[clang dependency scanning] C APIs for Current Working Directory Optimization

clang/include/clang-c/Dependencies.h

@@ -218,6 +218,20 @@ CINDEX_LINKAGE void
 clang_experimental_DependencyScannerServiceOptions_setCASOptions(
     CXDependencyScannerServiceOptions Opts, CXCASOptions);
 
+/**
+ * Set the working directory optimization option.
+ * The dependency scanner service option Opts will indicate to the scanner that
+ * the current working directory can or cannot be ignored when computing the
+ * pcms' context hashes. The scanner will then determine if it is safe to
+ * optimize each module and act accordingly.
+ *
+ * \param Value If it is non zero, the option is on. Otherwise the
+ * option is off.
+ */
+CINDEX_LINKAGE void
+clang_experimental_DependencyScannerServiceOptions_setCWDOptimization(
+    CXDependencyScannerServiceOptions Opts, int Value);
+
 /**
  * Specify a \c CXCASObjectStore in the given options. If an object store and
  * action cache are available, the scanner will produce cached commands.
@@ -578,6 +592,13 @@ CINDEX_LINKAGE const char *
 CINDEX_LINKAGE
 const char *clang_experimental_DepGraphModule_getCacheKey(CXDepGraphModule);
 
+/**
+ * \returns 1 if the scanner ignores the current working directory when
+ * computing the module's context hash. Otherwise returns 0.
+ */
+CINDEX_LINKAGE
+int clang_experimental_DepGraphModule_isCWDIgnored(CXDepGraphModule);
+
 /**
  * \returns the number \c CXDepGraphTUCommand objects in the graph.
  */

clang/include/clang/Tooling/DependencyScanning/DependencyScanningService.h

@@ -83,7 +83,11 @@ enum class ScanningOptimizations {
   IgnoreCWD = (1 << 4),
 
   DSS_LAST_BITMASK_ENUM(IgnoreCWD),
-  Default = All,
+
+  // The build system needs to be aware that the current working
+  // directory is ignored. Without a good way of notifying the build
+  // system, it is less risky to default to off.
+  Default = All & (~IgnoreCWD),
   FullIncludeTreeIrrelevant = HeaderSearch | VFS,
 };
 

clang/include/clang/Tooling/DependencyScanning/ModuleDepCollector.h

@@ -117,6 +117,9 @@ struct ModuleDeps {
   /// Whether this is a "system" module.
   bool IsSystem;
 
+  /// Whether current working directory is ignored.
+  bool IgnoreCWD;
+
   /// The path to the modulemap file which defines this module.
   ///
   /// This can be used to explicitly build this module. This file will

clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp

@@ -877,6 +877,7 @@ ModuleDepCollectorPP::handleTopLevelModule(const Module *M) {
       MD.ModuleCacheKey = Key->toString();
   }
 
+  MD.IgnoreCWD = IgnoreCWD;
   MDC.associateWithContextHash(CI, IgnoreCWD, MD);
 
   // Finish the compiler invocation. Requires dependencies and the context hash.

