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::Char16:
65   case PDB_BuiltinType::Char32:
66   case PDB_BuiltinType::UInt:
67   case PDB_BuiltinType::ULong:
68   case PDB_BuiltinType::HResult:
69   case PDB_BuiltinType::WCharT:
70     return lldb::eEncodingUint;
71   default:
72     return lldb::eEncodingInvalid;
73   }
74 }
75 
76 lldb::Encoding TranslateEnumEncoding(PDB_VariantType type) {
77   switch (type) {
78   case PDB_VariantType::Int8:
79   case PDB_VariantType::Int16:
80   case PDB_VariantType::Int32:
81   case PDB_VariantType::Int64:
82     return lldb::eEncodingSint;
83 
84   case PDB_VariantType::UInt8:
85   case PDB_VariantType::UInt16:
86   case PDB_VariantType::UInt32:
87   case PDB_VariantType::UInt64:
88     return lldb::eEncodingUint;
89 
90   default:
91     break;
92   }
93 
94   return lldb::eEncodingSint;
95 }
96 
97 CompilerType
98 GetBuiltinTypeForPDBEncodingAndBitSize(ClangASTContext &clang_ast,
99                                        const PDBSymbolTypeBuiltin &pdb_type,
100                                        Encoding encoding, uint32_t width) {
101   auto *ast = clang_ast.getASTContext();
102   if (!ast)
103     return CompilerType();
104 
105   switch (pdb_type.getBuiltinType()) {
106   default:
107     break;
108   case PDB_BuiltinType::None:
109     return CompilerType();
110   case PDB_BuiltinType::Void:
111     return clang_ast.GetBasicType(eBasicTypeVoid);
112   case PDB_BuiltinType::Bool:
113     return clang_ast.GetBasicType(eBasicTypeBool);
114   case PDB_BuiltinType::Long:
115     if (width == ast->getTypeSize(ast->LongTy))
116       return CompilerType(ast, ast->LongTy);
117     if (width == ast->getTypeSize(ast->LongLongTy))
118       return CompilerType(ast, ast->LongLongTy);
119     break;
120   case PDB_BuiltinType::ULong:
121     if (width == ast->getTypeSize(ast->UnsignedLongTy))
122       return CompilerType(ast, ast->UnsignedLongTy);
123     if (width == ast->getTypeSize(ast->UnsignedLongLongTy))
124       return CompilerType(ast, ast->UnsignedLongLongTy);
125     break;
126   case PDB_BuiltinType::WCharT:
127     if (width == ast->getTypeSize(ast->WCharTy))
128       return CompilerType(ast, ast->WCharTy);
129     break;
130   case PDB_BuiltinType::Char16:
131     return CompilerType(ast, ast->Char16Ty);
132   case PDB_BuiltinType::Char32:
133     return CompilerType(ast, ast->Char32Ty);
134   case PDB_BuiltinType::Float:
135     // Note: types `long double` and `double` have same bit size in MSVC and
136     // there is no information in the PDB to distinguish them. So when falling
137     // back to default search, the compiler type of `long double` will be
138     // represented by the one generated for `double`.
139     break;
140   }
141   // If there is no match on PDB_BuiltinType, fall back to default search by
142   // encoding and width only
143   return clang_ast.GetBuiltinTypeForEncodingAndBitSize(encoding, width);
144 }
145 
146 ConstString GetPDBBuiltinTypeName(const PDBSymbolTypeBuiltin &pdb_type,
147                                   CompilerType &compiler_type) {
148   PDB_BuiltinType kind = pdb_type.getBuiltinType();
149   switch (kind) {
150   default:
151     break;
152   case PDB_BuiltinType::Currency:
153     return ConstString("CURRENCY");
154   case PDB_BuiltinType::Date:
155     return ConstString("DATE");
156   case PDB_BuiltinType::Variant:
157     return ConstString("VARIANT");
158   case PDB_BuiltinType::Complex:
159     return ConstString("complex");
160   case PDB_BuiltinType::Bitfield:
161     return ConstString("bitfield");
162   case PDB_BuiltinType::BSTR:
163     return ConstString("BSTR");
164   case PDB_BuiltinType::HResult:
165     return ConstString("HRESULT");
166   case PDB_BuiltinType::BCD:
167     return ConstString("BCD");
168   case PDB_BuiltinType::Char16:
169     return ConstString("char16_t");
170   case PDB_BuiltinType::Char32:
171     return ConstString("char32_t");
