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   case PDB_BuiltinType::Char16:
63   case PDB_BuiltinType::Char32:
64     return lldb::eEncodingSint;
65   case PDB_BuiltinType::Bool:
66   case PDB_BuiltinType::UInt:
67   case PDB_BuiltinType::ULong:
68   case PDB_BuiltinType::HResult:
69     return lldb::eEncodingUint;
70   default:
71     return lldb::eEncodingInvalid;
72   }
73 }
74 
75 lldb::Encoding TranslateEnumEncoding(PDB_VariantType type) {
76   switch (type) {
77   case PDB_VariantType::Int8:
78   case PDB_VariantType::Int16:
79   case PDB_VariantType::Int32:
80   case PDB_VariantType::Int64:
81     return lldb::eEncodingSint;
82 
83   case PDB_VariantType::UInt8:
84   case PDB_VariantType::UInt16:
85   case PDB_VariantType::UInt32:
86   case PDB_VariantType::UInt64:
87     return lldb::eEncodingUint;
88 
89   default:
90     break;
91   }
92 
93   return lldb::eEncodingSint;
94 }
95 
96 CompilerType GetBuiltinTypeForPDBEncodingAndBitSize(
97     ClangASTContext *clang_ast, const PDBSymbolTypeBuiltin *pdb_type,
98     Encoding encoding, uint32_t width) {
99   if (!pdb_type)
100     return CompilerType();
101   if (!clang_ast)
102     return CompilerType();
103   auto *ast = clang_ast->getASTContext();
104   if (!ast)
105     return CompilerType();
106 
107   switch (pdb_type->getBuiltinType()) {
108   default: break;
109   case PDB_BuiltinType::None:
110     return CompilerType();
111   case PDB_BuiltinType::Void:
112     return clang_ast->GetBasicType(eBasicTypeVoid);
113   case PDB_BuiltinType::Bool:
114     return clang_ast->GetBasicType(eBasicTypeBool);
115   case PDB_BuiltinType::Long:
116     if (width == ast->getTypeSize(ast->LongTy))
117       return CompilerType(ast, ast->LongTy);
118     if (width == ast->getTypeSize(ast->LongLongTy))
119       return CompilerType(ast, ast->LongLongTy);
120     break;
121   case PDB_BuiltinType::ULong:
122     if (width == ast->getTypeSize(ast->UnsignedLongTy))
123       return CompilerType(ast, ast->UnsignedLongTy);
124     if (width == ast->getTypeSize(ast->UnsignedLongLongTy))
125       return CompilerType(ast, ast->UnsignedLongLongTy);
126     break;
127   case PDB_BuiltinType::WCharT:
128     if (width == ast->getTypeSize(ast->WCharTy))
129       return CompilerType(ast, ast->WCharTy);
130     break;
131   case PDB_BuiltinType::Char16:
132     return CompilerType(ast, ast->Char16Ty);
133   case PDB_BuiltinType::Char32:
134     return CompilerType(ast, ast->Char32Ty);
135   case PDB_BuiltinType::Float:
136     // Note: types `long double` and `double` have same bit size in MSVC and there
137     // is no information in the PDB to distinguish them. So when falling back
138     // to default search, the compiler type of `long double` will be represented by
139     // the one generated for `double`.