clang/test/ClangScanDeps/cwd-c-api.c

@@ -0,0 +1,42 @@
+// Test the current working directory C APIs.
+
+// RUN: rm -rf %t
+// RUN: split-file %s %t
+
+// RUN: c-index-test core -scan-deps -working-dir %S -- %clang \
+// RUN:   -c %t/main.c -fmodules -fmodules-cache-path=%t/module-cache \
+// RUN:   2>&1 > %t/no_cwd_opt.txt
+// RUN: cat %t/no_cwd_opt.txt | FileCheck %s --check-prefix=NO-CWD-OPT
+
+
+// RUN: c-index-test core -scan-deps -working-dir %S -optimize-cwd -- \
+// RUN:   %clang \
+// RUN:   -c %t/main.c -fmodules -fmodules-cache-path=%t/module-cache \
+// RUN:   2>&1 > %t/cwd_opt.txt
+// RUN: cat %t/cwd_opt.txt | FileCheck %s --check-prefix=CWD-OPT
+
+//--- module.modulemap
+module Mod { header "Mod.h" }
+
+//--- Mod.h
+int foo();
+
+//--- main.c
+#include "Mod.h"
+
+int main() {
+  return foo();
+}
+
+// NO-CWD-OPT: modules:
+// NO-CWD-OPT-NEXT:   module:
+// NO-CWD-OPT-NEXT:     name: Mod
+// NO-CWD-OPT-NEXT:     context-hash:{{.*}}
+// NO-CWD-OPT-NEXT:     cwd-ignored: 0
+
+
+// CWD-OPT: modules:
+// CWD-OPT-NEXT:   module:
+// CWD-OPT-NEXT:     name: Mod
+// CWD-OPT-NEXT:     context-hash:{{.*}}
+// CWD-OPT-NEXT:     cwd-ignored: 1

clang/test/ClangScanDeps/modules-context-hash-cwd.c

@@ -9,14 +9,14 @@
 // RUN: sed -e "s|DIR|%/t|g" %t/cdb3.json.in > %t/cdb3.json
 // RUN: sed -e "s|DIR|%/t|g" %t/cdb4.json.in > %t/cdb4.json
 // RUN: sed -e "s|DIR|%/t|g" %t/cdb5.json.in > %t/cdb5.json
-// RUN: clang-scan-deps -compilation-database %t/cdb0.json -format experimental-full > %t/result0.json
-// RUN: clang-scan-deps -compilation-database %t/cdb1.json -format experimental-full > %t/result1.json
+// RUN: clang-scan-deps -compilation-database %t/cdb0.json -format experimental-full -optimize-args=all > %t/result0.json
+// RUN: clang-scan-deps -compilation-database %t/cdb1.json -format experimental-full -optimize-args=all > %t/result1.json
 // It is not a typo to use cdb1.json for result2. We intend to use the same
 // compilation database, but different clang-scan-deps optimize-args options.
 // RUN: clang-scan-deps -compilation-database %t/cdb1.json -format experimental-full -optimize-args=header-search,system-warnings,vfs,canonicalize-macros > %t/result2.json
-// RUN: clang-scan-deps -compilation-database %t/cdb3.json -format experimental-full > %t/result3.json
-// RUN: clang-scan-deps -compilation-database %t/cdb4.json -format experimental-full > %t/result4.json
-// RUN: clang-scan-deps -compilation-database %t/cdb5.json -format experimental-full > %t/result5.json
+// RUN: clang-scan-deps -compilation-database %t/cdb3.json -format experimental-full -optimize-args=all > %t/result3.json
+// RUN: clang-scan-deps -compilation-database %t/cdb4.json -format experimental-full -optimize-args=all > %t/result4.json
+// RUN: clang-scan-deps -compilation-database %t/cdb5.json -format experimental-full -optimize-args=all > %t/result5.json
 // RUN: cat %t/result0.json %t/result1.json | FileCheck %s
 // RUN: cat %t/result0.json %t/result2.json | FileCheck %s -check-prefix=SKIPOPT
 // RUN: cat %t/result3.json %t/result4.json | FileCheck %s -check-prefix=RELPATH

clang/test/ClangScanDeps/modules-debug-dir.c

@@ -4,7 +4,7 @@
 // RUN: split-file %s %t
 // RUN: sed -e "s|DIR|%/t|g" %t/cdb.json.in > %t/cdb.json
 // RUN: clang-scan-deps -compilation-database %t/cdb.json -format \
-// RUN:   experimental-full > %t/result.json
+// RUN:   experimental-full -optimize-args=all > %t/result.json
 // RUN: cat %t/result.json | sed 's:\\\\\?:/:g' | FileCheck %s
 
