xref: /openbsd-src/gnu/llvm/clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp (revision 12c855180aad702bbcca06e0398d774beeafb155)
1e5dd7070Spatrick //===- ModuleDepCollector.cpp - Callbacks to collect deps -------*- C++ -*-===//
2e5dd7070Spatrick //
3*12c85518Srobert // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4*12c85518Srobert // See https://llvm.org/LICENSE.txt for license information.
5*12c85518Srobert // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6e5dd7070Spatrick //
7e5dd7070Spatrick //===----------------------------------------------------------------------===//
8e5dd7070Spatrick 
9e5dd7070Spatrick #include "clang/Tooling/DependencyScanning/ModuleDepCollector.h"
10e5dd7070Spatrick 
11*12c85518Srobert #include "clang/Basic/MakeSupport.h"
12e5dd7070Spatrick #include "clang/Frontend/CompilerInstance.h"
13e5dd7070Spatrick #include "clang/Lex/Preprocessor.h"
14e5dd7070Spatrick #include "clang/Tooling/DependencyScanning/DependencyScanningWorker.h"
15*12c85518Srobert #include "llvm/Support/BLAKE3.h"
16a9ac8606Spatrick #include "llvm/Support/StringSaver.h"
17*12c85518Srobert #include <optional>
18e5dd7070Spatrick 
19e5dd7070Spatrick using namespace clang;
20e5dd7070Spatrick using namespace tooling;
21e5dd7070Spatrick using namespace dependencies;
22e5dd7070Spatrick 
optimizeHeaderSearchOpts(HeaderSearchOptions & Opts,ASTReader & Reader,const serialization::ModuleFile & MF)23*12c85518Srobert static void optimizeHeaderSearchOpts(HeaderSearchOptions &Opts,
24*12c85518Srobert                                      ASTReader &Reader,
25*12c85518Srobert                                      const serialization::ModuleFile &MF) {
26*12c85518Srobert   // Only preserve search paths that were used during the dependency scan.
27*12c85518Srobert   std::vector<HeaderSearchOptions::Entry> Entries = Opts.UserEntries;
28*12c85518Srobert   Opts.UserEntries.clear();
29*12c85518Srobert 
30*12c85518Srobert   llvm::BitVector SearchPathUsage(Entries.size());
31*12c85518Srobert   llvm::DenseSet<const serialization::ModuleFile *> Visited;
32*12c85518Srobert   std::function<void(const serialization::ModuleFile *)> VisitMF =
33*12c85518Srobert       [&](const serialization::ModuleFile *MF) {
34*12c85518Srobert         SearchPathUsage |= MF->SearchPathUsage;
35*12c85518Srobert         Visited.insert(MF);
36*12c85518Srobert         for (const serialization::ModuleFile *Import : MF->Imports)
37*12c85518Srobert           if (!Visited.contains(Import))
38*12c85518Srobert             VisitMF(Import);
39*12c85518Srobert       };
40*12c85518Srobert   VisitMF(&MF);
41*12c85518Srobert 
42*12c85518Srobert   for (auto Idx : SearchPathUsage.set_bits())
43*12c85518Srobert     Opts.UserEntries.push_back(Entries[Idx]);
44*12c85518Srobert }
45*12c85518Srobert 
splitString(std::string S,char Separator)46*12c85518Srobert static std::vector<std::string> splitString(std::string S, char Separator) {
47*12c85518Srobert   SmallVector<StringRef> Segments;
48*12c85518Srobert   StringRef(S).split(Segments, Separator, /*MaxSplit=*/-1, /*KeepEmpty=*/false);
49*12c85518Srobert   std::vector<std::string> Result;
50*12c85518Srobert   Result.reserve(Segments.size());
51*12c85518Srobert   for (StringRef Segment : Segments)
52*12c85518Srobert     Result.push_back(Segment.str());
53*12c85518Srobert   return Result;
54*12c85518Srobert }
55*12c85518Srobert 
addOutputPaths(CompilerInvocation & CI,ModuleDeps & Deps)56*12c85518Srobert void ModuleDepCollector::addOutputPaths(CompilerInvocation &CI,
57*12c85518Srobert                                         ModuleDeps &Deps) {
58*12c85518Srobert   CI.getFrontendOpts().OutputFile =
59*12c85518Srobert       Consumer.lookupModuleOutput(Deps.ID, ModuleOutputKind::ModuleFile);
60*12c85518Srobert   if (!CI.getDiagnosticOpts().DiagnosticSerializationFile.empty())
61*12c85518Srobert     CI.getDiagnosticOpts().DiagnosticSerializationFile =
62*12c85518Srobert         Consumer.lookupModuleOutput(
63*12c85518Srobert             Deps.ID, ModuleOutputKind::DiagnosticSerializationFile);
64*12c85518Srobert   if (!CI.getDependencyOutputOpts().OutputFile.empty()) {
65*12c85518Srobert     CI.getDependencyOutputOpts().OutputFile =
66*12c85518Srobert         Consumer.lookupModuleOutput(Deps.ID, ModuleOutputKind::DependencyFile);
67*12c85518Srobert     CI.getDependencyOutputOpts().Targets =
68*12c85518Srobert         splitString(Consumer.lookupModuleOutput(
69*12c85518Srobert                         Deps.ID, ModuleOutputKind::DependencyTargets),
70*12c85518Srobert                     '\0');
71*12c85518Srobert     if (!CI.getDependencyOutputOpts().OutputFile.empty() &&
72*12c85518Srobert         CI.getDependencyOutputOpts().Targets.empty()) {
73*12c85518Srobert       // Fallback to -o as dependency target, as in the driver.
