[clang][deps] Further canonicalize implicit modules options in dep scan

[benlangmuir]

Summary

Disable or canonicalize compiler options that are not relevant in explicit module builds, similar to what we already did for the modules cache path. This reduces uninteresting differences between command-lines, which is particularly useful if there is a tool that can cache the compilations.

[jansvoboda11]

LGTM, thanks!

clang/lib/Tooling/DependencyScanning/DependencyScanningTool.cpp

 FullDependencies::getCommandLineWithoutModulePaths() const {
   std::vector<std::string> Args = OriginalCommandLine;
 
   Args.push_back("-fno-implicit-modules");
   Args.push_back("-fno-implicit-module-maps");
   for (const PrebuiltModuleDep &PMD : PrebuiltModuleDeps)
     Args.push_back("-fmodule-file=" + PMD.PCMFile);
 
-  // This argument is unused in explicit compiles.
-  llvm::erase_if(Args, [](const std::string &Arg) {
-    return Arg.find("-fmodules-cache-path=") == 0;
+  // These arguments are unused in explicit compiles.
+  llvm::erase_if(Args, [](StringRef Arg) {
+    if (Arg.consume_front("-fmodules-")) {
+      return Arg.startswith("cache-path=") ||
+             Arg.startswith("prune-interval=") ||
+             Arg.startswith("prune-after=") ||
+             Arg == "validate-once-per-build-session";
+    }
+    return Arg.startswith("-fbuild-session-file=");
   });
 
-  // TODO: Filter out the remaining implicit modules leftovers
-  // (e.g. "-fmodules-prune-interval=" or "-fmodules-prune-after=").
-
   return Args;
 }
 
 DependencyScanningTool::DependencyScanningTool(
     DependencyScanningService &Service)
     : Worker(Service) {}
 
 llvm::Expected<std::string> DependencyScanningTool::getDependencyFile(

clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp

CompilerInvocation ModuleDepCollector::makeInvocationForModuleBuildWithoutPaths(
   CI.getFrontendOpts().OutputFile.clear();
   CI.getCodeGenOpts().MainFileName.clear();
   CI.getCodeGenOpts().DwarfDebugFlags.clear();
 
   CI.getFrontendOpts().ProgramAction = frontend::GenerateModule;
   CI.getLangOpts()->ModuleName = Deps.ID.ModuleName;
   CI.getFrontendOpts().IsSystemModule = Deps.IsSystem;
 
+  // Disable implicit modules and canonicalize options that are only used by
+  // implicit modules.
   CI.getLangOpts()->ImplicitModules = false;
   CI.getHeaderSearchOpts().ImplicitModuleMaps = false;
   CI.getHeaderSearchOpts().ModuleCachePath.clear();
+  CI.getHeaderSearchOpts().ModulesValidateOncePerBuildSession = false;
+  CI.getHeaderSearchOpts().BuildSessionTimestamp = 0;
+  // The specific values we canonicalize to for pruning don't affect behaviour,
+  /// so use the default values so they will be dropped from the command-line.
+  CI.getHeaderSearchOpts().ModuleCachePruneInterval = 7 * 24 * 60 * 60;
+  CI.getHeaderSearchOpts().ModuleCachePruneAfter = 31 * 24 * 60 * 60;
 
   // Report the prebuilt modules this module uses.
   for (const auto &PrebuiltModule : Deps.PrebuiltModuleDeps)
     CI.getFrontendOpts().ModuleFiles.push_back(PrebuiltModule.PCMFile);
 
   CI.getFrontendOpts().ModuleMapFiles = Deps.ModuleMapFileDeps;
 
   Optimize(CI);

clang/test/ClangScanDeps/Inputs/removed-args/cdb.json.template

 [
   {
     "directory": "DIR",
-    "command": "clang -fsyntax-only DIR/tu.c -fmodules -fimplicit-module-maps -fmodules-cache-path=DIR/cache -include DIR/header.h -grecord-command-line -o DIR/tu.o",
+    "command": "clang -fsyntax-only DIR/tu.c -fmodules -fimplicit-module-maps -fmodules-validate-once-per-build-session -fbuild-session-file=DIR/build-session -fmodules-prune-interval=123 -fmodules-prune-after=123 -fmodules-cache-path=DIR/cache -include DIR/header.h -grecord-command-line -o DIR/tu.o",
     "file": "DIR/tu.c"
   }
 ]

clang/test/ClangScanDeps/removed-args.c

 // Some command-line arguments used for compiling translation units are not
 // compatible with the semantics of modules or are likely to differ between
 // identical modules discovered from different translation units. This test
 // checks such arguments are removed from the command-lines: '-include',
-// '-dwarf-debug-flag' and '-main-file-name'.
+// '-dwarf-debug-flag' and '-main-file-name'. Similarly, several arguments
+// such as '-fmodules-cache-path=' are only relevant for implicit modules, and
+// are removed to better-canonicalize the compilation.
 
