[PATCH/RFC] ARM: shmobile: Disallow PINCTRL without GPIOLIB

[Simon Horman]

On Tue, Mar 26, 2013 at 04:23:58PM +0100, Laurent Pinchart wrote:
> Hi Magnus,
> 
> On Tuesday 26 March 2013 13:04:25 Magnus Damm wrote:
> > On Thu, Mar 21, 2013 at 12:28 AM, Laurent Pinchart wrote:
> > > On Monday 18 March 2013 22:58:18 Magnus Damm wrote:
> > >> From: Magnus Damm <damm at opensource.se>
> > >> 
> > >> Modify mach-shmobile to only select PINCTRL in case of
> > >> ARCH_WANT_OPTIONAL_GPIOLIB is set.
> > >> 
> > >> This fixes a build error triggered when adding a new SoC
> > >> 
> > >> lacking GPIO software support (ARCH_WANT_OPTIONAL_GPIOLIB=n):
> > >>  CC      drivers/tty/vt/keyboard.o
> > >> 
> > >> In file included from drivers/pinctrl/core.c:30:0:
> > >> include/asm-generic/gpio.h: In function 'gpio_get_value_cansleep':
> > >> include/asm-generic/gpio.h:270:2: error: implicit declaration of function
> > >> '__gpio_get_value'
> > >> include/asm-generic/gpio.h: In function 'gpio_set_value_cansleep':
> > >> include/asm-generic/gpio.h:276:2: error: implicit declaration of function
> > >> '__gpio_set_value'
> > >> drivers/pinctrl/core.c: In function 'pinctrl_ready_for_gpio_range':
> > >> drivers/pinctrl/core.c:297:9: error: implicit declaration of function
> > >> 'gpio_to_chip'
> > >> drivers/pinctrl/core.c:297:27: warning: initialization makes pointer from
> > >> integer without a cast
> > >> drivers/pinctrl/core.c:304:45: error: dereferencing pointer to incomplete
> > >> type
> > >> drivers/pinctrl/core.c:305:26: error: dereferencing pointer to incomplete
> > >> type
> > >> drivers/pinctrl/core.c:305:39: error: dereferencing pointer to incomplete
> > >> type
> > >> make[2]: *** [drivers/pinctrl/core.o] Error 1
> > >> make[1]: *** [drivers/pinctrl] Error 2
> > >> make[1]: *** Waiting for unfinished jobs....
> > >> 
> > >>   LD      drivers/sh/built-in.o
> > >> 
> > >> Signed-off-by: Magnus Damm <damm at opensource.se>
> > >> 
> > >> ---
> > >> 
> > >>  Using renesas.git "next" 811689afc214564c4a5f238ecf4d8bdc0e52b615
> > >>  
> > >>  Trigger using the r8a73a4 patches that lack GPIO and PFC support.
> > >>  
> > >>  I am more than happy to replace this patch with something cleaner.
> > > 
> > > If I'm not mistaken your patch fixes the compilation breakage by
> > > unselecting PINCTRL and making it possible no to select GPIOLIB. I'm fine
> > > with that as an interim solution, but I wonder whether we shouldn't just
> > > force PINCTRL and GPIOLIB for ARCH_SHMOBILE at some point.
> > 
> > Thanks for your comments. I'm quite fine with any approach myself. My
> > main concern is to have to the mach-shmobile code in a state so it is
> > possible to add SoC support incrementally and start without any PFC or
> > GPIO and then add them one by one. Or add them together. Please note
> > that EMEV2 supports GPIO but not yet PINCTRL.
> 
> Right. In that case your patch is fine.
> 
> > About this issue, the fact that selecting PINCTRL without GPIO results
> > in compile error makes me think that something needs slight adjustment
> > in the PFC code.
> 
> The above errors come from the pinctrl core (although the PFC driver might not 
> compile either in this case). Linus, what's your opinion on this ? Do we want 
> to support systems with pinctrl but without gpiolib ?
> 
> > Or perhaps PINCTRL without GPIO isn't a valid combination? But if so, why do
> > we have separate Kconfig entries?

I am currently planning to apply this patch to the soc branch
of the renesas tree as Magnus's "[PATCH 00/04] ARM: shmobile: r8a73a4 SoC
support V3" series depends on it and I'd rather not block that series.

I realise that this patch is a temporary fix. A more permanent
solution can be based on top of it.