Re: [PATCH 2/2] iio: dac: adding support for Microchip MCP47FEB02

["Nuno Sá" <noname.nuno@gmail.com>]

On Mon, 2025-09-22 at 17:15 -0500, David Lechner wrote:
> On 9/22/25 3:10 PM, Nuno Sá wrote:
> > Hi Ariana,
> > 
> > Thanks for your patches. Some initial comments from me...
> > 
> > On Mon, 2025-09-22 at 14:30 +0300, Ariana Lazar wrote:
> 
> ...
> 
> > > +static IIO_DEVICE_ATTR(store_eeprom, 0200, NULL, mcp47feb02_store_eeprom,
> > > 0);
> > > +static struct attribute *mcp47feb02_attributes[] = {
> > > +	&iio_dev_attr_store_eeprom.dev_attr.attr,
> > > +	NULL,
> > > +};
> > > +
> > 
> > Not going to argue about the ABI for now but I don't think this is a
> > standard one? So
> > if acceptable you need an ABI doc.
> > 
> Here's a random idea. (I would wait for Jonathan to weigh in first before
> assuming it is an acceptable idea though :-p)
> 
> The config registers are pretty much going to be a one-time deal. So those
> could be written to only if they need it during probe.
> 
> For the voltage output registers, we could add extra out_voltageY channels
> that are the power-on output state channels. So writing to out_voltageY_raw
> wouldn't change any real output but would just be written to EEPROM. This
> way these voltages could be controlled independently from the real outputs
> and it uses existing ABI.
> 
> In any case, it would be interesting to hear more about how this chips are
> actually used to better understand this EEPROM feature.

I didn't really looked at the datasheet so this can be totally wrong. But we
have some LTC parts (mainly hwmon stuff) that are also packed with an EEPRON.
AFAIU, the usecase in there is to have some defaults you can program in the
chips (and there's a feature we can enable so the chip can save things into the
eeprom automatically). Now, in those drivers we don't really support doing
anything with the eeprom at runtime so I'm curious to see how this unfolds :)

- Nuno Sá