Re: [PATCH 5/6] dt-bindings: iio: temperature: ltc2983: document power supply

From: Jonathan Cameron <jic23@kernel.org>
To: Conor Dooley <conor@kernel.org>
Cc: "Nuno Sá" <noname.nuno@gmail.com>, "Nuno Sa" <nuno.sa@analog.com>,
	linux-iio@vger.kernel.org, devicetree@vger.kernel.org,
	"Lars-Peter Clausen" <lars@metafoo.de>,
	"Michael Hennerich" <Michael.Hennerich@analog.com>,
	"Rob Herring" <robh+dt@kernel.org>,
	"Krzysztof Kozlowski" <krzysztof.kozlowski+dt@linaro.org>,
	"Conor Dooley" <conor+dt@kernel.org>
Subject: Re: [PATCH 5/6] dt-bindings: iio: temperature: ltc2983: document power supply
Date: Thu, 22 Feb 2024 19:15:22 +0000	[thread overview]
Message-ID: <20240222191522.343f33a8@jic23-huawei> (raw)
In-Reply-To: <20240222-lance-sprinkled-04a7650ca316@spud>

On Thu, 22 Feb 2024 17:54:28 +0000
Conor Dooley <conor@kernel.org> wrote:

> On Thu, Feb 22, 2024 at 05:41:03PM +0100, Nuno Sá wrote:
> > On Thu, 2024-02-22 at 15:40 +0000, Conor Dooley wrote:  
> > > On Thu, Feb 22, 2024 at 01:55:56PM +0100, Nuno Sa wrote:  
> > > > Add a property for the VDD power supply regulator.
> > > > 
> > > > Signed-off-by: Nuno Sa <nuno.sa@analog.com>
> > > > ---
> > > >  Documentation/devicetree/bindings/iio/temperature/adi,ltc2983.yaml | 2 ++
> > > >  1 file changed, 2 insertions(+)
> > > > 
> > > > diff --git
> > > > a/Documentation/devicetree/bindings/iio/temperature/adi,ltc2983.yaml
> > > > b/Documentation/devicetree/bindings/iio/temperature/adi,ltc2983.yaml
> > > > index dbb85135fd66..8aae867a770a 100644
> > > > --- a/Documentation/devicetree/bindings/iio/temperature/adi,ltc2983.yaml
> > > > +++ b/Documentation/devicetree/bindings/iio/temperature/adi,ltc2983.yaml
> > > > @@ -57,6 +57,8 @@ properties:
> > > >    interrupts:
> > > >      maxItems: 1
> > > >  
> > > > +  vdd-supply: true  
> > > 
> > > Although technically an ABI break, should we make this supply required?
> > > It is, at the end of the day, required by the hardware for operation.
> > >   
> > 
> > I thought about it but then realized it could break some existing users which is
> > never a nice thing.  
> 
> Could you explain what scenario it actually breaks a system (not
> produces warnings with dtbs_check)?
> 
> If anything actually broke something, it would be the driver change that
> actually assumed that the regulator was present and refused to probe
> otherwise, right? In Linux at least, the regulator core will provide a
> dummy regulator if one doesn't exist - otherwise patch 6/6 would
> actually contain a DT ABI breakage, since it does:
> 
> 	ret = devm_regulator_get_enable(&spi->dev, "vdd");
> 	if (ret)
> 		return ret;
> 
> IMO making a new property required is only a meaningful break of the ABI
> when drivers reject probe when it is missing, but I must admit I don't
> know if other operating systems handle missing regulators as nicely as
> Linux does.

Agreed - adding a requirement on a supply to the dt-binding shouldn't
be a problem because of the dummy regulators.

Jonathan

> 
> > I recently (in another series - the IIO backend) went through some trouble to
> > actually not break ABI. Meaning, I had to do some not so neat hacking in the
> > driver because Rob was more comfortable with not breaking ABI in DT. So, I
> > assumed he would not like for me to break it in here.  
> 