Re: [PATCH] hwmon: (ina2xx) make regulator 'vs' support optional

[Guenter Roeck]

On Thu, Apr 03, 2025 at 05:29:26PM +0300, Ciprian Marian Costea wrote:
> On 4/3/2025 3:15 PM, Guenter Roeck wrote:
> > On 4/3/25 03:15, Ciprian Costea wrote:
> > > From: Ciprian Marian Costea <ciprianmarian.costea@xxxxxxxxxxx>
> > > 
> > > S32G2/S32G3 based boards which integrate the ina231 sensor do not have a
> > > dedicated voltage regulator.
> > > 
> > > Co-developed-by: Florin Buica <florin.buica@xxxxxxx>
> > > Signed-off-by: Florin Buica <florin.buica@xxxxxxx>
> > > Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@xxxxxxxxxxx>
> > > ---
> > >   drivers/hwmon/ina2xx.c | 4 ++--
> > >   1 file changed, 2 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/hwmon/ina2xx.c b/drivers/hwmon/ina2xx.c
> > > index 345fe7db9de9..ab4972f94a8c 100644
> > > --- a/drivers/hwmon/ina2xx.c
> > > +++ b/drivers/hwmon/ina2xx.c
> > > @@ -959,8 +959,8 @@ static int ina2xx_probe(struct i2c_client *client)
> > >           return PTR_ERR(data->regmap);
> > >       }
> > > -    ret = (dev, "vs");
> > > -    if (ret)
> > > +    ret = devm_regulator_get_enable_optional(dev, "vs");
> > 
> > devm_regulator_get_enable() should provide a dummy regulator if there is
> > no explicit regulator. Why does this not work ?
> > 
> > > +    if (ret < 0 && ret != -ENODEV)
> > 
> > Why this added check ?
> > 
> > I know it used to be necessary if regulator support is disabled,
> > but that is no longer the case.
> > 
> > Guenter
> > 
> 
> Hello Guenter,
> 
> I've just tested and devm_regulator_get_enable() does work as you've
> described, providing a dummy regulator.
> 
> But, according to the 'ti,ina2xx' binding [1] I see that the `vs-supply`
> property is not required. Hence wouldn't it be correct for `vs-supply` to be
> optional ? Using 'devm_regulator_get_enable_optional()'
> 
Yes, but the reasoning you provided is different and suggested that the
current code would not work. Since that is not the case, the change would
be purely cosmetic. Also, I still don't see why the -ENODEV check would be
necessary.

Guenter