https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91947
--- Comment #13 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> It's done the same way as we've always created the
> ${host_builddir}/c++config.h target, so I'm not sure why one is racy and the
> other isn't.
I didn't mean that the first one was not racy though ;-) But it has got only 3
>> while the new one has got 13 of them, so this can make a difference in
practice.
The issue happened on a Windows host for the 9 branch so you never know, but we
never had such a thing before you had backported the change onto that branch.
