Re: [GIT PULL] fbdev fixes and updates for v7.0-rc1

Re: [GIT PULL] fbdev fixes and updates for v7.0-rc1

[Linus Torvalds]

[ Adding Kconfig maintainers and linux-kbuild list ]

On Sat, 14 Feb 2026 at 13:30, Helge Deller <deller@kernel.org> wrote:
>
> Linus, I'm really sorry, but I messed up drivers/gpu/drm/Kconfig while
> trying to fix a merge conflict.
> My patch series should not have touched drivers/gpu/drm/Kconfig at all.
> That's purely my fault and not the fault of the patch author.

Humm. Funky how the Kconfig parts never complained about the
duplication of all those source lines, so the problem was basically
entirely hidden and things still "worked" even though that Kconfig
file had been so messed up.

I'm not sure if the Kconfig tools could perhaps warn about this kind
of duplication - we might have some of it intentionally - but it does
make me go "Hmm".

Nathan, Nicolas, comments? See that commit ca4ee40bf13d for the
partial revert, and notice how Kconfig is entirely happy both before
and after that..

                Linus

Re: [GIT PULL] fbdev fixes and updates for v7.0-rc1

[Nathan Chancellor]

Hi Linus,

On Sat, Feb 14, 2026 at 02:47:47PM -0800, Linus Torvalds wrote:
> [ Adding Kconfig maintainers and linux-kbuild list ]
> 
> On Sat, 14 Feb 2026 at 13:30, Helge Deller <deller@kernel.org> wrote:
> >
> > Linus, I'm really sorry, but I messed up drivers/gpu/drm/Kconfig while
> > trying to fix a merge conflict.
> > My patch series should not have touched drivers/gpu/drm/Kconfig at all.
> > That's purely my fault and not the fault of the patch author.
> 
> Humm. Funky how the Kconfig parts never complained about the
> duplication of all those source lines, so the problem was basically
> entirely hidden and things still "worked" even though that Kconfig
> file had been so messed up.
> 
> I'm not sure if the Kconfig tools could perhaps warn about this kind
> of duplication - we might have some of it intentionally - but it does
> make me go "Hmm".
> 
> Nathan, Nicolas, comments? See that commit ca4ee40bf13d for the
> partial revert, and notice how Kconfig is entirely happy both before
> and after that..

It seems like we should be able to check if we have seen an sourced
Kconfig already, presumably somewhere in or around zconf_nextfile() in
scripts/kconfig/lexer.l. Not sure how complicated it will be, I will see
if I can wire something like that up during the next development cycle
(amongst the other things on my plate). Given how wonky that looks in
menuconfig and the like, maybe it is worth making that a hard error.

Cheers,
Nathan