[Dependency Scanning] Add required additional C++ interop overlay module queries

Author: artemcm

include/swift/AST/ModuleDependencies.h

@@ -403,6 +403,9 @@ class ClangModuleDependencyStorage : public ModuleDependencyInfoStorageBase {
   /// CASID for the Root of ClangIncludeTree. Empty if not used.
   std::string CASClangIncludeTreeRootID;
 
+  /// Whether this is a "system" module.
+  bool IsSystem;
+
   ClangModuleDependencyStorage(const std::string &pcmOutputPath,
                                const std::string &mappedPCMPath,
                                const std::string &moduleMapFile,
@@ -412,15 +415,16 @@ class ClangModuleDependencyStorage : public ModuleDependencyInfoStorageBase {
                                const std::vector<std::string> &capturedPCMArgs,
                                const std::string &CASFileSystemRootID,
                                const std::string &clangIncludeTreeRoot,
-                               const std::string &moduleCacheKey)
+                               const std::string &moduleCacheKey,
+                               bool IsSystem)
       : ModuleDependencyInfoStorageBase(ModuleDependencyKind::Clang,
                                         moduleCacheKey),
         pcmOutputPath(pcmOutputPath), mappedPCMPath(mappedPCMPath),
         moduleMapFile(moduleMapFile), contextHash(contextHash),
         buildCommandLine(buildCommandLine), fileDependencies(fileDependencies),
         capturedPCMArgs(capturedPCMArgs),
         CASFileSystemRootID(CASFileSystemRootID),
-        CASClangIncludeTreeRootID(clangIncludeTreeRoot) {}
+        CASClangIncludeTreeRootID(clangIncludeTreeRoot), IsSystem(IsSystem) {}
 
   ModuleDependencyInfoStorageBase *clone() const override {
     return new ClangModuleDependencyStorage(*this);
@@ -549,11 +553,12 @@ class ModuleDependencyInfo {
       const std::vector<std::string> &capturedPCMArgs,
       const std::string &CASFileSystemRootID,
       const std::string &clangIncludeTreeRoot,
-      const std::string &moduleCacheKey) {
+      const std::string &moduleCacheKey,
+      bool IsSystem) {
     return ModuleDependencyInfo(std::make_unique<ClangModuleDependencyStorage>(
         pcmOutputPath, mappedPCMPath, moduleMapFile, contextHash,
         nonPathCommandLine, fileDependencies, capturedPCMArgs,
-        CASFileSystemRootID, clangIncludeTreeRoot, moduleCacheKey));
+        CASFileSystemRootID, clangIncludeTreeRoot, moduleCacheKey, IsSystem));
   }
 
   /// Describe a placeholder dependency swift module.

include/swift/ClangImporter/ClangImporter.h

@@ -665,6 +665,11 @@ bool requiresCPlusPlus(const clang::Module *module);
 /// (std_vector, std_iosfwd, etc).
 bool isCxxStdModule(const clang::Module *module);
 
+/// Returns true if the given module is one of the C++ standard library modules.
+/// This could be the top-level std module, or any of the libc++ split modules
+/// (std_vector, std_iosfwd, etc).
+bool isCxxStdModule(StringRef moduleName, bool IsSystem);
+
 /// Returns the pointee type if the given type is a C++ `const`
 /// reference type, `None` otherwise.
 std::optional<clang::QualType>

include/swift/DependencyScan/SerializedModuleDependencyCacheFormat.h

@@ -53,6 +53,8 @@ using ContextHashIDField = IdentifierIDField;
 
 /// A bit that indicates whether or not a module is a framework
 using IsFrameworkField = BCFixed<1>;
+/// A bit that indicates whether or not a module is a system module
+using IsSystemField = BCFixed<1>;
 
