Reland "[clang] Delay normalization of -fmodules-cache-path (llvm#150123)"

This reverts commit 613caa9, essentially
reapplying 4a4bdde after moving
`normalizeModuleCachePath` from clangFrontend to clangLex.

This PR is part of an effort to remove file system usage from the
command line parsing code. The reason for that is that it's impossible
to do file system access correctly without a configured VFS, and the VFS
can only be configured after the command line is parsed. I don't want to
intertwine command line parsing and VFS configuration, so I decided to
perform the file system access after the command line is parsed and the
VFS is configured - ideally right before the file system entity is used
for the first time.

This patch delays normalization of the module cache path until
`CompilerInstance` is asked for the cache path in the current
compilation context.

(cherry picked from commit 55bef46)

Author: jansvoboda11

clang/include/clang/Driver/Options.td

@@ -3399,7 +3399,8 @@ defm declspec : BoolOption<"f", "declspec",
 def fmodules_cache_path : Joined<["-"], "fmodules-cache-path=">, Group<i_Group>,
   Flags<[]>, Visibility<[ClangOption, CC1Option]>,
   MetaVarName<"<directory>">,
-  HelpText<"Specify the module cache path">;
+  HelpText<"Specify the module cache path">,
+  MarshallingInfoString<HeaderSearchOpts<"ModuleCachePath">>;
 def fmodules_user_build_path : Separate<["-"], "fmodules-user-build-path">, Group<i_Group>,
   Flags<[]>, Visibility<[ClangOption, CC1Option]>,
   MetaVarName<"<directory>">,

clang/include/clang/Lex/HeaderSearch.h

@@ -986,6 +986,9 @@ void ApplyHeaderSearchOptions(HeaderSearch &HS,
                               const LangOptions &Lang,
                               const llvm::Triple &triple);
 
+void normalizeModuleCachePath(FileManager &FileMgr, StringRef Path,
+                              SmallVectorImpl<char> &NormalizedPath);
+
 } // namespace clang
 
 #endif // LLVM_CLANG_LEX_HEADERSEARCH_H

clang/lib/Frontend/CompilerInstance.cpp

@@ -559,9 +559,16 @@ void CompilerInstance::createPreprocessor(TranslationUnitKind TUKind) {
 }
 
 std::string CompilerInstance::getSpecificModuleCachePath(StringRef ModuleHash) {
+  assert(FileMgr && "Specific module cache path requires a FileManager");
+
+  if (getHeaderSearchOpts().ModuleCachePath.empty())
+    return "";
+
   // Set up the module path, including the hash for the module-creation options.
-  SmallString<256> SpecificModuleCache(getHeaderSearchOpts().ModuleCachePath);
-  if (!SpecificModuleCache.empty() && !getHeaderSearchOpts().DisableModuleHash)
+  SmallString<256> SpecificModuleCache;
+  normalizeModuleCachePath(*FileMgr, getHeaderSearchOpts().ModuleCachePath,
+                           SpecificModuleCache);
+  if (!getHeaderSearchOpts().DisableModuleHash)
     llvm::sys::path::append(SpecificModuleCache, ModuleHash);
   return std::string(SpecificModuleCache);
 }

clang/lib/Frontend/CompilerInvocation.cpp

@@ -3577,9 +3577,6 @@ static void GenerateHeaderSearchArgs(const HeaderSearchOptions &Opts,
   if (Opts.UseLibcxx)
     GenerateArg(Consumer, OPT_stdlib_EQ, "libc++");
 
-  if (!Opts.ModuleCachePath.empty())
-    GenerateArg(Consumer, OPT_fmodules_cache_path, Opts.ModuleCachePath);
-
   for (const auto &File : Opts.PrebuiltModuleFiles)
     GenerateArg(Consumer, OPT_fmodule_file, File.first + "=" + File.second);
 
@@ -3682,8 +3679,7 @@ static void GenerateHeaderSearchArgs(const HeaderSearchOptions &Opts,
 }
 
 static bool ParseHeaderSearchArgs(HeaderSearchOptions &Opts, ArgList &Args,
-                                  DiagnosticsEngine &Diags,
-                                  const std::string &WorkingDir) {
+                                  DiagnosticsEngine &Diags) {
   unsigned NumErrorsBefore = Diags.getNumErrors();
 
   HeaderSearchOptions *HeaderSearchOpts = &Opts;
@@ -3696,17 +3692,6 @@ static bool ParseHeaderSearchArgs(HeaderSearchOptions &Opts, ArgList &Args,
   if (const Arg *A = Args.getLastArg(OPT_stdlib_EQ))
     Opts.UseLibcxx = (strcmp(A->getValue(), "libc++") == 0);
 
