1 //===--- ObjectFilePCHContainerOperations.cpp -----------------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 
9 #include "clang/CodeGen/ObjectFilePCHContainerOperations.h"
10 #include "CGDebugInfo.h"
11 #include "CodeGenModule.h"
12 #include "clang/AST/ASTContext.h"
13 #include "clang/AST/DeclObjC.h"
14 #include "clang/AST/Expr.h"
15 #include "clang/AST/RecursiveASTVisitor.h"
16 #include "clang/Basic/CodeGenOptions.h"
17 #include "clang/Basic/Diagnostic.h"
18 #include "clang/Basic/TargetInfo.h"
19 #include "clang/CodeGen/BackendUtil.h"
20 #include "clang/Frontend/CompilerInstance.h"
21 #include "clang/Lex/HeaderSearch.h"
22 #include "clang/Lex/Preprocessor.h"
23 #include "llvm/ADT/StringRef.h"
24 #include "llvm/Bitstream/BitstreamReader.h"
25 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
26 #include "llvm/IR/Constants.h"
27 #include "llvm/IR/DataLayout.h"
28 #include "llvm/IR/LLVMContext.h"
29 #include "llvm/IR/Module.h"
30 #include "llvm/MC/TargetRegistry.h"
31 #include "llvm/Object/COFF.h"
32 #include "llvm/Object/ObjectFile.h"
33 #include "llvm/Support/Path.h"
34 #include <memory>
35 #include <utility>
36 
37 using namespace clang;
38 
39 #define DEBUG_TYPE "pchcontainer"
40 
41 namespace {
42 class PCHContainerGenerator : public ASTConsumer {
43   DiagnosticsEngine &Diags;
44   const std::string MainFileName;
45   const std::string OutputFileName;
46   ASTContext *Ctx;
47   ModuleMap &MMap;
48   IntrusiveRefCntPtr<llvm::vfs::FileSystem> FS;
49   const HeaderSearchOptions &HeaderSearchOpts;
50   const PreprocessorOptions &PreprocessorOpts;
51   CodeGenOptions CodeGenOpts;
52   const TargetOptions TargetOpts;
53   LangOptions LangOpts;
54   std::unique_ptr<llvm::LLVMContext> VMContext;
55   std::unique_ptr<llvm::Module> M;
56   std::unique_ptr<CodeGen::CodeGenModule> Builder;
57   std::unique_ptr<raw_pwrite_stream> OS;
58   std::shared_ptr<PCHBuffer> Buffer;
59 
60   /// Visit every type and emit debug info for it.
61   struct DebugTypeVisitor : public RecursiveASTVisitor<DebugTypeVisitor> {
62     clang::CodeGen::CGDebugInfo &DI;
63     ASTContext &Ctx;
64     DebugTypeVisitor(clang::CodeGen::CGDebugInfo &DI, ASTContext &Ctx)
65         : DI(DI), Ctx(Ctx) {}
66 
67     /// Determine whether this type can be represented in DWARF.
68     static bool CanRepresent(const Type *Ty) {
69       return !Ty->isDependentType() && !Ty->isUndeducedType();
70     }
71 
72     bool VisitImportDecl(ImportDecl *D) {
73       if (!D->getImportedOwningModule())
74         DI.EmitImportDecl(*D);
75       return true;
76     }
77 
78     bool VisitTypeDecl(TypeDecl *D) {
79       // TagDecls may be deferred until after all decls have been merged and we
80       // know the complete type. Pure forward declarations will be skipped, but
81       // they don't need to be emitted into the module anyway.
82       if (auto *TD = dyn_cast<TagDecl>(D))
83         if (!TD->isCompleteDefinition())
84           return true;
85 
86       QualType QualTy = Ctx.getTypeDeclType(D);
87       if (!QualTy.isNull() && CanRepresent(QualTy.getTypePtr()))
88         DI.getOrCreateStandaloneType(QualTy, D->getLocation());
89       return true;
90     }
91 
92     bool VisitObjCInterfaceDecl(ObjCInterfaceDecl *D) {
93       QualType QualTy(D->getTypeForDecl(), 0);
94       if (!QualTy.isNull() && CanRepresent(QualTy.getTypePtr()))
95         DI.getOrCreateStandaloneType(QualTy, D->getLocation());
96       return true;
97     }
98 
99     bool VisitFunctionDecl(FunctionDecl *D) {
100       // Skip deduction guides.
