Re: [PATCH 1/2] dt-bindings: pinctrl: mediatek: deprecate custom drive strength property

[Rob Herring]

On Wed, Mar 29, 2023 at 09:58:57AM +0200, Krzysztof Kozlowski wrote:
> On 28/03/2023 15:41, AngeloGioacchino Del Regno wrote:
> > Il 28/03/23 15:06, Alexandre Mergnat ha scritto:
> >> Deprecate mediatek,drive-strength-adv which shall not exist, that was an
> >> unnecessary property that leaked upstream from downstream kernels and
> >> there's no reason to use it.
> >>
> >> The generic property drive-strength-microamp should be used instead.
> >>
> >> Signed-off-by: Alexandre Mergnat <amergnat@xxxxxxxxxxxx>
> >> ---
> >>   .../devicetree/bindings/pinctrl/mediatek,mt8183-pinctrl.yaml      | 8 ++++++--
> >>   .../devicetree/bindings/pinctrl/mediatek,mt8365-pinctrl.yaml      | 6 +++++-
> >>   2 files changed, 11 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/Documentation/devicetree/bindings/pinctrl/mediatek,mt8183-pinctrl.yaml b/Documentation/devicetree/bindings/pinctrl/mediatek,mt8183-pinctrl.yaml
> >> index c30cd0d010dd..b82a066b91ec 100644
> >> --- a/Documentation/devicetree/bindings/pinctrl/mediatek,mt8183-pinctrl.yaml
> >> +++ b/Documentation/devicetree/bindings/pinctrl/mediatek,mt8183-pinctrl.yaml
> > 
> > Thanks for doing MT8183!!!
> > 
> >> @@ -110,8 +110,13 @@ patternProperties:
> >>             drive-strength:
> > 
> > ..snip..
> > 
> >> diff --git a/Documentation/devicetree/bindings/pinctrl/mediatek,mt8365-pinctrl.yaml b/Documentation/devicetree/bindings/pinctrl/mediatek,mt8365-pinctrl.yaml
> >> index 4b96884a1afc..347f533776ba 100644
> >> --- a/Documentation/devicetree/bindings/pinctrl/mediatek,mt8365-pinctrl.yaml
> >> +++ b/Documentation/devicetree/bindings/pinctrl/mediatek,mt8365-pinctrl.yaml
> >> @@ -91,8 +91,13 @@ patternProperties:
> >>   
> >>             input-schmitt-disable: true
> >>   
> >> +          drive-strength-microamp:
> >> +            enum: [125, 250, 500, 1000]
> >> +
> >>             mediatek,drive-strength-adv:
> >> +            deprecated: true
> > 
> > In the case of MT8365, since there's *no* devicetree using this property, *at all*,
> > I think you can even just entirely remove this block, as that should not be
> > considered an ABI breakage in that case.
> > 
> > Krzysztof, can you please confirm?
> 
> If it is defined as ABI in a released kernel, then should be rather
> deprecated.

Yes, but if no one notices it's not an ABI. It's up to the platform 
maintainers really. I only care that the commits indicate they know what 
they are doing (usually not).

Rob