[PATCH 1/2 v3] ARM: s3c24xx: get rid of custom <mach/gpio.h>

[arnd@arndb.de (Arnd Bergmann)]

On Tuesday 14 January 2014 11:42:08 Linus Walleij wrote:
> On Wed, Jan 8, 2014 at 4:59 PM, Arnd Bergmann <arnd@arndb.de> wrote:
> 
> > I've asked the mighty grep and it said that these are the files
> > using the contents you moved around:
> >
> > drivers/gpio/gpio-samsung.c
> > drivers/leds/leds-s3c24xx.c
> > drivers/mmc/host/s3cmci.c
> > sound/soc/samsung/h1940_uda1380.c
> > sound/soc/samsung/neo1973_wm8753.c
> > sound/soc/samsung/rx1950_uda1380.c
> > sound/soc/samsung/s3c2412-i2s.c
> > sound/soc/samsung/s3c24xx-i2s.c
> > sound/soc/samsung/smartq_wm8987.c
> >
> > The symbols used in these drivers are:
> >
> > S3C....GP.()
> > S3C_GPIO_END
> > S3C_GPIO_PULL_.*
> > S3C_GPIO_SFN()
> > s3c_gpio_cfgall_range()
> > s3c_gpio_setpull()
> 
> Hm I will look closer at these. Maybe I can even fix some to just
> pass GPIO thru platform data and end-of-story them.
> Targeting v3.15.

Ok, sounds good. That will take care of the first two of this
list, which are the most common ones. The others are from
plat/gpio-cfg.h.

> But I guess you want me to move the
> <linux/platform_data/gpio-s3c*> back to <mach/*>
> before v3.14 so focusing on that now. (These patches
> cannot be just reverted as people have fixed other
> Samsung GPIO issues on top of them, so it needs to be done
> carefully.)

Yes, that would be good. Maybe <mach/gpio-samsung.h> for the
contents rather than putting them back into the common mach/gpio.h?

> > I think it would be worthwhile to go ahead and make all inclusions of
> > the plat/gpio-cfg.h file explicit, and remove it from mach/gpio.h, that
> > part should be simple enough and helpful at the same time.
> 
> <plat/gpio-cfg.h> is used by S5P as well, but I guess you've
> checked that.
> 
> If I take it driver-by-driver it should be possible to get rid of it
> or most of it I think, reducing it to the arch/arm/mach-* folder.
> 
> Certainly I should be attacking S3C and S5P simultaneously.

That makes sense, but note that there is also work going on to make
S5P64xx and S5PV210 multiplatform and DT-only and multiplatform,
and to deprecate S5PC100. If those things happen first, you no longer
have to worry about these platforms as they won't be using this
driver any more.

	Arnd