1 //===-- PDBASTParser.cpp ----------------------------------------*- 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 #include "PDBASTParser.h"
11 
12 #include "clang/AST/CharUnits.h"
13 #include "clang/AST/Decl.h"
14 #include "clang/AST/DeclCXX.h"
15 
16 #include "lldb/Symbol/ClangASTContext.h"
17 #include "lldb/Symbol/ClangUtil.h"
18 #include "lldb/Symbol/Declaration.h"
19 #include "lldb/Symbol/SymbolFile.h"
20 #include "lldb/Symbol/TypeSystem.h"
21 
22 #include "llvm/DebugInfo/PDB/IPDBLineNumber.h"
23 #include "llvm/DebugInfo/PDB/IPDBSourceFile.h"
24 #include "llvm/DebugInfo/PDB/PDBSymbol.h"
25 #include "llvm/DebugInfo/PDB/PDBSymbolData.h"
26 #include "llvm/DebugInfo/PDB/PDBSymbolFunc.h"
27 #include "llvm/DebugInfo/PDB/PDBSymbolTypeArray.h"
28 #include "llvm/DebugInfo/PDB/PDBSymbolTypeBuiltin.h"
29 #include "llvm/DebugInfo/PDB/PDBSymbolTypeEnum.h"
30 #include "llvm/DebugInfo/PDB/PDBSymbolTypeFunctionArg.h"
31 #include "llvm/DebugInfo/PDB/PDBSymbolTypeFunctionSig.h"
32 #include "llvm/DebugInfo/PDB/PDBSymbolTypePointer.h"
33 #include "llvm/DebugInfo/PDB/PDBSymbolTypeTypedef.h"
34 #include "llvm/DebugInfo/PDB/PDBSymbolTypeUDT.h"
35 
36 using namespace lldb;
37 using namespace lldb_private;
38 using namespace llvm::pdb;
39 
40 namespace {
41 int TranslateUdtKind(PDB_UdtType pdb_kind) {
42   switch (pdb_kind) {
43   case PDB_UdtType::Class:
44     return clang::TTK_Class;
45   case PDB_UdtType::Struct:
46     return clang::TTK_Struct;
47   case PDB_UdtType::Union:
48     return clang::TTK_Union;
49   case PDB_UdtType::Interface:
50     return clang::TTK_Interface;
51   }
52   return -1;
53 }
54 
55 lldb::Encoding TranslateBuiltinEncoding(PDB_BuiltinType type) {
56   switch (type) {
57   case PDB_BuiltinType::Float:
58     return lldb::eEncodingIEEE754;
59   case PDB_BuiltinType::Int:
60   case PDB_BuiltinType::Long:
61   case PDB_BuiltinType::Char:
62     return lldb::eEncodingSint;
63   case PDB_BuiltinType::Bool:
64   case PDB_BuiltinType::UInt:
65   case PDB_BuiltinType::ULong:
66   case PDB_BuiltinType::HResult:
67     return lldb::eEncodingUint;
68   default:
69     return lldb::eEncodingInvalid;
70   }
71 }
72 
73 lldb::Encoding TranslateEnumEncoding(PDB_VariantType type) {
74   switch (type) {
75   case PDB_VariantType::Int8:
76   case PDB_VariantType::Int16:
77   case PDB_VariantType::Int32:
78   case PDB_VariantType::Int64:
79     return lldb::eEncodingSint;
80 
81   case PDB_VariantType::UInt8:
82   case PDB_VariantType::UInt16:
83   case PDB_VariantType::UInt32:
84   case PDB_VariantType::UInt64:
85     return lldb::eEncodingUint;
86 
87   default:
88     break;
89   }
90 
91   return lldb::eEncodingSint;
92 }
93 
94 CompilerType GetBuiltinTypeForPDBEncodingAndBitSize(
95     ClangASTContext *clang_ast, const PDBSymbolTypeBuiltin *pdb_type,
96     Encoding encoding, uint32_t width) {
97   if (!pdb_type)
98     return CompilerType();
99   if (!clang_ast)
100     return CompilerType();
101   auto *ast = clang_ast->getASTContext();
102   if (!ast)
103     return CompilerType();
104 
105   switch (pdb_type->getBuiltinType()) {
106   default: break;
107   case PDB_BuiltinType::None:
108     return CompilerType();
109   case PDB_BuiltinType::Void:
110     return clang_ast->GetBasicType(eBasicTypeVoid);
111   case PDB_BuiltinType::Bool:
112     return clang_ast->GetBasicType(eBasicTypeBool);
113   case PDB_BuiltinType::Long:
114     if (width == ast->getTypeSize(ast->LongTy))
115       return CompilerType(ast, ast->LongTy);
116     if (width == ast->getTypeSize(ast->LongLongTy))
117       return CompilerType(ast, ast->LongLongTy);
118     break;
119   case PDB_BuiltinType::ULong:
120     if (width == ast->getTypeSize(ast->UnsignedLongTy))
121       return CompilerType(ast, ast->UnsignedLongTy);
122     if (width == ast->getTypeSize(ast->UnsignedLongLongTy))
123       return CompilerType(ast, ast->UnsignedLongLongTy);
124     break;
125   case PDB_BuiltinType::WCharT:
126     if (width == ast->getTypeSize(ast->WCharTy))
127       return CompilerType(ast, ast->WCharTy);
128     break;
129   case PDB_BuiltinType::Float:
130     // Note: types `long double` and `double` have same bit size in MSVC and there
131     // is no information in the PDB to distinguish them. So when falling back
132     // to default search, the compiler type of `long double` will be represented by
133     // the one generated for `double`.