74*12c85518Srobert       SmallString<128> Target;
75*12c85518Srobert       quoteMakeTarget(CI.getFrontendOpts().OutputFile, Target);
76*12c85518Srobert       CI.getDependencyOutputOpts().Targets.push_back(std::string(Target));
77*12c85518Srobert     }
78*12c85518Srobert   }
79*12c85518Srobert }
80*12c85518Srobert 
81*12c85518Srobert CompilerInvocation
makeInvocationForModuleBuildWithoutOutputs(const ModuleDeps & Deps,llvm::function_ref<void (CompilerInvocation &)> Optimize) const82*12c85518Srobert ModuleDepCollector::makeInvocationForModuleBuildWithoutOutputs(
83*12c85518Srobert     const ModuleDeps &Deps,
84*12c85518Srobert     llvm::function_ref<void(CompilerInvocation &)> Optimize) const {
85a9ac8606Spatrick   // Make a deep copy of the original Clang invocation.
86a9ac8606Spatrick   CompilerInvocation CI(OriginalInvocation);
87ec727ea7Spatrick 
88*12c85518Srobert   CI.resetNonModularOptions();
89*12c85518Srobert   CI.clearImplicitModuleBuildOptions();
90*12c85518Srobert 
91*12c85518Srobert   // Remove options incompatible with explicit module build or are likely to
92*12c85518Srobert   // differ between identical modules discovered from different translation
93*12c85518Srobert   // units.
94a9ac8606Spatrick   CI.getFrontendOpts().Inputs.clear();
95a9ac8606Spatrick   CI.getFrontendOpts().OutputFile.clear();
96ec727ea7Spatrick 
97*12c85518Srobert   // TODO: Figure out better way to set options to their default value.
98*12c85518Srobert   CI.getCodeGenOpts().MainFileName.clear();
99*12c85518Srobert   CI.getCodeGenOpts().DwarfDebugFlags.clear();
100*12c85518Srobert   if (!CI.getLangOpts()->ModulesCodegen) {
101*12c85518Srobert     CI.getCodeGenOpts().DebugCompilationDir.clear();
102*12c85518Srobert     CI.getCodeGenOpts().CoverageCompilationDir.clear();
103*12c85518Srobert   }
104*12c85518Srobert 
105*12c85518Srobert   // Map output paths that affect behaviour to "-" so their existence is in the
106*12c85518Srobert   // context hash. The final path will be computed in addOutputPaths.
107*12c85518Srobert   if (!CI.getDiagnosticOpts().DiagnosticSerializationFile.empty())
108*12c85518Srobert     CI.getDiagnosticOpts().DiagnosticSerializationFile = "-";
109*12c85518Srobert   if (!CI.getDependencyOutputOpts().OutputFile.empty())
110*12c85518Srobert     CI.getDependencyOutputOpts().OutputFile = "-";
111*12c85518Srobert   CI.getDependencyOutputOpts().Targets.clear();
112*12c85518Srobert 
113a9ac8606Spatrick   CI.getFrontendOpts().ProgramAction = frontend::GenerateModule;
114a9ac8606Spatrick   CI.getLangOpts()->ModuleName = Deps.ID.ModuleName;
115a9ac8606Spatrick   CI.getFrontendOpts().IsSystemModule = Deps.IsSystem;
116ec727ea7Spatrick 
117*12c85518Srobert   // Inputs
118*12c85518Srobert   InputKind ModuleMapInputKind(CI.getFrontendOpts().DashX.getLanguage(),
119*12c85518Srobert                                InputKind::Format::ModuleMap);
120*12c85518Srobert   CI.getFrontendOpts().Inputs.emplace_back(Deps.ClangModuleMapFile,
121*12c85518Srobert                                            ModuleMapInputKind);
122a9ac8606Spatrick 
123*12c85518Srobert   auto CurrentModuleMapEntry =
124*12c85518Srobert       ScanInstance.getFileManager().getFile(Deps.ClangModuleMapFile);
125*12c85518Srobert   assert(CurrentModuleMapEntry && "module map file entry not found");
126*12c85518Srobert 
127*12c85518Srobert   auto DepModuleMapFiles = collectModuleMapFiles(Deps.ClangModuleDeps);
128*12c85518Srobert   for (StringRef ModuleMapFile : Deps.ModuleMapFileDeps) {
129*12c85518Srobert     // TODO: Track these as `FileEntryRef` to simplify the equality check below.
130*12c85518Srobert     auto ModuleMapEntry = ScanInstance.getFileManager().getFile(ModuleMapFile);
131*12c85518Srobert     assert(ModuleMapEntry && "module map file entry not found");
132*12c85518Srobert 
133*12c85518Srobert     // Don't report module maps describing eagerly-loaded dependency. This
134*12c85518Srobert     // information will be deserialized from the PCM.
135*12c85518Srobert     // TODO: Verify this works fine when modulemap for module A is eagerly
136*12c85518Srobert     // loaded from A.pcm, and module map passed on the command line contains
137*12c85518Srobert     // definition of a submodule: "explicit module A.Private { ... }".
138*12c85518Srobert     if (EagerLoadModules && DepModuleMapFiles.contains(*ModuleMapEntry))
139*12c85518Srobert       continue;
140*12c85518Srobert 
141*12c85518Srobert     // Don't report module map file of the current module unless it also
142*12c85518Srobert     // describes a dependency (for symmetry).
143*12c85518Srobert     if (*ModuleMapEntry == *CurrentModuleMapEntry &&
144*12c85518Srobert         !DepModuleMapFiles.contains(*ModuleMapEntry))
145*12c85518Srobert       continue;
146*12c85518Srobert 
147*12c85518Srobert     CI.getFrontendOpts().ModuleMapFiles.emplace_back(ModuleMapFile);
148ec727ea7Spatrick   }
149ec727ea7Spatrick 
150*12c85518Srobert   // Report the prebuilt modules this module uses.
151*12c85518Srobert   for (const auto &PrebuiltModule : Deps.PrebuiltModuleDeps)
152*12c85518Srobert     CI.getFrontendOpts().ModuleFiles.push_back(PrebuiltModule.PCMFile);
153*12c85518Srobert 
154*12c85518Srobert   // Add module file inputs from dependencies.
155*12c85518Srobert   addModuleFiles(CI, Deps.ClangModuleDeps);
156*12c85518Srobert 
157*12c85518Srobert   // Remove any macro definitions that are explicitly ignored.
