bug of gpio transition in pinmux driver

[Linus Walleij]

On Wed, Nov 23, 2011 at 2:13 PM, Haojian Zhuang
<haojian.zhuang at gmail.com> wrote:

> There's a bug in pinmux driver. We can request gpio via
> pinmux_request_gpio(@gpio). @gpio is the gpio number.
>
> In pinmux_request_gpio():
>    pin = @gpio - range->base;
>    pin_request(pin);
>
> This pin is the index of pinctrl_pin_desc[]. What's the definition of
> pinctrl_pin_desc's index? It's the index of PAD, not GPIO. Since the
> goal of pinmux_request_gpio() is to avoid define too much gpio groups.
> We need to add the transition between gpio and pad. It's clear that
> the transition of "pin = @gpio - range->base" can't fit every silicon.
> Especially, one gpio can be routed to two pads in PXA silicon.

I think this issue is addressed in the patch named
"pinctrl: add a pin_base for sparse gpio-ranges"
from Chanho Park.

I have this patch included in linux-next, and the patch named
"pinctrl: GPIO direction support for muxing" is also
related.

You can also find this in my "for-next" branch of the
pinctrl git here:
http://git.linaro.org/gitweb?p=people/triad/linux-pinctrl.git;a=summary

Yours,
Linus Walleij