 // RUN: rm -rf %t && mkdir %t
 // RUN: cp %S/Inputs/removed-args/* %t
+// RUN: touch %t/build-session
 
 // RUN: sed "s|DIR|%/t|g" %S/Inputs/removed-args/cdb.json.template > %t/cdb.json
 // RUN: clang-scan-deps -compilation-database %t/cdb.json -format experimental-full > %t/result.json
 // RUN: cat %t/result.json | sed 's:\\\\\?:/:g' | FileCheck %s -DPREFIX=%/t
 //
 // CHECK:      {
 // CHECK-NEXT:   "modules": [
 // CHECK-NEXT:     {
 // CHECK-NEXT:       "clang-module-deps": [],
 // CHECK-NEXT:       "clang-modulemap-file": "[[PREFIX]]/module.modulemap",
 // CHECK-NEXT:       "command-line": [
 // CHECK-NEXT:         "-cc1"
 // CHECK-NOT:          "-dwarf-debug-flags"
 // CHECK-NOT:          "-main-file-name"
 // CHECK-NOT:          "-include"
+// CHECK-NOT:          "-fmodules-cache-path=
+// CHECK-NOT:          "-fmodules-validate-once-per-build-session"
+// CHECK-NOT:          "-fbuild-session-timestamp=
+// CHECK-NOT:          "-fmodules-prune-interval=
+// CHECK-NOT:          "-fmodules-prune-after=
 // CHECK:            ],
 // CHECK-NEXT:       "context-hash": "[[HASH_MOD_HEADER:.*]]",
 // CHECK-NEXT:       "file-deps": [
 // CHECK-NEXT:         "[[PREFIX]]/mod_header.h",
 // CHECK-NEXT:         "[[PREFIX]]/module.modulemap"
 // CHECK-NEXT:       ],
 // CHECK-NEXT:       "name": "ModHeader"
 // CHECK-NEXT:     },
 // CHECK-NEXT:     {
 // CHECK-NEXT:       "clang-module-deps": [],
 // CHECK-NEXT:       "clang-modulemap-file": "[[PREFIX]]/module.modulemap",
 // CHECK-NEXT:       "command-line": [
 // CHECK-NEXT:         "-cc1"
+// CHECK-NOT:          "-dwarf-debug-flags"
+// CHECK-NOT:          "-main-file-name"
 // CHECK-NOT:          "-include"
+// CHECK-NOT:          "-fmodules-cache-path=
+// CHECK-NOT:          "-fmodules-validate-once-per-build-session"
+// CHECK-NOT:          "-fbuild-session-timestamp=
+// CHECK-NOT:          "-fmodules-prune-interval=
+// CHECK-NOT:          "-fmodules-prune-after=
 // CHECK:            ],
 // CHECK-NEXT:       "context-hash": "[[HASH_MOD_TU:.*]]",
 // CHECK-NEXT:       "file-deps": [
 // CHECK-NEXT:         "[[PREFIX]]/mod_tu.h",
 // CHECK-NEXT:         "[[PREFIX]]/module.modulemap"
 // CHECK-NEXT:       ],
 // CHECK-NEXT:       "name": "ModTU"
 // CHECK-NEXT:     }
 // CHECK-NEXT:   ],
 // CHECK-NEXT:   "translation-units": [
 // CHECK-NEXT:     {
 // CHECK-NEXT:       "clang-context-hash": "[[HASH_TU:.*]]",
 // CHECK-NEXT:       "clang-module-deps": [
 // CHECK-NEXT:         {
 // CHECK-NEXT:           "context-hash": "[[HASH_MOD_HEADER]]",
 // CHECK-NEXT:           "module-name": "ModHeader"
 // CHECK-NEXT:         },
 // CHECK-NEXT:         {
 // CHECK-NEXT:           "context-hash": "[[HASH_MOD_TU]]",
 // CHECK-NEXT:           "module-name": "ModTU"
 // CHECK-NEXT:         }
 // CHECK-NEXT:       ]
+// CHECK-NEXT:       "command-line": [
+// CHECK-NEXT:         "-fsyntax-only",
+// CHECK-NOT:          "-fmodules-cache-path=
+// CHECK-NOT:          "-fmodules-validate-once-per-build-session"
+// CHECK-NOT:          "-fbuild-session-file=
+// CHECK-NOT:          "-fmodules-prune-interval=
+// CHECK-NOT:          "-fmodules-prune-after=
+// CHECK:            ],
 // CHECK:          }
 // CHECK-NEXT:   ]
 // CHECK-NEXT: }