140     break;
141   }
142   // If there is no match on PDB_BuiltinType, fall back to default search
143   // by encoding and width only
144   return clang_ast->GetBuiltinTypeForEncodingAndBitSize(encoding, width);
145 }
146 
147 ConstString GetPDBBuiltinTypeName(const PDBSymbolTypeBuiltin *pdb_type,
148                                   CompilerType &compiler_type) {
149   if (!pdb_type)
150     return compiler_type.GetTypeName();
151 
152   PDB_BuiltinType kind = pdb_type->getBuiltinType();
153   switch (kind) {
154   default: break;
155   case PDB_BuiltinType::Currency:
156     return ConstString("CURRENCY");
157   case PDB_BuiltinType::Date:
158     return ConstString("DATE");
159   case PDB_BuiltinType::Variant:
160     return ConstString("VARIANT");
161   case PDB_BuiltinType::Complex:
162     return ConstString("complex");
163   case PDB_BuiltinType::Bitfield:
164     return ConstString("bitfield");
165   case PDB_BuiltinType::BSTR:
166     return ConstString("BSTR");
167   case PDB_BuiltinType::HResult:
168     return ConstString("HRESULT");
169   case PDB_BuiltinType::BCD:
170     return ConstString("BCD");
171   case PDB_BuiltinType::Char16:
172     return ConstString("char16_t");
173   case PDB_BuiltinType::Char32:
174     return ConstString("char32_t");
175   case PDB_BuiltinType::None:
176     return ConstString("...");
177   }
178   return compiler_type.GetTypeName();
179 }
180 
181 bool GetDeclarationForSymbol(const PDBSymbol &symbol, Declaration &decl) {
182   auto &raw_sym = symbol.getRawSymbol();
183   auto first_line_up = raw_sym.getSrcLineOnTypeDefn();
184 
185   if (!first_line_up) {
186     auto lines_up = symbol.getSession().findLineNumbersByAddress(
187         raw_sym.getVirtualAddress(), raw_sym.getLength());
188     if (!lines_up)
189       return false;
190     first_line_up = lines_up->getNext();
191     if (!first_line_up)
192       return false;
193   }
194   uint32_t src_file_id = first_line_up->getSourceFileId();
195   auto src_file_up = symbol.getSession().getSourceFileById(src_file_id);
196   if (!src_file_up)
197     return false;
198 
199   FileSpec spec(src_file_up->getFileName(), /*resolve_path*/false);
200   decl.SetFile(spec);
201   decl.SetColumn(first_line_up->getColumnNumber());
202   decl.SetLine(first_line_up->getLineNumber());
203   return true;
204 }
205 }
206 
207 PDBASTParser::PDBASTParser(lldb_private::ClangASTContext &ast) : m_ast(ast) {}
208 
209 PDBASTParser::~PDBASTParser() {}
210 
211 // DebugInfoASTParser interface
212 
213 lldb::TypeSP PDBASTParser::CreateLLDBTypeFromPDBType(const PDBSymbol &type) {
214   // PDB doesn't maintain enough information to robustly rebuild the entire
215   // tree, and this is most problematic when it comes to figure out the
216   // right DeclContext to put a type in.  So for now, everything goes in
217   // the translation unit decl as a fully qualified type.
218   clang::DeclContext *tu_decl_ctx = m_ast.GetTranslationUnitDecl();
219   Declaration decl;
220 
221   switch (type.getSymTag()) {
222   case PDB_SymType::UDT: {
223     auto udt = llvm::dyn_cast<PDBSymbolTypeUDT>(&type);
224     assert(udt);
225     AccessType access = lldb::eAccessPublic;
226     PDB_UdtType udt_kind = udt->getUdtKind();
227     auto tag_type_kind = TranslateUdtKind(udt_kind);
228     if (tag_type_kind == -1)
229       return nullptr;
230 
231     if (udt_kind == PDB_UdtType::Class)
232       access = lldb::eAccessPrivate;
233 
234     CompilerType clang_type = m_ast.CreateRecordType(
235         tu_decl_ctx, access, udt->getName().c_str(), tag_type_kind,
236         lldb::eLanguageTypeC_plus_plus, nullptr);
237 
238     m_ast.SetHasExternalStorage(clang_type.GetOpaqueQualType(), true);
239 
240     return std::make_shared<lldb_private::Type>(
241         type.getSymIndexId(), m_ast.GetSymbolFile(),
242         ConstString(udt->getName()), udt->getLength(), nullptr,
243         LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl, clang_type,
244         lldb_private::Type::eResolveStateForward);
245   } break;
246   case PDB_SymType::Enum: {
247     auto enum_type = llvm::dyn_cast<PDBSymbolTypeEnum>(&type);
248     assert(enum_type);
249     auto underlying_type_up = enum_type->getUnderlyingType();
250     if (!underlying_type_up)
251       return nullptr;
252     lldb::Encoding encoding =
253         TranslateBuiltinEncoding(underlying_type_up->getBuiltinType());
254     // FIXME: Type of underlying builtin is always `Int`. We correct it with
255     // the very first enumerator's encoding if any.
