Adopt clang API for diagnosing invalid stat cache entries

Sources/SWBBuildSystem/BuildOperation.swift

@@ -1563,6 +1563,8 @@ internal final class OperationSystemAdaptor: SWBLLBuild.BuildSystemDelegate, Act
             // The build should be complete, validate the consistency of the target/task counts.
             self.validateTargetCompletion(buildSucceeded: buildSucceeded)
 
+            self.diagnoseInvalidNegativeStatCacheEntries(buildSucceeded: buildSucceeded)
+
             // If the build failed, make sure we flush any pending incremental build records.
             // Usually, driver instances are cleaned up and write out their incremental build records when a target finishes building. However, this won't necessarily be the case if the build fails. Ensure we write out any pending records before tearing down the graph so we don't use a stale record on a subsequent build.
             if !buildSucceeded {
@@ -1574,6 +1576,16 @@ internal final class OperationSystemAdaptor: SWBLLBuild.BuildSystemDelegate, Act
         }
     }
 
+    func diagnoseInvalidNegativeStatCacheEntries(buildSucceeded: Bool) {
+        let settings = operation.requestContext.getCachedSettings(operation.request.parameters)
+        guard settings.globalScope.evaluate(BuiltinMacros.VERIFY_CLANG_SCANNER_NEGATIVE_STAT_CACHE) || !buildSucceeded else {
+            return
+        }
+        for entry in dynamicOperationContext.clangModuleDependencyGraph.diagnoseInvalidNegativeStatCacheEntries() {
+            buildOutputDelegate.warning(Path(entry), "Clang reported an invalid negative stat cache entry for '\(entry)'; this may indicate a missing dependency which caused the file to be modified after being read by a dependent")
+        }
+    }
+
     /// Cleanup the compilation cache to reduce resource usage in environments not configured to preserve it.
     func cleanupCompilationCache() {
         let settings = operation.requestContext.getCachedSettings(operation.request.parameters)

Sources/SWBCSupport/CLibclang.cpp

@@ -1259,6 +1259,11 @@ extern "C" {
          */
         CXDiagnosticSet (*clang_experimental_DepGraph_getDiagnostics)(CXDepGraph);
 
+        /**
+         * Checks negatively cached paths in the stat cache against the current state of the filesystem and returns a list of discrepancies.
+         */
+        CXCStringArray (*clang_experimental_DependencyScannerService_getInvalidNegStatCachedPaths)(CXDependencyScannerService);
+
         // MARK: Driver API
 
         /**
@@ -1337,6 +1342,7 @@ struct LibclangWrapper {
     bool hasDependencyScanner;
     bool hasStructuredScanningDiagnostics;
     bool hasCAS;
+    bool hasNegativeStatCacheDiagnostics;
 
     LibclangWrapper(std::string path)
         : path(path),
@@ -1348,7 +1354,7 @@ struct LibclangWrapper {
 #else
           handle(dlopen(path.c_str(), RTLD_LAZY | RTLD_LOCAL)),
 #endif
-          isLeaked(false), hasRequiredAPI(true), hasDependencyScanner(true), hasStructuredScanningDiagnostics(true), hasCAS(true) {
+          isLeaked(false), hasRequiredAPI(true), hasDependencyScanner(true), hasStructuredScanningDiagnostics(true), hasCAS(true), hasNegativeStatCacheDiagnostics(false) {
 #if defined(_WIN32)
         DWORD cchLength = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, path.c_str(), -1, nullptr, 0);
         std::unique_ptr<wchar_t[]> wszPath(new wchar_t[cchLength]);
@@ -1488,6 +1494,11 @@ struct LibclangWrapper {
             hasCAS = false;
         }
 
+        LOOKUP_OPTIONAL(clang_experimental_DependencyScannerService_getInvalidNegStatCachedPaths);
+        if (fns.clang_experimental_DependencyScannerService_getInvalidNegStatCachedPaths) {
+            hasNegativeStatCacheDiagnostics = true;
+        }
+
         LOOKUP_OPTIONAL(clang_Driver_getExternalActionsForCommand_v0);
         LOOKUP_OPTIONAL(clang_Driver_ExternalActionList_dispose);
         LOOKUP_OPTIONAL(clang_install_aborting_llvm_fatal_error_handler);
@@ -1760,6 +1771,10 @@ extern "C" {
         return lib->wrapper->hasStructuredScanningDiagnostics;
     }
 
+    bool libclang_has_negative_stat_cache_diagnostics(libclang_t lib) {
+        return lib->wrapper->hasNegativeStatCacheDiagnostics;
+    }
+
     libclang_scanner_t libclang_scanner_create(libclang_t lib, libclang_casdatabases_t casdbs, libclang_casoptions_t casOpts) {
         return new libclang_scanner_t_{new LibclangScanner(
             lib->wrapper, LibclangFunctions::CXDependencyMode_Full,
@@ -2043,6 +2058,17 @@ extern "C" {
         return diagnostic_set;
     }
 
