1307f5ae8SZachary Turner //===-- CompileUnitIndex.cpp ------------------------------------*- C++ -*-===//
2307f5ae8SZachary Turner //
3*2946cd70SChandler Carruth // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4*2946cd70SChandler Carruth // See https://llvm.org/LICENSE.txt for license information.
5*2946cd70SChandler Carruth // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6307f5ae8SZachary Turner //
7307f5ae8SZachary Turner //===----------------------------------------------------------------------===//
8307f5ae8SZachary Turner 
9307f5ae8SZachary Turner #include "CompileUnitIndex.h"
10307f5ae8SZachary Turner 
11307f5ae8SZachary Turner #include "PdbIndex.h"
12307f5ae8SZachary Turner #include "PdbUtil.h"
13307f5ae8SZachary Turner 
14307f5ae8SZachary Turner #include "llvm/DebugInfo/CodeView/LazyRandomTypeCollection.h"
15307f5ae8SZachary Turner #include "llvm/DebugInfo/CodeView/SymbolDeserializer.h"
16307f5ae8SZachary Turner #include "llvm/DebugInfo/CodeView/TypeDeserializer.h"
17307f5ae8SZachary Turner #include "llvm/DebugInfo/MSF/MappedBlockStream.h"
18307f5ae8SZachary Turner #include "llvm/DebugInfo/PDB/Native/DbiModuleDescriptor.h"
19307f5ae8SZachary Turner #include "llvm/DebugInfo/PDB/Native/DbiStream.h"
20307f5ae8SZachary Turner #include "llvm/DebugInfo/PDB/Native/InfoStream.h"
21307f5ae8SZachary Turner #include "llvm/DebugInfo/PDB/Native/ModuleDebugStream.h"
22307f5ae8SZachary Turner #include "llvm/DebugInfo/PDB/Native/NamedStreamMap.h"
23307f5ae8SZachary Turner #include "llvm/DebugInfo/PDB/Native/TpiStream.h"
24307f5ae8SZachary Turner #include "llvm/Support/Path.h"
25307f5ae8SZachary Turner 
26307f5ae8SZachary Turner #include "lldb/Utility/LLDBAssert.h"
27307f5ae8SZachary Turner 
28307f5ae8SZachary Turner using namespace lldb;
29307f5ae8SZachary Turner using namespace lldb_private;
30307f5ae8SZachary Turner using namespace lldb_private::npdb;
31307f5ae8SZachary Turner using namespace llvm::codeview;
32307f5ae8SZachary Turner using namespace llvm::pdb;
33307f5ae8SZachary Turner 
34307f5ae8SZachary Turner static bool IsMainFile(llvm::StringRef main, llvm::StringRef other) {
35307f5ae8SZachary Turner   if (main == other)
36307f5ae8SZachary Turner     return true;
37307f5ae8SZachary Turner 
38307f5ae8SZachary Turner   // If the files refer to the local file system, we can just ask the file
39307f5ae8SZachary Turner   // system if they're equivalent.  But if the source isn't present on disk
40307f5ae8SZachary Turner   // then we still want to try.
41307f5ae8SZachary Turner   if (llvm::sys::fs::equivalent(main, other))
42307f5ae8SZachary Turner     return true;
43307f5ae8SZachary Turner 
44307f5ae8SZachary Turner   llvm::SmallString<64> normalized(other);
45307f5ae8SZachary Turner   llvm::sys::path::native(normalized);
46307f5ae8SZachary Turner   return main.equals_lower(normalized);
47307f5ae8SZachary Turner }
48307f5ae8SZachary Turner 
49307f5ae8SZachary Turner static void ParseCompile3(const CVSymbol &sym, CompilandIndexItem &cci) {
50307f5ae8SZachary Turner   cci.m_compile_opts.emplace();
51307f5ae8SZachary Turner   llvm::cantFail(
52307f5ae8SZachary Turner       SymbolDeserializer::deserializeAs<Compile3Sym>(sym, *cci.m_compile_opts));
53307f5ae8SZachary Turner }
54307f5ae8SZachary Turner 
55307f5ae8SZachary Turner static void ParseObjname(const CVSymbol &sym, CompilandIndexItem &cci) {
56307f5ae8SZachary Turner   cci.m_obj_name.emplace();
57307f5ae8SZachary Turner   llvm::cantFail(
58307f5ae8SZachary Turner       SymbolDeserializer::deserializeAs<ObjNameSym>(sym, *cci.m_obj_name));
59307f5ae8SZachary Turner }
60307f5ae8SZachary Turner 
61307f5ae8SZachary Turner static void ParseBuildInfo(PdbIndex &index, const CVSymbol &sym,
62307f5ae8SZachary Turner                            CompilandIndexItem &cci) {
63307f5ae8SZachary Turner   BuildInfoSym bis(SymbolRecordKind::BuildInfoSym);
64307f5ae8SZachary Turner   llvm::cantFail(SymbolDeserializer::deserializeAs<BuildInfoSym>(sym, bis));
65307f5ae8SZachary Turner 
66307f5ae8SZachary Turner   // S_BUILDINFO just points to an LF_BUILDINFO in the IPI stream.  Let's do
67307f5ae8SZachary Turner   // a little extra work to pull out the LF_BUILDINFO.