256     auto first_child = enum_type->findOneChild<PDBSymbolData>();
257     if (first_child) {
258       encoding = TranslateEnumEncoding(first_child->getValue().Type);
259     }
260     std::string name = enum_type->getName();
261     uint64_t bytes = enum_type->getLength();
262     CompilerType builtin_type;
263     if (bytes > 0)
264       builtin_type = GetBuiltinTypeForPDBEncodingAndBitSize(
265            &m_ast, underlying_type_up.get(), encoding, bytes * 8);
266     else
267       builtin_type = m_ast.GetBasicType(eBasicTypeInt);
268     // FIXME: PDB does not have information about scoped enumeration (Enum Class).
269     // Set it false for now.
270     bool isScoped = false;
271 
272     CompilerType ast_enum = m_ast.CreateEnumerationType(
273         name.c_str(), tu_decl_ctx, decl, builtin_type, isScoped);
274     auto enum_values = enum_type->findAllChildren<PDBSymbolData>();
275     if (enum_values) {
276       while (auto enum_value = enum_values->getNext()) {
277         if (enum_value->getDataKind() != PDB_DataKind::Constant)
278           continue;
279         AddEnumValue(ast_enum, *enum_value);
280       }
281     }
282     if (ClangASTContext::StartTagDeclarationDefinition(ast_enum))
283       ClangASTContext::CompleteTagDeclarationDefinition(ast_enum);
284 
285     GetDeclarationForSymbol(type, decl);
286     return std::make_shared<lldb_private::Type>(
287         type.getSymIndexId(), m_ast.GetSymbolFile(), ConstString(name), bytes,
288         nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl,
289         ast_enum, lldb_private::Type::eResolveStateFull);
290   } break;
291   case PDB_SymType::Typedef: {
292     auto type_def = llvm::dyn_cast<PDBSymbolTypeTypedef>(&type);
293     assert(type_def);
294     lldb_private::Type *target_type =
295         m_ast.GetSymbolFile()->ResolveTypeUID(type_def->getTypeId());
296     if (!target_type)
297       return nullptr;
298     std::string name = type_def->getName();
299     uint64_t bytes = type_def->getLength();
300     CompilerType target_ast_type = target_type->GetFullCompilerType();
301     CompilerDeclContext target_decl_ctx =
302         m_ast.GetSymbolFile()->GetDeclContextForUID(target_type->GetID());
303     CompilerType ast_typedef =
304         m_ast.CreateTypedefType(target_ast_type, name.c_str(), target_decl_ctx);
305     if (!ast_typedef)
306       return nullptr;
307 
308     return std::make_shared<lldb_private::Type>(
309         type_def->getSymIndexId(), m_ast.GetSymbolFile(), ConstString(name),
310         bytes, nullptr, target_type->GetID(),
311         lldb_private::Type::eEncodingIsTypedefUID, decl, ast_typedef,
312         lldb_private::Type::eResolveStateFull);
313   } break;
314   case PDB_SymType::Function:
315   case PDB_SymType::FunctionSig: {
316     std::string name;
317     PDBSymbolTypeFunctionSig *func_sig = nullptr;
318     if (auto pdb_func = llvm::dyn_cast<PDBSymbolFunc>(&type)) {
319       auto sig = pdb_func->getSignature();
320       if (!sig)
321         return nullptr;
322       func_sig = sig.release();
323       // Function type is named.
324       name = pdb_func->getName();
325     } else if (auto pdb_func_sig =
326               llvm::dyn_cast<PDBSymbolTypeFunctionSig>(&type)) {
327       func_sig = const_cast<PDBSymbolTypeFunctionSig*>(pdb_func_sig);
328     } else
329       llvm_unreachable("Unexpected PDB symbol!");
330 
331     auto arg_enum = func_sig->getArguments();
332     uint32_t num_args = arg_enum->getChildCount();
333     std::vector<CompilerType> arg_list;
334 
335     bool is_variadic = func_sig->isCVarArgs();
336     // Drop last variadic argument.
