On Mon, 10 Apr 2017 10:49:12 -0700 Richard Henderson <r...@twiddle.net> wrote:
> On 04/10/2017 10:44 AM, Greg Kurz wrote:
> > On Mon, 10 Apr 2017 18:29:57 +0100
> > Peter Maydell <peter.mayd...@linaro.org> wrote:
> >
> >> On 10 April 2017 at 18:26, Richard Henderson <r...@twiddle.net> wrote:
> >>> On 04/09/2017 11:06 PM, Nikunj A Dadhania wrote:
> >>>> --- a/cpus.c
> >>>> +++ b/cpus.c
> >>>> @@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
> >>>> } else if (use_icount) {
> >>>> error_setg(errp, "No MTTCG when icount is enabled");
> >>>> } else {
> >>>> -#ifndef TARGET_SUPPORT_MTTCG
> >>>> +#ifndef TARGET_SUPPORTS_MTTCG
> >>>
> >>>
> >>> This sort of thing is why glibc moved to using -Wundef.
> >>>
> >>> It would be a huge amount of work to convert our existing sources, but it
> >>> would probably pay off in the long run.
> >>
> >> We already build with -Wundef...
> >>
> >
> > From the gcc info page:
> >
> > '-Wundef'
> > Warn if an undefined identifier is evaluated in an '#if' directive.
> >
> > and BTW, isn't the purpose of #ifndef precisely to detect that the
> > identifier is undefined ?
>
> Yes, but it also has the typo problem above, whereas
>
> #if !TARGET_SUPPORTS_MTTCG
>
> plus -Wundef would have caught that error.
>
Turning such #ifdef/#ifndef to #if/#if ! would indeed make sense, but
we'd still have to be careful that every new addition follows the rule.
>
> r~
pgpKpjPhEc27_.pgp
Description: OpenPGP digital signature
