zturner created this revision. zturner added reviewers: aleksandr.urakov, labath, lemo. Herald added subscribers: erik.pilkington, JDevlieghere, aprantl.
This is an alternative to https://reviews.llvm.org/D54053 which uses a different approach. The goal of both is the same - to be able to improve the quality of the AST that we reconstruct when parsing the debug info. https://reviews.llvm.org/D54053 attempts to address this by demangling the unique name of each type, and using the structure of the demangler's AST to try to reconstruct a clang AST. However, there are some complications with this approach. The two biggest ones are: a) The mangling does not always provide enough information to disambiguate between two types, depending on where it occurs in the mangling. b) The mangling provides no way to differentiate outer classes from outer namespaces, so in `A::B::C`, we don't know if `A` and `B` are (class, class), (namespace, namespace), or (namespace, class). b) sounds like it could be an unimportant distinction, but since LLDB works by gradually building up an AST over time that grows as more and more debug info is parsed, you can very quickly end up in a situation where there are ambiguities in your AST. For example, you may decide that `B` is probably a namespace, so you create a `NamespaceDecl` for it in the AST, and then later someone instatiates a variable of type `A::B` and you have precise debug info telling you it's a class. This will create two decls at the same scope in the AST hierarchy with the same name, causing ambiguities and these will slowly build up over time leading to instability. The approach here is based off of the observation that the PDB contains information about nested classes in the parent -> child direction, just not the other way around. That is to say, if you have code such as: `struct A { struct B {}; };` Then the debug info record for `A` will tell you that it contains a nested type call `B`, along with an index for the full definition of `B` in the debug info. The problem we are facing all along is that if someone declares a variable of type `A::B`, they need the reverse mapping, and PDB doesn't offer that. So, the simple solution employed here is to simply pre-process all types up front and build the reverse mapping. This gives us perfect information about class hierarchy, and allows us to precisely determine if a part of a scope is a namespace (specifically, it will have no parent in the reverse mapping). But we can even re-purpose this pre-processing step for other things down the line. For example, we may wish to find all types name `Foo`, but maybe `Foo` is a template and the instantion is `Foo<int>`. We could use this pre-processing step to build this kind of hash table. And many other things as well. Note that the idea of demangling a type and using the structured demangler AST is not totally abandoned. For example, if you have a template instantiation named `Foo<int>`, the patch here will simply create a class with the name `Foo<int>`. In other words, we make no attempt to parse template parameters and create the appropriate instantiations in the AST. We also do not yet handle scoped classes (i.e. classes that are defined inside the body of a funtion). But we can handle those later. Note that I started adding a new kind of test, an ast test. I even retrofitted existing tests with ast testing functionality. I think this is a useful testing strategy to ensure we are generating correct ASTs from debug info. https://reviews.llvm.org/D54216 Files: lldb/lit/SymbolFile/NativePDB/Inputs/ast-reconstruction.lldbinit lldb/lit/SymbolFile/NativePDB/Inputs/function-types-classes.lldbinit lldb/lit/SymbolFile/NativePDB/Inputs/globals-classes.lldbinit lldb/lit/SymbolFile/NativePDB/ast-reconstruction.cpp lldb/lit/SymbolFile/NativePDB/function-types-classes.cpp lldb/lit/SymbolFile/NativePDB/global-classes.cpp lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.cpp lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.h lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h llvm/include/llvm/DebugInfo/CodeView/TypeRecord.h
