git: 269b7d8ac1e5 - stable/12 - tftpd: tests: raise targeted cstd to c11

[Kyle Evans]

On Tue, Dec 29, 2020 at 9:10 PM Ryan Libby <rlibby at freebsd.org> wrote:
>
> On Tue, Dec 29, 2020 at 5:21 PM Kyle Evans <kevans at freebsd.org> wrote:
> >
> > On Tue, Dec 29, 2020 at 7:00 PM Ryan Libby <rlibby at freebsd.org> wrote:
> > >
> > > On Tue, Dec 29, 2020 at 4:53 PM Kyle Evans <kevans at freebsd.org> wrote:
> > > >
> > > > On Tue, Dec 29, 2020 at 6:50 PM Kyle Evans <kevans at freebsd.org> wrote:
> > > > >
> > > > > On Tue, Dec 29, 2020 at 6:38 PM Ryan Libby <rlibby at freebsd.org> wrote:
> > > > > >
> > > > > > On Tue, Dec 29, 2020 at 3:32 PM Kyle Evans <kevans at freebsd.org> wrote:
> > > > > > >
> > > > > > > Hmm, I hadn't done this because we still have in-tree GCC4.2. Might want to go ahead and skip these tests entirely a level up for 4.2.
> > > > > > >
> > > > > >
> > > > > > Ah, thanks for the reminder.  I'll take another look re in-tree gcc.
> > > > > > Based on the other commits I've MFC'd, I'm pretty sure it's been
> > > > > > broken for months anyway, but I'll see what I can do if the fixes are
> > > > > > easy.
> > > > > >
> > > > >
> > > > > This should not be the case, we've been running arm/mips/etc. in CI up
> > > > > until the git transition.
> > > >
> > > > Sorry, arm is irrelevant for this branch- but mips/powerpc/sparc64.
> > >
> > > Perhaps only x86 was broken...
> > >
> > > How was tftpd/tests not already broken with alignas from r367305 which
> > > MFC'd r358556?
> > >
> >
> > That is an incredibly good question. The latest build for powerpc
> > shows that it did in-fact descend into and build libexec/tftpd/tests
> >
> > > In any case my plan right now is to try to remember the gcc 4.2
> > > voodoo, assess, and fix.  Please let me know if you think this needs
> > > immediate revert before then.
> > >
> >
> > Nah, I wouldn't expend the energy on a revert, tbh. If you don't get
> > it figured out by tomorrow or so, I'll re-test and smack it.
>
> Okay, what's going on is that the tftpd/tests/Makefile was already not
> building the test for gcc < 4.3, due to lack of __COUNTER__.  The use of
> _Alignas and -std=c11 would require gcc 4.7.  If we cared, we would bump
> the minimum gcc version required accordingly.  However, since there's
> no reason to build with gcc >= 4.3 and < 4.7, I don't think we care and
> I think the current state of affairs on stable/12 is acceptable.
>
> I checked powerpc.powerpc64 with in-tree gcc 4.2.1 and buildworld
> succeeded.  Hopefully ci will be up again soon too.
>
> I checked amd64 and it failed due to pkru.  I don't think we care to fix
> that either, but we technically could of course.  For amd64 the external
> toolchain with gcc 6.4 is broken at the moment but will be fixed after a
> few more MFCs.
>
> Thanks for reviewing this MFC!

Excellent, thanks for taking a look (and merging this back)!