[U-Boot] [PATCH 4/6] sunxi: GPIO: introduce sunxi_gpio_setup_dt_pins()

[Dr. Philipp Tomsich]

> On 14 Jul 2017, at 15:50, Simon Glass <sjg at chromium.org> wrote:
> 
> +Tom
> 
> On 7 July 2017 at 04:30, Andre Przywara <andre.przywara at arm.com <mailto:andre.przywara at arm.com>> wrote:
>> Hi Simon,
>> 
>> On 07/07/17 04:58, Simon Glass wrote:
>>> Hi Andre,
>>> 
>>> On 2 July 2017 at 18:59, Andre Przywara <andre.przywara at arm.com> wrote:
>>>> Instead of hard-coding GPIO pins used for a certain peripheral, we
>>>> should just use the pinctrl information from the DT.
>>>> The sun8i-emac driver has some simple implementation of that, so
>>>> let's just generalize this and copy the code to a more common
>>>> location.
>>>> On the way we add support for the new, generic pinctrl binding now
>>>> used by all Allwinner SoCs.
>>>> 
>>>> Signed-off-by: Andre Przywara <andre.przywara at arm.com>
>>>> ---
>>>> arch/arm/include/asm/arch-sunxi/gpio.h |  3 ++
>>>> arch/arm/mach-sunxi/pinmux.c           | 79 ++++++++++++++++++++++++++++++++++
>>>> 2 files changed, 82 insertions(+)
>>> 
>>> It looks like this should be done with a pinctrl driver, shouldn't it?
>> 
>> I knew you would say this ;-)
>> 
>> So technically you are right, but it looks like this is quite some work.
>> And as patch 5/6 demonstrates, this code here was actually already
>> mostly in, so this is basically just moving and generalisation.
>> 
>> The whole purpose of this series is to prepare for merging in the Linux
>> 4.13-rc1 DTs and still have working Ethernet, so I wanted to keep the
>> bar low on this.
>> 
>> I haven't looked in detail what it would take to have a proper Allwinner
>> pinctrl DM driver, but the binding Linux uses is not "self-contained",
>> as we still need SoC specific knowledge (namely into which actual mux
>> value the "emac" string, for instance, translates to; and this is per
>> SoC and pin). So that means pulling *a lot of* tables into U-Boot, which
>> I am not sure is worth at this point. We might get away with some
>> short-cuts, basically keeping our hard coded mux values around or just
>> defining those that we actually need.
>> But in general I consider this work more long term, and it should not
>> halt the more important goal of having the Linux and U-Boot DTs in sync.
> 
> With a due sense of foreboding and dread:
> 
> Reviewed-by: Simon Glass <sjg at chromium.org <mailto:sjg at chromium.org>>

I had submitted a DM driver to do this a long time ago, which included these
tables for the A64 (and had the necessary infrastructure to quickly add other
boards).

This was still tight enough, even to fit into an AArch64 SPL stage.

So I don’t see the purpose of not doing a DM pinctrl for sunxi, as the tables
can easily be shared between U-Boot and Linux—i.e. I had used the Linux
tables verbatim and only added a few lines of wrapper for U-Boot.

Regards,
Philipp.