[modules] Handle modules with nonstandard names in module.private.modulemaps

[graydon]

Summary

The module system supports accompanying a primary module (say Foo) with
an auxiliary "private" module (defined in an adjacent module.private.modulemap
file) that augments the primary module when associated private headers are
available. The feature is intended to be used to augment the primary
module with a submodule (say Foo.Private), however some users in the wild
are choosing to augment the primary module with an additional top-level module
with a "similar" name (in all cases so far: FooPrivate).

This "works" when a user of the module initially imports a private header,
such as '#import "Foo/something_private.h"' since the Foo import winds up
importing FooPrivate in passing. But if the import is subsequently recorded
in a PCH file, reloading the PCH will fail to validate because of a cross-check
that attempts to find the module.modulemap (or module.private.modulemap) using
HeaderSearch algorithm, applied to the "FooPrivate" name. Since it's stored in
Foo.framework/Modules, not FooPrivate.framework/Modules, the check fails and
the PCH is rejected.

This patch adds a compensatory workaround in the HeaderSearch algorithm
when searching (and failing to find) a module of the form FooPrivate: the
name used to derive filesystem paths is decoupled from the module name
being searched for, and if the initial search fails and the module is
named "FooPrivate", the filesystem search name is altered to remove the
"Private" suffix, and the algorithm is run a second time (still looking for
a module named FooPrivate, but looking in directories derived from Foo).

Accompanying this change is a new warning that triggers when a user loads
a module.private.modulemap that defines a top-level module with a different
name from the top-level module defined in its adjacent module.modulemap.

[manmanren]

Thanks,
Manman

[bruno]

Hi,

Thanks for working on this.

include/clang/Basic/DiagnosticLexKinds.td

647 ↗

(On Diff #81765)

It would be nice if we could also emit a FixIt to tell the user how should the modulemap be fixed.

lib/Lex/HeaderSearch.cpp

208 ↗

(On Diff #81765)

Remove the curly braces here

211 ↗

(On Diff #81765)

Remove this newline

216 ↗

(On Diff #81765)

Remove this newline

test/Modules/implicit-private-with-different-name.m

9 ↗

(On Diff #81765)

Can you check for the entire warning message?

[graydon]

Addressed review comments

[graydon]

Updates to address review comments

[bruno]

test/Modules/implicit-private-with-different-name.m

13 ↗

(On Diff #82161)

I think we can enhance the usability a bit here, how about:

warning: top-level module 'APrivate' in private module map, expected a submodule of 'A'
note: make 'APrivate' a submodule of 'A' to ensure it can be found by name

framework module APrivate {
                 ^
              A.Private

[graydon]

test/Modules/implicit-private-with-different-name.m

13 ↗

(On Diff #82161)

I'm not sure what you're requesting; as far as I can see that is exactly the diagnostic I'm emitting (along with the fixit)

[bruno]

Right, I missed it, LGTM

cfe/trunk/include/clang/Basic/DiagnosticGroups.td

: DiagGroup<"incompatible-pointer-types",
      IncompatibleFunctionPointerTypes]>;
 def IncompleteUmbrella : DiagGroup<"incomplete-umbrella">;
 def NonModularIncludeInFrameworkModule
   : DiagGroup<"non-modular-include-in-framework-module">;
 def NonModularIncludeInModule : DiagGroup<"non-modular-include-in-module",
                                           [NonModularIncludeInFrameworkModule]>;
 def IncompleteModule : DiagGroup<"incomplete-module",
     [IncompleteUmbrella, NonModularIncludeInModule]>;
+def PrivateModule : DiagGroup<"private-module">;
 
 def CXX11InlineNamespace : DiagGroup<"c++11-inline-namespace">;
 def InvalidNoreturn : DiagGroup<"invalid-noreturn">;
 def InvalidSourceEncoding : DiagGroup<"invalid-source-encoding">;
 def KNRPromotedParameter : DiagGroup<"knr-promoted-parameter">;
 def : DiagGroup<"init-self">;
 def : DiagGroup<"inline">;
 def : DiagGroup<"invalid-pch">;

