[PATCH v5 3/3] riscv: dts: starfive: add DeepComputing FML13V01 board device tree

[Alex Elder]

On 10/21/24 7:47 AM, Krzysztof Kozlowski wrote:
> On 21/10/2024 13:16, Conor Dooley wrote:
>> On Mon, Oct 21, 2024 at 09:17:59AM +0200, Krzysztof Kozlowski wrote:
>>> On Sun, Oct 20, 2024 at 09:49:59PM +0800, Guodong Xu wrote:
>>>> From: Sandie Cao <sandie.cao at deepcomputing.io>
>>>> +&camss {
>>>> +	status = "disabled";
>>>> +};
>>>> +
>>>> +&csi2rx {
>>>> +	status = "disabled";
>>>> +};
>>
>> You can drop these two, I marked them disabled in the common file
>> earlier this week.
>> 1
>>>> +
>>>> +&gmac0 {
>>>> +	status = "disabled";
>>>> +};
>>>> +
>>>> +&i2c0 {
>>>> +	status = "disabled";
>>>> +};
>>>> +
>>>> +&pwm {
>>>> +	status = "disabled";
>>>> +};
>>>> +
>>>> +&pwmdac {
>>>> +	status = "disabled";
>>>> +};
>>>> +
>>>> +&spi0 {
>>>> +	status = "disabled";
>>>
>>> If your board has to disable all these, then they should not have been
>>> enabled in DTSI in the first place. Only blocks present and working in
>>> the SoC (without amny external needs) should be enabled.
>>>
>>> I suggest to fix that aspect first.
>>
>> Eh, I don't think I agree. Having 5 disables here is a lesser evil than
>> reproducing 90% of jh7110-common.dtsi or shunting a bunch of stuff
>> around. Emil?
> 
> Why reproducing 90%? Only enable would be here, no? Or you want to say
> the common DTSI has things which do not exist?

For what it's worth, I agree with Krzysztof.  In the (long) cover
page we pointed this out, and offered to do it in a followup patch.
But if requested we can do it now.

So in v6, a new patch would be inserted before the other three,
and it would:
- Remove the status = "okay" lines for those nodes that are not enabled
   in this new platform, in "jh7110-common.dtsi"
- Add nodes where appropriate in:
     jh7110-milkv-mars.dts
     jh7110-pine64-star64.dts
     jh7110-starfive-visionfive-2.dtsi
   They'll look like this, to enable the ones disabled above, e.g.:
     &gmac0 {
         status = "okay";
     };

     &i2c0 {
         status = "okay";
     };

You guys should come to agreement, but I do think what Krzysztof says
is the right approach.  And unless convinced otherwise, this will be
what shows up in the next version of this series.

					-Alex

> Best regards,
> Krzysztof
>