> >
> > > > > + chan->overrange = fwnode_property_read_bool(child,
> > > > > + "adi,overrange");
> > > >
> > > > One line?
> > >
> > > It will pass the 80 col limit. AFAIR, Jonathan prefers to keep it when it
> > > does not hurt readability...
> >
> > I believe it will increase readability being located on one line.
>
> I mean, this is perfectly aligned with the open "(", so it's a pretty
> normal pattern. Anyways, I'm more than happy to move this into a one
> liner and just use the 100 limit. But let's see what Jonathan has to say
> because I do not want to move back and forward...
Here it happens to be particularly ugly because of the short first parameter,
so I'm fine with a longer line for this one.
> > > > > + st->regmap = devm_regmap_init(&spi->dev, NULL, st,
> > > > <c2688_regmap_config);
> > > >
> > > > I'm wondering why it's not a regmap SPI?
> > >
> > > The problem is on the read side... In the first transfer we write the
> > command/register
> > > to read, then we need to release the CS pin so that the device
> > executes the command,
> > > and only then we read the data. AFAIK, the regmap spi
> > implementation won't work like
> > > this. I think CS is kept asserted the whole time...
> >
> > I believe it's configurable, no? Like the cs_change flag somewhere.
> > Can you double check?
>
> Don't think we can... The read part just calls:
> https://elixir.bootlin.com/linux/latest/source/drivers/base/regmap/regmap-spi.c#L98
> and has no control over the spi transfer bits...
Feature to add then or a custom regmap_bus if you want to keep it in the driver.
J
>
> - Nuno Sá
