Re: [PATCH v1 10/10] kconfig: Introduce OVERLAY= for make checkconfig
From: Nicolas Schier
Date: Mon Nov 16 2026 - 10:54:00 EST
On Thu, Nov 12, 2026 at 10:31:22PM +0100, Ahmed S. Darwish wrote:
> Hi Nicolas,
>
> On Thu, 12 Nov 2026, Ahmed S. Darwish wrote:
> >
> > On Wed, 11 Nov 2026, Nicolas Schier wrote:
> > >
> > > Please add the optional overlay arguments right here:
> > >
> > > $(Q)$(checkconfig) check -r "$(RULES)" -c "$(KCONFIG_CONFIG)" $(addprefix -o , "$(OVERLAY)")
> > >
> >
> > Sure, will do.
> >
>
> I've just tried the suggested addprefix function:
>
> $ make V=1 RULES=preempt_rt checkconfig
> ...
> python3 -B ./tools/kconfig-check/check.py check -c ".config" -p "preempt_rt" -o ""
>
> Notice the breakage due to -o "".
>
> Unquoting OVERLAY will make it work:
>
> $(Q)$(checkconfig) check -r "$(RULES)" -c "$(KCONFIG_CONFIG)" $(addprefix -o , $(OVERLAY))
>
> But it breaks the consistent pattern of quoting all passed variables.
>
> So, can we keep it as it was originally done?
>
> Namely:
>
> checkconfig_overlay = $(if $(OVERLAY),-o "$(OVERLAY)")
> ...
>
> checkconfig:
> ...
> $(Q)$(checkconfig) check -r "$(RULES)" -c "$(KCONFIG_CONFIG)" $(checkconfig_overlay)
well, ok.
I still do not think that a top-level make target is a good idea. Any
important extension to checkconfig's arguments needs to be properly
mapped, and as you won't have the 'checkconfig' target in the middle of
a series of other targets, I still don't see a good reason for
implementing a top-level Makefile integration. Thus, I still recommend
to skip the top-level Makefile integration.
--
Nicolas