68307f5ae8SZachary Turner   LazyRandomTypeCollection &types = index.ipi().typeCollection();
69307f5ae8SZachary Turner   llvm::Optional<CVType> cvt = types.tryGetType(bis.BuildId);
70307f5ae8SZachary Turner 
71307f5ae8SZachary Turner   if (!cvt || cvt->kind() != LF_BUILDINFO)
72307f5ae8SZachary Turner     return;
73307f5ae8SZachary Turner 
74307f5ae8SZachary Turner   BuildInfoRecord bir;
75307f5ae8SZachary Turner   llvm::cantFail(TypeDeserializer::deserializeAs<BuildInfoRecord>(*cvt, bir));
76307f5ae8SZachary Turner   cci.m_build_info.assign(bir.ArgIndices.begin(), bir.ArgIndices.end());
77307f5ae8SZachary Turner }
78307f5ae8SZachary Turner 
79307f5ae8SZachary Turner static void ParseExtendedInfo(PdbIndex &index, CompilandIndexItem &item) {
80307f5ae8SZachary Turner   const CVSymbolArray &syms = item.m_debug_stream.getSymbolArray();
81307f5ae8SZachary Turner 
82307f5ae8SZachary Turner   // This is a private function, it shouldn't be called if the information
83307f5ae8SZachary Turner   // has already been parsed.
84307f5ae8SZachary Turner   lldbassert(!item.m_obj_name);
85307f5ae8SZachary Turner   lldbassert(!item.m_compile_opts);
86307f5ae8SZachary Turner   lldbassert(item.m_build_info.empty());
87307f5ae8SZachary Turner 
88307f5ae8SZachary Turner   // We're looking for 3 things.  S_COMPILE3, S_OBJNAME, and S_BUILDINFO.
89307f5ae8SZachary Turner   int found = 0;
90307f5ae8SZachary Turner   for (const CVSymbol &sym : syms) {
91307f5ae8SZachary Turner     switch (sym.kind()) {
92307f5ae8SZachary Turner     case S_COMPILE3:
93307f5ae8SZachary Turner       ParseCompile3(sym, item);
94307f5ae8SZachary Turner       break;
95307f5ae8SZachary Turner     case S_OBJNAME:
96307f5ae8SZachary Turner       ParseObjname(sym, item);
97307f5ae8SZachary Turner       break;
98307f5ae8SZachary Turner     case S_BUILDINFO:
99307f5ae8SZachary Turner       ParseBuildInfo(index, sym, item);
100307f5ae8SZachary Turner       break;
101307f5ae8SZachary Turner     default:
102307f5ae8SZachary Turner       continue;
103307f5ae8SZachary Turner     }
104307f5ae8SZachary Turner     if (++found >= 3)
105307f5ae8SZachary Turner       break;
106307f5ae8SZachary Turner   }
107307f5ae8SZachary Turner }
108307f5ae8SZachary Turner 
109307f5ae8SZachary Turner CompilandIndexItem::CompilandIndexItem(
1106284aee9SZachary Turner     PdbCompilandId id, llvm::pdb::ModuleDebugStreamRef debug_stream,
111307f5ae8SZachary Turner     llvm::pdb::DbiModuleDescriptor descriptor)
1126284aee9SZachary Turner     : m_id(id), m_debug_stream(std::move(debug_stream)),
113307f5ae8SZachary Turner       m_module_descriptor(std::move(descriptor)) {}
114307f5ae8SZachary Turner 
115307f5ae8SZachary Turner CompilandIndexItem &CompileUnitIndex::GetOrCreateCompiland(uint16_t modi) {
1166284aee9SZachary Turner   auto result = m_comp_units.try_emplace(modi, nullptr);
117307f5ae8SZachary Turner   if (!result.second)
118307f5ae8SZachary Turner     return *result.first->second;
119307f5ae8SZachary Turner 
120307f5ae8SZachary Turner   // Find the module list and load its debug information stream and cache it
121307f5ae8SZachary Turner   // since we need to use it for almost all interesting operations.
