Bootstrapped and regtested on x86_64-pc-linux-gnu, OK for trunk/14?
-- >8 --
While looking into PR118846 I noticed that we don't currently constrain
the linkage of functions involving CNTTPs of internal-linkage types. It
seems to me that this would be sensible to do.
PR c++/118849
gcc/cp/ChangeLog:
* decl2.cc (min_vis_expr_r): Constrain visibility according to
the type of decl_constant_var_p decls.
gcc/testsuite/ChangeLog:
* g++.dg/template/linkage6.C: New test.
Signed-off-by: Nathaniel Shead <nathanielosh...@gmail.com>
---
gcc/cp/decl2.cc | 4 +++-
gcc/testsuite/g++.dg/template/linkage6.C | 12 ++++++++++++
2 files changed, 15 insertions(+), 1 deletion(-)
create mode 100644 gcc/testsuite/g++.dg/template/linkage6.C
diff --git a/gcc/cp/decl2.cc b/gcc/cp/decl2.cc
index 4415cea93e0..9a76e00dcde 100644
--- a/gcc/cp/decl2.cc
+++ b/gcc/cp/decl2.cc
@@ -2820,7 +2820,9 @@ min_vis_expr_r (tree *tp, int */*walk_subtrees*/, void
*data)
/* The ODR allows definitions in different TUs to refer to distinct
constant variables with internal or no linkage, so such a reference
shouldn't affect visibility (PR110323). FIXME but only if the
- lvalue-rvalue conversion is applied. */;
+ lvalue-rvalue conversion is applied. We still want to restrict
+ visibility according to the type of the declaration however. */
+ tpvis = type_visibility (TREE_TYPE (t));
else if (! TREE_PUBLIC (t))
tpvis = VISIBILITY_ANON;
else
diff --git a/gcc/testsuite/g++.dg/template/linkage6.C
b/gcc/testsuite/g++.dg/template/linkage6.C
new file mode 100644
index 00000000000..fb589f67874
--- /dev/null
+++ b/gcc/testsuite/g++.dg/template/linkage6.C
@@ -0,0 +1,12 @@
+// { dg-do compile { target c++20 } }
+// { dg-final { scan-assembler-not "(weak|glob)\[^\n\]*_Z" { xfail
powerpc-*-aix* } } }
+
+namespace {
+ struct A {};
+}
+
+template <auto Q> void f() { }
+
+int main() {
+ f<A{}>();
+}
--
2.47.0
