Re: [RFC PATCH] i386: don't include lib/i386/reset.c in EFI builds

[Leif Lindholm]

On Fri, Jul 13, 2018 at 03:59:38PM +0200, Daniel Kiper wrote:
> On Fri, Jul 13, 2018 at 01:53:52PM +0100, Leif Lindholm wrote:
> > On Fri, Jul 13, 2018 at 01:27:08PM +0200, Daniel Kiper wrote:
> > > > > > (i386_)ieee1275 implements its own grub_reboot(), so that should be
> > > > > > fine. (This does mean that i386_ieee1275 may currently be unable to
> > > > > > load the reboot module on master.)
> > > > >
> > > > > Hmmm... So, it looks that your solution is safer. Then
> > > > >
> > > > > Reviewed-by: Daniel Kiper <address@hidden>
> > > > >
> > > > > If there are no objections I will apply this in a week or so.
> > > >
> > > > In that case, I think it may be worth extending the test to
> > > >
> > > > #if !defined (GRUB_MACHINE_EFI) && !defined (GRUB_MACHINE_IEEE1275)
> > > >
> > > > I had not noticed that bit when I sent the original patch.
> > > >
> > > > But this is theorising based on looking at source code without
> > > > testing.
> > >
> > > Do you think about lib/ieee1275/reboot.c? It is used on PowerPC and SPARC 
> > > only.
> > > So, It seems to me that we do not need "!defined (GRUB_MACHINE_IEEE1275)" 
> > > here.
> >
> > Oh, right.
> >
> > Then I think we still have a problem with I386_IEEE1275, but am less
> > sure how to deal with it.
> 
> I have just build the i386-ieee1275 platform. It looks that the platform
> uses standard i386 reboot mechanism. So, I would put #ifndef GRUB_MACHINE_EFI
> like it was in original patch.

Yes, you are correct. This is handled (as you said) by i386_ieee1275
not including lib/ieee1275/reboot.c.

And indeed, since these are both in the reboot module (which I had
somehow managed to miss), it would have triggered a build-time fault
if it had been an issue.

Apologies for the noise.

/
    Leif