158*12c85518Srobert   if (!CI.getHeaderSearchOpts().ModulesIgnoreMacros.empty()) {
159*12c85518Srobert     llvm::erase_if(
160*12c85518Srobert         CI.getPreprocessorOpts().Macros,
161*12c85518Srobert         [&CI](const std::pair<std::string, bool> &Def) {
162*12c85518Srobert           StringRef MacroDef = Def.first;
163*12c85518Srobert           return CI.getHeaderSearchOpts().ModulesIgnoreMacros.contains(
164*12c85518Srobert               llvm::CachedHashString(MacroDef.split('=').first));
165*12c85518Srobert         });
166*12c85518Srobert     // Remove the now unused option.
167*12c85518Srobert     CI.getHeaderSearchOpts().ModulesIgnoreMacros.clear();
168*12c85518Srobert   }
169*12c85518Srobert 
170*12c85518Srobert   Optimize(CI);
171a9ac8606Spatrick 
172a9ac8606Spatrick   return CI;
173a9ac8606Spatrick }
174a9ac8606Spatrick 
collectModuleMapFiles(ArrayRef<ModuleID> ClangModuleDeps) const175*12c85518Srobert llvm::DenseSet<const FileEntry *> ModuleDepCollector::collectModuleMapFiles(
176*12c85518Srobert     ArrayRef<ModuleID> ClangModuleDeps) const {
177*12c85518Srobert   llvm::DenseSet<const FileEntry *> ModuleMapFiles;
178*12c85518Srobert   for (const ModuleID &MID : ClangModuleDeps) {
179*12c85518Srobert     ModuleDeps *MD = ModuleDepsByID.lookup(MID);
180*12c85518Srobert     assert(MD && "Inconsistent dependency info");
181*12c85518Srobert     // TODO: Track ClangModuleMapFile as `FileEntryRef`.
182*12c85518Srobert     auto FE = ScanInstance.getFileManager().getFile(MD->ClangModuleMapFile);
183*12c85518Srobert     assert(FE && "Missing module map file that was previously found");
184*12c85518Srobert     ModuleMapFiles.insert(*FE);
185*12c85518Srobert   }
186*12c85518Srobert   return ModuleMapFiles;
187a9ac8606Spatrick }
188a9ac8606Spatrick 
addModuleMapFiles(CompilerInvocation & CI,ArrayRef<ModuleID> ClangModuleDeps) const189*12c85518Srobert void ModuleDepCollector::addModuleMapFiles(
190*12c85518Srobert     CompilerInvocation &CI, ArrayRef<ModuleID> ClangModuleDeps) const {
191*12c85518Srobert   if (EagerLoadModules)
192*12c85518Srobert     return; // Only pcm is needed for eager load.
193a9ac8606Spatrick 
194*12c85518Srobert   for (const ModuleID &MID : ClangModuleDeps) {
195*12c85518Srobert     ModuleDeps *MD = ModuleDepsByID.lookup(MID);
196*12c85518Srobert     assert(MD && "Inconsistent dependency info");
197*12c85518Srobert     CI.getFrontendOpts().ModuleMapFiles.push_back(MD->ClangModuleMapFile);
198*12c85518Srobert   }
199a9ac8606Spatrick }
200a9ac8606Spatrick 
addModuleFiles(CompilerInvocation & CI,ArrayRef<ModuleID> ClangModuleDeps) const201*12c85518Srobert void ModuleDepCollector::addModuleFiles(
202*12c85518Srobert     CompilerInvocation &CI, ArrayRef<ModuleID> ClangModuleDeps) const {
203*12c85518Srobert   for (const ModuleID &MID : ClangModuleDeps) {
204*12c85518Srobert     std::string PCMPath =
205*12c85518Srobert         Consumer.lookupModuleOutput(MID, ModuleOutputKind::ModuleFile);
206*12c85518Srobert     if (EagerLoadModules)
207*12c85518Srobert       CI.getFrontendOpts().ModuleFiles.push_back(std::move(PCMPath));
208*12c85518Srobert     else
209*12c85518Srobert       CI.getHeaderSearchOpts().PrebuiltModuleFiles.insert(
210*12c85518Srobert           {MID.ModuleName, std::move(PCMPath)});
211*12c85518Srobert   }
212a9ac8606Spatrick }
213a9ac8606Spatrick 
needsModules(FrontendInputFile FIF)214*12c85518Srobert static bool needsModules(FrontendInputFile FIF) {
215*12c85518Srobert   switch (FIF.getKind().getLanguage()) {
216*12c85518Srobert   case Language::Unknown:
217*12c85518Srobert   case Language::Asm:
218*12c85518Srobert   case Language::LLVM_IR:
219*12c85518Srobert     return false;
220*12c85518Srobert   default:
221*12c85518Srobert     return true;
222ec727ea7Spatrick   }
223*12c85518Srobert }
224ec727ea7Spatrick 
applyDiscoveredDependencies(CompilerInvocation & CI)225*12c85518Srobert void ModuleDepCollector::applyDiscoveredDependencies(CompilerInvocation &CI) {
226*12c85518Srobert   CI.clearImplicitModuleBuildOptions();
227*12c85518Srobert 
228*12c85518Srobert   if (llvm::any_of(CI.getFrontendOpts().Inputs, needsModules)) {
229*12c85518Srobert     Preprocessor &PP = ScanInstance.getPreprocessor();
230*12c85518Srobert     if (Module *CurrentModule = PP.getCurrentModuleImplementation())
231*12c85518Srobert       if (OptionalFileEntryRef CurrentModuleMap =
232*12c85518Srobert               PP.getHeaderSearchInfo()
233*12c85518Srobert                   .getModuleMap()
234*12c85518Srobert                   .getModuleMapFileForUniquing(CurrentModule))
235*12c85518Srobert         CI.getFrontendOpts().ModuleMapFiles.emplace_back(
236*12c85518Srobert             CurrentModuleMap->getName());
237*12c85518Srobert 
238*12c85518Srobert     SmallVector<ModuleID> DirectDeps;
239*12c85518Srobert     for (const auto &KV : ModularDeps)
240*12c85518Srobert       if (KV.second->ImportedByMainFile)
241*12c85518Srobert         DirectDeps.push_back(KV.second->ID);
242*12c85518Srobert 
243*12c85518Srobert     // TODO: Report module maps the same way it's done for modular dependencies.