101       if (isa<CXXDeductionGuideDecl>(D))
102         return true;
103 
104       if (isa<CXXMethodDecl>(D))
105         // This is not yet supported. Constructing the `this' argument
106         // mandates a CodeGenFunction.
107         return true;
108 
109       SmallVector<QualType, 16> ArgTypes;
110       for (auto *i : D->parameters())
111         ArgTypes.push_back(i->getType());
112       QualType RetTy = D->getReturnType();
113       QualType FnTy = Ctx.getFunctionType(RetTy, ArgTypes,
114                                           FunctionProtoType::ExtProtoInfo());
115       if (CanRepresent(FnTy.getTypePtr()))
116         DI.EmitFunctionDecl(D, D->getLocation(), FnTy);
117       return true;
118     }
119 
120     bool VisitObjCMethodDecl(ObjCMethodDecl *D) {
121       if (!D->getClassInterface())
122         return true;
123 
124       bool selfIsPseudoStrong, selfIsConsumed;
125       SmallVector<QualType, 16> ArgTypes;
126       ArgTypes.push_back(D->getSelfType(Ctx, D->getClassInterface(),
127                                         selfIsPseudoStrong, selfIsConsumed));
128       ArgTypes.push_back(Ctx.getObjCSelType());
129       for (auto *i : D->parameters())
130         ArgTypes.push_back(i->getType());
131       QualType RetTy = D->getReturnType();
132       QualType FnTy = Ctx.getFunctionType(RetTy, ArgTypes,
133                                           FunctionProtoType::ExtProtoInfo());
134       if (CanRepresent(FnTy.getTypePtr()))
135         DI.EmitFunctionDecl(D, D->getLocation(), FnTy);
136       return true;
137     }
138   };
139 
140 public:
141   PCHContainerGenerator(CompilerInstance &CI, const std::string &MainFileName,
142                         const std::string &OutputFileName,
143                         std::unique_ptr<raw_pwrite_stream> OS,
144                         std::shared_ptr<PCHBuffer> Buffer)
145       : Diags(CI.getDiagnostics()), MainFileName(MainFileName),
146         OutputFileName(OutputFileName), Ctx(nullptr),
147         MMap(CI.getPreprocessor().getHeaderSearchInfo().getModuleMap()),
148         FS(&CI.getVirtualFileSystem()),
149         HeaderSearchOpts(CI.getHeaderSearchOpts()),
150         PreprocessorOpts(CI.getPreprocessorOpts()),
151         TargetOpts(CI.getTargetOpts()), LangOpts(CI.getLangOpts()),
152         OS(std::move(OS)), Buffer(std::move(Buffer)) {
153     // The debug info output isn't affected by CodeModel and
154     // ThreadModel, but the backend expects them to be nonempty.
155     CodeGenOpts.CodeModel = "default";
156     LangOpts.setThreadModel(LangOptions::ThreadModelKind::Single);
157     CodeGenOpts.DebugTypeExtRefs = true;
158     // When building a module MainFileName is the name of the modulemap file.
159     CodeGenOpts.MainFileName =
160         LangOpts.CurrentModule.empty() ? MainFileName : LangOpts.CurrentModule;
161     CodeGenOpts.setDebugInfo(llvm::codegenoptions::FullDebugInfo);
162     CodeGenOpts.setDebuggerTuning(CI.getCodeGenOpts().getDebuggerTuning());
163     CodeGenOpts.DebugPrefixMap =
164         CI.getInvocation().getCodeGenOpts().DebugPrefixMap;
165     CodeGenOpts.DebugStrictDwarf = CI.getCodeGenOpts().DebugStrictDwarf;
166   }
167 
168   ~PCHContainerGenerator() override = default;
169 
170   void Initialize(ASTContext &Context) override {
171     assert(!Ctx && "initialized multiple times");
172 
173     Ctx = &Context;
174     VMContext.reset(new llvm::LLVMContext());
175     M.reset(new llvm::Module(MainFileName, *VMContext));
176     M->setDataLayout(Ctx->getTargetInfo().getDataLayoutString());
177     Builder.reset(new CodeGen::CodeGenModule(
178         *Ctx, FS, HeaderSearchOpts, PreprocessorOpts, CodeGenOpts, *M, Diags));
179 
180     // Prepare CGDebugInfo to emit debug info for a clang module.
181     auto *DI = Builder->getModuleDebugInfo();
182     StringRef ModuleName = llvm::sys::path::filename(MainFileName);
183     DI->setPCHDescriptor(
184         {ModuleName, "", OutputFileName, ASTFileSignature::createDISentinel()});
185     DI->setModuleMap(MMap);
186   }
187 
188   bool HandleTopLevelDecl(DeclGroupRef D) override {
189     if (Diags.hasErrorOccurred())
190       return true;
191 
192     // Collect debug info for all decls in this group.