122307f5ae8SZachary Turner   const DbiModuleList &modules = m_index.dbi().modules();
123307f5ae8SZachary Turner   llvm::pdb::DbiModuleDescriptor descriptor = modules.getModuleDescriptor(modi);
124307f5ae8SZachary Turner   uint16_t stream = descriptor.getModuleStreamIndex();
125307f5ae8SZachary Turner   std::unique_ptr<llvm::msf::MappedBlockStream> stream_data =
126307f5ae8SZachary Turner       m_index.pdb().createIndexedStream(stream);
127307f5ae8SZachary Turner   llvm::pdb::ModuleDebugStreamRef debug_stream(descriptor,
128307f5ae8SZachary Turner                                                std::move(stream_data));
129307f5ae8SZachary Turner   cantFail(debug_stream.reload());
130307f5ae8SZachary Turner 
131307f5ae8SZachary Turner   std::unique_ptr<CompilandIndexItem> &cci = result.first->second;
132307f5ae8SZachary Turner 
133307f5ae8SZachary Turner   cci = llvm::make_unique<CompilandIndexItem>(
1346284aee9SZachary Turner       PdbCompilandId{modi}, std::move(debug_stream), std::move(descriptor));
135307f5ae8SZachary Turner   ParseExtendedInfo(m_index, *cci);
136307f5ae8SZachary Turner 
137307f5ae8SZachary Turner   cci->m_strings.initialize(debug_stream.getSubsectionsArray());
138307f5ae8SZachary Turner   PDBStringTable &strings = cantFail(m_index.pdb().getStringTable());
139307f5ae8SZachary Turner   cci->m_strings.setStrings(strings.getStringTable());
140307f5ae8SZachary Turner 
141307f5ae8SZachary Turner   // We want the main source file to always comes first.  Note that we can't
142307f5ae8SZachary Turner   // just push_back the main file onto the front because `GetMainSourceFile`
143307f5ae8SZachary Turner   // computes it in such a way that it doesn't own the resulting memory.  So we
144307f5ae8SZachary Turner   // have to iterate the module file list comparing each one to the main file
145307f5ae8SZachary Turner   // name until we find it, and we can cache that one since the memory is backed
146307f5ae8SZachary Turner   // by a contiguous chunk inside the mapped PDB.
