2015-02-13 17:25 GMT+01:00 Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>:
> Hi Maxime,
>
> Am Freitag, den 13.02.2015, 16:59 +0100 schrieb Maxime Coquelin:
>> Hi Philipp,
>>
>> 2015-02-13 12:47 GMT+01:00 Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>:
>> > Hi Maxime,
>> >
>> > Am Donnerstag, den 12.02.2015, 18:46 +0100 schrieb Maxime Coquelin:
>> > [...]
>> >> + soc {
>> >> + reset_ahb1: reset@40023810 {
>> >> + #reset-cells = <1>;
>> >> + compatible = "st,stm32-reset";
>> >> + reg = <0x40023810 0x4>;
>> >> + };
>> >> +
>> >> + reset_ahb2: reset@40023814 {
>> >> + #reset-cells = <1>;
>> >> + compatible = "st,stm32-reset";
>> >> + reg = <0x40023814 0x4>;
>> >> + };
>> >> +
>> >> + reset_ahb3: reset@40023818 {
>> >> + #reset-cells = <1>;
>> >> + compatible = "st,stm32-reset";
>> >> + reg = <0x40023818 0x4>;
>> >> + };
>> >> +
>> >> + reset_apb1: reset@40023820 {
>> >> + #reset-cells = <1>;
>> >> + compatible = "st,stm32-reset";
>> >> + reg = <0x40023820 0x4>;
>> >> + };
>> >> +
>> >> + reset_apb2: reset@40023824 {
>> >> + #reset-cells = <1>;
>> >> + compatible = "st,stm32-reset";
>> >> + reg = <0x40023824 0x4>;
>> >> + };
>> >
>> > These are mostly consecutive, single registers. I wonder if these are
>> > part of the same IP block and thus should be grouped together into the
>> > same reset controller node?
>>
>> What I could to is to have two instances. One for AHB and one for APB domain.
>> Doing this, I will have one instance per domain, and only consecutive registers.
>> Is it fine for you?
>
> Looking at
> http://www.st.com/web/en/resource/technical/document/reference_manual/DM00031020.pdf
> Table 34 (RCC register map and reset values), I'd say there is a single
> "Reset and Clock Control" device at 0x40023800 - 0x40023884:
>
> soc {
> rcc: rcc@40023800 {
> #clock-cells = <1>;
> #reset-cells = <1>;
> compatible = "st,stm32-rcc";
> reg = <0x40023800 0x84>;
> };
>
> ...
>
> If you really want to describe the reset controller parts (offsets +0x10
> to +0x24) in a separate node, I won't argue against it too long,
> although this is a somewhat arbitrary decision.
>
> In any case, the whole register at offset +0x1c is reserved, so there is
> no reason to split the reset controller. It is ok to have unused ranges
> as is already the case with reserved bits inside the used registers.
Ok. I understand your point.
But it will be more difficult at usage, because the node referencing
the fourth reset bit of apb2 register will have to pass 164 as
parameter.
It is error prone IMHO.
Other solution would be to add some defines for each reset line in the
DT-Bindings, as we do today for STi platform.
But it is giving an unneeded constraint between DT and reset trees.
Br,
Maxime
>
> regards
> Philipp
>
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
