[clang][deps] Collect precompiled deps from submodules too

[jansvoboda11]

Summary

In this patch, the dependency scanner starts collecting precompiled dependencies from all encountered submodules, not only from top-level modules.

[jansvoboda11]

clang/test/ClangScanDeps/modules-pch-common-via-submodule.c

85

Previously, we'd miss this module included from mod_tu_sub.h, resulting in a compilation error for ModTU.

[dexonsmith]

LGTM.

clang/test/ClangScanDeps/modules-pch-common-via-submodule.c

85

Can you add a comment to the top of the testcase explaining what's interesting about it? (Something similar to what you have here.)

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

 //===- ModuleDepCollector.h - Callbacks to collect deps ---------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
private:
   llvm::DenseSet<const Module *> DirectModularDeps;
   /// Working set of direct modular dependencies that have already been built.
   llvm::DenseSet<const Module *> DirectPrebuiltModularDeps;
 
   void handleImport(const Module *Imported);
 
   /// Adds direct modular dependencies that have already been built to the
   /// ModuleDeps instance.
-  void addDirectPrebuiltModuleDeps(const Module *M, ModuleDeps &MD);
+  void
+  addAllSubmodulePrebuiltDeps(const Module *M, ModuleDeps &MD,
+                              llvm::DenseSet<const Module *> &SeenSubmodules);
+  void addModulePrebuiltDeps(const Module *M, ModuleDeps &MD,
+                             llvm::DenseSet<const Module *> &SeenSubmodules);
 
   /// Traverses the previously collected direct modular dependencies to discover
   /// transitive modular dependencies and fills the parent \c ModuleDepCollector
   /// with both.
   ModuleID handleTopLevelModule(const Module *M);
   void addAllSubmoduleDeps(const Module *M, ModuleDeps &MD,
                            llvm::DenseSet<const Module *> &AddedModules);
   void addModuleDep(const Module *M, ModuleDeps &MD,

clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp

 //===- ModuleDepCollector.cpp - Callbacks to collect deps -------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
MDC.Instance.getASTReader()->visitInputFiles(
           return;
         }
         MD.FileDeps.insert(IF.getFile()->getName());
       });
 
   // Add direct prebuilt module dependencies now, so that we can use them when
   // creating a CompilerInvocation and computing context hash for this
   // ModuleDeps instance.
-  addDirectPrebuiltModuleDeps(M, MD);
+  llvm::DenseSet<const Module *> SeenModules;
+  addAllSubmodulePrebuiltDeps(M, MD, SeenModules);
 
   MD.Invocation = MDC.makeInvocationForModuleBuildWithoutPaths(MD);
   MD.ID.ContextHash = MD.Invocation.getModuleHash();
 
   llvm::DenseSet<const Module *> AddedModules;
   addAllSubmoduleDeps(M, MD, AddedModules);
 
   return MD.ID;
 }
 
-void ModuleDepCollectorPP::addDirectPrebuiltModuleDeps(const Module *M,
-                                                       ModuleDeps &MD) {
+void ModuleDepCollectorPP::addAllSubmodulePrebuiltDeps(
+    const Module *M, ModuleDeps &MD,
+    llvm::DenseSet<const Module *> &SeenSubmodules) {
+  addModulePrebuiltDeps(M, MD, SeenSubmodules);
+
+  for (const Module *SubM : M->submodules())
+    addAllSubmodulePrebuiltDeps(SubM, MD, SeenSubmodules);
+}
+
+void ModuleDepCollectorPP::addModulePrebuiltDeps(
+    const Module *M, ModuleDeps &MD,
+    llvm::DenseSet<const Module *> &SeenSubmodules) {
   for (const Module *Import : M->Imports)
     if (Import->getTopLevelModule() != M->getTopLevelModule())
       if (MDC.isPrebuiltModule(Import))
+        if (SeenSubmodules.insert(Import->getTopLevelModule()).second)
           MD.PrebuiltModuleDeps.emplace_back(Import);
 }
 
 void ModuleDepCollectorPP::addAllSubmoduleDeps(
     const Module *M, ModuleDeps &MD,
     llvm::DenseSet<const Module *> &AddedModules) {
   addModuleDep(M, MD, AddedModules);
 
   for (const Module *SubM : M->submodules())

clang/test/ClangScanDeps/Inputs/modules-pch-common-via-submodule/cdb_pch.json

+[
+  {
+    "directory": "DIR",
+    "command": "clang -x c-header DIR/pch.h -fmodules -gmodules -fimplicit-module-maps -fmodules-cache-path=DIR/cache -o DIR/pch.h.gch",
+    "file": "DIR/pch.h"
+  }
+]

clang/test/ClangScanDeps/Inputs/modules-pch-common-via-submodule/cdb_tu.json

+[
+  {
+    "directory": "DIR",
+    "command": "clang -fsyntax-only DIR/tu.c -fmodules -gmodules -fimplicit-module-maps -fmodules-cache-path=DIR/cache -include DIR/pch.h -o DIR/tu.o",
+    "file": "DIR/tu.c"
+  }
+]

clang/test/ClangScanDeps/Inputs/modules-pch-common-via-submodule/mod_tu_sub.h

+#include "mod_common.h"

clang/test/ClangScanDeps/Inputs/modules-pch-common-via-submodule/module.modulemap

