[clang][deps] Lazy dependency directives (llvm#86347)

Since b4c83a1, `Preprocessor` and
`Lexer` are aware of the concept of scanning dependency directives. This
makes it possible to scan for them on-demand rather than eagerly on the
first filesystem operation (open, or even just stat).

This might improve performance, but is also necessary for the "PCH as
module" mode. Some precompiled header sources use the ".pch" file
extension, which means they were not getting scanned for dependency
directives. This was okay when the PCH was the main input file in a
separate scan step, because there we just lex the file in a
scanning-specific frontend action. But when such source gets treated as
a module implicitly loaded from a TU, it will get compiled as any other
module - with Sema - which will result in compilation errors. (See
attached test case.)

rdar://107663951
(cherry picked from commit b768a8c)

Author: jansvoboda11

clang/include/clang/Tooling/DependencyScanning/DependencyScanningFilesystem.h

@@ -255,6 +255,8 @@ class EntryRef {
   /// The underlying cached entry.
   const CachedFileSystemEntry &Entry;
 
+  friend class DependencyScanningWorkerFilesystem;
+
 public:
   EntryRef(StringRef Name, const CachedFileSystemEntry &Entry)
       : Filename(Name), Entry(Entry) {}
@@ -316,14 +318,15 @@ class DependencyScanningWorkerFilesystem
   ///
   /// Attempts to use the local and shared caches first, then falls back to
   /// using the underlying filesystem.
-  llvm::ErrorOr<EntryRef>
-  getOrCreateFileSystemEntry(StringRef Filename,
-                             bool DisableDirectivesScanning = false);
+  llvm::ErrorOr<EntryRef> getOrCreateFileSystemEntry(StringRef Filename);
 
-private:
-  /// Check whether the file should be scanned for preprocessor directives.
-  bool shouldScanForDirectives(StringRef Filename);
+  /// Ensure the directive tokens are populated for this file entry.
+  ///
+  /// Returns true if the directive tokens are populated for this file entry,
+  /// false if not (i.e. this entry is not a file or its scan fails).
+  bool ensureDirectiveTokensArePopulated(EntryRef Entry);
 
+private:
   /// For a filename that's not yet associated with any entry in the caches,
   /// uses the underlying filesystem to either look up the entry based in the
   /// shared cache indexed by unique ID, or creates new entry from scratch.
@@ -333,11 +336,6 @@ class DependencyScanningWorkerFilesystem
   computeAndStoreResult(StringRef OriginalFilename,
                         StringRef FilenameForLookup);
 
-  /// Scan for preprocessor directives for the given entry if necessary and
-  /// returns a wrapper object with reference semantics.
-  EntryRef scanForDirectivesIfNecessary(const CachedFileSystemEntry &Entry,
-                                        StringRef Filename, bool Disable);
-
   /// Represents a filesystem entry that has been stat-ed (and potentially read)
   /// and that's about to be inserted into the cache as `CachedFileSystemEntry`.
   struct TentativeEntry {

clang/lib/Tooling/DependencyScanning/DependencyScanningFilesystem.cpp

@@ -46,24 +46,25 @@ DependencyScanningWorkerFilesystem::readFile(StringRef Filename) {
   return TentativeEntry(Stat, std::move(Buffer), std::move(CASContents));
 }
 
-EntryRef DependencyScanningWorkerFilesystem::scanForDirectivesIfNecessary(
-    const CachedFileSystemEntry &Entry, StringRef Filename, bool Disable) {
-  if (Entry.isError() || Entry.isDirectory() || Disable ||
-      !shouldScanForDirectives(Filename))
-    return EntryRef(Filename, Entry);
+bool DependencyScanningWorkerFilesystem::ensureDirectiveTokensArePopulated(
+    EntryRef Ref) {
+  auto &Entry = Ref.Entry;
+
+  if (Entry.isError() || Entry.isDirectory())
+    return false;
 
   CachedFileContents *Contents = Entry.getCachedContents();
   assert(Contents && "contents not initialized");
 
   // Double-checked locking.
   if (Contents->DepDirectives.load())
-    return EntryRef(Filename, Entry);
+    return true;
 
   std::lock_guard<std::mutex> GuardLock(Contents->ValueLock);
 
   // Double-checked locking.
   if (Contents->DepDirectives.load())
-    return EntryRef(Filename, Entry);
+    return true;
 
   SmallVector<dependency_directives_scan::Directive, 64> Directives;
   // Scan the file for preprocessor directives that might affect the
@@ -74,16 +75,16 @@ EntryRef DependencyScanningWorkerFilesystem::scanForDirectivesIfNecessary(
     Contents->DepDirectiveTokens.clear();
     // FIXME: Propagate the diagnostic if desired by the client.
     Contents->DepDirectives.store(new std::optional<DependencyDirectivesTy>());
-    return EntryRef(Filename, Entry);
+    return false;
   }
 
   // This function performed double-checked locking using `DepDirectives`.
   // Assigning it must be the last thing this function does, otherwise other
-  // threads may skip the
-  // critical section (`DepDirectives != nullptr`), leading to a data race.
+  // threads may skip the critical section (`DepDirectives != nullptr`), leading
+  // to a data race.
   Contents->DepDirectives.store(
       new std::optional<DependencyDirectivesTy>(std::move(Directives)));
-  return EntryRef(Filename, Entry);
+  return true;
 }
 