134     break;
135   }
136   // If there is no match on PDB_BuiltinType, fall back to default search
137   // by encoding and width only
138   return clang_ast->GetBuiltinTypeForEncodingAndBitSize(encoding, width);
139 }
140 
141 ConstString GetPDBBuiltinTypeName(const PDBSymbolTypeBuiltin *pdb_type,
142                                   CompilerType &compiler_type) {
143   if (!pdb_type)
144     return compiler_type.GetTypeName();
145 
146   PDB_BuiltinType kind = pdb_type->getBuiltinType();
147   switch (kind) {
148   default: break;
149   case PDB_BuiltinType::Currency:
150     return ConstString("CURRENCY");
151   case PDB_BuiltinType::Date:
152     return ConstString("DATE");
153   case PDB_BuiltinType::Variant:
154     return ConstString("VARIANT");
155   case PDB_BuiltinType::Complex:
156     return ConstString("complex");
157   case PDB_BuiltinType::Bitfield:
158     return ConstString("bitfield");
159   case PDB_BuiltinType::BSTR:
160     return ConstString("BSTR");
161   case PDB_BuiltinType::HResult:
162     return ConstString("HRESULT");
163   case PDB_BuiltinType::BCD:
164     return ConstString("BCD");
165   case PDB_BuiltinType::None:
166     return ConstString("...");
167   }
168   return compiler_type.GetTypeName();
169 }
170 
171 bool GetDeclarationForSymbol(const PDBSymbol &symbol, Declaration &decl) {
172   auto &raw_sym = symbol.getRawSymbol();
173   auto lines_up = symbol.getSession().findLineNumbersByAddress(
174       raw_sym.getVirtualAddress(), raw_sym.getLength());
175   if (!lines_up)
176     return false;
177   auto first_line_up = lines_up->getNext();
178   if (!first_line_up)
179     return false;
180 
181   uint32_t src_file_id = first_line_up->getSourceFileId();
182   auto src_file_up = symbol.getSession().getSourceFileById(src_file_id);
183   if (!src_file_up)
184     return false;
185 
186   FileSpec spec(src_file_up->getFileName(), /*resolve_path*/false);
187   decl.SetFile(spec);
188   decl.SetColumn(first_line_up->getColumnNumber());
189   decl.SetLine(first_line_up->getLineNumber());
190   return true;
191 }
192 }
193 
194 PDBASTParser::PDBASTParser(lldb_private::ClangASTContext &ast) : m_ast(ast) {}
195 
196 PDBASTParser::~PDBASTParser() {}
197 
198 // DebugInfoASTParser interface
199 
200 lldb::TypeSP PDBASTParser::CreateLLDBTypeFromPDBType(const PDBSymbol &type) {
201   // PDB doesn't maintain enough information to robustly rebuild the entire
202   // tree, and this is most problematic when it comes to figure out the
203   // right DeclContext to put a type in.  So for now, everything goes in
204   // the translation unit decl as a fully qualified type.
