Re: [PATCH 1/2] dt-bindings: hwmon: tmp102: Add TMP110 and TMP113 devices

[Flaviu Nistor <flaviu.nistor@gmail.com>]

On Mon, Sep 29, 2025 at 08:07:43PM +0300, Flaviu Nistor wrote:
>On Mon, Sep 24, 2025 at 19:12:26 +0100, Conor Dooley wrote:
>>On Wed, Sep 24, 2025 at 08:55:39AM -0700, Guenter Roeck wrote:
>>> On 9/15/25 10:18, Conor Dooley wrote:
>>> > On Mon, Sep 15, 2025 at 08:08:18PM +0300, Flaviu Nistor wrote:
>>> > > Add a compatible string for TMP110 and TMP113 devices.
>>> > > 
>>> > > Signed-off-by: Flaviu Nistor <flaviu.nistor@gmail.com>
>>> > > ---
>>> > >   Documentation/devicetree/bindings/hwmon/ti,tmp102.yaml | 2 ++
>>> > >   1 file changed, 2 insertions(+)
>>> > > 
>>> > > diff --git a/Documentation/devicetree/bindings/hwmon/ti,tmp102.yaml b/Documentation/devicetree/bindings/hwmon/ti,tmp102.yaml
>>> > > index 96b2e4969f78..840b5306a8cf 100644
>>> > > --- a/Documentation/devicetree/bindings/hwmon/ti,tmp102.yaml
>>> > > +++ b/Documentation/devicetree/bindings/hwmon/ti,tmp102.yaml
>>> > > @@ -13,6 +13,8 @@ properties:
>>> > >     compatible:
>>> > >       enum:
>>> > >         - ti,tmp102
>>> > > +      - ti,tmp110
>>> > > +      - ti,tmp113
>>> > 
>>> > The driver has no match data and no compatible based decisions added in
>>> > your patch. Why is a fallback to tmp102 not suitable?
>>> > 
>>> 
>>> That should work for tmp110. However, tmp113 does have additional registers, and it
>>> might be desirable to report the content of those using debugfs. Fallback for that chip
>>> would work for now, but that might have to change later if support for the additional
>>> registers is added to the driver.
>>
>> In that case, only the driver would need to change not the binding.
>> That's fine though, of course.
>>
>
> I am a little confused and I would appreciate some advice if I should leave
> the series as it is or send a v2 only for documentation and Kconfig updates,
> droping the binding update?
>
>>> 
>>> Guenter
>>> 
>>> > >     interrupts:
>>> > >       maxItems: 1
>>> > > -- 
>>> > > 2.43.0
>>> > > 
>>>

After some time (probably to much) I went through the patch again and
based on Rob Herring response:
https://lore.kernel.org/all/20250922162004.GA199858-robh@kernel.org/
and a similar example for sensirion sh20, sh21 and sh25 sensors, 
where individual compatible string were added in the binding and the
driver even if there is no match data:
https://lore.kernel.org/all/20250908-sht2x-v4-3-bc15f68af7de@gmail.com/
https://lore.kernel.org/all/20250908-sht2x-v4-4-bc15f68af7de@gmail.com/
I think that the patch can stay as it is for the binding, so different
compatible string can be use in the device tree in order for the real
hardware to be describe.

Flaviu