Index: llvm/include/llvm/DebugInfo/CodeView/TypeRecord.h =================================================================== --- llvm/include/llvm/DebugInfo/CodeView/TypeRecord.h +++ llvm/include/llvm/DebugInfo/CodeView/TypeRecord.h @@ -429,6 +429,10 @@ return (Options & ClassOptions::ForwardReference) != ClassOptions::None; } + bool containsNestedClass() const { + return (Options & ClassOptions::ContainsNestedClass) != ClassOptions::None; + } + bool isScoped() const { return (Options & ClassOptions::Scoped) != ClassOptions::None; } Index: lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h =================================================================== --- lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h +++ lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h @@ -158,6 +158,11 @@ void DumpClangAST(Stream &s) override; private: + std::pair<clang::DeclContext *, std::string> + CreateDeclInfoForType(const llvm::codeview::TagRecord &record, + llvm::codeview::TypeIndex ti); + + void PreprocessTpiStream(); size_t FindTypesByName(llvm::StringRef name, uint32_t max_matches, TypeMap &types); @@ -176,10 +181,9 @@ const llvm::codeview::ArrayRecord &ar); lldb::TypeSP CreateProcedureType(PdbSymUid type_uid, const llvm::codeview::ProcedureRecord &pr); - lldb::TypeSP - CreateClassStructUnion(PdbSymUid type_uid, llvm::StringRef name, size_t size, - clang::TagTypeKind ttk, - clang::MSInheritanceAttr::Spelling inheritance); + lldb::TypeSP CreateClassStructUnion( + PdbSymUid type_uid, const llvm::codeview::TagRecord &record, size_t size, + clang::TagTypeKind ttk, clang::MSInheritanceAttr::Spelling inheritance); lldb::FunctionSP GetOrCreateFunction(PdbSymUid func_uid, const SymbolContext &sc); @@ -205,6 +209,8 @@ llvm::DenseMap<clang::TagDecl *, DeclStatus> m_decl_to_status; llvm::DenseMap<lldb::user_id_t, clang::TagDecl *> m_uid_to_decl; + llvm::DenseMap<llvm::codeview::TypeIndex, llvm::codeview::TypeIndex> + m_parent_types; llvm::DenseMap<lldb::user_id_t, lldb::VariableSP> m_global_vars; llvm::DenseMap<lldb::user_id_t, lldb::FunctionSP> m_functions; Index: lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp =================================================================== --- lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp +++ lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp @@ -13,13 +13,16 @@ #include "clang/AST/CharUnits.h" #include "clang/AST/Decl.h" #include "clang/AST/DeclCXX.h" +#include "clang/AST/Type.h" #include "lldb/Core/Module.h" #include "lldb/Core/PluginManager.h" #include "lldb/Core/StreamBuffer.h" +#include "lldb/Core/StreamFile.h" #include "lldb/Symbol/ClangASTContext.h" #include "lldb/Symbol/ClangASTImporter.h" #include "lldb/Symbol/ClangExternalASTSourceCommon.h" +#include "lldb/Symbol/ClangUtil.h" #include "lldb/Symbol/CompileUnit.h" #include "lldb/Symbol/LineTable.h" #include "lldb/Symbol/ObjectFile.h" @@ -43,14 +46,14 @@ #include "llvm/DebugInfo/PDB/Native/SymbolStream.h" #include "llvm/DebugInfo/PDB/Native/TpiStream.h" #include "llvm/DebugInfo/PDB/PDBTypes.h" +#include "llvm/Demangle/MicrosoftDemangle.h" #include "llvm/Object/COFF.h" #include "llvm/Support/Allocator.h" #include "llvm/Support/BinaryStreamReader.h" +#include "llvm/Support/Error.h" #include "llvm/Support/ErrorOr.h" #include "llvm/Support/MemoryBuffer.h" -#include "Plugins/Language/CPlusPlus/CPlusPlusNameParser.h" - #include "PdbSymUid.h" #include "PdbUtil.h" #include "UdtRecordCompleter.h" @@ -527,9 +530,60 @@ TypeSystem *ts = GetTypeSystemForLanguage(eLanguageTypeC_plus_plus); m_clang = llvm::dyn_cast_or_null<ClangASTContext>(ts); m_importer = llvm::make_unique<ClangASTImporter>(); + + PreprocessTpiStream(); lldbassert(m_clang); } +void SymbolFileNativePDB::PreprocessTpiStream() { + LazyRandomTypeCollection &types = m_index->tpi().typeCollection(); + + for (auto ti = types.getFirst(); ti; ti = types.getNext(*ti)) { + CVType type = types.getType(*ti); + if (!IsTagRecord(type)) + continue; + + CVTagRecord tag = CVTagRecord::create(type); + // We're looking for LF_NESTTYPE records in the field list, so ignore + // forward references (no field list), and anything without a nested class + // (since there won't be any LF_NESTTYPE records). + if (tag.asTag().isForwardRef() || !tag.asTag().containsNestedClass()) + continue; + + struct ProcessTpiStream : public TypeVisitorCallbacks { + ProcessTpiStream(PdbIndex &index, TypeIndex parent, + llvm::DenseMap<TypeIndex, TypeIndex> &parents) + : index(index), parents(parents), parent(parent) {} + + PdbIndex &index; + llvm::DenseMap<TypeIndex, TypeIndex> &parents; + TypeIndex parent; + + llvm::Error visitKnownMember(CVMemberRecord &CVR, + NestedTypeRecord &Record) override { + parents[Record.Type] = parent; + CVType child = index.tpi().getType(Record.Type); + if (!IsForwardRefUdt(child)) + return llvm::ErrorSuccess(); + llvm::Expected<TypeIndex> full_decl = + index.tpi().findFullDeclForForwardRef(Record.Type); + if (!full_decl) { + llvm::consumeError(full_decl.takeError()); + return llvm::ErrorSuccess(); + } + parents[*full_decl] = parent; + return llvm::ErrorSuccess(); + } + }; + + CVType field_list = m_index->tpi().getType(tag.asTag().FieldList); + ProcessTpiStream process(*m_index, *ti, m_parent_types); + llvm::Error error = visitMemberRecordStream(field_list.data(), process); + if (error) + llvm::consumeError(std::move(error)); + } +} + uint32_t SymbolFileNativePDB::GetNumCompileUnits() { const DbiModuleList &modules = m_index->dbi().modules(); uint32_t count = modules.getModuleCount(); @@ -730,16 +784,69 @@ ct, Type::eResolveStateFull); } +static std::string RenderDemanglerNode(llvm::ms_demangle::Node *n) { + OutputStream OS; + initializeOutputStream(nullptr, nullptr, OS, 1024); + n->output(OS, llvm::ms_demangle::OF_Default); + OS << '\0'; + return {OS.getBuffer()}; +} + +std::pair<clang::DeclContext *, std::string> +SymbolFileNativePDB::CreateDeclInfoForType(const TagRecord &record, + TypeIndex ti) { + llvm::ms_demangle::Demangler demangler; + StringView sv(record.UniqueName.begin(), record.UniqueName.size()); + llvm::ms_demangle::TagTypeNode *ttn = demangler.parseTagUniqueName(sv); + llvm::ms_demangle::IdentifierNode *idn = + ttn->QualifiedName->getUnqualifiedIdentifier(); + std::string uname = RenderDemanglerNode(idn); + + llvm::ms_demangle::NodeArrayNode *name_components = + ttn->QualifiedName->Components; + llvm::ArrayRef<llvm::ms_demangle::Node *> scopes(name_components->Nodes, + name_components->Count - 1); + + clang::DeclContext *context = m_clang->GetTranslationUnitDecl(); + + // If this type doesn't have a parent type in the debug info, then the best we + // can do is to say that it's either a series of namespaces (if the scope is + // non-empty), or the translation unit (if the scope is empty). + auto parent_iter = m_parent_types.find(ti); + if (parent_iter == m_parent_types.end()) { + if (scopes.empty()) + return {context, uname}; + + for (llvm::ms_demangle::Node *scope : scopes) { + auto *nii = static_cast<llvm::ms_demangle::NamedIdentifierNode *>(scope); + std::string str = RenderDemanglerNode(nii); + context = m_clang->GetUniqueNamespaceDeclaration(str.c_str(), context); + } + return {context, uname}; + } + + // Otherwise, all we need to do is get the parent type of this type and + // recurse into our lazy type creation / AST reconstruction logic to get an + // LLDB TypeSP for the parent. This will cause the AST to automatically get + // the right DeclContext created for any parent. + TypeSP parent = GetOrCreateType(parent_iter->second); + if (!parent) + return {context, uname}; + CompilerType parent_ct = parent->GetForwardCompilerType(); + clang::QualType qt = ClangUtil::GetCanonicalQualType(parent_ct); + context = clang::TagDecl::castToDeclContext(qt->getAsTagDecl()); + return {context, uname}; +} + lldb::TypeSP