205   clang::DeclContext *tu_decl_ctx = m_ast.GetTranslationUnitDecl();
206   Declaration decl;
207 
208   switch (type.getSymTag()) {
209   case PDB_SymType::UDT: {
210     auto udt = llvm::dyn_cast<PDBSymbolTypeUDT>(&type);
211     assert(udt);
212     AccessType access = lldb::eAccessPublic;
213     PDB_UdtType udt_kind = udt->getUdtKind();
214     auto tag_type_kind = TranslateUdtKind(udt_kind);
215     if (tag_type_kind == -1)
216       return nullptr;
217 
218     if (udt_kind == PDB_UdtType::Class)
219       access = lldb::eAccessPrivate;
220 
221     CompilerType clang_type = m_ast.CreateRecordType(
222         tu_decl_ctx, access, udt->getName().c_str(), tag_type_kind,
223         lldb::eLanguageTypeC_plus_plus, nullptr);
224 
225     m_ast.SetHasExternalStorage(clang_type.GetOpaqueQualType(), true);
226 
227     return std::make_shared<lldb_private::Type>(
228         type.getSymIndexId(), m_ast.GetSymbolFile(),
229         ConstString(udt->getName()), udt->getLength(), nullptr,
230         LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl, clang_type,
231         lldb_private::Type::eResolveStateForward);
232   } break;
233   case PDB_SymType::Enum: {
234     auto enum_type = llvm::dyn_cast<PDBSymbolTypeEnum>(&type);
235     assert(enum_type);
236     auto underlying_type_up = enum_type->getUnderlyingType();
237     if (!underlying_type_up)
238       return nullptr;
239     lldb::Encoding encoding =
240         TranslateBuiltinEncoding(underlying_type_up->getBuiltinType());
241     // FIXME: Type of underlying builtin is always `Int`. We correct it with
242     // the very first enumerator's encoding if any.
243     auto first_child = enum_type->findOneChild<PDBSymbolData>();
244     if (first_child) {
245       encoding = TranslateEnumEncoding(first_child->getValue().Type);
246     }
247     std::string name = enum_type->getName();
248     uint64_t bytes = enum_type->getLength();
249     CompilerType builtin_type;
250     if (bytes > 0)
251       builtin_type = GetBuiltinTypeForPDBEncodingAndBitSize(
252            &m_ast, underlying_type_up.get(), encoding, bytes * 8);
253     else
254       builtin_type = m_ast.GetBasicType(eBasicTypeInt);
255     // FIXME: PDB does not have information about scoped enumeration (Enum Class).
256     // Set it false for now.
257     bool isScoped = false;
258 
259     CompilerType ast_enum = m_ast.CreateEnumerationType(
260         name.c_str(), tu_decl_ctx, decl, builtin_type, isScoped);
261     auto enum_values = enum_type->findAllChildren<PDBSymbolData>();
262     if (enum_values) {
263       while (auto enum_value = enum_values->getNext()) {
264         if (enum_value->getDataKind() != PDB_DataKind::Constant)
265           continue;
266         AddEnumValue(ast_enum, *enum_value);
267       }
268     }
269     if (ClangASTContext::StartTagDeclarationDefinition(ast_enum))
270       ClangASTContext::CompleteTagDeclarationDefinition(ast_enum);
271 
272     return std::make_shared<lldb_private::Type>(
273         type.getSymIndexId(), m_ast.GetSymbolFile(), ConstString(name), bytes,
274         nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl,
275         ast_enum, lldb_private::Type::eResolveStateFull);
276   } break;
277   case PDB_SymType::Typedef: {
278     auto type_def = llvm::dyn_cast<PDBSymbolTypeTypedef>(&type);
279     assert(type_def);
280     lldb_private::Type *target_type =
281         m_ast.GetSymbolFile()->ResolveTypeUID(type_def->getTypeId());
282     if (!target_type)
283       return nullptr;
284     std::string name = type_def->getName();
285     uint64_t bytes = type_def->getLength();
286     CompilerType target_ast_type = target_type->GetFullCompilerType();
287     CompilerDeclContext target_decl_ctx =
288         m_ast.GetSymbolFile()->GetDeclContextForUID(target_type->GetID());
289     CompilerType ast_typedef =
290         m_ast.CreateTypedefType(target_ast_type, name.c_str(), target_decl_ctx);
291     return std::make_shared<lldb_private::Type>(
292         type_def->getSymIndexId(), m_ast.GetSymbolFile(), ConstString(name),
293         bytes, nullptr, target_type->GetID(),
294         lldb_private::Type::eEncodingIsTypedefUID, decl, ast_typedef,
295         lldb_private::Type::eResolveStateFull);
296   } break;
297   case PDB_SymType::Function:
298   case PDB_SymType::FunctionSig: {
299     std::string name;
300     PDBSymbolTypeFunctionSig *func_sig = nullptr;
301     if (auto pdb_func = llvm::dyn_cast<PDBSymbolFunc>(&type)) {
302       auto sig = pdb_func->getSignature();
303       if (!sig)
304         return nullptr;
305       func_sig = sig.release();
306       // Function type is named.
