On 3/16/26 10:41 AM, Marek Polacek wrote:
Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
OK, but I'm surprised we even build a PARM_DECL for (void). Why does
that happen?
-- >8 --
Annotations on void parameter don't make much sense so they
should be rejected, even if the wording isn't currently so
clear on that.
PR c++/123618
gcc/cp/ChangeLog:
* tree.cc (handle_annotation_attribute): Reject annotations on void
parameters.
gcc/testsuite/ChangeLog:
* g++.dg/reflect/annotations14.C: New test.
---
gcc/cp/tree.cc | 6 ++++++
gcc/testsuite/g++.dg/reflect/annotations14.C | 5 +++++
2 files changed, 11 insertions(+)
create mode 100644 gcc/testsuite/g++.dg/reflect/annotations14.C
diff --git a/gcc/cp/tree.cc b/gcc/cp/tree.cc
index 20288ed04eb..7296f2b5c0d 100644
--- a/gcc/cp/tree.cc
+++ b/gcc/cp/tree.cc
@@ -5912,6 +5912,12 @@ handle_annotation_attribute (tree *node, tree ARG_UNUSED
(name),
*no_add_attrs = true;
return NULL_TREE;
}
+ if (TREE_CODE (*node) == PARM_DECL && VOID_TYPE_P (TREE_TYPE (*node)))
+ {
+ error ("annotation on void parameter");
+ *no_add_attrs = true;
+ return NULL_TREE;
+ }
if (!type_dependent_expression_p (TREE_VALUE (args)))
{
if (!structural_type_p (TREE_TYPE (TREE_VALUE (args))))
diff --git a/gcc/testsuite/g++.dg/reflect/annotations14.C
b/gcc/testsuite/g++.dg/reflect/annotations14.C
new file mode 100644
index 00000000000..d22a9fff3ff
--- /dev/null
+++ b/gcc/testsuite/g++.dg/reflect/annotations14.C
@@ -0,0 +1,5 @@
+// PR c++/123618
+// { dg-do compile { target c++26 } }
+// { dg-additional-options "-freflection" }
+
+void f([[=1]] void) {} // { dg-error "annotation on void parameter" }
base-commit: 2894493adf7434a83b9ead2617468e817b337126
