Re: [PATCH 0/9] google: Add support for Laguna CPM

[Krzysztof Kozlowski]

On 03/11/2026 03:06, Brian Norris wrote:
> On Wed, Oct 28, 2026 at 02:24:35PM +0100, Krzysztof Kozlowski wrote:
>> Either you expect review or not. RFC means no.
> 
> OK, PATCH.
> 
>> If that's patch, then at
>> the top level of cover letter, as the most important thing, you explain
>> the dependency and merging proposal of this.
> 
> OK, I'll try to move it up. To be fair though, every change that has any
> dependency is clearly noted in its below-the-fold notes.
> 
>> You combined here like 5
>> different subsystens...
> 
> What do you recommend here? I believe the general principles of DT
> binding development suggest there should be a single binding doc for
> describing the "Central Power Manager" (and the services it supports).
> CPM supports functionality covered by many subsystems.

Standard upstreaming process for new SoCs.
https://lore.kernel.org/linux-samsung-soc/CADrjBPq_0nUYRABKpskRF_dhHu+4K=duPVZX==0pr+cjSL_caQ@xxxxxxxxxxxxxx/T/#m2d9130a1342ab201ab49670fa6c858ee3724c83c

and the perfect example:
https://lore.kernel.org/all/20231121-topic-sm8650-upstream-dt-v3-0-db9d0507ffd3@xxxxxxxxxx/

> 
> While in theory, bindings can be written without corresponding software
> support (drivers), in practice it usually works out better to develop
> them together, to prove that the binding describes the hardware/firmware
> sufficiently.
> 
> And even if the binding goes separately, I don't think I can 100% split
> the driver support by subsystem, as they all require the same underlying
> CPM helpers, and those helpers don't have much meaning without their
> clients. So some multi-subsystem work is still necessary IMO.
> 
> So I don't really see how I can address your complaint. But I'm open to
You don't know how to split patchset per respective subsystems? Somehow
Linaro, Qualcomm and many other people managed to do that, so just ask
them. Linaro folks might be easiest to reach, through Peter Griffin for
example.


Best regards,
Krzysztof