244*12c85518Srobert     addModuleMapFiles(CI, DirectDeps);
245*12c85518Srobert 
246*12c85518Srobert     addModuleFiles(CI, DirectDeps);
247*12c85518Srobert 
248*12c85518Srobert     for (const auto &KV : DirectPrebuiltModularDeps)
249*12c85518Srobert       CI.getFrontendOpts().ModuleFiles.push_back(KV.second.PCMFile);
250*12c85518Srobert   }
251*12c85518Srobert }
252*12c85518Srobert 
getModuleContextHash(const ModuleDeps & MD,const CompilerInvocation & CI,bool EagerLoadModules)253*12c85518Srobert static std::string getModuleContextHash(const ModuleDeps &MD,
254*12c85518Srobert                                         const CompilerInvocation &CI,
255*12c85518Srobert                                         bool EagerLoadModules) {
256*12c85518Srobert   llvm::HashBuilder<llvm::TruncatedBLAKE3<16>,
257*12c85518Srobert                     llvm::support::endianness::native>
258*12c85518Srobert       HashBuilder;
259*12c85518Srobert   SmallString<32> Scratch;
260*12c85518Srobert 
261*12c85518Srobert   // Hash the compiler version and serialization version to ensure the module
262*12c85518Srobert   // will be readable.
263*12c85518Srobert   HashBuilder.add(getClangFullRepositoryVersion());
264*12c85518Srobert   HashBuilder.add(serialization::VERSION_MAJOR, serialization::VERSION_MINOR);
265*12c85518Srobert 
266*12c85518Srobert   // Hash the BuildInvocation without any input files.
267*12c85518Srobert   SmallVector<const char *, 32> DummyArgs;
268*12c85518Srobert   CI.generateCC1CommandLine(DummyArgs, [&](const Twine &Arg) {
269*12c85518Srobert     Scratch.clear();
270*12c85518Srobert     StringRef Str = Arg.toStringRef(Scratch);
271*12c85518Srobert     HashBuilder.add(Str);
272*12c85518Srobert     return "<unused>";
273*12c85518Srobert   });
274*12c85518Srobert 
275*12c85518Srobert   // Hash the module dependencies. These paths may differ even if the invocation
276*12c85518Srobert   // is identical if they depend on the contents of the files in the TU -- for
277*12c85518Srobert   // example, case-insensitive paths to modulemap files. Usually such a case
278*12c85518Srobert   // would indicate a missed optimization to canonicalize, but it may be
279*12c85518Srobert   // difficult to canonicalize all cases when there is a VFS.
280*12c85518Srobert   for (const auto &ID : MD.ClangModuleDeps) {
281*12c85518Srobert     HashBuilder.add(ID.ModuleName);
282*12c85518Srobert     HashBuilder.add(ID.ContextHash);
283*12c85518Srobert   }
284*12c85518Srobert 
285*12c85518Srobert   HashBuilder.add(EagerLoadModules);
286*12c85518Srobert 
287*12c85518Srobert   llvm::BLAKE3Result<16> Hash = HashBuilder.final();
288*12c85518Srobert   std::array<uint64_t, 2> Words;
289*12c85518Srobert   static_assert(sizeof(Hash) == sizeof(Words), "Hash must match Words");
290*12c85518Srobert   std::memcpy(Words.data(), Hash.data(), sizeof(Hash));
291*12c85518Srobert   return toString(llvm::APInt(sizeof(Words) * 8, Words), 36, /*Signed=*/false);
292*12c85518Srobert }
293*12c85518Srobert 
associateWithContextHash(const CompilerInvocation & CI,ModuleDeps & Deps)294*12c85518Srobert void ModuleDepCollector::associateWithContextHash(const CompilerInvocation &CI,
295*12c85518Srobert                                                   ModuleDeps &Deps) {
296*12c85518Srobert   Deps.ID.ContextHash = getModuleContextHash(Deps, CI, EagerLoadModules);
297*12c85518Srobert   bool Inserted = ModuleDepsByID.insert({Deps.ID, &Deps}).second;
298*12c85518Srobert   (void)Inserted;
299*12c85518Srobert   assert(Inserted && "duplicate module mapping");
300ec727ea7Spatrick }
301ec727ea7Spatrick 
FileChanged(SourceLocation Loc,FileChangeReason Reason,SrcMgr::CharacteristicKind FileType,FileID PrevFID)302e5dd7070Spatrick void ModuleDepCollectorPP::FileChanged(SourceLocation Loc,
303e5dd7070Spatrick                                        FileChangeReason Reason,
304e5dd7070Spatrick                                        SrcMgr::CharacteristicKind FileType,
305e5dd7070Spatrick                                        FileID PrevFID) {
306e5dd7070Spatrick   if (Reason != PPCallbacks::EnterFile)
307e5dd7070Spatrick     return;
308e5dd7070Spatrick 
309ec727ea7Spatrick   // This has to be delayed as the context hash can change at the start of
310ec727ea7Spatrick   // `CompilerInstance::ExecuteAction`.
311ec727ea7Spatrick   if (MDC.ContextHash.empty()) {
312*12c85518Srobert     MDC.ContextHash = MDC.ScanInstance.getInvocation().getModuleHash();
313ec727ea7Spatrick     MDC.Consumer.handleContextHash(MDC.ContextHash);
314ec727ea7Spatrick   }
315ec727ea7Spatrick 
316*12c85518Srobert   SourceManager &SM = MDC.ScanInstance.getSourceManager();
317e5dd7070Spatrick 
318e5dd7070Spatrick   // Dependency generation really does want to go all the way to the
319e5dd7070Spatrick   // file entry for a source location to find out what is depended on.