337     if (is_variadic)
338       --num_args;
339     for (uint32_t arg_idx = 0; arg_idx < num_args; arg_idx++) {
340       auto arg = arg_enum->getChildAtIndex(arg_idx);
341       if (!arg)
342         break;
343       lldb_private::Type *arg_type =
344           m_ast.GetSymbolFile()->ResolveTypeUID(arg->getSymIndexId());
345       // If there's some error looking up one of the dependent types of this
346       // function signature, bail.
347       if (!arg_type)
348         return nullptr;
349       CompilerType arg_ast_type = arg_type->GetFullCompilerType();
350       arg_list.push_back(arg_ast_type);
351     }
352     lldbassert(arg_list.size() <= num_args);
353 
354     auto pdb_return_type = func_sig->getReturnType();
355     lldb_private::Type *return_type =
356         m_ast.GetSymbolFile()->ResolveTypeUID(pdb_return_type->getSymIndexId());
357     // If there's some error looking up one of the dependent types of this
358     // function signature, bail.
359     if (!return_type)
360       return nullptr;
361     CompilerType return_ast_type = return_type->GetFullCompilerType();
362     uint32_t type_quals = 0;
363     if (func_sig->isConstType())
364       type_quals |= clang::Qualifiers::Const;
365     if (func_sig->isVolatileType())
366       type_quals |= clang::Qualifiers::Volatile;
367     CompilerType func_sig_ast_type = m_ast.CreateFunctionType(
368         return_ast_type, arg_list.data(), arg_list.size(), is_variadic,
369         type_quals);
370 
371     GetDeclarationForSymbol(type, decl);
372     return std::make_shared<lldb_private::Type>(
373         type.getSymIndexId(), m_ast.GetSymbolFile(), ConstString(name), 0,
374         nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl,
375         func_sig_ast_type, lldb_private::Type::eResolveStateFull);
376   } break;
377   case PDB_SymType::ArrayType: {
378     auto array_type = llvm::dyn_cast<PDBSymbolTypeArray>(&type);
379     assert(array_type);
380     uint32_t num_elements = array_type->getCount();
381     uint32_t element_uid = array_type->getElementTypeId();
382     uint32_t bytes = array_type->getLength();
383 
384     // If array rank > 0, PDB gives the element type at N=0. So element type
385     // will parsed in the order N=0, N=1,..., N=rank sequentially.
386     lldb_private::Type *element_type =
387         m_ast.GetSymbolFile()->ResolveTypeUID(element_uid);
388     if (!element_type)
389       return nullptr;
390 
391     CompilerType element_ast_type = element_type->GetForwardCompilerType();
392     // If element type is UDT, it needs to be complete.
393     if (ClangASTContext::IsCXXClassType(element_ast_type) &&
394         element_ast_type.GetCompleteType() == false) {
395       if (ClangASTContext::StartTagDeclarationDefinition(element_ast_type)) {
396         ClangASTContext::CompleteTagDeclarationDefinition(element_ast_type);
397       } else {
398         // We are not able to start defintion.