-  // Canonicalize -fmodules-cache-path before storing it.
-  SmallString<128> P(Args.getLastArgValue(OPT_fmodules_cache_path));
-  if (!(P.empty() || llvm::sys::path::is_absolute(P))) {
-    if (WorkingDir.empty())
-      llvm::sys::fs::make_absolute(P);
-    else
-      llvm::sys::fs::make_absolute(WorkingDir, P);
-  }
-  llvm::sys::path::remove_dots(P);
-  Opts.ModuleCachePath = std::string(P);
-
   // Only the -fmodule-file=<name>=<file> form.
   for (const auto *A : Args.filtered(OPT_fmodule_file)) {
     StringRef Val = A->getValue();
@@ -5561,8 +5546,7 @@ bool CompilerInvocation::CreateFromArgsImpl(
   InputKind DashX = Res.getFrontendOpts().DashX;
   ParseTargetArgs(Res.getTargetOpts(), Args, Diags);
   llvm::Triple T(Res.getTargetOpts().Triple);
-  ParseHeaderSearchArgs(Res.getHeaderSearchOpts(), Args, Diags,
-                        Res.getFileSystemOpts().WorkingDir);
+  ParseHeaderSearchArgs(Res.getHeaderSearchOpts(), Args, Diags);
   if (Res.getFrontendOpts().GenReducedBMI ||
       Res.getFrontendOpts().ProgramAction ==
           frontend::GenerateReducedModuleInterface ||

clang/lib/Lex/HeaderSearch.cpp

@@ -2183,3 +2183,10 @@ std::string HeaderSearch::suggestPathToFileForDiagnostics(
   }
   return path::convert_to_slash(Filename);
 }
+
+void clang::normalizeModuleCachePath(FileManager &FileMgr, StringRef Path,
+                                     SmallVectorImpl<char> &NormalizedPath) {
+  NormalizedPath.assign(Path.begin(), Path.end());
+  FileMgr.makeAbsolutePath(NormalizedPath);
+  llvm::sys::path::remove_dots(NormalizedPath);
+}

clang/lib/Serialization/ASTWriter.cpp

@@ -57,6 +57,7 @@
 #include "clang/Basic/TargetInfo.h"
 #include "clang/Basic/TargetOptions.h"
 #include "clang/Basic/Version.h"
+#include "clang/Frontend/CompilerInstance.h"
 #include "clang/Lex/HeaderSearch.h"
 #include "clang/Lex/HeaderSearchOptions.h"
 #include "clang/Lex/MacroInfo.h"
@@ -1731,9 +1732,13 @@ void ASTWriter::WriteControlBlock(Preprocessor &PP, StringRef isysroot) {
   const HeaderSearchOptions &HSOpts =
       PP.getHeaderSearchInfo().getHeaderSearchOpts();
 
+  SmallString<256> HSOpts_ModuleCachePath;
+  normalizeModuleCachePath(PP.getFileManager(), HSOpts.ModuleCachePath,
+                           HSOpts_ModuleCachePath);
+
   AddString(HSOpts.Sysroot, Record);
   AddString(HSOpts.ResourceDir, Record);
-  AddString(HSOpts.ModuleCachePath, Record);
+  AddString(HSOpts_ModuleCachePath, Record);
   AddString(HSOpts.ModuleUserBuildPath, Record);
   Record.push_back(HSOpts.DisableModuleHash);
   Record.push_back(HSOpts.ImplicitModuleMaps);

clang/lib/Tooling/DependencyScanning/DependencyScanningWorker.cpp

@@ -609,9 +609,10 @@ class DependencyScanningAction : public tooling::ToolAction {
 
     // Use the dependency scanning optimized file system if requested to do so.
     if (DepFS) {
-      StringRef ModulesCachePath =
-          ScanInstance.getHeaderSearchOpts().ModuleCachePath;
-
+      SmallString<256> ModulesCachePath;
+      normalizeModuleCachePath(
+          *FileMgr, ScanInstance.getHeaderSearchOpts().ModuleCachePath,
+          ModulesCachePath);
       DepFS->resetBypassedPathPrefix();
       if (!ModulesCachePath.empty())
         DepFS->setBypassedPathPrefix(ModulesCachePath);

clang/test/Modules/modules-cache-path-canonicalization-output.c

@@ -0,0 +1,18 @@
+// This checks that implicitly-built modules produce identical PCM
+// files regardless of the spelling of the same module cache path.
+
+// RUN: rm -rf %t
+// RUN: split-file %s %t
+
+// RUN: %clang_cc1 -fmodules -fimplicit-module-maps -fsyntax-only %t/tu.c \
+// RUN:   -fmodules-cache-path=%t/cache -fdisable-module-hash
+// RUN: mv %t/cache/M.pcm %t/M.pcm
+// RUN: %clang_cc1 -fmodules -fimplicit-module-maps -fsyntax-only %t/tu.c \
+// RUN:   -fmodules-cache-path=%t/./cache -fdisable-module-hash
+// RUN: diff %t/./cache/M.pcm %t/M.pcm
+
+//--- tu.c
+#include "M.h"
+//--- M.h
+//--- module.modulemap
+module M { header "M.h" }