320e5dd7070Spatrick   // We do not want #line markers to affect dependency generation!
321*12c85518Srobert   if (std::optional<StringRef> Filename =
322a9ac8606Spatrick           SM.getNonBuiltinFilenameForID(SM.getFileID(SM.getExpansionLoc(Loc))))
323*12c85518Srobert     MDC.addFileDep(llvm::sys::path::remove_leading_dotslash(*Filename));
324e5dd7070Spatrick }
325e5dd7070Spatrick 
InclusionDirective(SourceLocation HashLoc,const Token & IncludeTok,StringRef FileName,bool IsAngled,CharSourceRange FilenameRange,OptionalFileEntryRef File,StringRef SearchPath,StringRef RelativePath,const Module * Imported,SrcMgr::CharacteristicKind FileType)326e5dd7070Spatrick void ModuleDepCollectorPP::InclusionDirective(
327e5dd7070Spatrick     SourceLocation HashLoc, const Token &IncludeTok, StringRef FileName,
328*12c85518Srobert     bool IsAngled, CharSourceRange FilenameRange, OptionalFileEntryRef File,
329e5dd7070Spatrick     StringRef SearchPath, StringRef RelativePath, const Module *Imported,
330e5dd7070Spatrick     SrcMgr::CharacteristicKind FileType) {
331e5dd7070Spatrick   if (!File && !Imported) {
332e5dd7070Spatrick     // This is a non-modular include that HeaderSearch failed to find. Add it
333e5dd7070Spatrick     // here as `FileChanged` will never see it.
334*12c85518Srobert     MDC.addFileDep(FileName);
335ec727ea7Spatrick   }
336ec727ea7Spatrick   handleImport(Imported);
337e5dd7070Spatrick }
338e5dd7070Spatrick 
moduleImport(SourceLocation ImportLoc,ModuleIdPath Path,const Module * Imported)339ec727ea7Spatrick void ModuleDepCollectorPP::moduleImport(SourceLocation ImportLoc,
340ec727ea7Spatrick                                         ModuleIdPath Path,
341ec727ea7Spatrick                                         const Module *Imported) {
342*12c85518Srobert   if (MDC.ScanInstance.getPreprocessor().isInImportingCXXNamedModules()) {
343*12c85518Srobert     P1689ModuleInfo RequiredModule;
344*12c85518Srobert     RequiredModule.ModuleName = Path[0].first->getName().str();
345*12c85518Srobert     RequiredModule.Type = P1689ModuleInfo::ModuleType::NamedCXXModule;
346*12c85518Srobert     MDC.RequiredStdCXXModules.push_back(RequiredModule);
347*12c85518Srobert     return;
348*12c85518Srobert   }
349*12c85518Srobert 
350ec727ea7Spatrick   handleImport(Imported);
351ec727ea7Spatrick }
352ec727ea7Spatrick 
handleImport(const Module * Imported)353ec727ea7Spatrick void ModuleDepCollectorPP::handleImport(const Module *Imported) {
354e5dd7070Spatrick   if (!Imported)
355e5dd7070Spatrick     return;
356e5dd7070Spatrick 
357a9ac8606Spatrick   const Module *TopLevelModule = Imported->getTopLevelModule();
358a9ac8606Spatrick 
359a9ac8606Spatrick   if (MDC.isPrebuiltModule(TopLevelModule))
360*12c85518Srobert     MDC.DirectPrebuiltModularDeps.insert(
361*12c85518Srobert         {TopLevelModule, PrebuiltModuleDep{TopLevelModule}});
362a9ac8606Spatrick   else
363a9ac8606Spatrick     DirectModularDeps.insert(TopLevelModule);
364e5dd7070Spatrick }
365e5dd7070Spatrick 
EndOfMainFile()366e5dd7070Spatrick void ModuleDepCollectorPP::EndOfMainFile() {
367*12c85518Srobert   FileID MainFileID = MDC.ScanInstance.getSourceManager().getMainFileID();
368*12c85518Srobert   MDC.MainFile = std::string(MDC.ScanInstance.getSourceManager()
369*12c85518Srobert                                  .getFileEntryForID(MainFileID)
370*12c85518Srobert                                  ->getName());
371e5dd7070Spatrick 
372*12c85518Srobert   auto &PP = MDC.ScanInstance.getPreprocessor();
373*12c85518Srobert   if (PP.isInNamedModule()) {
374*12c85518Srobert     P1689ModuleInfo ProvidedModule;
375*12c85518Srobert     ProvidedModule.ModuleName = PP.getNamedModuleName();
376*12c85518Srobert     ProvidedModule.Type = P1689ModuleInfo::ModuleType::NamedCXXModule;
377*12c85518Srobert     ProvidedModule.IsStdCXXModuleInterface = PP.isInNamedInterfaceUnit();
378*12c85518Srobert     // Don't put implementation (non partition) unit as Provide.
379*12c85518Srobert     // Put the module as required instead. Since the implementation
380*12c85518Srobert     // unit will import the primary module implicitly.