399         return nullptr;
400       }
401     }
402     CompilerType array_ast_type =
403         m_ast.CreateArrayType(element_ast_type, num_elements, /*is_gnu_vector*/false);
404     TypeSP type_sp = std::make_shared<lldb_private::Type>(
405         array_type->getSymIndexId(), m_ast.GetSymbolFile(), ConstString(),
406         bytes, nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID,
407         decl, array_ast_type, lldb_private::Type::eResolveStateFull);
408     type_sp->SetEncodingType(element_type);
409     return type_sp;
410   } break;
411   case PDB_SymType::BuiltinType: {
412     auto *builtin_type = llvm::dyn_cast<PDBSymbolTypeBuiltin>(&type);
413     assert(builtin_type);
414     PDB_BuiltinType builtin_kind = builtin_type->getBuiltinType();
415     if (builtin_kind == PDB_BuiltinType::None)
416       return nullptr;
417 
418     uint64_t bytes = builtin_type->getLength();
419     Encoding encoding = TranslateBuiltinEncoding(builtin_kind);
420     CompilerType builtin_ast_type = GetBuiltinTypeForPDBEncodingAndBitSize(
421         &m_ast, builtin_type, encoding, bytes * 8);
422 
423     if (builtin_type->isConstType())
424       builtin_ast_type = builtin_ast_type.AddConstModifier();
425 
426     if (builtin_type->isVolatileType())
427       builtin_ast_type = builtin_ast_type.AddVolatileModifier();
428 
429     auto type_name = GetPDBBuiltinTypeName(builtin_type, builtin_ast_type);
430 
431     return std::make_shared<lldb_private::Type>(
432         builtin_type->getSymIndexId(), m_ast.GetSymbolFile(), type_name,
433         bytes, nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID,
434         decl, builtin_ast_type, lldb_private::Type::eResolveStateFull);
435   } break;
436   case PDB_SymType::PointerType: {
437     auto *pointer_type = llvm::dyn_cast<PDBSymbolTypePointer>(&type);
438     assert(pointer_type);
439     Type *pointee_type = m_ast.GetSymbolFile()->ResolveTypeUID(
440         pointer_type->getPointeeType()->getSymIndexId());
441     if (!pointee_type)
442       return nullptr;
443 
444     CompilerType pointer_ast_type;
445     pointer_ast_type = pointee_type->GetFullCompilerType();
446     if (pointer_type->isReference())
447       pointer_ast_type = pointer_ast_type.GetLValueReferenceType();
448     else if (pointer_type->getRawSymbol().isRValueReference())
449       pointer_ast_type = pointer_ast_type.GetRValueReferenceType();
450     else
451       pointer_ast_type = pointer_ast_type.GetPointerType();
452 
453     if (pointer_type->isConstType())
454       pointer_ast_type = pointer_ast_type.AddConstModifier();
455 
456     if (pointer_type->isVolatileType())
457       pointer_ast_type = pointer_ast_type.AddVolatileModifier();
458 
459     if (pointer_type->getRawSymbol().isRestrictedType())
460       pointer_ast_type = pointer_ast_type.AddRestrictModifier();
461 
462     return std::make_shared<lldb_private::Type>(
463         pointer_type->getSymIndexId(), m_ast.GetSymbolFile(), ConstString(),
464         pointer_type->getLength(), nullptr, LLDB_INVALID_UID,
465         lldb_private::Type::eEncodingIsUID, decl, pointer_ast_type,
466         lldb_private::Type::eResolveStateFull);
467   } break;
468   default: break;
469   }
470   return nullptr;
471 }
472 
473 bool PDBASTParser::AddEnumValue(CompilerType enum_type,
474                                 const PDBSymbolData &enum_value) const {
475   Declaration decl;
476   Variant v = enum_value.getValue();
477   std::string name = enum_value.getName();
478   int64_t raw_value;
479   switch (v.Type) {
480   case PDB_VariantType::Int8:
481     raw_value = v.Value.Int8;
482     break;
483   case PDB_VariantType::Int16:
484     raw_value = v.Value.Int16;
485     break;
486   case PDB_VariantType::Int32:
487     raw_value = v.Value.Int32;
488     break;
489   case PDB_VariantType::Int64:
490     raw_value = v.Value.Int64;
491     break;
492   case PDB_VariantType::UInt8:
493     raw_value = v.Value.UInt8;
494     break;
495   case PDB_VariantType::UInt16:
496     raw_value = v.Value.UInt16;
497     break;
498   case PDB_VariantType::UInt32:
499     raw_value = v.Value.UInt32;
500     break;
501   case PDB_VariantType::UInt64:
502     raw_value = v.Value.UInt64;
503     break;
504   default:
505     return false;
506   }
507   CompilerType underlying_type =
508       m_ast.GetEnumerationIntegerType(enum_type.GetOpaqueQualType());
509   uint32_t byte_size = m_ast.getASTContext()->getTypeSize(
510       ClangUtil::GetQualType(underlying_type));
511   return m_ast.AddEnumerationValueToEnumerationType(
512       enum_type.GetOpaqueQualType(), underlying_type, decl, name.c_str(),
513       raw_value, byte_size * 8);
514 }
515