307       name = pdb_func->getName();
308     } else if (auto pdb_func_sig =
309               llvm::dyn_cast<PDBSymbolTypeFunctionSig>(&type)) {
310       func_sig = const_cast<PDBSymbolTypeFunctionSig*>(pdb_func_sig);
311     } else
312       llvm_unreachable("Unexpected PDB symbol!");
313 
314     auto arg_enum = func_sig->getArguments();
315     uint32_t num_args = arg_enum->getChildCount();
316     std::vector<CompilerType> arg_list;
317 
318     bool is_variadic = func_sig->isCVarArgs();
319     // Drop last variadic argument.
320     if (is_variadic)
321       --num_args;
322     for (uint32_t arg_idx = 0; arg_idx < num_args; arg_idx++) {
323       auto arg = arg_enum->getChildAtIndex(arg_idx);
324       if (!arg)
325         break;
326       lldb_private::Type *arg_type =
327           m_ast.GetSymbolFile()->ResolveTypeUID(arg->getSymIndexId());
328       // If there's some error looking up one of the dependent types of this
329       // function signature, bail.
330       if (!arg_type)
331         return nullptr;
332       CompilerType arg_ast_type = arg_type->GetFullCompilerType();
333       arg_list.push_back(arg_ast_type);
334     }
335     lldbassert(arg_list.size() <= num_args);
336 
337     auto pdb_return_type = func_sig->getReturnType();
338     lldb_private::Type *return_type =
339         m_ast.GetSymbolFile()->ResolveTypeUID(pdb_return_type->getSymIndexId());
340     // If there's some error looking up one of the dependent types of this
341     // function signature, bail.
342     if (!return_type)
343       return nullptr;
344     CompilerType return_ast_type = return_type->GetFullCompilerType();
345     uint32_t type_quals = 0;
346     if (func_sig->isConstType())
347       type_quals |= clang::Qualifiers::Const;
348     if (func_sig->isVolatileType())
349       type_quals |= clang::Qualifiers::Volatile;
350     CompilerType func_sig_ast_type = m_ast.CreateFunctionType(
351         return_ast_type, arg_list.data(), arg_list.size(), is_variadic,
352         type_quals);
353 
354     GetDeclarationForSymbol(type, decl);
355     return std::make_shared<lldb_private::Type>(
356         type.getSymIndexId(), m_ast.GetSymbolFile(), ConstString(name), 0,
357         nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl,
358         func_sig_ast_type, lldb_private::Type::eResolveStateFull);
359   } break;
360   case PDB_SymType::ArrayType: {
361     auto array_type = llvm::dyn_cast<PDBSymbolTypeArray>(&type);
362     assert(array_type);
363     uint32_t num_elements = array_type->getCount();
364     uint32_t element_uid = array_type->getElementType()->getSymIndexId();
365     uint32_t bytes = array_type->getLength();
366 
367     lldb_private::Type *element_type =
368         m_ast.GetSymbolFile()->ResolveTypeUID(element_uid);
369     if (!element_type)
370       return nullptr;
371     CompilerType element_ast_type = element_type->GetFullCompilerType();
372     CompilerType array_ast_type =
373         m_ast.CreateArrayType(element_ast_type, num_elements, false);
374     return std::make_shared<lldb_private::Type>(
375         array_type->getSymIndexId(), m_ast.GetSymbolFile(), ConstString(),
376         bytes, nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID,
377         decl, array_ast_type, lldb_private::Type::eResolveStateFull);
378   } break;
379   case PDB_SymType::BuiltinType: {
380     auto *builtin_type = llvm::dyn_cast<PDBSymbolTypeBuiltin>(&type);
381     assert(builtin_type);
382     PDB_BuiltinType builtin_kind = builtin_type->getBuiltinType();
383     if (builtin_kind == PDB_BuiltinType::None)
384       return nullptr;
385 