193     for (auto *I : D)
194       if (!I->isFromASTFile()) {
195         DebugTypeVisitor DTV(*Builder->getModuleDebugInfo(), *Ctx);
196         DTV.TraverseDecl(I);
197       }
198     return true;
199   }
200 
201   void HandleTopLevelDeclInObjCContainer(DeclGroupRef D) override {
202     HandleTopLevelDecl(D);
203   }
204 
205   void HandleTagDeclDefinition(TagDecl *D) override {
206     if (Diags.hasErrorOccurred())
207       return;
208 
209     if (D->isFromASTFile())
210       return;
211 
212     // Anonymous tag decls are deferred until we are building their declcontext.
213     if (D->getName().empty())
214       return;
215 
216     // Defer tag decls until their declcontext is complete.
217     auto *DeclCtx = D->getDeclContext();
218     while (DeclCtx) {
219       if (auto *D = dyn_cast<TagDecl>(DeclCtx))
220         if (!D->isCompleteDefinition())
221           return;
222       DeclCtx = DeclCtx->getParent();
223     }
224 
225     DebugTypeVisitor DTV(*Builder->getModuleDebugInfo(), *Ctx);
226     DTV.TraverseDecl(D);
227     Builder->UpdateCompletedType(D);
228   }
229 
230   void HandleTagDeclRequiredDefinition(const TagDecl *D) override {
231     if (Diags.hasErrorOccurred())
232       return;
233 
234     if (const RecordDecl *RD = dyn_cast<RecordDecl>(D))
235       Builder->getModuleDebugInfo()->completeRequiredType(RD);
236   }
237 
238   void HandleImplicitImportDecl(ImportDecl *D) override {
239     if (!D->getImportedOwningModule())
240       Builder->getModuleDebugInfo()->EmitImportDecl(*D);
241   }
242 
243   /// Emit a container holding the serialized AST.
244   void HandleTranslationUnit(ASTContext &Ctx) override {
245     assert(M && VMContext && Builder);
246     // Delete these on function exit.
247     std::unique_ptr<llvm::LLVMContext> VMContext = std::move(this->VMContext);
248     std::unique_ptr<llvm::Module> M = std::move(this->M);
249     std::unique_ptr<CodeGen::CodeGenModule> Builder = std::move(this->Builder);
250 
251     if (Diags.hasErrorOccurred())
252       return;
253 
254     M->setTargetTriple(Ctx.getTargetInfo().getTriple().getTriple());
255     M->setDataLayout(Ctx.getTargetInfo().getDataLayoutString());
256 
257     // PCH files don't have a signature field in the control block,
258     // but LLVM detects DWO CUs by looking for a non-zero DWO id.
259     // We use the lower 64 bits for debug info.
260 
261     uint64_t Signature =
262         Buffer->Signature ? Buffer->Signature.truncatedValue() : ~1ULL;
263 
264     Builder->getModuleDebugInfo()->setDwoId(Signature);
265 
266     // Finalize the Builder.
267     if (Builder)
268       Builder->Release();
269 
270     // Ensure the target exists.
271     std::string Error;
272     auto Triple = Ctx.getTargetInfo().getTriple();
273     if (!llvm::TargetRegistry::lookupTarget(Triple.getTriple(), Error))
274       llvm::report_fatal_error(llvm::Twine(Error));
275 
276     // Emit the serialized Clang AST into its own section.
277     assert(Buffer->IsComplete && "serialization did not complete");
278     auto &SerializedAST = Buffer->Data;
279     auto Size = SerializedAST.size();
280 
281     if (Triple.isOSBinFormatWasm()) {
282       // Emit __clangast in custom section instead of named data segment
283       // to find it while iterating sections.
284       // This could be avoided if all data segements (the wasm sense) were
285       // represented as their own sections (in the llvm sense).