 DependencyScanningFilesystemSharedCache::
@@ -167,34 +168,11 @@ DependencyScanningFilesystemSharedCache::CacheShard::
   return *EntriesByFilename.insert({Filename, &Entry}).first->getValue();
 }
 
-/// Whitelist file extensions that should be minimized, treating no extension as
-/// a source file that should be minimized.
-///
-/// This is kinda hacky, it would be better if we knew what kind of file Clang
-/// was expecting instead.
-static bool shouldScanForDirectivesBasedOnExtension(StringRef Filename) {
-  StringRef Ext = llvm::sys::path::extension(Filename);
-  if (Ext.empty())
-    return true; // C++ standard library
-  return llvm::StringSwitch<bool>(Ext)
-      .CasesLower(".c", ".cc", ".cpp", ".c++", ".cxx", true)
-      .CasesLower(".h", ".hh", ".hpp", ".h++", ".hxx", true)
-      .CasesLower(".m", ".mm", true)
-      .CasesLower(".i", ".ii", ".mi", ".mmi", true)
-      .CasesLower(".def", ".inc", true)
-      .Default(false);
-}
-
 static bool shouldCacheStatFailures(StringRef Filename) {
   StringRef Ext = llvm::sys::path::extension(Filename);
   if (Ext.empty())
     return false; // This may be the module cache directory.
-  // Only cache stat failures on files that are not expected to change during
-  // the build.
-  StringRef FName = llvm::sys::path::filename(Filename);
-  if (FName == "module.modulemap" || FName == "module.map")
-    return true;
-  return shouldScanForDirectivesBasedOnExtension(Filename);
+  return true;
 }
 
 DependencyScanningWorkerFilesystem::DependencyScanningWorkerFilesystem(
@@ -207,11 +185,6 @@ DependencyScanningWorkerFilesystem::DependencyScanningWorkerFilesystem(
   updateWorkingDirForCacheLookup();
 }
 
-bool DependencyScanningWorkerFilesystem::shouldScanForDirectives(
-    StringRef Filename) {
-  return shouldScanForDirectivesBasedOnExtension(Filename);
-}
-
 const CachedFileSystemEntry &
 DependencyScanningWorkerFilesystem::getOrEmplaceSharedEntryForUID(
     TentativeEntry TEntry) {
@@ -265,7 +238,7 @@ DependencyScanningWorkerFilesystem::computeAndStoreResult(
 
 llvm::ErrorOr<EntryRef>
 DependencyScanningWorkerFilesystem::getOrCreateFileSystemEntry(
-    StringRef OriginalFilename, bool DisableDirectivesScanning) {
+    StringRef OriginalFilename) {
   StringRef FilenameForLookup;
   SmallString<256> PathBuf;
   if (llvm::sys::path::is_absolute_gnu(OriginalFilename)) {
@@ -282,15 +255,11 @@ DependencyScanningWorkerFilesystem::getOrCreateFileSystemEntry(
   assert(llvm::sys::path::is_absolute_gnu(FilenameForLookup));
   if (const auto *Entry =
           findEntryByFilenameWithWriteThrough(FilenameForLookup))
-    return scanForDirectivesIfNecessary(*Entry, OriginalFilename,
-                                        DisableDirectivesScanning)
-        .unwrapError();
+    return EntryRef(OriginalFilename, *Entry).unwrapError();
   auto MaybeEntry = computeAndStoreResult(OriginalFilename, FilenameForLookup);
   if (!MaybeEntry)
     return MaybeEntry.getError();
-  return scanForDirectivesIfNecessary(*MaybeEntry, OriginalFilename,
-                                      DisableDirectivesScanning)
-      .unwrapError();
+  return EntryRef(OriginalFilename, *MaybeEntry).unwrapError();
 }
 
 llvm::ErrorOr<llvm::vfs::Status>

clang/lib/Tooling/DependencyScanning/DependencyScanningWorker.cpp

@@ -507,7 +507,8 @@ class DependencyScanningAction : public tooling::ToolAction {
           -> std::optional<ArrayRef<dependency_directives_scan::Directive>> {
         if (llvm::ErrorOr<EntryRef> Entry =
                 LocalDepFS->getOrCreateFileSystemEntry(File.getName()))
-          return Entry->getDirectiveTokens();
+          if (LocalDepFS->ensureDirectiveTokensArePopulated(*Entry))
+            return Entry->getDirectiveTokens();
         return std::nullopt;
       };
     }

clang/test/ClangScanDeps/modules-extension.c

@@ -0,0 +1,33 @@
+// RUN: rm -rf %t
+// RUN: split-file %s %t
+
+// This test checks that source files with uncommon extensions still undergo
+// dependency directives scan. If header.pch would not and b.h would, the scan
+// would fail when parsing `void function(B)` and not knowing the symbol B.
+
+//--- module.modulemap
+module __PCH { header "header.pch" }
+module B { header "b.h" }
+
+//--- header.pch
+#include "b.h"
+void function(B);
+
+//--- b.h
+typedef int B;
+
+//--- tu.c
+int main() {
+  function(0);
+  return 0;
+}
+
+//--- cdb.json.in
+[{
+  "directory": "DIR",
+  "file": "DIR/tu.c",
+  "command": "clang -c DIR/tu.c -fmodules -fmodules-cache-path=DIR/cache -fimplicit-module-maps -include DIR/header.pch"
+}]
+
+// RUN: sed -e "s|DIR|%/t|g" %t/cdb.json.in > %t/cdb.json
+// RUN: clang-scan-deps -compilation-database %t/cdb.json -format experimental-full > %t/deps.json