cfe/trunk/include/clang/Basic/DiagnosticLexKinds.td

 def err_mmap_explicit_top_level : Error<
   "'explicit' is not permitted on top-level modules">;
 def err_mmap_nested_submodule_id : Error<
   "qualified module name can only be used to define modules at the top level">;
 def err_mmap_expected_feature : Error<"expected a feature name">;
 def err_mmap_expected_attribute : Error<"expected an attribute name">;
 def warn_mmap_unknown_attribute : Warning<"unknown attribute '%0'">,
   InGroup<IgnoredAttributes>;
+def warn_mmap_mismatched_top_level_private : Warning<
+  "top-level module '%0' in private module map, expected a submodule of '%1'">,
+  InGroup<PrivateModule>;
+def note_mmap_rename_top_level_private_as_submodule : Note<
+  "make '%0' a submodule of '%1' to ensure it can be found by name">,
+  InGroup<PrivateModule>;
 
 def warn_auto_module_import : Warning<
   "treating #%select{include|import|include_next|__include_macros}0 as an "
   "import of module '%1'">, InGroup<AutoImport>, DefaultIgnore;
 def note_implicit_top_level_module_import_here : Note<
   "submodule of top-level module '%0' implicitly imported here">;
 def warn_uncovered_module_header : Warning<
   "umbrella header for module '%0' does not include header '%1'">, 

cfe/trunk/include/clang/Lex/HeaderSearch.h

public:
   ///
   /// \param Modules Will be filled with the set of known, top-level modules.
   void collectAllModules(SmallVectorImpl<Module *> &Modules);
 
   /// \brief Load all known, top-level system modules.
   void loadTopLevelSystemModules();
 
 private:
+
+  /// \brief Lookup a module with the given module name and search-name.
+  ///
+  /// \param ModuleName The name of the module we're looking for.
+  ///
+  /// \param SearchName The "search-name" to derive filesystem paths from
+  /// when looking for the module map; this is usually equal to ModuleName,
+  /// but for compatibility with some buggy frameworks, additional attempts
+  /// may be made to find the module under a related-but-different search-name.
+  ///
+  /// \returns The module named ModuleName.
+  Module *lookupModule(StringRef ModuleName, StringRef SearchName);
+
   /// \brief Retrieve a module with the given name, which may be part of the
   /// given framework.
   ///
   /// \param Name The name of the module to retrieve.
   ///
   /// \param Dir The framework directory (e.g., ModuleName.framework).
   ///
   /// \param IsSystem Whether the framework directory is part of the system

cfe/trunk/lib/Lex/HeaderSearch.cpp

std::string HeaderSearch::getModuleFileName(StringRef ModuleName,
   return Result.str().str();
 }
 
 Module *HeaderSearch::lookupModule(StringRef ModuleName, bool AllowSearch) {
   // Look in the module map to determine if there is a module by this name.
   Module *Module = ModMap.findModule(ModuleName);
   if (Module || !AllowSearch || !HSOpts->ImplicitModuleMaps)
     return Module;
-  
+  StringRef SearchName = ModuleName;
+  Module = lookupModule(ModuleName, SearchName);
+
+  // The facility for "private modules" -- adjacent, optional module maps named
+  // module.private.modulemap that are supposed to define private submodules --
+  // is sometimes misused by frameworks that name their associated private
+  // module FooPrivate, rather than as a submodule named Foo.Private as
+  // intended. Here we compensate for such cases by looking in directories named
+  // Foo.framework, when we previously looked and failed to find a
+  // FooPrivate.framework.
+  if (!Module && SearchName.consume_back("Private"))
+    Module = lookupModule(ModuleName, SearchName);
+  return Module;
+}
+
+Module *HeaderSearch::lookupModule(StringRef ModuleName, StringRef SearchName) {
+  Module *Module = nullptr;
+
   // Look through the various header search paths to load any available module
   // maps, searching for a module map that describes this module.
   for (unsigned Idx = 0, N = SearchDirs.size(); Idx != N; ++Idx) {
     if (SearchDirs[Idx].isFramework()) {
-      // Search for or infer a module map for a framework.
+      // Search for or infer a module map for a framework. Here we use
+      // SearchName rather than ModuleName, to permit finding private modules
+      // named FooPrivate in buggy frameworks named Foo.
       SmallString<128> FrameworkDirName;
       FrameworkDirName += SearchDirs[Idx].getFrameworkDir()->getName();
-      llvm::sys::path::append(FrameworkDirName, ModuleName + ".framework");
-      if (const DirectoryEntry *FrameworkDir 
+      llvm::sys::path::append(FrameworkDirName, SearchName + ".framework");
+      if (const DirectoryEntry *FrameworkDir
             = FileMgr.getDirectory(FrameworkDirName)) {
         bool IsSystem
           = SearchDirs[Idx].getDirCharacteristic() != SrcMgr::C_User;
         Module = loadFrameworkModule(ModuleName, FrameworkDir, IsSystem);
         if (Module)
           break;
       }
     }

cfe/trunk/lib/Lex/ModuleMap.cpp

if (Attrs.IsSystem || IsSystem)
     ActiveModule->IsSystem = true;
   if (Attrs.IsExternC)
     ActiveModule->IsExternC = true;
   if (Attrs.NoUndeclaredIncludes ||
       (!ActiveModule->Parent && ModuleName == "Darwin"))
     ActiveModule->NoUndeclaredIncludes = true;
   ActiveModule->Directory = Directory;
 
