On 09/06/26 10:28, [email protected] wrote: > Thank you for your contribution! Sashiko AI review found 2 potential issue(s) > to consider: > - [High] Adding `vdd-supply`, `vlogic-supply`, and unconditionally requiring > `vref-supply` breaks backward compatibility for existing device trees. > - [Medium] Making `vlogic-supply` unconditionally required forces device > trees to document a non-existent hardware connection for single-channel > devices. > -- > > commit 8357cd3efa5f819678f24f0cbf0a39208f4078bc > Author: Rodrigo Alencar <[email protected]> > Subject: dt-bindings: iio: dac: ad5696: rework on power supplies > > This commit reworks the power supply bindings for the ad5696 DAC series. It > adds vdd-supply, vlogic-supply, and vref-supply, deprecates vcc-supply, and > updates the required properties. > > > diff --git a/Documentation/devicetree/bindings/iio/dac/adi,ad5696.yaml > > b/Documentation/devicetree/bindings/iio/dac/adi,ad5696.yaml > > index 2dcc049f30e91..e5fbaec4adf71 100644 > > --- a/Documentation/devicetree/bindings/iio/dac/adi,ad5696.yaml > > +++ b/Documentation/devicetree/bindings/iio/dac/adi,ad5696.yaml > > [ ... ] > > > @@ -55,8 +65,21 @@ properties: > > required: > > - compatible > > - reg > > + - vdd-supply > > + - vlogic-supply > > [Severity: High] > Does adding vdd-supply and vlogic-supply to the global required list break > backward compatibility for existing device trees? Older device trees that > omitted these will now fail dtbs_check.
Yes, for correctness, it breaks the dtbs_check. In the driver side we would rely on the dummy/stub regulator in case those are not declared. > [Severity: Medium] > Does making vlogic-supply unconditionally required force device trees to > document a non-existent hardware connection? Single-channel devices in this > family (such as adi,ad5693, adi,ad5693r, and adi,ad5311r) physically do not > have a dedicated VLOGIC pin. VLOGIC does exist for those parts. However, as mentioned before, for some tiny packages it is internally connected to VDD: - 10-Lead MSOP package for AD5693R/AD5691R exposes the VLOGIC pin - 8-Lead LFCSP package for AD5693R/AD5692R/AD5691R/AD5693 does NOT expose the VLOGIC pin I think we agreed to make it required and user can just repeat the vdd-supply value or not have it at all. Driver would rely on the dummy/stub regulator. > > > > allOf: > > + - if: > > + properties: > > + compatible: > > + contains: > > + anyOf: > > + - const: adi,ad5693 > > + - const: adi,ad5694 > > + - const: adi,ad5696 > > + then: > > + required: > > + - vref-supply > > [Severity: High] > Does unconditionally requiring vref-supply break older device trees that > correctly utilized the formerly valid vcc-supply? > > To preserve DT ABI compatibility, should the schema allow either vcc-supply or > vref-supply to satisfy the reference voltage requirement? Same here. For correctness, it breaks DT ABI. > -- > Sashiko AI review ยท > https://sashiko.dev/#/patchset/[email protected]?part=2 -- Kind regards, Rodrigo Alencar
