RE: [PATCH] iio: dac: max5821: convert device register to device managed function

["Sa, Nuno" <Nuno.Sa@analog.com>]

> From: Alexandru Ardelean <ardeleanalex@gmail.com>
> Sent: Tuesday, July 20, 2021 9:23 AM
> To: Théo Borém Fabris <theobf@usp.br>
> Cc: Jonathan Cameron <jic23@kernel.org>; Lars-Peter Clausen
> <lars@metafoo.de>; linux-iio <linux-iio@vger.kernel.org>; LKML
> <linux-kernel@vger.kernel.org>
> Subject: Re: [PATCH] iio: dac: max5821: convert device register to
> device managed function
> 
> On Mon, Jul 19, 2021 at 6:25 PM Théo Borém Fabris <theobf@usp.br>
> wrote:
> >
> > Hi, Alexandru.
> >
> > Em seg., 19 de jul. de 2021 às 04:33, Alexandru Ardelean
> > <ardeleanalex@gmail.com> escreveu:
> > >
> > > On Sun, Jul 18, 2021 at 11:42 PM Théo Borém Fabris
> <theobf@usp.br> wrote:
> > > >
> > > > Add a device managed hook, via devm_add_action_or_reset()
> and
> > > > max5821_regulator_disable(), to disable voltage regulator on
> device
> > > > detach.
> > > > Replace iio_device_register() by devm_iio_device_register() and
> remove
> > > > the max5821_remove() function used to unregister the device
> and disable the
> > > > voltage regulator.
> > > > Remove i2c_set_clientdata() from the probe function, since
> > > > i2c_get_clientdata() is not used anymore.
> > >
> > > Looks good overall.
> > > A few comments inline.
> > >
> > > >
> > > > Signed-off-by: Théo Borém Fabris <theobf@usp.br>
> > > > ---
> > > >  drivers/iio/dac/max5821.c | 30 ++++++++++++++++--------------
> > > >  1 file changed, 16 insertions(+), 14 deletions(-)
> > > >
> > > > diff --git a/drivers/iio/dac/max5821.c
> b/drivers/iio/dac/max5821.c
> > > > index bd6e75699a63..44c04ae70b32 100644
> > > > --- a/drivers/iio/dac/max5821.c
> > > > +++ b/drivers/iio/dac/max5821.c
> > > > @@ -294,6 +294,13 @@ static const struct iio_info max5821_info
> = {
> > > >         .write_raw = max5821_write_raw,
> > > >  };
> > > >
> > > > +static void max5821_regulator_disable(void *data)
> > > > +{
> > > > +       struct regulator *rdata = data;
> > > > +
> > > > +       regulator_disable(rdata);
> > >
> > > This can be simplified a bit:
> > >
> > > static void max5821_regulator_disable(void *reg)
> > > {
> > >       regulator_disable(reg);
> > > }
> > >
> > > I used to do explicit casting, but then I also figured that it's not
> necessary.
> > >
> > Ok.
> >
> > > > +}
> > > > +
> > > >  static int max5821_probe(struct i2c_client *client,
> > > >                         const struct i2c_device_id *id)
> > > >  {
> > > > @@ -306,7 +313,6 @@ static int max5821_probe(struct i2c_client
> *client,
> > > >         if (!indio_dev)
> > > >                 return -ENOMEM;
> > > >         data = iio_priv(indio_dev);
> > > > -       i2c_set_clientdata(client, indio_dev);
> > > >         data->client = client;
> > > >         mutex_init(&data->lock);
> > > >
> > > > @@ -331,6 +337,14 @@ static int max5821_probe(struct
> i2c_client *client,
> > > >                 goto error_free_reg;
> > > >         }
> > > >
> > > > +       ret = devm_add_action_or_reset(&client->dev,
> max5821_regulator_disable,
> > > > +                                      data->vref_reg);
> > > > +       if (ret) {
> > > > +               dev_err(&client->dev,
> > > > +                       "Failed to add action to managed regulator: %d\n",
> ret);
> > > > +               goto error_disable_reg;
> > >
> > > return ret;
> > >
> > > devm_add_action_or_reset() should call
> max5821_regulator_disable() in
> > > case of error
> > >
> > Ok.
> >
> > > > +       }
> > > > +
> > > >         ret = regulator_get_voltage(data->vref_reg);
> > > >         if (ret < 0) {
> > > >                 dev_err(&client->dev,
> > > > @@ -346,7 +360,7 @@ static int max5821_probe(struct i2c_client
> *client,
> > > >         indio_dev->modes = INDIO_DIRECT_MODE;
> > > >         indio_dev->info = &max5821_info;
> > > >
> > > > -       return iio_device_register(indio_dev);
> > > > +       return devm_iio_device_register(&client->dev, indio_dev);
> > > >
> > > >  error_disable_reg:
> > >
> > > This entire goto block should be removed.
> > > The idea of using only devm_ functions is to not have these goto
> statements.
> > >
> > I thought the action added via devm_add_action (and devres_add)
> was called only
> > on driver detach, thus the error_disable_reg label would be
> necessary
> > to handle the
> 
> devm_add_action() yes
> this is devm_add_action_or_reset() which looks like this:
> 
> static inline int devm_add_action_or_reset(struct device *dev,
>                        void (*action)(void *), void *data)
> {
>     int ret;
> 
>     ret = devm_add_action(dev, action, data);
>     if (ret)
>         action(data);
> 
>     return ret;
> }
> 
> it can be found in "include/linux/device.h"
> 
> > possible error on regulator_get_voltage. Could you please clarify for
> > me when does
> > a driver detach happen?
> 
> a driver detach happens when:
> * the kmod is unloaded (assuming the driver is running as a kmod)
> * manually unbinding the driver from sysfs ;
>    a quick article about this [it's for USB, but other interfaces use
>    the same mechanism]:
> 
> https://urldefense.com/v3/__https://lwn.net/Articles/143397/__;!!A
> 3Ni8CS0y2Y!rw-
> qxijVlohZWCG9q_wWFPGASqW7JWcfS48uJSwY8coKiBPoA9Z0G2WSCZ
> clCw$
>    there should be something under /sys/bus/spi/drivers/xxxx/unbind
> * when the system powers down (reboots)
> * maybe there is some other new method to do this [that I don't know
> about]

If the device get's deleted with 'device_del()'...

- Nuno Sá