SymbolFileNativePDB::CreateClassStructUnion( - PdbSymUid type_uid, llvm::StringRef name, size_t size, + PdbSymUid type_uid, const llvm::codeview::TagRecord &record, size_t size, clang::TagTypeKind ttk, clang::MSInheritanceAttr::Spelling inheritance) { - // Ignore unnamed-tag UDTs. - name = DropNameScope(name); - if (name.empty()) - return nullptr; - - clang::DeclContext *decl_context = m_clang->GetTranslationUnitDecl(); + const PdbTypeSymId &tid = type_uid.asTypeSym(); + TypeIndex ti(tid.index); + clang::DeclContext *decl_context = nullptr; + std::string uname; + std::tie(decl_context, uname) = CreateDeclInfoForType(record, ti); lldb::AccessType access = (ttk == clang::TTK_Class) ? lldb::eAccessPrivate : lldb::eAccessPublic; @@ -749,8 +856,9 @@ metadata.SetIsDynamicCXXType(false); CompilerType ct = - m_clang->CreateRecordType(decl_context, access, name.str().c_str(), ttk, + m_clang->CreateRecordType(decl_context, access, uname.c_str(), ttk, lldb::eLanguageTypeC_plus_plus, &metadata); + lldbassert(ct.IsValid()); clang::CXXRecordDecl *record_decl = @@ -771,7 +879,7 @@ // FIXME: Search IPI stream for LF_UDT_MOD_SRC_LINE. Declaration decl; return std::make_shared<Type>(type_uid.toOpaqueId(), m_clang->GetSymbolFile(), - ConstString(name), size, nullptr, + ConstString(uname), size, nullptr, LLDB_INVALID_UID, Type::eEncodingIsUID, decl, ct, Type::eResolveStateForward); } @@ -782,14 +890,13 @@ clang::MSInheritanceAttr::Spelling inheritance = GetMSInheritance(m_index->tpi().typeCollection(), cr); - return CreateClassStructUnion(type_uid, cr.getName(), cr.getSize(), ttk, - inheritance); + return CreateClassStructUnion(type_uid, cr, cr.getSize(), ttk, inheritance); } lldb::TypeSP SymbolFileNativePDB::CreateTagType(PdbSymUid type_uid, const UnionRecord &ur) { return CreateClassStructUnion( - type_uid, ur.getName(), ur.getSize(), clang::TTK_Union, + type_uid, ur, ur.getSize(), clang::TTK_Union, clang::MSInheritanceAttr::Spelling::Keyword_single_inheritance); } Index: lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.h =================================================================== --- lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.h +++ lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.h @@ -13,14 +13,57 @@ #include "lldb/lldb-enumerations.h" #include "llvm/DebugInfo/CodeView/SymbolRecord.h" +#include "llvm/DebugInfo/CodeView/TypeRecord.h" #include "llvm/DebugInfo/PDB/PDBTypes.h" #include <tuple> #include <utility> namespace lldb_private { namespace npdb { +struct CVTagRecord { + enum Kind { Class, Struct, Union, Enum }; + + static CVTagRecord create(llvm::codeview::CVType type); + + Kind kind() const { return m_kind; } + + const llvm::codeview::TagRecord &asTag() const { + if (m_kind == Struct || m_kind == Class) + return cvclass; + if (m_kind == Enum) + return cvenum; + return cvunion; + } + + const llvm::codeview::ClassRecord &asClass() const { + assert(m_kind == Struct || m_kind == Class); + return cvclass; + } + + const llvm::codeview::EnumRecord &asEnum() const { + assert(m_kind == Enum); + return cvenum; + } + + const llvm::codeview::UnionRecord &asUnion() const { + assert(m_kind == Union); + return cvunion; + } + +private: + CVTagRecord(llvm::codeview::ClassRecord &&c); + CVTagRecord(llvm::codeview::UnionRecord &&u); + CVTagRecord(llvm::codeview::EnumRecord &&e); + Kind m_kind; + union { + llvm::codeview::ClassRecord cvclass; + llvm::codeview::EnumRecord cvenum; + llvm::codeview::UnionRecord cvunion; + }; +}; + struct SegmentOffset { SegmentOffset() = default; SegmentOffset(uint16_t s, uint32_t o) : segment(s), offset(o) {} @@ -56,6 +99,7 @@ } bool IsForwardRefUdt(llvm::codeview::CVType cvt); +bool IsTagRecord(llvm::codeview::CVType cvt); lldb::AccessType TranslateMemberAccess(llvm::codeview::MemberAccess access); llvm::codeview::TypeIndex GetFieldListIndex(llvm::codeview::CVType cvt); Index: lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.cpp =================================================================== --- lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.cpp +++ lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.cpp @@ -21,6 +21,36 @@ using namespace llvm::codeview; using namespace llvm::pdb; +CVTagRecord CVTagRecord::create(CVType type) { + assert(IsTagRecord(type) && "type is not a tag record!"); + switch (type.kind()) { + case LF_CLASS: + case LF_STRUCTURE: + case LF_INTERFACE: { + ClassRecord cr; + llvm::cantFail(TypeDeserializer::deserializeAs<ClassRecord>(type, cr)); + return CVTagRecord(std::move(cr)); + } + case LF_UNION: { + UnionRecord ur; + return CVTagRecord(std::move(ur)); + } + case LF_ENUM: { + EnumRecord er; + return CVTagRecord(std::move(er)); + } + default: + llvm_unreachable("Unreachable!"); + } +} + +CVTagRecord::CVTagRecord(ClassRecord &&c) + : cvclass(std::move(c)), + m_kind(cvclass.Kind == TypeRecordKind::Struct ? Struct : Class) {} +CVTagRecord::CVTagRecord(UnionRecord &&u) + : cvunion(std::move(u)), m_kind(Union) {} +CVTagRecord::CVTagRecord(EnumRecord &&e) : cvenum(std::move(e)), m_kind(Enum) {} + PDB_SymType lldb_private::npdb::CVSymToPDBSym(SymbolKind kind) { switch (kind) { case S_COMPILE3: @@ -306,6 +336,18 @@ } } +bool lldb_private::npdb::IsTagRecord(llvm::codeview::CVType cvt) { + switch (cvt.kind()) { + case LF_CLASS: + case LF_STRUCTURE: + case LF_UNION: + case LF_ENUM: + return true; + default: + return false; + } +} + lldb::AccessType lldb_private::npdb::TranslateMemberAccess(MemberAccess access) { switch (access) { Index: lldb/lit/SymbolFile/NativePDB/global-classes.cpp =================================================================== --- lldb/lit/SymbolFile/NativePDB/global-classes.cpp +++ lldb/lit/SymbolFile/NativePDB/global-classes.cpp @@ -270,6 +270,94 @@ // CHECK-NEXT: (long long &) m = {{.*}} // CHECK-NEXT: } +// CHECK: Dumping clang ast for 1 modules. +// CHECK: TranslationUnitDecl {{.*}} +// CHECK: |-CXXRecordDecl {{.*}} class ClassWithPadding definition +// CHECK: | |-FieldDecl {{.*}} a 'char' +// CHECK: | |-FieldDecl {{.*}} b 'short' +// CHECK: | |-FieldDecl {{.*}} c 'char [2]' +// CHECK: | |-FieldDecl {{.*}} d 'int' +// CHECK: | |-FieldDecl {{.*}} e 'char' +// CHECK: | |-FieldDecl {{.*}} f 'int' +// CHECK: | |-FieldDecl {{.*}} g 'long long' +// CHECK: | |-FieldDecl {{.*}} h 'char [3]' +// CHECK: | |-FieldDecl {{.*}} i 'long long' +// CHECK: | |-FieldDecl {{.*}} j 'char [2]' +// CHECK: | |-FieldDecl {{.*}} k 'long long' +// CHECK: | |-FieldDecl {{.*}} l 'char' +// CHECK: | `-FieldDecl {{.*}} m 'long long' +// CHECK: |-CXXRecordDecl {{.*}} class ClassNoPadding definition +// CHECK: | |-FieldDecl {{.*}} a 'unsigned char' +// CHECK: | |-FieldDecl {{.*}} b 'char' +// CHECK: | |-FieldDecl {{.*}} c 'bool' +// CHECK: | |-FieldDecl {{.*}} d 'bool' +// CHECK: | |-FieldDecl {{.*}} e 'short' +// CHECK: | |-FieldDecl {{.*}} f 'unsigned short' +// CHECK: | |-FieldDecl {{.*}} g 'unsigned int' +// CHECK: | |-FieldDecl {{.*}} h 'int' +// CHECK: | |-FieldDecl {{.*}} i 'unsigned long' +// CHECK: | |-FieldDecl {{.*}} j 'long' +// CHECK: | |-FieldDecl {{.*}} k 'float' +// CHECK: | |-FieldDecl {{.*}} l 'EnumType' +// CHECK: | |-FieldDecl {{.*}} m 'double' +// CHECK: | |-FieldDecl {{.*}} n 'unsigned long long' +// CHECK: | |-FieldDecl {{.*}} o 'long long' +// CHECK: | `-FieldDecl {{.*}} p 'int [5]' +// CHECK: |-EnumDecl {{.*}} EnumType +// CHECK: | |-EnumConstantDecl {{.*}} A 'unsigned int' +// CHECK: | `-EnumConstantDecl {{.*}} B 'unsigned int' +// CHECK: |-CXXRecordDecl {{.*}} struct DerivedClass definition +// CHECK: | |-public 'BaseClass<int>' +// CHECK: | `-FieldDecl {{.*}} DerivedMember 'int' +// CHECK: |-CXXRecordDecl {{.*}} struct BaseClass<int> definition +// CHECK: | `-FieldDecl {{.*}} BaseMember 'int' +// CHECK: |-CXXRecordDecl {{.*}} struct EBO definition +// CHECK: | |-public 'EmptyBase' +// CHECK: | `-FieldDecl {{.*}} Member 'int' +// CHECK: |-CXXRecordDecl {{.*}} struct EmptyBase definition +// CHECK: |-CXXRecordDecl {{.