172   case PDB_BuiltinType::None:
173     return ConstString("...");
174   }
175   return compiler_type.GetTypeName();
176 }
177 
178 bool GetDeclarationForSymbol(const PDBSymbol &symbol, Declaration &decl) {
179   auto &raw_sym = symbol.getRawSymbol();
180   auto first_line_up = raw_sym.getSrcLineOnTypeDefn();
181 
182   if (!first_line_up) {
183     auto lines_up = symbol.getSession().findLineNumbersByAddress(
184         raw_sym.getVirtualAddress(), raw_sym.getLength());
185     if (!lines_up)
186       return false;
187     first_line_up = lines_up->getNext();
188     if (!first_line_up)
189       return false;
190   }
191   uint32_t src_file_id = first_line_up->getSourceFileId();
192   auto src_file_up = symbol.getSession().getSourceFileById(src_file_id);
193   if (!src_file_up)
194     return false;
195 
196   FileSpec spec(src_file_up->getFileName(), /*resolve_path*/ false);
197   decl.SetFile(spec);
198   decl.SetColumn(first_line_up->getColumnNumber());
199   decl.SetLine(first_line_up->getLineNumber());
200   return true;
201 }
202 } // namespace
203 
204 PDBASTParser::PDBASTParser(lldb_private::ClangASTContext &ast) : m_ast(ast) {}
205 
206 PDBASTParser::~PDBASTParser() {}
207 
208 // DebugInfoASTParser interface
209 
210 lldb::TypeSP PDBASTParser::CreateLLDBTypeFromPDBType(const PDBSymbol &type) {
211   // PDB doesn't maintain enough information to robustly rebuild the entire
212   // tree, and this is most problematic when it comes to figure out the right
213   // DeclContext to put a type in.  So for now, everything goes in the
214   // translation unit decl as a fully qualified type.
215   clang::DeclContext *tu_decl_ctx = m_ast.GetTranslationUnitDecl();
216   Declaration decl;
217 
218   switch (type.getSymTag()) {
219   case PDB_SymType::UDT: {
220     auto udt = llvm::dyn_cast<PDBSymbolTypeUDT>(&type);
221     assert(udt);
222     AccessType access = lldb::eAccessPublic;
223     PDB_UdtType udt_kind = udt->getUdtKind();
224     auto tag_type_kind = TranslateUdtKind(udt_kind);
225     if (tag_type_kind == -1)
226       return nullptr;
227 
228     if (udt_kind == PDB_UdtType::Class)
229       access = lldb::eAccessPrivate;
230 
231     CompilerType clang_type = m_ast.CreateRecordType(
232         tu_decl_ctx, access, udt->getName().c_str(), tag_type_kind,
233         lldb::eLanguageTypeC_plus_plus, nullptr);
234 
235     m_ast.SetHasExternalStorage(clang_type.GetOpaqueQualType(), true);
236 
237     return std::make_shared<lldb_private::Type>(
238         type.getSymIndexId(), m_ast.GetSymbolFile(),
239         ConstString(udt->getName()), udt->getLength(), nullptr,
240         LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl, clang_type,
241         lldb_private::Type::eResolveStateForward);
242   } break;
243   case PDB_SymType::Enum: {
244     auto enum_type = llvm::dyn_cast<PDBSymbolTypeEnum>(&type);
245     assert(enum_type);
246     auto underlying_type_up = enum_type->getUnderlyingType();
247     if (!underlying_type_up)
248       return nullptr;
249     lldb::Encoding encoding =
250         TranslateBuiltinEncoding(underlying_type_up->getBuiltinType());
251     // FIXME: Type of underlying builtin is always `Int`. We correct it with
252     // the very first enumerator's encoding if any.
253     auto first_child = enum_type->findOneChild<PDBSymbolData>();
254     if (first_child) {
255       encoding = TranslateEnumEncoding(first_child->getValue().Type);
256     }
257     std::string name = enum_type->getName();
258     uint64_t bytes = enum_type->getLength();
259     CompilerType builtin_type;
260     if (bytes > 0)
261       builtin_type = GetBuiltinTypeForPDBEncodingAndBitSize(
262           m_ast, *underlying_type_up, encoding, bytes * 8);
263     else
264       builtin_type = m_ast.GetBasicType(eBasicTypeInt);
265     // FIXME: PDB does not have information about scoped enumeration (Enum
266     // Class). Set it false for now.
