Re: [PATCH v3 2/3] ARM: dts: microchip: sama7g5: Add flexcom 10 node

[claudiu beznea]

On 16.02.2024 11:35, Conor Dooley wrote:
> On Fri, Feb 16, 2024 at 10:24:13AM +0200, claudiu beznea wrote:
>> On 16.02.2024 09:56, Alexandre Belloni wrote:
>>> On 16/02/2024 06:58:10+0000, Mihai.Sain@xxxxxxxxxxxxx wrote:
>>>>> diff --git a/arch/arm/boot/dts/microchip/sama7g5.dtsi b/arch/arm/boot/dts/microchip/sama7g5.dtsi
>>>>> index 269e0a3ca269..c030b318985a 100644
>>>>> --- a/arch/arm/boot/dts/microchip/sama7g5.dtsi
>>>>> +++ b/arch/arm/boot/dts/microchip/sama7g5.dtsi
>>>>> @@ -958,6 +958,30 @@ i2c9: i2c@600 {
>>>>>  			};
>>>>>  		};
>>>>>  
>>>>> +		flx10: flexcom@e2820000 {
>>>>> +			compatible = "atmel,sama5d2-flexcom";
>>>>
>>>> My comment here was ignored:
>>>> https://lore.kernel.org/all/20240214-robe-pregnancy-a1b056c9fe14@spud/
>>>>
>>>> The SAMA7G5 has the same flexcom controller as SAMA5D2 MPU.
>>>>
>>>
>>> Still, it needs its own compatible plus a fallback to
>>> atmel,sama5d2-flexcom
>>
>> I agree with this. Though, flexcom documentation is subject to YAML
>> conversion (a patch has been re-posted these days [1] and *maybe* it will
>> be integrated this time). And there are multiple SoC DTs that need to be
>> updated with their own flexcom compatible (lan966x, sam9x60, sama7g5).
>>
>> To avoid conflicting with the work at [1] and postponing this series we may
>> do the update after the [1] is done.
>>
>> Let me know your thoughts. Either way is fine by me.
> 
> I'd be inclined to say that if we are gonna take a shortcut here, then
> this patch should add a specific compatible so that when the yaml
> conversion goes through you'll get a warning about this being
> undocumented rather than silence.

All the flexcom nodes from all flexcom capable SoCs (including SAMA7G5)
have the same compatible introduced by Mihai.

I don't like the idea of updating only the DTSes, either update all DTSes
and documentation or do it as it is already done (with sama5d2 compatible).

> 
> A resend on the flexcom patch is required though, the rebase was not
> done correctly, so maybe Balakrishnan could "atmel,sama7g5-flexcom"
> add with a fallback to "atmel,sama5d2-flexcom" while they're fixing
> it up?

I agree, and DTSes should also be updated along with documentation.

With this we can go forward with this patch and avoid conflicting with work
that is currently in progress for flexcom.

Thank you,
Claudiu Beznea

> 
> Cheers,
> Conor.