381*12c85518Srobert     if (PP.isInImplementationUnit())
382*12c85518Srobert       MDC.RequiredStdCXXModules.push_back(ProvidedModule);
383*12c85518Srobert     else
384*12c85518Srobert       MDC.ProvidedStdCXXModule = ProvidedModule;
385*12c85518Srobert   }
386a9ac8606Spatrick 
387*12c85518Srobert   if (!MDC.ScanInstance.getPreprocessorOpts().ImplicitPCHInclude.empty())
388*12c85518Srobert     MDC.addFileDep(MDC.ScanInstance.getPreprocessorOpts().ImplicitPCHInclude);
389*12c85518Srobert 
390*12c85518Srobert   for (const Module *M :
391*12c85518Srobert        MDC.ScanInstance.getPreprocessor().getAffectingClangModules())
392*12c85518Srobert     if (!MDC.isPrebuiltModule(M))
393*12c85518Srobert       DirectModularDeps.insert(M);
394*12c85518Srobert 
395*12c85518Srobert   for (const Module *M : DirectModularDeps)
396*12c85518Srobert     handleTopLevelModule(M);
397*12c85518Srobert 
398*12c85518Srobert   MDC.Consumer.handleDependencyOutputOpts(*MDC.Opts);
399*12c85518Srobert 
400*12c85518Srobert   if (MDC.IsStdModuleP1689Format)
401*12c85518Srobert     MDC.Consumer.handleProvidedAndRequiredStdCXXModules(
402*12c85518Srobert         MDC.ProvidedStdCXXModule, MDC.RequiredStdCXXModules);
403*12c85518Srobert 
404*12c85518Srobert   for (auto &&I : MDC.ModularDeps)
405*12c85518Srobert     MDC.Consumer.handleModuleDependency(*I.second);
406*12c85518Srobert 
407*12c85518Srobert   for (auto &&I : MDC.FileDeps)
408*12c85518Srobert     MDC.Consumer.handleFileDependency(I);
409*12c85518Srobert 
410*12c85518Srobert   for (auto &&I : MDC.DirectPrebuiltModularDeps)
411*12c85518Srobert     MDC.Consumer.handlePrebuiltModuleDependency(I.second);
412*12c85518Srobert }
413*12c85518Srobert 
414*12c85518Srobert std::optional<ModuleID>
handleTopLevelModule(const Module * M)415*12c85518Srobert ModuleDepCollectorPP::handleTopLevelModule(const Module *M) {
416*12c85518Srobert   assert(M == M->getTopLevelModule() && "Expected top level module!");
417*12c85518Srobert 
418a9ac8606Spatrick   // A top-level module might not be actually imported as a module when
419a9ac8606Spatrick   // -fmodule-name is used to compile a translation unit that imports this
420a9ac8606Spatrick   // module. In that case it can be skipped. The appropriate header
421a9ac8606Spatrick   // dependencies will still be reported as expected.
422a9ac8606Spatrick   if (!M->getASTFile())
423*12c85518Srobert     return {};
424e5dd7070Spatrick 
425a9ac8606Spatrick   // If this module has been handled already, just return its ID.
426*12c85518Srobert   auto ModI = MDC.ModularDeps.insert({M, nullptr});
427a9ac8606Spatrick   if (!ModI.second)
428*12c85518Srobert     return ModI.first->second->ID;
429e5dd7070Spatrick 
430*12c85518Srobert   ModI.first->second = std::make_unique<ModuleDeps>();
431*12c85518Srobert   ModuleDeps &MD = *ModI.first->second;
432e5dd7070Spatrick 
433a9ac8606Spatrick   MD.ID.ModuleName = M->getFullModuleName();
434a9ac8606Spatrick   MD.ImportedByMainFile = DirectModularDeps.contains(M);
435a9ac8606Spatrick   MD.IsSystem = M->IsSystem;
436a9ac8606Spatrick 
437*12c85518Srobert   ModuleMap &ModMapInfo =
438*12c85518Srobert       MDC.ScanInstance.getPreprocessor().getHeaderSearchInfo().getModuleMap();
439*12c85518Srobert 
440*12c85518Srobert   OptionalFileEntryRef ModuleMap = ModMapInfo.getModuleMapFileForUniquing(M);
441*12c85518Srobert 
442*12c85518Srobert   if (ModuleMap) {
443*12c85518Srobert     SmallString<128> Path = ModuleMap->getNameAsRequested();
444*12c85518Srobert     ModMapInfo.canonicalizeModuleMapPath(Path);
445*12c85518Srobert     MD.ClangModuleMapFile = std::string(Path);
446*12c85518Srobert   }
447a9ac8606Spatrick 
448e5dd7070Spatrick   serialization::ModuleFile *MF =
449*12c85518Srobert       MDC.ScanInstance.getASTReader()->getModuleManager().lookup(
450*12c85518Srobert           M->getASTFile());
451*12c85518Srobert   MDC.ScanInstance.getASTReader()->visitInputFiles(
452e5dd7070Spatrick       *MF, true, true, [&](const serialization::InputFile &IF, bool isSystem) {
453a9ac8606Spatrick         // __inferred_module.map is the result of the way in which an implicit
454a9ac8606Spatrick         // module build handles inferred modules. It adds an overlay VFS with
455a9ac8606Spatrick         // this file in the proper directory and relies on the rest of Clang to
456a9ac8606Spatrick         // handle it like normal. With explicitly built modules we don't need
457a9ac8606Spatrick         // to play VFS tricks, so replace it with the correct module map.
458a9ac8606Spatrick         if (IF.getFile()->getName().endswith("__inferred_module.map")) {
459*12c85518Srobert           MDC.addFileDep(MD, ModuleMap->getName());
460a9ac8606Spatrick           return;
461a9ac8606Spatrick         }
462*12c85518Srobert         MDC.addFileDep(MD, IF.getFile()->getName());
463e5dd7070Spatrick       });
464e5dd7070Spatrick 
465*12c85518Srobert   llvm::DenseSet<const Module *> SeenDeps;
466*12c85518Srobert   addAllSubmodulePrebuiltDeps(M, MD, SeenDeps);
467*12c85518Srobert   addAllSubmoduleDeps(M, MD, SeenDeps);
468*12c85518Srobert   addAllAffectingClangModules(M, MD, SeenDeps);
469a9ac8606Spatrick 
470*12c85518Srobert   MDC.ScanInstance.getASTReader()->visitTopLevelModuleMaps(
471*12c85518Srobert       *MF, [&](FileEntryRef FE) {
472*12c85518Srobert         if (FE.getNameAsRequested().endswith("__inferred_module.map"))
473*12c85518Srobert           return;
474*12c85518Srobert         MD.ModuleMapFileDeps.emplace_back(FE.getNameAsRequested());
475*12c85518Srobert       });
476a9ac8606Spatrick 
477*12c85518Srobert   CompilerInvocation CI = MDC.makeInvocationForModuleBuildWithoutOutputs(
478*12c85518Srobert       MD, [&](CompilerInvocation &BuildInvocation) {
479*12c85518Srobert         if (MDC.OptimizeArgs)
480*12c85518Srobert           optimizeHeaderSearchOpts(BuildInvocation.getHeaderSearchOpts(),
481*12c85518Srobert                                    *MDC.ScanInstance.getASTReader(), *MF);
482*12c85518Srobert       });
483*12c85518Srobert 
484*12c85518Srobert   MDC.associateWithContextHash(CI, MD);
485*12c85518Srobert 
486*12c85518Srobert   // Finish the compiler invocation. Requires dependencies and the context hash.
