Re: [PATCH 1/2] arm64: dts: mt8173: elm: Set uart0 to mmio32 iotype

[Enric Balletbo i Serra <enric.balletbo@collabora.com>]

Hi Hsin-Yi,

On 10/9/20 9:57, Hsin-Yi Wang wrote:
> On Thu, Sep 10, 2020 at 3:20 PM Enric Balletbo i Serra
> <enric.balletbo@collabora.com> wrote:
>>
>> Hi Hsin-Yi,
>>
>> On 10/9/20 7:46, Hsin-Yi Wang wrote:
>>> Set uart0 iotype to mmio32 to make earlycon work with stdout-path.
>>>
>>> Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
>>> ---
>>>  arch/arm64/boot/dts/mediatek/mt8173-elm.dtsi | 2 ++
>>>  1 file changed, 2 insertions(+)
>>>
>>> diff --git a/arch/arm64/boot/dts/mediatek/mt8173-elm.dtsi b/arch/arm64/boot/dts/mediatek/mt8173-elm.dtsi
>>> index a5a12b2599a4a..d54e62f72c65d 100644
>>> --- a/arch/arm64/boot/dts/mediatek/mt8173-elm.dtsi
>>> +++ b/arch/arm64/boot/dts/mediatek/mt8173-elm.dtsi
>>> @@ -1160,6 +1160,8 @@ &thermal {
>>>  };
>>>
>>>  &uart0 {
>>> +     reg-io-width = <4>;
>>> +     reg-shift = <2>;
>>
>> I am wondering if these properties are common enough to go to mt8173.dtsi
>> instead of here.
>>
> Since stdout-path is set on elm.dtsi (using uart0), so I add this
> attribute here.
> 

Why needs to be defined in the same place that stdout-path is set?

My guess is that reg-io-width and reg-shift are SoC specific, as they define how
the IO access to the register is done. In fact, I think that these properties
should be added in all the uarts, not only the one that has defined the stdout-path.

Cheers,
  Enric