 /// Arrays of various identifiers, distinguished for readability
 using IdentifierIDArryField = llvm::BCArray<IdentifierIDField>;
@@ -193,7 +195,8 @@ using ClangModuleDetailsLayout =
                    FlagIDArrayIDField,        // capturedPCMArgs
                    IdentifierIDField,         // CASFileSystemRootID
                    IdentifierIDField,         // clangIncludeTreeRoot
-                   IdentifierIDField          // moduleCacheKey
+                   IdentifierIDField,         // moduleCacheKey
+                   IsSystemField              // isSystem
                    >;
 } // namespace graph_block
 

lib/ClangImporter/ClangImporter.cpp

@@ -7894,14 +7894,20 @@ bool importer::requiresCPlusPlus(const clang::Module *module) {
 }
 
 bool importer::isCxxStdModule(const clang::Module *module) {
-  if (module->getTopLevelModuleName() == "std")
+  return isCxxStdModule(module->getTopLevelModuleName(),
+                        module->getTopLevelModule()->IsSystem);
+}
+
+bool importer::isCxxStdModule(StringRef moduleName, bool IsSystem) {
+  if (moduleName == "std")
     return true;
   // In recent libc++ versions the module is split into multiple top-level
   // modules (std_vector, std_utility, etc).
-  if (module->getTopLevelModule()->IsSystem &&
-      module->getTopLevelModuleName().starts_with("std_"))
+  if (IsSystem && moduleName.starts_with("std_")) {
+    if (moduleName == "std_errno_h")
+      return false;
     return true;
-
+  }
   return false;
 }
 

lib/ClangImporter/ClangModuleDependencyScanner.cpp

@@ -289,7 +289,7 @@ ModuleDependencyVector ClangImporter::bridgeClangModuleDependencies(
     auto dependencies = ModuleDependencyInfo::forClangModule(
         pcmPath, mappedPCMPath, clangModuleDep.ClangModuleMapFile,
         clangModuleDep.ID.ContextHash, swiftArgs, fileDeps, capturedPCMArgs,
-        RootID, IncludeTree, /*module-cache-key*/ "");
+        RootID, IncludeTree, /*module-cache-key*/ "", clangModuleDep.IsSystem);
     for (const auto &moduleName : clangModuleDep.ClangModuleDeps) {
       dependencies.addModuleImport(moduleName.ModuleName, &alreadyAddedModules);
       // It is safe to assume that all dependencies of a Clang module are Clang modules.

lib/DependencyScan/ModuleDependencyCacheSerialization.cpp

@@ -574,12 +574,12 @@ bool ModuleDependenciesCacheDeserializer::readGraph(SwiftDependencyScanningServi
       cache.configureForContextHash(getContextHash());
       unsigned pcmOutputPathID, mappedPCMPathID, moduleMapPathID, contextHashID,
           commandLineArrayID, fileDependenciesArrayID, capturedPCMArgsArrayID,
-          CASFileSystemRootID, clangIncludeTreeRootID, moduleCacheKeyID;
+          CASFileSystemRootID, clangIncludeTreeRootID, moduleCacheKeyID, isSystem;
       ClangModuleDetailsLayout::readRecord(
           Scratch, pcmOutputPathID, mappedPCMPathID, moduleMapPathID,
           contextHashID, commandLineArrayID, fileDependenciesArrayID,
           capturedPCMArgsArrayID, CASFileSystemRootID, clangIncludeTreeRootID,
-          moduleCacheKeyID);
+          moduleCacheKeyID, isSystem);
       auto pcmOutputPath = getIdentifier(pcmOutputPathID);
       if (!pcmOutputPath)
         llvm::report_fatal_error("Bad pcm output path");
@@ -615,7 +615,7 @@ bool ModuleDependenciesCacheDeserializer::readGraph(SwiftDependencyScanningServi
       auto moduleDep = ModuleDependencyInfo::forClangModule(
           *pcmOutputPath, *mappedPCMPath, *moduleMapPath, *contextHash,
           *commandLineArgs, *fileDependencies, *capturedPCMArgs,
-          *rootFileSystemID, *clangIncludeTreeRoot, *moduleCacheKey);
+          *rootFileSystemID, *clangIncludeTreeRoot, *moduleCacheKey, isSystem);
 