487*12c85518Srobert   MDC.addOutputPaths(CI, MD);
488*12c85518Srobert 
489*12c85518Srobert   MD.BuildArguments = CI.getCC1CommandLine();
490a9ac8606Spatrick 
491a9ac8606Spatrick   return MD.ID;
492a9ac8606Spatrick }
493a9ac8606Spatrick 
forEachSubmoduleSorted(const Module * M,llvm::function_ref<void (const Module *)> F)494*12c85518Srobert static void forEachSubmoduleSorted(const Module *M,
495*12c85518Srobert                                    llvm::function_ref<void(const Module *)> F) {
496*12c85518Srobert   // Submodule order depends on order of header includes for inferred submodules
497*12c85518Srobert   // we don't care about the exact order, so sort so that it's consistent across
498*12c85518Srobert   // TUs to improve sharing.
499*12c85518Srobert   SmallVector<const Module *> Submodules(M->submodule_begin(),
500*12c85518Srobert                                          M->submodule_end());
501*12c85518Srobert   llvm::stable_sort(Submodules, [](const Module *A, const Module *B) {
502*12c85518Srobert     return A->Name < B->Name;
503*12c85518Srobert   });
504*12c85518Srobert   for (const Module *SubM : Submodules)
505*12c85518Srobert     F(SubM);
506*12c85518Srobert }
507*12c85518Srobert 
addAllSubmodulePrebuiltDeps(const Module * M,ModuleDeps & MD,llvm::DenseSet<const Module * > & SeenSubmodules)508*12c85518Srobert void ModuleDepCollectorPP::addAllSubmodulePrebuiltDeps(
509*12c85518Srobert     const Module *M, ModuleDeps &MD,
510*12c85518Srobert     llvm::DenseSet<const Module *> &SeenSubmodules) {
511*12c85518Srobert   addModulePrebuiltDeps(M, MD, SeenSubmodules);
512*12c85518Srobert 
513*12c85518Srobert   forEachSubmoduleSorted(M, [&](const Module *SubM) {
514*12c85518Srobert     addAllSubmodulePrebuiltDeps(SubM, MD, SeenSubmodules);
515*12c85518Srobert   });
516*12c85518Srobert }
517*12c85518Srobert 
addModulePrebuiltDeps(const Module * M,ModuleDeps & MD,llvm::DenseSet<const Module * > & SeenSubmodules)518*12c85518Srobert void ModuleDepCollectorPP::addModulePrebuiltDeps(
519*12c85518Srobert     const Module *M, ModuleDeps &MD,
520*12c85518Srobert     llvm::DenseSet<const Module *> &SeenSubmodules) {
521a9ac8606Spatrick   for (const Module *Import : M->Imports)
522a9ac8606Spatrick     if (Import->getTopLevelModule() != M->getTopLevelModule())
523*12c85518Srobert       if (MDC.isPrebuiltModule(Import->getTopLevelModule()))
524*12c85518Srobert         if (SeenSubmodules.insert(Import->getTopLevelModule()).second)
525*12c85518Srobert           MD.PrebuiltModuleDeps.emplace_back(Import->getTopLevelModule());
526e5dd7070Spatrick }
527e5dd7070Spatrick 
addAllSubmoduleDeps(const Module * M,ModuleDeps & MD,llvm::DenseSet<const Module * > & AddedModules)528ec727ea7Spatrick void ModuleDepCollectorPP::addAllSubmoduleDeps(
529ec727ea7Spatrick     const Module *M, ModuleDeps &MD,
530ec727ea7Spatrick     llvm::DenseSet<const Module *> &AddedModules) {
531ec727ea7Spatrick   addModuleDep(M, MD, AddedModules);
532e5dd7070Spatrick 
533*12c85518Srobert   forEachSubmoduleSorted(M, [&](const Module *SubM) {
534ec727ea7Spatrick     addAllSubmoduleDeps(SubM, MD, AddedModules);
535*12c85518Srobert   });
536e5dd7070Spatrick }
537e5dd7070Spatrick 
addModuleDep(const Module * M,ModuleDeps & MD,llvm::DenseSet<const Module * > & AddedModules)538ec727ea7Spatrick void ModuleDepCollectorPP::addModuleDep(
539ec727ea7Spatrick     const Module *M, ModuleDeps &MD,
540ec727ea7Spatrick     llvm::DenseSet<const Module *> &AddedModules) {
541e5dd7070Spatrick   for (const Module *Import : M->Imports) {
542a9ac8606Spatrick     if (Import->getTopLevelModule() != M->getTopLevelModule() &&
543a9ac8606Spatrick         !MDC.isPrebuiltModule(Import)) {
544*12c85518Srobert       if (auto ImportID = handleTopLevelModule(Import->getTopLevelModule()))
545ec727ea7Spatrick         if (AddedModules.insert(Import->getTopLevelModule()).second)
546*12c85518Srobert           MD.ClangModuleDeps.push_back(*ImportID);
547*12c85518Srobert     }
548*12c85518Srobert   }
549*12c85518Srobert }
550*12c85518Srobert 
addAllAffectingClangModules(const Module * M,ModuleDeps & MD,llvm::DenseSet<const Module * > & AddedModules)551*12c85518Srobert void ModuleDepCollectorPP::addAllAffectingClangModules(
552*12c85518Srobert     const Module *M, ModuleDeps &MD,
553*12c85518Srobert     llvm::DenseSet<const Module *> &AddedModules) {
554*12c85518Srobert   addAffectingClangModule(M, MD, AddedModules);
555*12c85518Srobert 
556*12c85518Srobert   for (const Module *SubM : M->submodules())
557*12c85518Srobert     addAllAffectingClangModules(SubM, MD, AddedModules);