 //--- cdb.json.in

clang/test/Index/Core/scan-deps-by-mod-name.m

@@ -13,6 +13,7 @@
 // CHECK-NEXT:   module:
 // CHECK-NEXT:     name: ModA
 // CHECK-NEXT:     context-hash: [[HASH_MOD_A:[A-Z0-9]+]]
+// CHECK-NEXT:     cwd-ignored: 0
 // CHECK-NEXT:     module-map-path: [[PREFIX]]/Inputs/module/module.modulemap
 // CHECK-NEXT:     module-deps:
 // CHECK-NEXT:     file-deps:

clang/test/Index/Core/scan-deps-cas.m

@@ -32,6 +32,7 @@
 // CHECK-NEXT:   module:
 // CHECK-NEXT:     name: ModA
 // CHECK-NEXT:     context-hash: [[HASH_MOD_A:[A-Z0-9]+]]
+// CHECK-NEXT:     cwd-ignored: 0
 // CHECK-NEXT:     module-map-path: [[PREFIX]]/Inputs/module/module.modulemap
 // CHECK-NEXT:     cache-key: [[CASFS_MODA_CACHE_KEY:llvmcas://[[:xdigit:]]+]]
 // CHECK-NEXT:     module-deps:
@@ -70,6 +71,7 @@
 // INCLUDE_TREE-NEXT:   module:
 // INCLUDE_TREE-NEXT:     name: ModA
 // INCLUDE_TREE-NEXT:     context-hash: [[HASH_MOD_A:[A-Z0-9]+]]
+// INCLUDE_TREE-NEXT:     cwd-ignored: 0
 // INCLUDE_TREE-NEXT:     module-map-path: [[PREFIX]]/Inputs/module/module.modulemap
 // INCLUDE_TREE-NEXT:     include-tree-id: [[ModA_INCLUDE_TREE_ID:llvmcas://[[:xdigit:]]+]]
 // INCLUDE_TREE-NEXT:     cache-key: [[ModA_CACHE_KEY:llvmcas://[[:xdigit:]]+]]

clang/test/Index/Core/scan-deps.m

@@ -21,6 +21,7 @@
 // CHECK-NEXT:   module:
 // CHECK-NEXT:     name: ModA
 // CHECK-NEXT:     context-hash: [[HASH_MOD_A:[A-Z0-9]+]]
+// CHECK-NEXT:     cwd-ignored: 0
 // CHECK-NEXT:     module-map-path: [[PREFIX]]/Inputs/module/module.modulemap
 // CHECK-NEXT:     module-deps:
 // CHECK-NEXT:     file-deps:

clang/tools/c-index-test/core_main.cpp

@@ -161,7 +161,11 @@ static cl::opt<bool> TestCASCancellation(
     "test-cas-cancellation",
     cl::desc(
         "perform extra CAS API invocation and cancel it for testing purposes"));
-}
+static cl::opt<bool> OptimizeCWD(
+    "optimize-cwd",
+    cl::desc(
+        "instruct the scanner to ignore current working directory if safe."));
+} // namespace options
 } // anonymous namespace
 
 static void printSymbolInfo(SymbolInfo SymInfo, raw_ostream &OS);
@@ -714,6 +718,10 @@ static int scanDeps(ArrayRef<const char *> Args, std::string WorkingDirectory,
   clang_experimental_DependencyScannerServiceOptions_setDependencyMode(
       Opts, CXDependencyMode_Full);
 