       // Add dependencies of this module
       for (const auto &moduleName : *currentModuleImports)
@@ -1064,7 +1064,7 @@ void ModuleDependenciesCacheSerializer::writeModuleInfo(
         getArrayID(moduleID, ModuleIdentifierArrayKind::CapturedPCMArgs),
         getIdentifier(clangDeps->CASFileSystemRootID),
         getIdentifier(clangDeps->CASClangIncludeTreeRootID),
-        getIdentifier(clangDeps->moduleCacheKey));
+        getIdentifier(clangDeps->moduleCacheKey), clangDeps->IsSystem);
 
     break;
   }

lib/DependencyScan/ModuleDependencyScanner.cpp

@@ -21,6 +21,7 @@
 #include "swift/AST/TypeCheckRequests.h"
 #include "swift/Basic/FileTypes.h"
 #include "swift/Basic/PrettyStackTrace.h"
+#include "swift/ClangImporter/ClangImporter.h"
 #include "swift/DependencyScan/ModuleDependencyScanner.h"
 #include "swift/Frontend/ModuleInterfaceLoader.h"
 #include "swift/Serialization/SerializedModuleLoader.h"
@@ -774,24 +775,46 @@ void ModuleDependencyScanner::resolveSwiftOverlayDependencies(
         // FIXME: Once all clients know to fetch these dependencies from
         // `swiftOverlayDependencies`, the goal is to no longer have them in
         // `directDependencies` so the following will need to go away.
-	directDependencies.insert({moduleName, cachedInfo->getKind()});
+        directDependencies.insert({moduleName, cachedInfo->getKind()});
       } else {
         // Cache discovered module dependencies.
         cache.recordDependencies(lookupResult.value());
         if (!lookupResult.value().empty()) {
-          swiftOverlayDependencies.insert(
-                                          {moduleName, lookupResult.value()[0].first.Kind});
+          swiftOverlayDependencies.insert({moduleName, lookupResult.value()[0].first.Kind});
           // FIXME: Once all clients know to fetch these dependencies from
           // `swiftOverlayDependencies`, the goal is to no longer have them in
           // `directDependencies` so the following will need to go away.
-	  directDependencies.insert(
-                                    {moduleName, lookupResult.value()[0].first.Kind});
-	}
+          directDependencies.insert({moduleName, lookupResult.value()[0].first.Kind});
+        }
       }
     }
   };
   for (const auto &clangDep : clangDependencies)
     recordResult(clangDep);
+
+  // C++ Interop requires additional handling
+  if (ScanCompilerInvocation.getLangOptions().EnableCXXInterop) {
+    for (const auto &clangDepName : clangDependencies) {
+      // If this Clang module is a part of the C++ stdlib, and we haven't loaded
+      // the overlay for it so far, it is a split libc++ module (e.g.
+      // std_vector). Load the CxxStdlib overlay explicitly.
+      const auto &clangDepInfo =
+          cache.findDependency(clangDepName, ModuleDependencyKind::Clang)
+              .value()
+              ->getAsClangModule();
+      if (importer::isCxxStdModule(clangDepName, clangDepInfo->IsSystem) &&
+          !swiftOverlayDependencies.contains(
+              {clangDepName, ModuleDependencyKind::SwiftInterface}) &&
+          !swiftOverlayDependencies.contains(
+              {clangDepName, ModuleDependencyKind::SwiftBinary})) {
+        ScanningThreadPool.async(
+            scanForSwiftDependency,
+            getModuleImportIdentifier(ScanASTContext.Id_CxxStdlib.str()));
+        ScanningThreadPool.wait();
+        recordResult(ScanASTContext.Id_CxxStdlib.str().str());
+      }
+    }
+  }
 }
 
 void ModuleDependencyScanner::discoverCrossImportOverlayDependencies(