558*12c85518Srobert }
559*12c85518Srobert 
addAffectingClangModule(const Module * M,ModuleDeps & MD,llvm::DenseSet<const Module * > & AddedModules)560*12c85518Srobert void ModuleDepCollectorPP::addAffectingClangModule(
561*12c85518Srobert     const Module *M, ModuleDeps &MD,
562*12c85518Srobert     llvm::DenseSet<const Module *> &AddedModules) {
563*12c85518Srobert   for (const Module *Affecting : M->AffectingClangModules) {
564*12c85518Srobert     assert(Affecting == Affecting->getTopLevelModule() &&
565*12c85518Srobert            "Not quite import not top-level module");
566*12c85518Srobert     if (Affecting != M->getTopLevelModule() &&
567*12c85518Srobert         !MDC.isPrebuiltModule(Affecting)) {
568*12c85518Srobert       if (auto ImportID = handleTopLevelModule(Affecting))
569*12c85518Srobert         if (AddedModules.insert(Affecting).second)
570*12c85518Srobert           MD.ClangModuleDeps.push_back(*ImportID);
571e5dd7070Spatrick     }
572e5dd7070Spatrick   }
573e5dd7070Spatrick }
574e5dd7070Spatrick 
ModuleDepCollector(std::unique_ptr<DependencyOutputOptions> Opts,CompilerInstance & ScanInstance,DependencyConsumer & C,CompilerInvocation OriginalCI,bool OptimizeArgs,bool EagerLoadModules,bool IsStdModuleP1689Format)575ec727ea7Spatrick ModuleDepCollector::ModuleDepCollector(
576*12c85518Srobert     std::unique_ptr<DependencyOutputOptions> Opts,
577*12c85518Srobert     CompilerInstance &ScanInstance, DependencyConsumer &C,
578*12c85518Srobert     CompilerInvocation OriginalCI, bool OptimizeArgs, bool EagerLoadModules,
579*12c85518Srobert     bool IsStdModuleP1689Format)
580*12c85518Srobert     : ScanInstance(ScanInstance), Consumer(C), Opts(std::move(Opts)),
581*12c85518Srobert       OriginalInvocation(std::move(OriginalCI)), OptimizeArgs(OptimizeArgs),
582*12c85518Srobert       EagerLoadModules(EagerLoadModules),
583*12c85518Srobert       IsStdModuleP1689Format(IsStdModuleP1689Format) {}
584e5dd7070Spatrick 
attachToPreprocessor(Preprocessor & PP)585e5dd7070Spatrick void ModuleDepCollector::attachToPreprocessor(Preprocessor &PP) {
586*12c85518Srobert   PP.addPPCallbacks(std::make_unique<ModuleDepCollectorPP>(*this));
587e5dd7070Spatrick }
588e5dd7070Spatrick 
attachToASTReader(ASTReader & R)589e5dd7070Spatrick void ModuleDepCollector::attachToASTReader(ASTReader &R) {}
590a9ac8606Spatrick 
isPrebuiltModule(const Module * M)591a9ac8606Spatrick bool ModuleDepCollector::isPrebuiltModule(const Module *M) {
592a9ac8606Spatrick   std::string Name(M->getTopLevelModuleName());
593a9ac8606Spatrick   const auto &PrebuiltModuleFiles =
594*12c85518Srobert       ScanInstance.getHeaderSearchOpts().PrebuiltModuleFiles;
595a9ac8606Spatrick   auto PrebuiltModuleFileIt = PrebuiltModuleFiles.find(Name);
596a9ac8606Spatrick   if (PrebuiltModuleFileIt == PrebuiltModuleFiles.end())
597a9ac8606Spatrick     return false;
598a9ac8606Spatrick   assert("Prebuilt module came from the expected AST file" &&
599a9ac8606Spatrick          PrebuiltModuleFileIt->second == M->getASTFile()->getName());
600a9ac8606Spatrick   return true;
601a9ac8606Spatrick }
602*12c85518Srobert 
makeAbsoluteAndPreferred(CompilerInstance & CI,StringRef Path,SmallVectorImpl<char> & Storage)603*12c85518Srobert static StringRef makeAbsoluteAndPreferred(CompilerInstance &CI, StringRef Path,
604*12c85518Srobert                                           SmallVectorImpl<char> &Storage) {
605*12c85518Srobert   if (llvm::sys::path::is_absolute(Path) &&
606*12c85518Srobert       !llvm::sys::path::is_style_windows(llvm::sys::path::Style::native))
607*12c85518Srobert     return Path;
608*12c85518Srobert   Storage.assign(Path.begin(), Path.end());
609*12c85518Srobert   CI.getFileManager().makeAbsolutePath(Storage);
610*12c85518Srobert   llvm::sys::path::make_preferred(Storage);
611*12c85518Srobert   return StringRef(Storage.data(), Storage.size());
612*12c85518Srobert }
613*12c85518Srobert 
addFileDep(StringRef Path)614*12c85518Srobert void ModuleDepCollector::addFileDep(StringRef Path) {
615*12c85518Srobert   llvm::SmallString<256> Storage;
616*12c85518Srobert   Path = makeAbsoluteAndPreferred(ScanInstance, Path, Storage);
617*12c85518Srobert   FileDeps.push_back(std::string(Path));
618*12c85518Srobert }
619*12c85518Srobert 
addFileDep(ModuleDeps & MD,StringRef Path)620*12c85518Srobert void ModuleDepCollector::addFileDep(ModuleDeps &MD, StringRef Path) {
621*12c85518Srobert   llvm::SmallString<256> Storage;
622*12c85518Srobert   Path = makeAbsoluteAndPreferred(ScanInstance, Path, Storage);
623*12c85518Srobert   MD.FileDeps.insert(Path);
624*12c85518Srobert }
625