Re: [PATCH] gpio: GPIO_MVEBU should not default to y when compile-testing

[Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>]

Hi Andrew,

On Tue, Dec 3, 2024 at 9:59 PM Andrew Lunn <andrew@xxxxxxx> wrote:
> On Tue, Dec 03, 2024 at 09:29:02PM +0100, Bartosz Golaszewski wrote:
> > On Tue, Dec 3, 2024 at 9:13 PM Andrew Lunn <andrew@xxxxxxx> wrote:
> > >
> > > On Tue, Dec 03, 2024 at 05:23:16PM +0100, Geert Uytterhoeven wrote:
> > > > Merely enabling compile-testing should not enable additional
> > > > functionality.
> > >
> > > Sorry for being dumb, but i don't actually see what is wrong here.
> > >
> > > There are 4 GPIO drivers which have
> > >
> > >         def_bool y

None of these depend on || COMPILE_TEST.

> > > COMPILE_TEST is about building as much as possible, in order to find
> > > build bugs. So i don't get what you mean by additional
> > > functionality. No additional functionality within the MVEBU driver
> > > gets enabled by COMPILE_TEST.
> >
> > No, Geert is right. I have been yelled at by Linus Torvalds already
> > for too eagerly enabling too many options. COMPILE_TEST is really
> > about making it possible to build more things with make allmodconfig,
> > not necessarily defaulting to y for everything.
>
> So are you saying COMPILE_TEST should allow something to be enabled,
> but should not actually enable it?

It is meant for relaxing dependencies, i.e. to allow you to do
build-testing of drivers that cannot possibly work on the system you
are compiling a kernel for.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds