Re: [PATCH v3 5/5] iio: adc: ad7192: Add AD7194 support

[Alisa-Dariana Roman <alisadariana@xxxxxxxxx>]

Hello and thank you for the feedback!

On 09.02.2024 00:27, David Lechner wrote:
> On Thu, Feb 8, 2024 at 11:25 AM Alisa-Dariana Roman
> <alisadariana@xxxxxxxxx> wrote:
> > Unlike the other AD719Xs, AD7194 has configurable differential
> > channels. The default configuration for these channels can be changed
> > from the devicetree.

...

> > +static const struct iio_info ad7194_info = {
> > +       .read_raw = ad7192_read_raw,
> > +       .write_raw = ad7192_write_raw,
> > +       .write_raw_get_fmt = ad7192_write_raw_get_fmt,
> > +       .read_avail = ad7192_read_avail,
> > +       .validate_trigger = ad_sd_validate_trigger,
> > +       .update_scan_mode = ad7192_update_scan_mode,
> > +};
>
> Isn't this identical to ad7192_info and ad7195_info now that .attrs is
> removed? It seems like we could consolidate here.

Those are not exactly identical since: 92 has bridge switch attribute, 
95 has bridge switch and ac excitation attributes and 94 has no custom 
attributes. I used a different info structure for 94 in order to avoid 
showing extra attributes.

> > +
> >   static const struct iio_info ad7195_info = {
> >          .read_raw = ad7192_read_raw,
> >          .write_raw = ad7192_write_raw,
> > @@ -1009,6 +1049,80 @@ static const struct iio_chan_spec ad7193_channels[] = {
> >          IIO_CHAN_SOFT_TIMESTAMP(14),
> >   };
> >
> > +static struct iio_chan_spec ad7194_channels[] = {
> > +       AD7193_DIFF_CHANNEL(0, 1, 2, 0x001),
> > +       AD7193_DIFF_CHANNEL(1, 3, 4, 0x023),
> > +       AD7193_DIFF_CHANNEL(2, 5, 6, 0x045),
> > +       AD7193_DIFF_CHANNEL(3, 7, 8, 0x067),
> > +       AD7193_DIFF_CHANNEL(4, 9, 10, 0x089),
> > +       AD7193_DIFF_CHANNEL(5, 11, 12, 0x0AB),
> > +       AD7193_DIFF_CHANNEL(6, 13, 14, 0x0CD),
> > +       AD7193_DIFF_CHANNEL(7, 15, 16, 0x0EF),
> > +       AD719x_TEMP_CHANNEL(8, AD7194_CH_TEMP),
> > +       AD7193_CHANNEL(9, 1, AD7194_CH_AIN1),
> > +       AD7193_CHANNEL(10, 2, AD7194_CH_AIN2),
> > +       AD7193_CHANNEL(11, 3, AD7194_CH_AIN3),
> > +       AD7193_CHANNEL(12, 4, AD7194_CH_AIN4),
> > +       AD7193_CHANNEL(13, 5, AD7194_CH_AIN5),
> > +       AD7193_CHANNEL(14, 6, AD7194_CH_AIN6),
> > +       AD7193_CHANNEL(15, 7, AD7194_CH_AIN7),
> > +       AD7193_CHANNEL(16, 8, AD7194_CH_AIN8),
> > +       AD7193_CHANNEL(17, 9, AD7194_CH_AIN9),
> > +       AD7193_CHANNEL(18, 10, AD7194_CH_AIN10),
> > +       AD7193_CHANNEL(19, 11, AD7194_CH_AIN11),
> > +       AD7193_CHANNEL(20, 12, AD7194_CH_AIN12),
> > +       AD7193_CHANNEL(21, 13, AD7194_CH_AIN13),
> > +       AD7193_CHANNEL(22, 14, AD7194_CH_AIN14),
> > +       AD7193_CHANNEL(23, 15, AD7194_CH_AIN15),
> > +       AD7193_CHANNEL(24, 16, AD7194_CH_AIN16),
>
> Shouldn't these be differential channels since they are
> pseudo-differential inputs measuring the difference between AINx and
> AINCOM?
>
> > +       IIO_CHAN_SOFT_TIMESTAMP(25),
> > +};
>
> i.e. like this (where AINCOM is voltage0 AINx is voltagex)
>
> static struct iio_chan_spec ad7194_channels[] = {
>         AD7193_DIFF_CHANNEL(0, 1, 0, AD7194_CH_AIN1),
>         AD7193_DIFF_CHANNEL(1, 2, 0, AD7194_CH_AIN2),
>         AD7193_DIFF_CHANNEL(2, 3, 0, AD7194_CH_AIN3),
>         AD7193_DIFF_CHANNEL(3, 4, 0, AD7194_CH_AIN4),
>         AD7193_DIFF_CHANNEL(4, 5, 0, AD7194_CH_AIN5),
>         AD7193_DIFF_CHANNEL(5, 6, 0, AD7194_CH_AIN6),
>         AD7193_DIFF_CHANNEL(6, 7, 0, AD7194_CH_AIN7),
>         AD7193_DIFF_CHANNEL(7, 8, 0, AD7194_CH_AIN8),
>         AD7193_DIFF_CHANNEL(8, 9, 0, AD7194_CH_AIN9),
>         AD7193_DIFF_CHANNEL(9, 10, 0, AD7194_CH_AIN10),
>         AD7193_DIFF_CHANNEL(10, 11, 0, AD7194_CH_AIN11),
>         AD7193_DIFF_CHANNEL(11, 12, 0, AD7194_CH_AIN12),
>         AD7193_DIFF_CHANNEL(12, 13, 0, AD7194_CH_AIN13),
>         AD7193_DIFF_CHANNEL(13, 14, 0, AD7194_CH_AIN14),
>         AD7193_DIFF_CHANNEL(14, 15, 0, AD7194_CH_AIN15),
>         AD7193_DIFF_CHANNEL(15, 16, 0, AD7194_CH_AIN16),
>         AD719x_TEMP_CHANNEL(16, AD7194_CH_TEMP),
>         IIO_CHAN_SOFT_TIMESTAMP(17),
> };

