[PATCH] pinctrl: sunxi: Fix recursive dependency

[arnd@arndb.de (Arnd Bergmann)]

On Monday 19 May 2014 16:29:13 Maxime Ripard wrote:
> Hi
> 
> On Mon, May 19, 2014 at 01:30:54PM +0200, Arnd Bergmann wrote:
> > On Saturday 17 May 2014 18:21:55 Maxime Ripard wrote:
> > > Fix the following configuration error:
> > > drivers/pinctrl/sunxi/Kconfig:3:error: recursive dependency detected!
> > > drivers/pinctrl/sunxi/Kconfig:3:        symbol PINCTRL_SUNXI is selected by PINCTRL_SUN4I_A10
> > > drivers/pinctrl/sunxi/Kconfig:9:        symbol PINCTRL_SUN4I_A10 default value contains PINCTRL_SUNXI
> > > 
> > > Drop the select on PINCTRL_SUNXI, and make PINCTRL_SUNXI enabled by default instead.
> > > 
> > > Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
> > 
> > I stubled over the same thing already, but came to a different conclusion.
> > 
> > >  drivers/pinctrl/sunxi/Kconfig | 8 +-------
> > >  1 file changed, 1 insertion(+), 7 deletions(-)
> > > 
> > > diff --git a/drivers/pinctrl/sunxi/Kconfig b/drivers/pinctrl/sunxi/Kconfig
> > > index 2007327e0c48..8411857d2a45 100644
> > > --- a/drivers/pinctrl/sunxi/Kconfig
> > > +++ b/drivers/pinctrl/sunxi/Kconfig
> > > @@ -1,32 +1,26 @@
> > >  if ARCH_SUNXI
> > >  
> > >  config PINCTRL_SUNXI
> > > -       bool
> > > +       def_bool ARCH_SUNXI
> > >         select PINMUX
> > >         select GENERIC_PINCONF
> > >  
> > >  config PINCTRL_SUN4I_A10
> > >         def_bool PINCTRL_SUNXI || MACH_SUN4I
> > > -       select PINCTRL_SUNXI
> > >  
> > 
> > This means you won't be able to stop selecting all drivers dropping
> > the 'select PINCTRL_SUNXI' later. How about this one?
> 
> I'm not sure what you mean here.
> 
> We won't have to select PINCTRL_SUNXI anymore with this patch either,
> since it's already enabled whenever ARCH_SUNXI is enabled.
> 
> And the drivers won't have to be selected either, because they will
> either default to PINCTRL_SUNXI, or to the MACH_SUN*I definition.
> 
> As I was saying in the commit log, this is just a temporary solution,
> and eventually, and the select PINCTRL_SUNXI should be removed
> eventually.

The difference is that in your patch, PINCTRL_SUNXI defaults to 'y'
if ARCH_SUNXI is set, so you have to remove both the 'select' and
the 'def_bool' and then put back the 'select PINCTRL_SUNXI' for the
individual drivers. I guess it's not a big difference, you just end up
doing one patch extra.

	Arnd