+module ModCommon {
+  header "mod_common.h"
+}
+
+module ModTU {
+  header "mod_tu.h"
+
+  module ModTUSub {
+    header "mod_tu_sub.h"
+  }
+}

clang/test/ClangScanDeps/Inputs/modules-pch-common-via-submodule/pch.h

+#include "mod_common.h"

clang/test/ClangScanDeps/Inputs/modules-pch-common-via-submodule/tu.c

+#include "mod_tu.h"

clang/test/ClangScanDeps/modules-pch-common-via-submodule.c

+// RUN: rm -rf %t && mkdir %t
+// RUN: cp %S/Inputs/modules-pch-common-via-submodule/* %t
+
+// Scan dependencies of the PCH:
+//
+// RUN: sed "s|DIR|%/t|g" %S/Inputs/modules-pch-common-via-submodule/cdb_pch.json > %t/cdb.json
+// RUN: echo -%t > %t/result_pch.json
+// RUN: clang-scan-deps -compilation-database %t/cdb.json -format experimental-full \
+// RUN:   -generate-modules-path-args -module-files-dir %t/build >> %t/result_pch.json
+// RUN: cat %t/result_pch.json | sed 's:\\\\\?:/:g' | FileCheck %s -check-prefix=CHECK-PCH
+//
+// CHECK-PCH:      -[[PREFIX:.*]]
+// CHECK-PCH-NEXT: {
+// CHECK-PCH-NEXT:   "modules": [
+// CHECK-PCH-NEXT:     {
+// CHECK-PCH-NEXT:       "clang-module-deps": [],
+// CHECK-PCH-NEXT:       "clang-modulemap-file": "[[PREFIX]]/module.modulemap",
+// CHECK-PCH-NEXT:       "command-line": [
+// CHECK-PCH-NEXT:         "-cc1"
+// CHECK-PCH:              "-emit-module"
+// CHECK-PCH:              "-fmodules"
+// CHECK-PCH:              "-fmodule-name=ModCommon"
+// CHECK-PCH:              "-fno-implicit-modules"
+// CHECK-PCH:            ],
+// CHECK-PCH-NEXT:       "context-hash": "[[HASH_MOD_COMMON:.*]]",
+// CHECK-PCH-NEXT:       "file-deps": [
+// CHECK-PCH-NEXT:         "[[PREFIX]]/mod_common.h",
+// CHECK-PCH-NEXT:         "[[PREFIX]]/module.modulemap"
+// CHECK-PCH-NEXT:       ],
+// CHECK-PCH-NEXT:       "name": "ModCommon"
+// CHECK-PCH-NEXT:     }
+// CHECK-PCH-NEXT:   ],
+// CHECK-PCH-NEXT:   "translation-units": [
+// CHECK-PCH-NEXT:     {
+// CHECK-PCH-NEXT:       "clang-context-hash": "[[HASH_PCH:.*]]",
+// CHECK-PCH-NEXT:       "clang-module-deps": [
+// CHECK-PCH-NEXT:         {
+// CHECK-PCH-NEXT:           "context-hash": "[[HASH_MOD_COMMON]]",
+// CHECK-PCH-NEXT:           "module-name": "ModCommon"
+// CHECK-PCH-NEXT:         }
+// CHECK-PCH-NEXT:       ],
+// CHECK-PCH-NEXT:       "command-line": [
+// CHECK-PCH-NEXT:         "-fno-implicit-modules"
+// CHECK-PCH-NEXT:         "-fno-implicit-module-maps"
+// CHECK-PCH-DAG:          "-fmodule-file=[[PREFIX]]/build/[[HASH_MOD_COMMON]]/ModCommon-{{.*}}.pcm"
+// CHECK-PCH-NEXT:         "-fmodule-map-file=[[PREFIX]]/module.modulemap"
+// CHECK-PCH-NEXT:       ],
+// CHECK-PCH-NEXT:       "file-deps": [
+// CHECK-PCH-NEXT:         "[[PREFIX]]/pch.h"
+// CHECK-PCH-NEXT:       ],
+// CHECK-PCH-NEXT:       "input-file": "[[PREFIX]]/pch.h"
+// CHECK-PCH-NEXT:     }
+// CHECK-PCH-NEXT:   ]
+// CHECK-PCH-NEXT: }
+
+// Explicitly build the PCH:
+//
+// RUN: tail -n +2 %t/result_pch.json > %t/result_pch_stripped.json
+// RUN: %python %S/../../utils/module-deps-to-rsp.py %t/result_pch_stripped.json \
+// RUN:   --module-name=ModCommon > %t/mod_common.cc1.rsp
+// RUN: %python %S/../../utils/module-deps-to-rsp.py %t/result_pch_stripped.json \
+// RUN:   --tu-index=0 > %t/pch.rsp
+//
