On Fri, Aug 14, 2026 at 12:31 PM Matt Turner <[email protected]> wrote:
>
> The test carried two dg-do compile directives: one restricting it to x86
> and, added later, one excluding only hppa. DejaGnu honors the last one, so
> the x86 restriction never had any effect and the test has run everywhere
> since it was added, which is why hppa needed excluding at all.
>
> Require fstack_protector instead, so the test runs wherever the feature
> exists. This fixes it on alpha, which does not support -fstack-protector
> and warns that the option is unsupported. The hppa exclusion becomes
> redundant, as check_effective_target_fstack_protector already returns 0
> there.
>
> gcc/testsuite/ChangeLog:
>
> * g++.dg/no-stack-protector-attr.C: Remove the dead dg-do directive
> and the hppa exclusion, and require fstack_protector.
Ok.
> ---
> gcc/testsuite/g++.dg/no-stack-protector-attr.C | 5 ++---
> 1 file changed, 2 insertions(+), 3 deletions(-)
>
> diff --git ./gcc/testsuite/g++.dg/no-stack-protector-attr.C
> ./gcc/testsuite/g++.dg/no-stack-protector-attr.C
> index 3314a94bd7b..a285c551795 100644
> --- ./gcc/testsuite/g++.dg/no-stack-protector-attr.C
> +++ ./gcc/testsuite/g++.dg/no-stack-protector-attr.C
> @@ -1,11 +1,10 @@
> /* PR c/94722 */
> /* Test that stack protection is disabled via no_stack_protector attribute.
> */
>
> -/* { dg-do compile { target i?86-*-* x86_64-*-* } } */
> +/* { dg-do compile } */
> +/* { dg-require-effective-target fstack_protector } */
> /* { dg-options "-O2 -fstack-protector-all" } */
>
> -/* { dg-do compile { target { ! hppa*-*-* } } } */
> -
> int __attribute__((no_stack_protector)) c()
> {
> int a;
> --
> 2.54.0
>