I tried to follow the existing style of the driver: for each 
pseudo-differential channel(AINx - AINCOM) there is an iio channel like 
this in_voltagex_raw; and for each differential channel(AINx - AINy) 
there is an iio channel like this in_voltagex-in_voltagey_raw. AD7194 
has 16 pseudo-differential channels/8 fully differential channels so I 
thought the (AINx - AINCOM) channels should be static and only the 8 
differential ones could be configured by the user from the devicetree by 
choosing the input pins.

The existing style of the driver, AD7192 has 4 pseudo differential 
channels and 2 (non configurable) differential channels:
static const struct iio_chan_spec ad7192_channels[] = {
	AD719x_DIFF_CHANNEL(0, 1, 2, AD7192_CH_AIN1P_AIN2M),
	AD719x_DIFF_CHANNEL(1, 3, 4, AD7192_CH_AIN3P_AIN4M),
	AD719x_TEMP_CHANNEL(2, AD7192_CH_TEMP),
	AD719x_DIFF_CHANNEL(3, 2, 2, AD7192_CH_AIN2P_AIN2M),
	AD719x_CHANNEL(4, 1, AD7192_CH_AIN1),
	AD719x_CHANNEL(5, 2, AD7192_CH_AIN2),
	AD719x_CHANNEL(6, 3, AD7192_CH_AIN3),
	AD719x_CHANNEL(7, 4, AD7192_CH_AIN4),
	IIO_CHAN_SOFT_TIMESTAMP(8),
};

Would it be better to respect the existing style or to do like you 
suggested and have a total of 16 differential channels that are 
configurable from the device tree?

Kind regards,
Alisa-Dariana Roman