Re: [PATCH v6 2/3] dt-bindings: power: reset: fix minimum reg count for pmk8350-pon

[Rakesh Kota]

On Tue, Jul 28, 2026 at 12:20:12PM +0200, Krzysztof Kozlowski wrote:
> On 28/07/2026 12:12, Rakesh Kota wrote:
> > On Mon, Jul 27, 2026 at 08:36:38AM +0200, Krzysztof Kozlowski wrote:
> >> On Fri, Jul 24, 2026 at 03:44:12PM +0530, Rakesh Kota wrote:
> >>> The GEN3 PON peripheral requires two register address spaces: "hlos"
> >>> for the primary PON block and "pbs" for the address needs to be
> >>> specified to facilitate software debouncing. Set minItems: 2 for reg
> >>
> >> That's ABI break and I do not underestand if this was broken before or
> >> not. Driver ignores the PBS, so everything was working fine, no? Then
> >> why PBS is required now? Is it actually working or not?
> >>
> > 
> > Not an ABI break. All in-tree GEN3 DTs using qcom,pmk8350-pon have
> > always specified both "hlos" and "pbs" reg entries — pmk8350.dtsi,
> 
> That's not related to ABI.

my earlier reply focused on justifying this as "not an ABI break,"
but that was never your question. Since the driver works fine without
PBS, there's no functional bug being fixed here, so the ABI-break
framing doesn't apply either way.

> What about my DTS?

you're right, Whether an in-tree DTS omits PBS today isn't
the point - tightening minItems retroactively is an ABI concern
regardless.

> > pmk8550.dtsi, pmk8850.dtsi, sc8280xp-pmics.dtsi. No issues were seen
> > precisely because PBS was never omitted on any GEN3 platform.
> > 
> > minItems: 2 simply enforces in the schema what has always been present
> > in practice, ensuring future GEN3 DTs cannot accidentally omit the PBS
> > address space which is a GEN3 hardware requirement.
> This does not answer my questions.

Hopefully the above does now - to summarize: PBS is not functionally
required (driver skips it fine), so this patch isn't fixing a bug,
and tightening the schema risks breaking a DTB.

Given that, should I post a v7 dropping this patch from the series ?

regards
Rakesh Kota	
> 
> Best regards,
> Krzysztof