147307f5ae8SZachary Turner   llvm::SmallString<64> main_file = GetMainSourceFile(*cci);
148307f5ae8SZachary Turner   std::string s = main_file.str();
149307f5ae8SZachary Turner   llvm::sys::path::native(main_file);
150307f5ae8SZachary Turner 
151307f5ae8SZachary Turner   uint32_t file_count = modules.getSourceFileCount(modi);
152307f5ae8SZachary Turner   cci->m_file_list.reserve(file_count);
153307f5ae8SZachary Turner   bool found_main_file = false;
154307f5ae8SZachary Turner   for (llvm::StringRef file : modules.source_files(modi)) {
155307f5ae8SZachary Turner     if (!found_main_file && IsMainFile(main_file, file)) {
156307f5ae8SZachary Turner       cci->m_file_list.insert(cci->m_file_list.begin(), file);
157307f5ae8SZachary Turner       found_main_file = true;
158307f5ae8SZachary Turner       continue;
159307f5ae8SZachary Turner     }
160307f5ae8SZachary Turner     cci->m_file_list.push_back(file);
161307f5ae8SZachary Turner   }
162307f5ae8SZachary Turner 
163307f5ae8SZachary Turner   return *cci;
164307f5ae8SZachary Turner }
165307f5ae8SZachary Turner 
166307f5ae8SZachary Turner const CompilandIndexItem *CompileUnitIndex::GetCompiland(uint16_t modi) const {
1676284aee9SZachary Turner   auto iter = m_comp_units.find(modi);
168307f5ae8SZachary Turner   if (iter == m_comp_units.end())
169307f5ae8SZachary Turner     return nullptr;
170307f5ae8SZachary Turner   return iter->second.get();
171307f5ae8SZachary Turner }
172307f5ae8SZachary Turner 
173307f5ae8SZachary Turner CompilandIndexItem *CompileUnitIndex::GetCompiland(uint16_t modi) {
1746284aee9SZachary Turner   auto iter = m_comp_units.find(modi);
175307f5ae8SZachary Turner   if (iter == m_comp_units.end())
176307f5ae8SZachary Turner     return nullptr;
177307f5ae8SZachary Turner   return iter->second.get();
178307f5ae8SZachary Turner }
179307f5ae8SZachary Turner 
180307f5ae8SZachary Turner llvm::SmallString<64>
181307f5ae8SZachary Turner CompileUnitIndex::GetMainSourceFile(const CompilandIndexItem &item) const {
182307f5ae8SZachary Turner   // LF_BUILDINFO contains a list of arg indices which point to LF_STRING_ID
183307f5ae8SZachary Turner   // records in the IPI stream.  The order of the arg indices is as follows:
184307f5ae8SZachary Turner   // [0] - working directory where compiler was invoked.
185307f5ae8SZachary Turner   // [1] - absolute path to compiler binary
186307f5ae8SZachary Turner   // [2] - source file name
187307f5ae8SZachary Turner   // [3] - path to compiler generated PDB (the /Zi PDB, although this entry gets
188307f5ae8SZachary Turner   //       added even when using /Z7)
189307f5ae8SZachary Turner   // [4] - full command line invocation.
190307f5ae8SZachary Turner   //
191307f5ae8SZachary Turner   // We need to form the path [0]\[2] to generate the full path to the main
192307f5ae8SZachary Turner   // file.source
193307f5ae8SZachary Turner   if (item.m_build_info.size() < 3)
194307f5ae8SZachary Turner     return {""};
195307f5ae8SZachary Turner 
196307f5ae8SZachary Turner   LazyRandomTypeCollection &types = m_index.ipi().typeCollection();
197307f5ae8SZachary Turner 
198307f5ae8SZachary Turner   StringIdRecord working_dir;
199307f5ae8SZachary Turner   StringIdRecord file_name;
200307f5ae8SZachary Turner   CVType dir_cvt = types.getType(item.m_build_info[0]);
201307f5ae8SZachary Turner   CVType file_cvt = types.getType(item.m_build_info[2]);
202307f5ae8SZachary Turner   llvm::cantFail(
203307f5ae8SZachary Turner       TypeDeserializer::deserializeAs<StringIdRecord>(dir_cvt, working_dir));
204307f5ae8SZachary Turner   llvm::cantFail(
205307f5ae8SZachary Turner       TypeDeserializer::deserializeAs<StringIdRecord>(file_cvt, file_name));
206307f5ae8SZachary Turner 
207b3130b4fSZachary Turner   llvm::sys::path::Style style = working_dir.String.startswith("/")
208b3130b4fSZachary Turner                                      ? llvm::sys::path::Style::posix
209b3130b4fSZachary Turner                                      : llvm::sys::path::Style::windows;
210b3130b4fSZachary Turner   if (llvm::sys::path::is_absolute(file_name.String, style))
211b3130b4fSZachary Turner     return file_name.String;
212b3130b4fSZachary Turner 
213307f5ae8SZachary Turner   llvm::SmallString<64> absolute_path = working_dir.String;
214307f5ae8SZachary Turner   llvm::sys::path::append(absolute_path, file_name.String);
215307f5ae8SZachary Turner   return absolute_path;
216307f5ae8SZachary Turner }
217