Re: [linux-sunxi] [RFC PATCH] drivers: pinctrl: add driver for Allwinner A64 SoC

[Maxime Ripard]

Hi Andre,

On Tue, Jan 05, 2016 at 12:00:11PM +0000, Andre Przywara wrote:
> Hi Maxime,
> 
> On 04/01/16 20:30, Maxime Ripard wrote:
> > Hi Andre,
> > 
> > On Mon, Jan 04, 2016 at 10:29:06AM +0000, Andre Przywara wrote:
> >>>> diff --git a/drivers/pinctrl/sunxi/Makefile b/drivers/pinctrl/sunxi/Makefile
> >>>> index e080290..130e7bc 100644
> >>>> --- a/drivers/pinctrl/sunxi/Makefile
> >>>> +++ b/drivers/pinctrl/sunxi/Makefile
> >>>> @@ -12,5 +12,6 @@ obj-$(CONFIG_PINCTRL_SUN7I_A20)               += pinctrl-sun7i-a20.o
> >>>>  obj-$(CONFIG_PINCTRL_SUN8I_A23)                += pinctrl-sun8i-a23.o
> >>>>  obj-$(CONFIG_PINCTRL_SUN8I_A23_R)      += pinctrl-sun8i-a23-r.o
> >>>>  obj-$(CONFIG_PINCTRL_SUN8I_A33)                += pinctrl-sun8i-a33.o
> >>>> +obj-$(CONFIG_PINCTRL_A64)              += pinctrl-a64.o
> >>>
> >>> Shouldn't this follow pinctrl config name like other sunXi SOCs?
> >>> This should be PINCTRL_SUN??_A64.
> >>
> >> I never really got the reason we use those sunxi names in addition to
> >> the SoC name in the first place, maybe apart from copying from some
> >> Allwinner code.
> >> Since I decided to not look at Allwinner's BSP at all (if avoidable), I
> >> also thought it would be time to drop that sunxi naming, which looks
> >> redundant to me.
> >> Is there any reason why we would need this (beside having a rather
> >> unique identifier prefix)?
> > 
> > It's mostly historical.
> > 
> > 
> > Back when we started this, There was a few SoCs already out: A10,
> > A10s, A12 and A13, which was very similar to the Cortex-A naming
> > scheme (and I think the Cortex-A12 was also announced at the time).
> > 
> > We couldn't really use the SoC family either, since there was already
> > multiple SoCs that were part of the same family (the A10s, A12 and
> > A13, part of the sun5i family).
> > 
> > In order to avoid any confusion, we chose to go with both to uniquely
> > and without any confusion possible, and we just went on with that
> > naming scheme for consistency.
> 
> I see, thanks for the explanation.
> I was wondering since we now move to a new architecture as well to avoid
> this historic "ballast", but I have no problems with adding "_sun50i_"
> to the identifiers and file names.
> To me as only a casual sunxi user I found it mostly hard to memorize the
> connections between the sunxi numbering and the SoC names (I just know
> that the A20 is sun7i ;-). So for finding a specific dts for instance,
> you have to start with the sunxi number to get it TAB completed ...

I guess it doesn't really matter for arm64, since it seems like
there's a sub-folder per SoC family.

However, having an a12.dtsi in arch/arm/boot/dts will probably bring a
lot of confusion :)

I still think that we should maintain the current compatible scheme
though, in order to keep consistency.

> With that being said: Would you prefer to have a sun50i prefix? I see
> that having just "a64" on itself is not very specific.

I'd say that having a DTSI in arch/arm64/boot/dts/sunxi/a64.dtsi would
not bring a lot of confusion. If you're there, you know what you're
dealing with already.

If you feel like it makes the life easier to new or casual users, go
ahead.

Maxime

-- 
Maxime Ripard, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

Attachment: signature.asc
Description: Digital signature