+  if (!ActiveModule->Parent) {
+    StringRef MapFileName(ModuleMapFile->getName());
+    if (MapFileName.endswith("module.private.modulemap") ||
+        MapFileName.endswith("module_private.map")) {
+      // Adding a top-level module from a private modulemap is likely a
+      // user error; we check to see if there's another top-level module
+      // defined in the non-private map in the same dir, and if so emit a
+      // warning.
+      for (auto E = Map.module_begin(); E != Map.module_end(); ++E) {
+        auto const *M = E->getValue();
+        if (!M->Parent &&
+            M->Directory == ActiveModule->Directory &&
+            M->Name != ActiveModule->Name) {
+          Diags.Report(ActiveModule->DefinitionLoc,
+                       diag::warn_mmap_mismatched_top_level_private)
+            << ActiveModule->Name << M->Name;
+          // The pattern we're defending against here is typically due to
+          // a module named FooPrivate which is supposed to be a submodule
+          // called Foo.Private. Emit a fixit in that case.
+          auto D =
+            Diags.Report(ActiveModule->DefinitionLoc,
+                         diag::note_mmap_rename_top_level_private_as_submodule);
+          D << ActiveModule->Name << M->Name;
+          StringRef Bad(ActiveModule->Name);
+          if (Bad.consume_back("Private")) {
+            SmallString<128> Fixed = Bad;
+            Fixed.append(".Private");
+            D << FixItHint::CreateReplacement(ActiveModule->DefinitionLoc,
+                                              Fixed);
+          }
+          break;
+        }
+      }
+    }
+  }
+
   bool Done = false;
   do {
     switch (Tok.Kind) {
     case MMToken::EndOfFile:
     case MMToken::RBrace:
       Done = true;
       break;
 

cfe/trunk/test/Modules/Inputs/implicit-private-with-different-name/A.framework/Headers/a.h

+extern int APUBLIC;

cfe/trunk/test/Modules/Inputs/implicit-private-with-different-name/A.framework/Headers/aprivate.h

+extern int APRIVATE;

cfe/trunk/test/Modules/Inputs/implicit-private-with-different-name/A.framework/Modules/module.modulemap

+framework module A {
+  header "a.h"
+  export *
+}

cfe/trunk/test/Modules/Inputs/implicit-private-with-different-name/A.framework/Modules/module.private.modulemap

+framework module APrivate {
+  header "aprivate.h"
+  export *
+}

cfe/trunk/test/Modules/implicit-private-with-different-name.m

+// RUN: rm -rf %t
+
+// Build PCH using A, with adjacent private module APrivate, which winds up being implicitly referenced
+// RUN: %clang_cc1 -verify -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -F %S/Inputs/implicit-private-with-different-name -emit-pch -o %t-A.pch %s
+
+// Use the PCH with no explicit way to resolve PrivateA, still pick it up through MODULE_DIRECTORY reference in PCH control block
+// RUN: %clang_cc1 -verify -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -F %S/Inputs/implicit-private-with-different-name -include-pch %t-A.pch %s -fsyntax-only
+
+// Check the fixit
+// RUN: %clang_cc1  -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -F %S/Inputs/implicit-private-with-different-name -include-pch %t-A.pch %s -fsyntax-only -fdiagnostics-parseable-fixits %s 2>&1 | FileCheck %s
+
+// expected-warning@Inputs/implicit-private-with-different-name/A.framework/Modules/module.private.modulemap:1{{top-level module 'APrivate' in private module map, expected a submodule of 'A'}}
+// expected-note@Inputs/implicit-private-with-different-name/A.framework/Modules/module.private.modulemap:1{{make 'APrivate' a submodule of 'A' to ensure it can be found by name}}
+// CHECK: fix-it:"{{.*}}/Inputs/implicit-private-with-different-name/A.framework/Modules/module.private.modulemap":{1:18-1:26}:"A.Private"
+
+#ifndef HEADER
+#define HEADER
+#import "A/aprivate.h"
+const int *y = &APRIVATE;
+#endif