On Fri, Aug 28, 2009 at 11:57 AM, Florian Fainelli[off-list ref] wrote:

Le Friday 28 August 2009 11:26:58 Manuel Lauss, vous avez écrit :

quoted

I wrote:

quoted

Ralf Baechle wrote:

quoted

  CC      drivers/input/keyboard/gpio_keys.o
/home/ralf/src/linux/linux-mips/drivers/input/keyboard/gpio_keys.c: In
function ‘gpio_keys_probe’:
/home/ralf/src/linux/linux-mips/drivers/input/keyboard/gpio_keys.c:123:
error: implicit declaration of function ‘gpio_request’
/home/ralf/src/linux/linux-mips/drivers/input/keyboard/gpio_keys.c:135:
error: implicit declaration of function ‘gpio_free’ make[5]: ***
[drivers/input/keyboard/gpio_keys.o] Error 1
make[4]: *** [drivers/input/keyboard] Error 2
make[3]: *** [drivers/input] Error 2
make[2]: *** [drivers] Error 2
make[1]: *** [sub-make] Error 2

Either something like the patch below, or adding stubs for
gpio_request/gpio_free to asm/mach-au1x00/gpio-au1000.h in the
CONFIG_GPIOLIB=n case should fix it.

Florian, Ralf, I prefer the latter approach;  saves everyone from
having to add #ifdef CONFIG_GPIOLIB around gpio_request() calls.

Here's an untested patch.  What do you think?  If it works for you, please
add it to your patchqueue!

Thanks!

---

From: Manuel Lauss <manuel.lauss@gmail.com>
Subject: [PATCH] Alchemy: add gpio_request/gpio_free stubs for
CONFIG_GPIOLIB=n

Some drivers use gpio_request/gpio_free regardless of whether
gpiolib is actually built;  add stubs to work around the ensuing
compile failures.

This is better, though fixing the gpio keyboard driver might probably be a good approach.

As I wrote above, I don't think adding "#ifdef CONFIG_GPIOLIB" to all
in-kernel users
of gpio_request() is an acceptable solution; one would be to
unconditionally enable GPIOLIB
on Alchemy (but I like speedy gpio calls to bitbang busses...),
another this patch.

Thanks for testing!
      Manuel Lauss