On Wed, 2 May 2012 01:26:12 +0200, Linus Walleij <linus.walleij@linaro.org> wrote:
> On Mon, Apr 30, 2012 at 8:52 AM, Sachin Kamat <sachin.kamat@linaro.org> wrote:
> 
> > +#if defined(CONFIG_ARCH_EXYNOS4) || defined(CONFIG_ARCH_EXYNOS5)
> > +#endif
> > -static struct samsung_gpio_chip exynos4_gpios_1[] = {
> >  #ifdef CONFIG_ARCH_EXYNOS4
> > +static struct samsung_gpio_chip exynos4_gpios_1[] = {
> > -#endif
> > +#endif
> > -static struct samsung_gpio_chip exynos4_gpios_2[] = {
> >  #ifdef CONFIG_ARCH_EXYNOS4
> > +static struct samsung_gpio_chip exynos4_gpios_2[] = {
> > -#endif
> > +#endif
> > -static struct samsung_gpio_chip exynos4_gpios_3[] = {
> >  #ifdef CONFIG_ARCH_EXYNOS4
> > +static struct samsung_gpio_chip exynos4_gpios_3[] = {
> > -#endif
> > +#endif
> >  #ifdef CONFIG_ARCH_EXYNOS5
> > +#if defined(CONFIG_CPU_EXYNOS4210) || defined(CONFIG_SOC_EXYNOS5250)
> > +#endif
> > +#if defined(CONFIG_CPU_EXYNOS4210) || defined(CONFIG_SOC_EXYNOS5250)
> > +#endif
> 
> We really want to get rid of this kind of stuff from all drivers, #ifdefs are
> declared ugly already in Documentation/CodingStyle.
> 
> Any chance you could solve this problem by reworking the driver to
> pass some flag in platform data tell which exynos it's for
> and jist adapt at runtime instead of the compile-time quirkiness?
> 
> Besides looking better, it helps us to get to a single zImage for the
> exynoses too..

I don't see any single-zImage issues here.  There are no #else clauses
in the #ifdef blocks so all it does it compile out unused code when
exynos4 & 5 is not enabled. I'm going to apply it.

g.