Re: [PATCH 0/3] pinctrl: meson-g12a: add pinctrl driver support

[Martin Blumenstingl]

Hi Yixun,

On Thu, Jul 5, 2018 at 4:53 AM Yixun Lan <yixun.lan@xxxxxxxxxxx> wrote:
>
> HI Neil
>
> On 07/04/18 22:57, Neil Armstrong wrote:
> > Hi Yixun,
> >
> > On 05/07/2018 00:45, Yixun Lan wrote:
> >> This patch series try to add pinctrl driver support for
> >> the Meson-G12A SoC.
> >
> > Thanks for submitting these patches.
> >
> > Can you explicit this patchset with more details on the G12A SoC family ?
> > It's relationship with AXG and the differences in term of pinmuxing with the other SoC families ?
> >
> I thought this was already discussed while we doing pinctrl driver for
> Meson-AXG SoC.
>
> Anyway, here it is:
>
> Starting from Meson-AXG SoC, the pinctrl controller block using 4
> continues bit to specific pin mux function, while comparing to old
> generation SoC which kind of using various length bits for the pin mux
> definition. The new design would greatly simplify the software model..
>
> for detail example, one 32bit register can describe 8 pins, and each of
> them has 0-7 value to set, start from value 0 to 7.
>
> partition the register into 8 parts:
> bit[3:0]
> bit[7:4]
> bit[11:8]
> bit[15:12]
> bit[19:16]
> bit[23:20]
> bit[27:24]
> bit[31:28]
>
>
> for each value:
> value == 0, means the pin is GPIO
> value = {1, 2, ... 7 } is one of specific PIN function
OK, so AXG and G12A use the same register layout -> thus the same
pinmux ops are re-used



> I could put this info into cover-letter or commit message?
if you have to resend this series anyways then it would be great if
you could add it to the commit description

> > Why is there a GPIOE bank within the AO controller ?
> >
>
> It actually sit in the AO domain, although it's sounds strange from the
> naming..
>
> I'm not sure if it's good idea to append a AO suffix? since the
> documentation just use the plain GPIOE
I am fine with plain GPIOE if that's what your internal documentation uses
it would be great if you could add a comment (or at least a note in
the commit message) indicating that this is how the hardware is
designed (initially I thought this was a bug since I have no
documentation for the G12A chipset)

Regards
Martin