Re: [PATCH v2 6/8] dt-bindings: soc: qcom: pmic-glink: Move X1E80100 out of fallbacks

[Krzysztof Kozlowski <krzk@kernel.org>]

On 03/06/2025 08:59, Fenglin Wu wrote:
> 
> On 6/3/2025 2:47 PM, Krzysztof Kozlowski wrote:
>> On 03/06/2025 08:42, Fenglin Wu wrote:
>>> On 6/2/2025 3:40 PM, Krzysztof Kozlowski wrote:
>>>> On 30/05/2025 09:35, Fenglin Wu via B4 Relay wrote:
>>>>> From: Fenglin Wu <fenglin.wu@oss.qualcomm.com>
>>>>>
>>>>> Move X1E80100 out of the fallbacks of SM8550 in pmic-glink support.
>>>> Why?
>>>>
>>>> Do not describe what you do here, it's obvious. We see it from the diff.
>>>>
>>>>
>>>> Best regards,
>>>> Krzysztof
>>> Previously, in qcom_battmgr driver, x1e80100 was specified with a match
>>> data the same as sc8280xp, also sm8550 was treated a fallback of sm8350
>>> without the need of a match data.
>>>
>>> In ucsi_glink driver, sm8550 had a match data and x1e80100 was treated
>>> as a fallback of sm8550. There was no issues to make x1e80100 as a
>>> fallback of sm8550 from both qcom_battmgr and ucsi_glink driver perspective.
>>>
>>> In patch [5/8] in this series, in qcom_battmgr driver, it added charge
>>> control functionality for sm8550 and x1e80100 differently hence
>>> different match data was specified for them, and it makes x1e80100 ad
>>> sm8550 incompatible and they need to be treated differently.
>> So you break ABI and that's your problem to fix. You cannot make devices
>> incompatible without good justification.
> 
> I would say x1e80100 and sm8550 are different and incompatible from a 
> battery management firmware support perspective. The x1e80100 follows 
> the sc8280xp as a compute platform, whereas the sm8550 follows the 
> sm8350 as a mobile platform.

Not correct arguments for compatibility.

> 
> The difference between them was initially ignored because the sm8550 
> could use everything that the sm8350 has, and no match data needed to be 
> specified for it. However, now the sm8550 has new features that the 
> sm8350 doesn't have, requiring us to treat it differently, thus the 
> incompatibility was acknowledged.

So they are perfectly compatible.

I really do not understand what we are discussing here. Explain in
simple terms of DT spec: what is incompatible that SW cannot use one
interface to handle the other?


Best regards,
Krzysztof