*}} struct PaddedBases definition +// CHECK: | |-public 'BaseClass<char>' +// CHECK: | |-public 'BaseClass<short>' +// CHECK: | |-public 'BaseClass<int>' +// CHECK: | `-FieldDecl {{.*}} DerivedMember 'long long' +// CHECK: |-CXXRecordDecl {{.*}} struct BaseClass<char> definition +// CHECK: | `-FieldDecl {{.*}} BaseMember 'int' +// CHECK: |-CXXRecordDecl {{.*}} struct BaseClass<short> definition +// CHECK: | `-FieldDecl {{.*}} BaseMember 'int' +// CHECK: |-CXXRecordDecl {{.*}} struct <unnamed-type-UnnamedClassInstance> definition +// CHECK: | |-FieldDecl {{.*}} x 'int' +// CHECK: | `-FieldDecl {{.*}} EBOC 'EBO' +// CHECK: |-CXXRecordDecl {{.*}} struct Pointers definition +// CHECK: | |-FieldDecl {{.*}} a 'void *' +// CHECK: | |-FieldDecl {{.*}} b 'char *' +// CHECK: | |-FieldDecl {{.*}} c 'bool *' +// CHECK: | |-FieldDecl {{.*}} e 'short *' +// CHECK: | |-FieldDecl {{.*}} f 'unsigned short *' +// CHECK: | |-FieldDecl {{.*}} g 'unsigned int *' +// CHECK: | |-FieldDecl {{.*}} h 'int *' +// CHECK: | |-FieldDecl {{.*}} i 'unsigned long *' +// CHECK: | |-FieldDecl {{.*}} j 'long *' +// CHECK: | |-FieldDecl {{.*}} k 'float *' +// CHECK: | |-FieldDecl {{.*}} l 'EnumType *' +// CHECK: | |-FieldDecl {{.*}} m 'double *' +// CHECK: | |-FieldDecl {{.*}} n 'unsigned long long *' +// CHECK: | `-FieldDecl {{.*}} o 'long long *' +// CHECK: |-CXXRecordDecl {{.*}} struct References definition +// CHECK: | |-FieldDecl {{.*}} a 'char &' +// CHECK: | |-FieldDecl {{.*}} b 'bool &' +// CHECK: | |-FieldDecl {{.*}} c 'short &' +// CHECK: | |-FieldDecl {{.*}} d 'unsigned short &' +// CHECK: | |-FieldDecl {{.*}} e 'unsigned int &' +// CHECK: | |-FieldDecl {{.*}} f 'int &' +// CHECK: | |-FieldDecl {{.*}} g 'unsigned long &' +// CHECK: | |-FieldDecl {{.*}} h 'long &' +// CHECK: | |-FieldDecl {{.*}} i 'float &' +// CHECK: | |-FieldDecl {{.*}} j 'EnumType &' +// CHECK: | |-FieldDecl {{.*}} k 'double &' +// CHECK: | |-FieldDecl {{.*}} l 'unsigned long long &' +// CHECK: | `-FieldDecl {{.*}} m 'long long &' +// CHECK: `-<undeserialized declarations> + int main(int argc, char **argv) { return 0; } \ No newline at end of file Index: lldb/lit/SymbolFile/NativePDB/function-types-classes.cpp =================================================================== --- lldb/lit/SymbolFile/NativePDB/function-types-classes.cpp +++ lldb/lit/SymbolFile/NativePDB/function-types-classes.cpp @@ -2,7 +2,7 @@ // REQUIRES: lld // Test that we can display function signatures with class types. -// RUN: clang-cl /Z7 /GS- /GR- /c -Xclang -fkeep-static-consts /Fo%t.obj -- %s +// RUN: clang-cl /Z7 /GS- /GR- /c -fstandalone-debug -Xclang -fkeep-static-consts /Fo%t.obj -- %s // RUN: lld-link /DEBUG /nodefaultlib /entry:main /OUT:%t.exe /PDB:%t.pdb -- %t.obj // RUN: env LLDB_USE_NATIVE_PDB_READER=1 lldb -f %t.exe -s \ // RUN: %p/Inputs/function-types-classes.lldbinit | FileCheck %s @@ -66,6 +66,13 @@ }; }; +// clang (incorrectly) doesn't emit debug information for outer classes +// unless they are instantiated. They should also be emitted if there +// is an inner class which is instantiated. +A::C ForceInstantiateAC; +B ForceInstantiateB; +B::A ForceInstantiateBA; + template<typename T> struct TC {}; @@ -81,38 +88,45 @@ // classes nested in namespaces and inner classes -// FIXME: LLDB with native pdb plugin doesn't currently resolve nested names -// correctly, because it requires creating clang::NamespaceDecl or -// clang::RecordDecl for the outer namespace or classes. PDB doesn't contain -// sufficient information to distinguish namespace scopes from nested class -// scopes, so the best we can hope for is a heuristic reconstruction of the -// clang AST based on demangling the type's unique name. However, this is -// as-yet unimplemented in the native PDB plugin, so for now all of these will -// all just look like `S` when LLDB prints them. auto e = &three<A::B::S*, B::A::S*, A::C::S&>; -// CHECK: (S *(*)(S *, S &)) e = {{.*}} +// CHECK: (A::B::S *(*)(B::A::S *, A::C::S &)) e = {{.*}} auto f = &three<A::C::S&, A::B::S*, B::A::S*>; -// CHECK: (S &(*)(S *, S *)) f = {{.