On Sun, 4 Jul 2021 17:10:23 +0100 Jonathan Cameron <jic23@xxxxxxxxxx> wrote: > On Sat, 3 Jul 2021 18:55:40 +0200 > Andreas Kemnade <andreas@xxxxxxxxxxxx> wrote: > > > On Sat, 3 Jul 2021 18:39:40 +0200 > > Andreas Kemnade <andreas@xxxxxxxxxxxx> wrote: > > > > > Hi, > > > > > > On Sat, 3 Jul 2021 16:59:50 +0100 > > > Jonathan Cameron <jic23@xxxxxxxxxx> wrote: > > > > > > > On Sat, 3 Jul 2021 10:42:20 +0200 > > > > Andreas Kemnade <andreas@xxxxxxxxxxxx> wrote: > > > > > > > > > Add devicetree support so that consumers can reference the channels > > > > > via devicetree, especially the power subdevice can make use of that > > > > > to provide voltage_now properties. > > > > > > > > Does the mapping vary from board to board? Often these mappings are > > > > internal to the chip so might as well be provided hard coded in the > > > > relevant drivers rather than via DT. See drivers that have iio_map > > > > structure arrays. > > > > > > > Most things are internal to the chip, but > > > AIN1/AIN0 are external and could be connected to anything. > > > > > hmm, iio_map stuff looks nice, so before messing with devicetree, > > I could solve 90% of the problem by just using iio_map? For my use > > cases it is enough to have the internal stuff at the moment. That would > > simplify stuff a lot. > > > > So I could go forward with the iio_map stuff now, and if there is a use > > case for AIN1/0, the devicetree stuff can be added later? > > I was just thinking the same. I 'think' that it will first try to find > a mapping via device tree and then use the iio_map stuff. > > So you can probably get away with a mixture of the two. > Worth testing that works though (hook up iio-hwmon to AIN0 perhaps whilst > also using the iio_map approach). > > I might be completely wrong though and am not aware of anyone currently > doing this... > I tested that approach, It works, so I will first post a series with just the iio_map stuff and later the devicetree stuff. Regards, Andreas
