[clang] Thread Safety Analysis: Support attributes on function pointers (PR #191187)

Aaron Puchert via cfe-commits cfe-commits at lists.llvm.org
Sun Apr 19 15:39:59 PDT 2026


================
@@ -674,15 +674,28 @@ void Parser::ParseGNUAttributeArgs(
   // These may refer to the function arguments, but need to be parsed early to
   // participate in determining whether it's a redeclaration.
   std::optional<ParseScope> PrototypeScope;
-  if (normalizeAttrName(AttrName->getName()) == "enable_if" &&
-      D && D->isFunctionDeclarator()) {
-    const DeclaratorChunk::FunctionTypeInfo& FTI = D->getFunctionTypeInfo();
-    PrototypeScope.emplace(this, Scope::FunctionPrototypeScope |
-                                     Scope::FunctionDeclarationScope |
-                                     Scope::DeclScope);
-    for (unsigned i = 0; i != FTI.NumParams; ++i)
-      Actions.ActOnReenterCXXMethodParameter(
-          getCurScope(), dyn_cast_or_null<ParmVarDecl>(FTI.Params[i].Param));
+  if (D && (normalizeAttrName(AttrName->getName()) == "enable_if" ||
+            IsAttributeLateParsedStandard(*AttrName))) {
+    // Find the innermost function chunk to make its parameters available for
+    // attribute argument parsing. This is necessary for attributes like thread
+    // safety annotations on function pointers which reference their parameters.
+    for (unsigned i = 0; i < D->getNumTypeObjects(); ++i) {
+      if (D->getTypeObject(i).Kind == DeclaratorChunk::Function) {
+        const DeclaratorChunk::FunctionTypeInfo &FTI = D->getTypeObject(i).Fun;
----------------
aaronpuchert wrote:

This also looks to me like we actually want a type attribute.

https://github.com/llvm/llvm-project/pull/191187


More information about the cfe-commits mailing list