svn commit: r367647 - head/tests/sys/vm

[Kyle Evans]

On Fri, Nov 13, 2020 at 12:50 PM Adrian Chadd <adrian at freebsd.org> wrote:
>
> Author: adrian
> Date: Fri Nov 13 18:50:24 2020
> New Revision: 367647
> URL: https://svnweb.freebsd.org/changeset/base/367647
>
> Log:
>   [tests] Fix unused variable warning in gcc
>
>   Reviewed by:  markj, imp, cem,
>   Approved by:  markj
>   Differential Revision:        https://reviews.freebsd.org/D26792
>
> Modified:
>   head/tests/sys/vm/page_fault_signal.c
>
> Modified: head/tests/sys/vm/page_fault_signal.c
> ==============================================================================
> --- head/tests/sys/vm/page_fault_signal.c       Fri Nov 13 18:34:13 2020        (r367646)
> +++ head/tests/sys/vm/page_fault_signal.c       Fri Nov 13 18:50:24 2020        (r367647)
> @@ -115,7 +114,7 @@ ATF_TC_BODY(page_fault_signal__segv_accerr_2, tc)
>         ATF_REQUIRE(p != MAP_FAILED);
>         if (sigsetjmp(sig_env, 1) == 0) {
>                 setup_signals();
> -               dummy = *p;
> +               (void)*(volatile int *)p;
>         }
>         (void)munmap(p, sz);
>         ATF_CHECK_EQ(SIGSEGV, last_sig);

A minor nit, we could/should probably add an explicit atf_tc_fail()
after the access. While it seems unlikely that the compiler might
elide it and it would still fail anyways because last_sig would
certainly not be SIGSEGV, it'd be good to catch it earlier with an
explicit fail. I don't recall how good atf-c's diagnostics are
otherwise, if it'd be clear that last_sig == 0 or not without
requiring further triage.

Thanks,

Kyle Evans