+  if (options::OptimizeCWD)
+    clang_experimental_DependencyScannerServiceOptions_setCWDOptimization(Opts,
+                                                                          1);
+
   if (DBs)
     clang_experimental_DependencyScannerServiceOptions_setCASDatabases(Opts,
                                                                        DBs);
@@ -806,6 +814,7 @@ static int scanDeps(ArrayRef<const char *> Args, std::string WorkingDirectory,
       const char *Name = clang_experimental_DepGraphModule_getName(Mod);
       const char *ContextHash =
           clang_experimental_DepGraphModule_getContextHash(Mod);
+      int CwdIgnored = clang_experimental_DepGraphModule_isCWDIgnored(Mod);
       const char *ModuleMapPath =
           clang_experimental_DepGraphModule_getModuleMapPath(Mod);
       const char *ModuleIncludeTreeID =
@@ -823,6 +832,7 @@ static int scanDeps(ArrayRef<const char *> Args, std::string WorkingDirectory,
       llvm::outs() << "  module:\n"
                    << "    name: " << Name << "\n"
                    << "    context-hash: " << ContextHash << "\n"
+                   << "    cwd-ignored: " << CwdIgnored << "\n"
                    << "    module-map-path: "
                    << (ModuleMapPath ? ModuleMapPath : "<none>") << "\n";
       if (ModuleIncludeTreeID)

clang/tools/libclang/CDependencies.cpp

@@ -34,6 +34,7 @@ namespace {
 struct DependencyScannerServiceOptions {
   ScanningOutputFormat ConfiguredFormat = ScanningOutputFormat::Full;
   CASOptions CASOpts;
+  ScanningOptimizations OptimizeArgs = ScanningOptimizations::Default;
   std::shared_ptr<cas::ObjectStore> CAS;
   std::shared_ptr<cas::ActionCache> Cache;
 
@@ -115,6 +116,15 @@ void clang_experimental_DependencyScannerServiceOptions_setCASOptions(
   unwrap(Opts)->CASOpts = *cas::unwrap(CASOpts);
 }
 
+void clang_experimental_DependencyScannerServiceOptions_setCWDOptimization(
+    CXDependencyScannerServiceOptions Opts, int Value) {
+  auto Mask =
+      Value != 0 ? ScanningOptimizations::All : ScanningOptimizations::None;
+  auto OptArgs = unwrap(Opts)->OptimizeArgs;
+  unwrap(Opts)->OptimizeArgs = (OptArgs & ~ScanningOptimizations::IgnoreCWD) |
+                               (ScanningOptimizations::IgnoreCWD & Mask);
+}
+
 void clang_experimental_DependencyScannerServiceOptions_setObjectStore(
     CXDependencyScannerServiceOptions Opts, CXCASObjectStore CAS) {
   unwrap(Opts)->CAS = cas::unwrap(CAS)->CAS;
@@ -171,7 +181,8 @@ clang_experimental_DependencyScannerService_create_v1(
   }
   return wrap(new DependencyScanningService(
       ScanningMode::DependencyDirectivesScan, Format, unwrap(Opts)->CASOpts,
-      std::move(CAS), std::move(Cache), std::move(FS)));
+      std::move(CAS), std::move(Cache), std::move(FS),
+      unwrap(Opts)->OptimizeArgs));
 }
 
 void clang_experimental_DependencyScannerService_dispose_v0(
@@ -630,6 +641,10 @@ clang_experimental_DepGraphModule_getCacheKey(CXDepGraphModule CXDepMod) {
   return nullptr;
 }
 
+int clang_experimental_DepGraphModule_isCWDIgnored(CXDepGraphModule CXDepMod) {
+  return unwrap(CXDepMod)->ModDeps->IgnoreCWD;
+}
+
 size_t clang_experimental_DepGraph_getNumTUCommands(CXDepGraph Graph) {
   TranslationUnitDeps &TUDeps = unwrap(Graph)->TUDeps;
   return TUDeps.Commands.size();

clang/tools/libclang/libclang.map

@@ -569,6 +569,12 @@ LLVM_19 {
     clang_Cursor_getBinaryOpcodeStr;
 };
 
+LLVM_21 {
+  global:
+    clang_experimental_DependencyScannerServiceOptions_setCWDOptimization;
+    clang_experimental_DepGraphModule_isCWDIgnored;
+};
+
 # Example of how to add a new symbol version entry.  If you do add a new symbol
 # version, please update the example to depend on the version you added.
 # LLVM_X {