286       // TODO: https://github.com/WebAssembly/tool-conventions/issues/138
287       llvm::NamedMDNode *MD =
288           M->getOrInsertNamedMetadata("wasm.custom_sections");
289       llvm::Metadata *Ops[2] = {
290           llvm::MDString::get(*VMContext, "__clangast"),
291           llvm::MDString::get(*VMContext,
292                               StringRef(SerializedAST.data(), Size))};
293       auto *NameAndContent = llvm::MDTuple::get(*VMContext, Ops);
294       MD->addOperand(NameAndContent);
295     } else {
296       auto Int8Ty = llvm::Type::getInt8Ty(*VMContext);
297       auto *Ty = llvm::ArrayType::get(Int8Ty, Size);
298       auto *Data = llvm::ConstantDataArray::getString(
299           *VMContext, StringRef(SerializedAST.data(), Size),
300           /*AddNull=*/false);
301       auto *ASTSym = new llvm::GlobalVariable(
302           *M, Ty, /*constant*/ true, llvm::GlobalVariable::InternalLinkage,
303           Data, "__clang_ast");
304       // The on-disk hashtable needs to be aligned.
305       ASTSym->setAlignment(llvm::Align(8));
306 
307       // Mach-O also needs a segment name.
308       if (Triple.isOSBinFormatMachO())
309         ASTSym->setSection("__CLANG,__clangast");
310       // COFF has an eight character length limit.
311       else if (Triple.isOSBinFormatCOFF())
312         ASTSym->setSection("clangast");
313       else
314         ASTSym->setSection("__clangast");
315     }
316 
317     LLVM_DEBUG({
318       // Print the IR for the PCH container to the debug output.
319       llvm::SmallString<0> Buffer;
320       clang::EmitBackendOutput(
321           Diags, HeaderSearchOpts, CodeGenOpts, TargetOpts, LangOpts,
322           Ctx.getTargetInfo().getDataLayoutString(), M.get(),
323           BackendAction::Backend_EmitLL, FS,
324           std::make_unique<llvm::raw_svector_ostream>(Buffer));
325       llvm::dbgs() << Buffer;
326     });
327 
328     // Use the LLVM backend to emit the pch container.
329     clang::EmitBackendOutput(Diags, HeaderSearchOpts, CodeGenOpts, TargetOpts,
330                              LangOpts,
331                              Ctx.getTargetInfo().getDataLayoutString(), M.get(),
332                              BackendAction::Backend_EmitObj, FS, std::move(OS));
333 
334     // Free the memory for the temporary buffer.
335     llvm::SmallVector<char, 0> Empty;
336     SerializedAST = std::move(Empty);
337   }
338 };
339 
340 } // anonymous namespace
341 
342 std::unique_ptr<ASTConsumer>
343 ObjectFilePCHContainerWriter::CreatePCHContainerGenerator(
344     CompilerInstance &CI, const std::string &MainFileName,
345     const std::string &OutputFileName,
346     std::unique_ptr<llvm::raw_pwrite_stream> OS,
347     std::shared_ptr<PCHBuffer> Buffer) const {
348   return std::make_unique<PCHContainerGenerator>(
349       CI, MainFileName, OutputFileName, std::move(OS), Buffer);
350 }
351 
352 ArrayRef<StringRef> ObjectFilePCHContainerReader::getFormats() const {
353   static StringRef Formats[] = {"obj", "raw"};
354   return Formats;
355 }
356 
357 StringRef
358 ObjectFilePCHContainerReader::ExtractPCH(llvm::MemoryBufferRef Buffer) const {
359   StringRef PCH;
360   auto OFOrErr = llvm::object::ObjectFile::createObjectFile(Buffer);
361   if (OFOrErr) {
362     auto &OF = OFOrErr.get();
363     bool IsCOFF = isa<llvm::object::COFFObjectFile>(*OF);
364     // Find the clang AST section in the container.
365     for (auto &Section : OF->sections()) {
366       StringRef Name;
367       if (Expected<StringRef> NameOrErr = Section.getName())
368         Name = *NameOrErr;
369       else
370         consumeError(NameOrErr.takeError());
371 
372       if ((!IsCOFF && Name == "__clangast") || (IsCOFF && Name == "clangast")) {
373         if (Expected<StringRef> E = Section.getContents())
374           return *E;
375         else {
376           handleAllErrors(E.takeError(), [&](const llvm::ErrorInfoBase &EIB) {
377             EIB.log(llvm::errs());
378           });
379           return "";
380         }
381       }
382     }
383   }
384   handleAllErrors(OFOrErr.takeError(), [&](const llvm::ErrorInfoBase &EIB) {
385     if (EIB.convertToErrorCode() ==
386         llvm::object::object_error::invalid_file_type)
387       // As a fallback, treat the buffer as a raw AST.
388       PCH = Buffer.getBuffer();
389     else
390       EIB.log(llvm::errs());
391   });
392   return PCH;
393 }
394