+// RUN: %clang @%t/mod_common.cc1.rsp
+// RUN: %clang -x c-header %t/pch.h -fmodules -gmodules -fimplicit-module-maps \
+// RUN:   -fmodules-cache-path=%t/cache -o %t/pch.h.gch @%t/pch.rsp
+
+// Scan dependencies of the TU:
+//
+// RUN: sed "s|DIR|%/t|g" %S/Inputs/modules-pch-common-via-submodule/cdb_tu.json > %t/cdb.json
+// RUN: echo -%t > %t/result_tu.json
+// RUN: clang-scan-deps -compilation-database %t/cdb.json -format experimental-full \
+// RUN:   -generate-modules-path-args -module-files-dir %t/build >> %t/result_tu.json
+// RUN: cat %t/result_tu.json | sed 's:\\\\\?:/:g' | FileCheck %s -check-prefix=CHECK-TU
+//
+// CHECK-TU:      -[[PREFIX:.*]]
+// CHECK-TU-NEXT: {
+// CHECK-TU-NEXT:   "modules": [
+// CHECK-TU-NEXT:     {
+// CHECK-TU-NEXT:       "clang-module-deps": [],
+// CHECK-TU-NEXT:       "clang-modulemap-file": "[[PREFIX]]/module.modulemap",
+// CHECK-TU-NEXT:       "command-line": [
+// CHECK-TU-NEXT:         "-cc1"
+// CHECK-TU:              "-emit-module"
+// CHECK-TU:              "-fmodule-file=[[PREFIX]]/build/[[HASH_MOD_COMMON:.*]]/ModCommon-{{.*}}.pcm"
+// CHECK-TU:              "-fmodules"
+// CHECK-TU:              "-fmodule-name=ModTU"
+// CHECK-TU:              "-fno-implicit-modules"
+// CHECK-TU:            ],
+// CHECK-TU-NEXT:       "context-hash": "[[HASH_MOD_TU:.*]]",
+// CHECK-TU-NEXT:       "file-deps": [
+// CHECK-TU-NEXT:         "[[PREFIX]]/mod_tu.h",
+// CHECK-TU-NEXT:         "[[PREFIX]]/mod_tu_sub.h",
+// CHECK-TU-NEXT:         "[[PREFIX]]/module.modulemap"
+// CHECK-TU-NEXT:       ],
+// CHECK-TU-NEXT:       "name": "ModTU"
+// CHECK-TU-NEXT:     }
+// CHECK-TU-NEXT:   ],
+// CHECK-TU-NEXT:   "translation-units": [
+// CHECK-TU-NEXT:     {
+// CHECK-TU-NEXT:       "clang-context-hash": "[[HASH_TU:.*]]",
+// CHECK-TU-NEXT:       "clang-module-deps": [
+// CHECK-TU-NEXT:         {
+// CHECK-TU-NEXT:           "context-hash": "[[HASH_MOD_TU]]"
+// CHECK-TU-NEXT:           "module-name": "ModTU"
+// CHECK-TU-NEXT:         }
+// CHECK-TU-NEXT:       ],
+// CHECK-TU-NEXT:       "command-line": [
+// CHECK-TU-NEXT:         "-fno-implicit-modules",
+// CHECK-TU-NEXT:         "-fno-implicit-module-maps",
+// CHECK-TU-NEXT:         "-fmodule-file=[[PREFIX]]/build/[[HASH_MOD_TU:.*]]/ModTU-{{.*}}.pcm",
+// CHECK-TU-NEXT:         "-fmodule-map-file=[[PREFIX]]/module.modulemap"
+// CHECK-TU-NEXT:       ],
+// CHECK-TU-NEXT:       "file-deps": [
+// CHECK-TU-NEXT:         "[[PREFIX]]/tu.c",
+// CHECK-TU-NEXT:         "[[PREFIX]]/pch.h.gch"
+// CHECK-TU-NEXT:       ],
+// CHECK-TU-NEXT:       "input-file": "[[PREFIX]]/tu.c"
+// CHECK-TU-NEXT:     }
+// CHECK-TU-NEXT:   ]
+// CHECK-TU-NEXT: }
+
+// Explicitly build the TU:
+//
+// RUN: tail -n +2 %t/result_tu.json > %t/result_tu_stripped.json
+// RUN: %python %S/../../utils/module-deps-to-rsp.py %t/result_tu_stripped.json \
+// RUN:   --module-name=ModTU > %t/mod_tu.cc1.rsp
+// RUN: %python %S/../../utils/module-deps-to-rsp.py %t/result_tu_stripped.json \
+// RUN:   --tu-index=0 > %t/tu.rsp
+//
+// RUN: %clang @%t/mod_tu.cc1.rsp
+// RUN: %clang -fsyntax-only %t/tu.c -fmodules -gmodules -fimplicit-module-maps \
+// RUN:   -fmodules-cache-path=%t/cache -include %t/pch.h -o %t/tu.o @%t/tu.rsp