386     uint64_t bytes = builtin_type->getLength();
387     Encoding encoding = TranslateBuiltinEncoding(builtin_kind);
388     CompilerType builtin_ast_type = GetBuiltinTypeForPDBEncodingAndBitSize(
389         &m_ast, builtin_type, encoding, bytes * 8);
390 
391     Type::EncodingDataType encoding_data_type = Type::eEncodingIsUID;
392     if (builtin_type->isConstType()) {
393       encoding_data_type = Type::eEncodingIsConstUID;
394       builtin_ast_type = builtin_ast_type.AddConstModifier();
395     }
396     if (builtin_type->isVolatileType()) {
397       encoding_data_type = Type::eEncodingIsVolatileUID;
398       builtin_ast_type = builtin_ast_type.AddVolatileModifier();
399     }
400     auto type_name = GetPDBBuiltinTypeName(builtin_type, builtin_ast_type);
401 
402     return std::make_shared<lldb_private::Type>(
403         builtin_type->getSymIndexId(), m_ast.GetSymbolFile(), type_name,
404         bytes, nullptr, LLDB_INVALID_UID, encoding_data_type,
405         decl, builtin_ast_type, lldb_private::Type::eResolveStateFull);
406   } break;
407   case PDB_SymType::PointerType: {
408     auto *pointer_type = llvm::dyn_cast<PDBSymbolTypePointer>(&type);
409     assert(pointer_type);
410     Type *pointee_type = m_ast.GetSymbolFile()->ResolveTypeUID(
411         pointer_type->getPointeeType()->getSymIndexId());
412     if (!pointee_type)
413       return nullptr;
414 
415     CompilerType pointer_ast_type;
416     Type::EncodingDataType encoding_data_type = Type::eEncodingIsPointerUID;
417     if (pointer_type->isReference()) {
418       encoding_data_type = Type::eEncodingIsLValueReferenceUID;
419       pointer_ast_type =
420           pointee_type->GetFullCompilerType().GetLValueReferenceType();
421     } else
422       pointer_ast_type = pointee_type->GetFullCompilerType().GetPointerType();
423 
424     return std::make_shared<lldb_private::Type>(
425         pointer_type->getSymIndexId(), m_ast.GetSymbolFile(), ConstString(),
426         pointer_type->getLength(), nullptr, LLDB_INVALID_UID,
427         encoding_data_type, decl, pointer_ast_type,
428         lldb_private::Type::eResolveStateFull);
429   } break;
430   default: break;
431   }
432   return nullptr;
433 }
434 
435 bool PDBASTParser::AddEnumValue(CompilerType enum_type,
436                                 const PDBSymbolData &enum_value) const {
437   Declaration decl;
438   Variant v = enum_value.getValue();
439   std::string name = enum_value.getName();
440   int64_t raw_value;
441   switch (v.Type) {
442   case PDB_VariantType::Int8:
443     raw_value = v.Value.Int8;
444     break;
445   case PDB_VariantType::Int16:
446     raw_value = v.Value.Int16;
447     break;
448   case PDB_VariantType::Int32:
449     raw_value = v.Value.Int32;
450     break;
451   case PDB_VariantType::Int64:
452     raw_value = v.Value.Int64;
453     break;
454   case PDB_VariantType::UInt8:
455     raw_value = v.Value.UInt8;
456     break;
457   case PDB_VariantType::UInt16:
458     raw_value = v.Value.UInt16;
459     break;
460   case PDB_VariantType::UInt32:
461     raw_value = v.Value.UInt32;
462     break;
463   case PDB_VariantType::UInt64:
464     raw_value = v.Value.UInt64;
465     break;
466   default:
467     return false;
468   }
469   CompilerType underlying_type =
470       m_ast.GetEnumerationIntegerType(enum_type.GetOpaqueQualType());
471   uint32_t byte_size = m_ast.getASTContext()->getTypeSize(
472       ClangUtil::GetQualType(underlying_type));
473   return m_ast.AddEnumerationValueToEnumerationType(
474       enum_type.GetOpaqueQualType(), underlying_type, decl, name.c_str(),
475       raw_value, byte_size * 8);
476 }
477