267     bool isScoped = false;
268 
269     CompilerType ast_enum = m_ast.CreateEnumerationType(
270         name.c_str(), tu_decl_ctx, decl, builtin_type, isScoped);
271     auto enum_values = enum_type->findAllChildren<PDBSymbolData>();
272     if (enum_values) {
273       while (auto enum_value = enum_values->getNext()) {
274         if (enum_value->getDataKind() != PDB_DataKind::Constant)
275           continue;
276         AddEnumValue(ast_enum, *enum_value);
277       }
278     }
279     if (ClangASTContext::StartTagDeclarationDefinition(ast_enum))
280       ClangASTContext::CompleteTagDeclarationDefinition(ast_enum);
281 
282     GetDeclarationForSymbol(type, decl);
283     return std::make_shared<lldb_private::Type>(
284         type.getSymIndexId(), m_ast.GetSymbolFile(), ConstString(name), bytes,
285         nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl,
286         ast_enum, lldb_private::Type::eResolveStateFull);
287   } break;
288   case PDB_SymType::Typedef: {
289     auto type_def = llvm::dyn_cast<PDBSymbolTypeTypedef>(&type);
290     assert(type_def);
291     lldb_private::Type *target_type =
292         m_ast.GetSymbolFile()->ResolveTypeUID(type_def->getTypeId());
293     if (!target_type)
294       return nullptr;
295     std::string name = type_def->getName();
296     uint64_t bytes = type_def->getLength();
297     CompilerType target_ast_type = target_type->GetFullCompilerType();
298     CompilerDeclContext target_decl_ctx =
299         m_ast.GetSymbolFile()->GetDeclContextForUID(target_type->GetID());
300     CompilerType ast_typedef =
301         m_ast.CreateTypedefType(target_ast_type, name.c_str(), target_decl_ctx);
302     if (!ast_typedef)
303       return nullptr;
304 
305     return std::make_shared<lldb_private::Type>(
306         type_def->getSymIndexId(), m_ast.GetSymbolFile(), ConstString(name),
307         bytes, nullptr, target_type->GetID(),
308         lldb_private::Type::eEncodingIsTypedefUID, decl, ast_typedef,
309         lldb_private::Type::eResolveStateFull);
310   } break;
311   case PDB_SymType::Function:
312   case PDB_SymType::FunctionSig: {
313     std::string name;
314     PDBSymbolTypeFunctionSig *func_sig = nullptr;
315     if (auto pdb_func = llvm::dyn_cast<PDBSymbolFunc>(&type)) {
316       if (pdb_func->isCompilerGenerated())
317         return nullptr;
318 
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 =
368         m_ast.CreateFunctionType(return_ast_type, arg_list.data(),
369                                  arg_list.size(), is_variadic, 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 = m_ast.CreateArrayType(
403         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, bytes,
433         nullptr, LLDB_INVALID_UID, lldb_private::Type::eEncodingIsUID, decl,
434         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->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->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:
469     break;
470   }
471   return nullptr;
472 }
473 
474 bool PDBASTParser::AddEnumValue(CompilerType enum_type,
475                                 const PDBSymbolData &enum_value) const {
476   Declaration decl;
477   Variant v = enum_value.getValue();
478   std::string name = enum_value.getName();
479   int64_t raw_value;
480   switch (v.Type) {
481   case PDB_VariantType::Int8:
482     raw_value = v.Value.Int8;
483     break;
484   case PDB_VariantType::Int16:
485     raw_value = v.Value.Int16;
486     break;
487   case PDB_VariantType::Int32:
488     raw_value = v.Value.Int32;
489     break;
490   case PDB_VariantType::Int64:
491     raw_value = v.Value.Int64;
492     break;
493   case PDB_VariantType::UInt8:
494     raw_value = v.Value.UInt8;
495     break;
496   case PDB_VariantType::UInt16:
497     raw_value = v.Value.UInt16;
498     break;
499   case PDB_VariantType::UInt32:
500     raw_value = v.Value.UInt32;
501     break;
502   case PDB_VariantType::UInt64:
503     raw_value = v.Value.UInt64;
504     break;
505   default:
506     return false;
507   }
508   CompilerType underlying_type =
509       m_ast.GetEnumerationIntegerType(enum_type.GetOpaqueQualType());
510   uint32_t byte_size = m_ast.getASTContext()->getTypeSize(
511       ClangUtil::GetQualType(underlying_type));
512   return m_ast.AddEnumerationValueToEnumerationType(
513       enum_type.GetOpaqueQualType(), underlying_type, decl, name.c_str(),
514       raw_value, byte_size * 8);
515 }
516