraverseDecl(D);
221     Builder->UpdateCompletedType(D);
222   }
223 
224   void HandleTagDeclRequiredDefinition(const TagDecl *D) override {
225     if (Diags.hasErrorOccurred())
226       return;
227 
228     if (const RecordDecl *RD = dyn_cast<RecordDecl>(D))
229       Builder->getModuleDebugInfo()->completeRequiredType(RD);
230   }
231 
232   /// Emit a container holding the serialized AST.
233   void HandleTranslationUnit(ASTContext &Ctx) override {
234     assert(M && VMContext && Builder);
235     // Delete these on function exit.
236     std::unique_ptr<llvm::LLVMContext> VMContext = std::move(this->VMContext);
237     std::unique_ptr<llvm::Module> M = std::move(this->M);
238     std::unique_ptr<CodeGen::CodeGenModule> Builder = std::move(this->Builder);
239 
240     if (Diags.hasErrorOccurred())
241       return;
242 
243     M->setTargetTriple(Ctx.getTargetInfo().getTriple().getTriple());
244     M->setDataLayout(Ctx.getTargetInfo().getDataLayout());
245 
246     // PCH files don't have a signature field in the control block,
247     // but LLVM detects DWO CUs by looking for a non-zero DWO id.
248     // We use the lower 64 bits for debug info.
249     uint64_t Signature =
250         Buffer->Signature
251             ? (uint64_t)Buffer->Signature[1] << 32 | Buffer->Signature[0]
252             : ~1ULL;
253     Builder->getModuleDebugInfo()->setDwoId(Signature);
254 
255     // Finalize the Builder.
256     if (Builder)
257       Builder->Release();
258 
259     // Ensure the target exists.
260     std::string Error;
261     auto Triple = Ctx.getTargetInfo().getTriple();
262     if (!llvm::TargetRegistry::lookupTarget(Triple.getTriple(), Error))
263       llvm::report_fatal_error(Error);
264 
265     // Emit the serialized Clang AST into its own section.
266     assert(Buffer->IsComplete && "serialization did not complete");
267     auto &SerializedAST = Buffer->Data;
268     auto Size = SerializedAST.size();
269     auto Int8Ty = llvm::Type::getInt8Ty(*VMContext);
270     auto *Ty = llvm::ArrayType::get(Int8Ty, Size);
271     auto *Data = llvm::ConstantDataArray::getString(
272         *VMContext, StringRef(SerializedAST.data(), Size),
273         /*AddNull=*/false);
274     auto *ASTSym = new llvm::GlobalVariable(
275         *M, Ty, /*constant*/ true, llvm::GlobalVariable::InternalLinkage, Data,
276         "__clang_ast");
277     // The on-disk hashtable needs to be aligned.
278     ASTSym->setAlignment(8);
279 
280     // Mach-O also needs a segment name.
281     if (Triple.isOSBinFormatMachO())
282       ASTSym->setSection("__CLANG,__clangast");
283     // COFF has an eight character length limit.
284     else if (Triple.isOSBinFormatCOFF())
285       ASTSym->setSection("clangast");
286     else
287       ASTSym->setSection("__clangast");
288 
289     DEBUG({
290       // Print the IR for the PCH container to the debug output.
291       llvm::SmallString<0> Buffer;
292       clang::EmitBackendOutput(
293           Diags, HeaderSearchOpts, CodeGenOpts, TargetOpts, LangOpts,
294           Ctx.getTargetInfo().getDataLayout(), M.get(),
295           BackendAction::Backend_EmitLL,
296           llvm::make_unique<llvm::raw_svector_ostream>(Buffer));
297       llvm::dbgs() << Buffer;
298     });
299 
300     // Use the LLVM backend to emit the pch container.
301     clang::EmitBackendOutput(Diags, HeaderSearchOpts, CodeGenOpts, TargetOpts,
302                              LangOpts, Ctx.getTargetInfo().getDataLayout(),
303                              M.get(), BackendAction::Backend_EmitObj,
304                              std::move(OS));
305 
306     // Free the memory for the temporary buffer.
307     llvm::SmallVector<char, 0> Empty;
308     SerializedAST = std::move(Empty);
309   }
310 };
311 
312 } // anonymous namespace
313 
314 std::unique_ptr<ASTConsumer>
315 ObjectFilePCHContainerWriter::CreatePCHContainerGenerator(
316     CompilerInstance &CI, const std::string &MainFileName,
317     const std::string &OutputFileName,
318     std::unique_ptr<llvm::raw_pwrite_stream> OS,
319     std::shared_ptr<PCHBuffer> Buffer) const {
320   return llvm::make_unique<PCHContainerGenerator>(
321       CI, MainFileName, OutputFileName, std::move(OS), Buffer);
322 }
323 
324 StringRef
325 ObjectFilePCHContainerReader::ExtractPCH(llvm::MemoryBufferRef Buffer) const {
326   StringRef PCH;
327   auto OFOrErr = llvm::object::ObjectFile::createObjectFile(Buffer);
328   if (OFOrErr) {
329     auto &OF = OFOrErr.get();
330     bool IsCOFF = isa<llvm::object::COFFObjectFile>(*OF);
331     // Find the clang AST section in the container.
332     for (auto &Section : OF->sections()) {
333       StringRef Name;
334       Section.getName(Name);
335       if ((!IsCOFF && Name == "__clangast") || (IsCOFF && Name == "clangast")) {
336         Section.getContents(PCH);
337         return PCH;
338       }
339     }
340   }
341   handleAllErrors(OFOrErr.takeError(), [&](const llvm::ErrorInfoBase &EIB) {
342     if (EIB.convertToErrorCode() ==
343         llvm::object::object_error::invalid_file_type)
344       // As a fallback, treat the buffer as a raw AST.
345       PCH = Buffer.getBuffer();
346     else
347       EIB.log(llvm::errs());
348   });
349   return PCH;
350 }
351