Author: Danny Milosavljevic <dannym@friendly-machines.com>
Date: 2025-10-20
Subject: Make clang importer heed CPLUS_INCLUDE_PATH

--- a/include/swift/ClangImporter/ClangImporter.h	1970-01-01 00:00:01.000000000 +0000
+++ b/include/swift/ClangImporter/ClangImporter.h	2025-10-21 14:31:53.328441417 +0000
@@ -787,6 +787,9 @@
   SmallVector<std::pair<std::string, std::string>, 1> overridenFiles;
 
   bool requiresBuiltinHeadersInSystemModules;
+
+  /// Extra arguments to pass to the Clang invocation (e.g., -isystem paths).
+  std::vector<std::string> extraArgs;
 };
 
 /// On Linux, some platform libraries (glibc, libstdc++) are not modularized.
--- a/lib/ClangImporter/ClangImporter.cpp	1970-01-01 00:00:01.000000000 +0000
+++ b/lib/ClangImporter/ClangImporter.cpp	2025-10-21 14:34:20.744234737 +0000
@@ -1326,6 +1326,12 @@
   importer->requiresBuiltinHeadersInSystemModules =
       fileMapping.requiresBuiltinHeadersInSystemModules;
 
+  // Add extra arguments from file mapping (e.g., C++ stdlib include paths).
+  ctx.ClangImporterOpts.ExtraArgs.insert(
+      ctx.ClangImporterOpts.ExtraArgs.end(),
+      fileMapping.extraArgs.begin(),
+      fileMapping.extraArgs.end());
+
   // Create a new Clang compiler invocation.
   {
     if (auto ClangArgs = importer->getClangCC1Arguments(ctx, VFS))
--- a/lib/ClangImporter/ClangIncludePaths.cpp	1970-01-01 00:00:01.000000000 +0000
+++ b/lib/ClangImporter/ClangIncludePaths.cpp	2025-10-21 14:32:26.864394627 +0000
@@ -21,6 +21,7 @@
 #include "clang/Driver/ToolChain.h"
 #include "clang/Frontend/CompilerInstance.h"
 #include "llvm/WindowsDriver/MSVCPaths.h"
+#include <cstdlib>
 
 using namespace swift;
 
@@ -306,9 +307,71 @@
                                      {"cstdlib", "string", "vector"}, vfs)) {
     cxxStdlibDir = dir.value();
   } else {
-    if (!suppressDiagnostic)
-      ctx.Diags.diagnose(SourceLoc(), diag::libstdcxx_not_found, triple.str());
-    return;
+    // FALLBACK: Try reading CPLUS_INCLUDE_PATH from environment.
+    // This is needed because Swift doesn't go through the full clang driver
+    // command construction that processes environment variables into -cxx-isystem
+    // arguments. When AddClangCXXStdlibIncludeArgs fails to find GCC (e.g., due
+    // to --gcc-toolchain pointing to wrong location or GCC detection failing),
+    // parsedStdlibArgs will be empty and findFirstIncludeDir returns nullopt.
+    // In Guix, CPLUS_INCLUDE_PATH is set by the build environment.
+    const char *cplusIncludePath = std::getenv("CPLUS_INCLUDE_PATH");
+    if (cplusIncludePath) {
+      llvm::SmallVector<llvm::StringRef, 8> envPaths;
+      llvm::StringRef(cplusIncludePath).split(envPaths, ':', -1, false);
+
+      for (const auto &envPath : envPaths) {
+        Path dir(envPath.str());
+        // Check if this directory contains C++ stdlib headers
+        Path testFile(dir);
+        llvm::sys::path::append(testFile, "cstdlib");
+        if (vfs->exists(testFile)) {
+          llvm::sys::path::remove_dots(dir, /*remove_dot_dot=*/true);
+          cxxStdlibDir = dir;
+          break;
+        }
+      }
+    }
+
+    if (cxxStdlibDir.empty()) {
+      if (!suppressDiagnostic)
+        ctx.Diags.diagnose(SourceLoc(), diag::libstdcxx_not_found, triple.str());
+      return;
+    }
+  }
+
+  // ADD INCLUDE PATHS FOR ARCH-SPECIFIC HEADERS:
+  // GCC's libstdc++ has architecture-specific headers in subdirectories.
+  // Standard clang driver adds THREE separate -internal-isystem include paths:
+  //   1. /path/to/gcc/include/c++            (base: cstdlib, string, vector, etc.)
+  //   2. /path/to/gcc/include/c++/TRIPLE     (arch: bits/c++config.h, ext/*, etc.)
+  //   3. /path/to/gcc/include/c++/backward   (backward compatibility headers)
+  // This allows clang to find both generic headers (cstddef) and arch-specific
+  // headers (bits/c++config.h) in their respective directories.
+  //
+  // We add these three paths as -isystem arguments via fileMapping.extraArgs.
+  // This is equivalent to what clang's AddClangCXXStdlibIncludeArgs does when
+  // it calls addSystemInclude three times in Gnu.cpp:addLibStdCXXIncludePaths.
+  // Using -isystem instead of VFS redirects ensures error messages reference
+  // real filesystem paths that users can verify exist.
+
+  // 1. Base directory (already found above in cxxStdlibDir)
+  fileMapping.extraArgs.push_back("-isystem");
+  fileMapping.extraArgs.push_back(std::string(cxxStdlibDir));
+
+  // 2. Architecture-specific subdirectory (e.g., x86_64-unknown-linux-gnu)
+  Path archSpecificDir(cxxStdlibDir);
+  llvm::sys::path::append(archSpecificDir, triple.str());
+  if (vfs->exists(archSpecificDir)) {
+    fileMapping.extraArgs.push_back("-isystem");
+    fileMapping.extraArgs.push_back(std::string(archSpecificDir));
+  }
+
+  // 3. Backward compatibility directory
+  Path backwardDir(cxxStdlibDir);
+  llvm::sys::path::append(backwardDir, "backward");
+  if (vfs->exists(backwardDir)) {
+    fileMapping.extraArgs.push_back("-isystem");
+    fileMapping.extraArgs.push_back(std::string(backwardDir));
   }
 
   Path actualModuleMapPath;