+    void libclang_scanner_diagnose_invalid_negative_stat_cache_entries(libclang_scanner_t scanner, void (^path_callback)(const char *)) {
+        auto lib = scanner->scanner->lib;
+        if (!lib->fns.clang_experimental_DependencyScannerService_getInvalidNegStatCachedPaths) {
+            return;
+        }
+        LibclangFunctions::CXCStringArray paths = lib->fns.clang_experimental_DependencyScannerService_getInvalidNegStatCachedPaths(scanner->scanner->service);
+        for (size_t i = 0; i < paths.Count; ++i) {
+            path_callback(paths.Strings[i]);
+        }
+    }
+
     bool libclang_scanner_scan_dependencies(
         libclang_scanner_t scanner, int argc, char *const *argv, const char *workingDirectory,
         __attribute__((noescape)) module_lookup_output_t module_lookup_output,

Sources/SWBCSupport/CLibclang.h

@@ -107,6 +107,9 @@ CSUPPORT_EXPORT bool libclang_has_scanner(libclang_t lib);
 /// Whether libclang supports reporting structured scanning diagnostics.
 CSUPPORT_EXPORT bool libclang_has_structured_scanner_diagnostics(libclang_t lib);
 
+/// Whether libclang supports reporting negative stat caching diagnostics.
+CSUPPORT_EXPORT bool libclang_has_negative_stat_cache_diagnostics(libclang_t lib);
+
 /// Create a new scanner instance with optional CAS databases.
 CSUPPORT_EXPORT libclang_scanner_t libclang_scanner_create(libclang_t lib, libclang_casdatabases_t, libclang_casoptions_t);
 
@@ -184,6 +187,9 @@ typedef size_t (^module_lookup_output_t)(
     const char *module_name, const char *context_hash,
     clang_output_kind_t kind, char *output, size_t max_len);
 
+/// Reports invalid entries in the scanner's negative stat cache.
+CSUPPORT_EXPORT void libclang_scanner_diagnose_invalid_negative_stat_cache_entries(libclang_scanner_t scanner, void (^path_callback)(const char *));
+
 /// Scan the given Clang "cc1" invocation, looking for dependencies.
 ///
 /// NOTE: This function is thread-safe.

Sources/SWBCore/LibclangVendored/Libclang.swift

@@ -91,6 +91,10 @@ public final class Libclang {
         libclang_has_structured_scanner_diagnostics(lib)
     }
 
+    public var supportsNegativeStatCacheDiagnostics: Bool {
+        libclang_has_negative_stat_cache_diagnostics(lib)
+    }
+
     public var supportsCASPruning: Bool {
         libclang_has_cas_pruning_feature(lib)
     }
@@ -274,6 +278,17 @@ public final class DependencyScanner {
         return fileDeps
     }
 
+    public func diagnoseInvalidNegativeStatCacheEntries() -> [String] {
+        var entries: [String] = []
+        libclang_scanner_diagnose_invalid_negative_stat_cache_entries(scanner, { cString in
+            guard let cString else {
+                return
+            }
+            entries.append(String(cString: cString))
+        })
+        return entries
+    }
+
     public func generateReproducer(
         commandLine: [String],
         workingDirectory: String

Sources/SWBCore/Settings/BuiltinMacros.swift

@@ -1165,6 +1165,7 @@ public final class BuiltinMacros {
     public static let VERSION_INFO_STRING = BuiltinMacros.declareStringMacro("VERSION_INFO_STRING")
     public static let VERSION_INFO_SUFFIX = BuiltinMacros.declareStringMacro("VERSION_INFO_SUFFIX")
     public static let ValidateForStore = BuiltinMacros.declareBooleanMacro("ValidateForStore")
+    public static let VERIFY_CLANG_SCANNER_NEGATIVE_STAT_CACHE = BuiltinMacros.declareBooleanMacro("VERIFY_CLANG_SCANNER_NEGATIVE_STAT_CACHE")
     public static let WARNING_CFLAGS = BuiltinMacros.declareStringListMacro("WARNING_CFLAGS")
     public static let WARNING_LDFLAGS = BuiltinMacros.declareStringListMacro("WARNING_LDFLAGS")
     public static let WATCHKIT_2_SUPPORT_FOLDER_PATH = BuiltinMacros.declareStringMacro("WATCHKIT_2_SUPPORT_FOLDER_PATH")
@@ -2402,6 +2403,7 @@ public final class BuiltinMacros {
         VERSION_INFO_STRING,
         VERSION_INFO_SUFFIX,
         ValidateForStore,
+        VERIFY_CLANG_SCANNER_NEGATIVE_STAT_CACHE,
         WARNING_CFLAGS,
         WARNING_LDFLAGS,
         WATCHKIT_2_SUPPORT_FOLDER_PATH,

Sources/SWBTaskExecution/DynamicTaskSpecs/ClangModuleDependencyGraph.swift

@@ -560,6 +560,17 @@ package final class ClangModuleDependencyGraph {
         return clangWithScanner.casDBs
     }
 
+    package func diagnoseInvalidNegativeStatCacheEntries() -> [String] {
+        registryQueue.blocking_sync {
+            self.scannerRegistry.values.flatMap { libClangWithScanner in
+                guard libClangWithScanner.scanner.libclang.supportsNegativeStatCacheDiagnostics else {
+                    return Array<String>()
+                }
+                return libClangWithScanner.scanner.diagnoseInvalidNegativeStatCacheEntries()
+            }
+        }
+    }
+
     package func generateReproducer(forFailedDependency dependency: DependencyInfo,
                                     libclangPath: Path, casOptions: CASOptions?) throws -> String? {
         let clangWithScanner = try libclangWithScanner(