On 10/29/24 09:32, Grant Peltier wrote:
On Mon, Oct 28, 2024 at 11:18:39AM +0100, Geert Uytterhoeven wrote:
v4:
- Revert devicetree property name to "vout-voltage-divider" and refactor
property description and driver implementation to match existing
vout-voltage-divider implementation in max20730 as no suitable generic
voltage divider schema exists.
Can you please elaborate (or point to the email that did so, in case
I missed it)?
In reply to v2, Günter wrote:
"I would prefer, in the order of preference,
1) an applicable generic property definition
2) a definition that is already used elsewhere
3) a new chips specific definition"
https://lore.kernel.org/all/3f460b62-4cd1-49dd-a98b-1fbcfdbd3af0@xxxxxxxxxxxx
Thanks!
Gr{oetje,eeting}s,
Geert
Hi Geert,
After looking through existing bindings definitions, the only generic
voltage divider property that I found was one that is intended to scale
voltages for IIO io-channels. However, the use case here is to scale
particular hwmon PMBus telemetry, which most closely matches the existing
implementation of the vout-voltage-divider property in the max20730
driver so I opted to copy that implementation based on Guenter's
preferences. Is there something that I missed or more information that
you are looking for?
I had a look myself. I agree with Grant, the existing bindings are really
not a good match. They are directly tied to iio, the implementation in
drivers/iio/afe/iio-rescale.c is completely iio specific, and I see
no means to use it for anything but iio. The problem is that it is
a "compatible" property, not parameter property (or set of properties)
as I would have expected [sorry if I use the wrong terminology].
We could possibly use the actual properties, "output-ohms" and "full-ohms",
but I am not really sure if that makes too much sense.
Note that there are a variety of voltage divider properties.
gw,voltage-divider-ohms
lltc,fb-voltage-divider
mps,fb-voltage-divider
vout-voltage-divider
All those specify a pair of resistors. I think that model is much better
suited to the problem at hand than a "compatible".
I agree though that this should be explained in the patch description.
Guenter