*}} +// CHECK: (A::C::S &(*)(A::B::S *, B::A::S *)) f = {{.*}} auto g = &three<B::A::S*, A::C::S&, A::B::S*>; -// CHECK: (S *(*)(S &, S *)) g = {{.*}} +// CHECK: (B::A::S *(*)(A::C::S &, A::B::S *)) g = {{.*}} // parameter types that are themselves template instantiations. auto h = &four<TC<void>, TC<int>, TC<TC<int>>, TC<A::B::S>>; -// Note the awkward space in TC<TC<int> >. This is because this is how template -// instantiations are emitted by the compiler, as the fully instantiated name. -// Only via reconstruction of the AST through the mangled type name (see above -// comment) can we hope to do better than this). -// CHECK: (TC<void> (*)(TC<int>, TC<TC<int> >, S>)) h = {{.*}} +// CHECK: (TC<void> (*)(TC<int>, TC<struct TC<int>>, TC<struct A::B::S>)) h = {{.*}} auto i = &nullary<A::B::S>; -// CHECK: (S (*)()) i = {{.*}} +// CHECK: (A::B::S (*)()) i = {{.*}} // Make sure we can handle types that don't have complete debug info. struct Incomplete; auto incomplete = &three<Incomplete*, Incomplete**, const Incomplete*>; // CHECK: (Incomplete *(*)(Incomplete **, const Incomplete *)) incomplete = {{.*}} +// CHECK: TranslationUnitDecl {{.*}} +// CHECK: |-CXXRecordDecl {{.*}} class C definition +// CHECK: |-CXXRecordDecl {{.*}} union U definition +// CHECK: |-EnumDecl {{.*}} E +// CHECK: |-CXXRecordDecl {{.*}} struct S definition +// CHECK: |-CXXRecordDecl {{.*}} struct B +// CHECK: | |-CXXRecordDecl {{.*}} struct A +// CHECK: | | |-CXXRecordDecl {{.*}} struct S +// CHECK: |-NamespaceDecl {{.*}} A +// CHECK: | |-CXXRecordDecl {{.*}} struct C +// CHECK: | | |-CXXRecordDecl {{.*}} struct S +// CHECK: | `-NamespaceDecl {{.*}} B +// CHECK: | `-CXXRecordDecl {{.*}} struct S definition +// CHECK: |-CXXRecordDecl {{.*}} struct TC<int> definition +// CHECK: |-CXXRecordDecl {{.*}} struct TC<struct TC<int>> definition +// CHECK: |-CXXRecordDecl {{.*}} struct TC<struct A::B::S> definition +// CHECK: |-CXXRecordDecl {{.*}} struct TC<void> definition +// CHECK: |-CXXRecordDecl {{.*}} struct Incomplete + int main(int argc, char **argv) { return 0; } Index: lldb/lit/SymbolFile/NativePDB/ast-reconstruction.cpp =================================================================== --- /dev/null +++ lldb/lit/SymbolFile/NativePDB/ast-reconstruction.cpp @@ -0,0 +1,131 @@ +// clang-format off +// REQUIRES: lld + +// Test various interesting cases for AST reconstruction. +// RUN: clang-cl /Z7 /GS- /GR- /c /Fo%t.obj -- %s +// RUN: lld-link /DEBUG /nodefaultlib /entry:main /OUT:%t.exe /PDB:%t.pdb -- %t.obj +// RUN: env LLDB_USE_NATIVE_PDB_READER=1 lldb -f %t.exe -s \ +// RUN: %p/Inputs/ast-reconstruction.lldbinit 2>&1 | FileCheck %s + +// Test trivial versions of each tag type. +class TrivialC {}; +struct TrivialS {}; +union TrivialU {}; +enum TrivialE {TE_A}; + +// Test reconstruction of DeclContext hierarchies. +namespace A { + namespace B { + template<typename T> + struct C { + T ABCMember; + }; + + // Let's try a template specialization with a different implementation + template<> + struct C<void> { + void *ABCSpecializationMember; + }; + } + + // Let's make sure we can distinguish classes and namespaces. Also let's try + // a non-type template parameter. + template<int N> + struct C { + class D { + int ACDMember = 0; + C<N - 1> *CPtr = nullptr; + }; + }; + + struct D { + // Let's make a nested class with the same name as another nested class + // elsewhere, and confirm that they appear in the right DeclContexts in + // the AST. + struct E { + int ADDMember; + }; + }; +} + + +// Let's try an anonymous namespace. +namespace { + template<typename T> + struct Anonymous { + int AnonymousMember; + // And a nested class within an anonymous namespace + struct D { + int AnonymousDMember; + }; + }; +} + +TrivialC TC; +TrivialS TS; +TrivialU TU; +TrivialE TE; + +A::B::C<int> ABCInt; +A::B::C<float> ABCFloat; +A::B::C<void> ABCVoid; + +A::C<0> AC0; +A::C<-1> ACNeg1; + +A::C<0>::D AC0D; +A::C<-1>::D ACNeg1D; +A::D AD; +A::D::E ADE; + +// FIXME: Anonymous namespaces aren't working correctly. +Anonymous<int> AnonInt; +Anonymous<A::B::C<void>> AnonABCVoid; +Anonymous<A::B::C<int>>::D AnonABCVoidD; + +// FIXME: Enum size isn't being correctly determined. +// FIXME: Can't read memory for variable values. + +// CHECK: (TrivialC) TC = {} +// CHECK: (TrivialS) TS = {} +// CHECK: (TrivialU) TU = {} +// CHECK: (TrivialE) TE = <Unable to determine byte size.> +// CHECK: (A::B::C<int>) ABCInt = (ABCMember = <read memory from {{.*}} failed>) +// CHECK: (A::B::C<float>) ABCFloat = (ABCMember = <read memory from {{.*}} failed>) +// CHECK: (A::B::C<void>) ABCVoid = (ABCSpecializationMember = <read memory from {{.*}} failed>) +// CHECK: (A::C<0>) AC0 = {} +// CHECK: (A::C<-1>) ACNeg1 = {} +// CHECK: (A::C<0>::D) AC0D = (ACDMember = <read memory from {{.*}} failed>, CPtr = <read memory from {{.*}} failed>) +// CHECK: (A::C<-1>::D) ACNeg1D = (ACDMember = <read memory from {{.*}} failed>, CPtr = <read memory from {{.*}} failed>) +// CHECK: (A::D) AD = {} +// CHECK: (A::D::E) ADE = (ADDMember = <read memory from {{.*}} failed>) +// CHECK: Dumping clang ast for 1 modules. +// CHECK: TranslationUnitDecl {{.*}} +// CHECK: |-CXXRecordDecl {{.*}} class TrivialC definition +// CHECK: |-CXXRecordDecl {{.*}} struct TrivialS definition +// CHECK: |-CXXRecordDecl {{.*}} union TrivialU definition +// CHECK: |-EnumDecl {{.*}} TrivialE +// CHECK: |-NamespaceDecl {{.*}} A +// CHECK: | |-NamespaceDecl {{.*}} B +// CHECK: | | |-CXXRecordDecl {{.*}} struct C<int> definition +// CHECK: | | | `-FieldDecl {{.*}} ABCMember 'int' +// CHECK: | | |-CXXRecordDecl {{.*}} struct C<float> definition +// CHECK: | | | `-FieldDecl {{.*}} ABCMember 'float' +// CHECK: | | `-CXXRecordDecl {{.*}} struct C<void> definition +// CHECK: | | `-FieldDecl {{.*}} ABCSpecializationMember 'void *' +// CHECK: | |-CXXRecordDecl {{.*}} struct C<0> definition +// CHECK: | | `-CXXRecordDecl {{.*}} class D definition +// CHECK: | | |-FieldDecl {{.*}} ACDMember 'int' +// CHECK: | | `-FieldDecl {{.*}} CPtr 'A::C<-1> *' +// CHECK: | |-CXXRecordDecl {{.*}} struct C<-1> definition +// CHECK: | | `-CXXRecordDecl {{.*}} class D definition +// CHECK: | | |-FieldDecl {{.*}} ACDMember 'int' +// CHECK: | | `-FieldDecl {{.*}} CPtr 'A::C<-2> *' +// CHECK: | |-CXXRecordDecl {{.*}} struct C<-2> +// CHECK: | `-CXXRecordDecl {{.*}} struct D definition +// CHECK: | `-CXXRecordDecl {{.*}} struct E definition +// CHECK: | `-FieldDecl {{.*}} ADDMember 'int' + +int main(int argc, char **argv) { + return 0; +} \ No newline at end of file Index: lldb/lit/SymbolFile/NativePDB/Inputs/globals-classes.lldbinit =================================================================== --- lldb/lit/SymbolFile/NativePDB/Inputs/globals-classes.lldbinit +++ lldb/lit/SymbolFile/NativePDB/Inputs/globals-classes.lldbinit @@ -11,4 +11,6 @@ target variable -T PointersInstance target variable -T ReferencesInstance +target modules dump ast + quit \ No newline at end of file Index: lldb/lit/SymbolFile/NativePDB/Inputs/function-types-classes.lldbinit =================================================================== --- lldb/lit/SymbolFile/NativePDB/Inputs/function-types-classes.lldbinit +++ lldb/lit/SymbolFile/NativePDB/Inputs/function-types-classes.lldbinit @@ -9,4 +9,6 @@ target variable i target variable incomplete +target modules dump ast + quit Index: lldb/lit/SymbolFile/NativePDB/Inputs/ast-reconstruction.lldbinit =================================================================== --- /dev/null +++ lldb/lit/SymbolFile/NativePDB/Inputs/ast-reconstruction.lldbinit @@ -0,0 +1,20 @@ +target variable TC +target variable TS +target variable TU +target variable TE + +target variable ABCInt +target variable ABCFloat +target variable ABCVoid + +target variable AC0 +target variable ACNeg1 + +target variable AC0D +target variable ACNeg1D +target variable AD +target variable ADE + +target modules dump ast + +quit
_______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits