On Wed, Mar 23, 2022 at 03:57:44PM +0100, Linus Walleij wrote: > On Mon, Mar 21, 2022 at 8:10 PM Rob Herring <robh@xxxxxxxxxx> wrote: > > > > +properties: > > > + compatible: > > > + description: One compatible per product using this chip. Each product > > > + need deliberate custom values for things such as LRA resonance > > > + frequency and these are not stored in the device tree, rather we > > > + let the operating system look up the appropriate parameters from a > > > + table. > > > + enum: > > > + - immersion,isa1200-janice > > > + - immersion,isa1200-gavini > > > > Same device, different boards. I think I would put necessary properties > > in the DT. > > That will be all of these (from the driver): > > +struct isa1200_config { > + u8 ldo_voltage; > + bool pwm_in; > + bool erm; > + u8 clkdiv; > + u8 plldiv; > + u8 freq; > + u8 duty; > + u8 period; > +}; Could be all, but in your 2 cases some of these values are the same. > > Example: > > +/* Configuration for Janice, Samsung Galaxy S Advance GT-I9070 */ > +static const struct isa1200_config isa1200_janice = { > + .ldo_voltage = ISA1200_LDO_VOLTAGE_30V, > + .pwm_in = false, > + .clkdiv = ISA1200_HCTRL0_DIV_256, > + .plldiv = 2, > + .freq = 0, > + .duty = 0x3b, > + .period = 0x77, > +}; > > This is derived from the compatible rather than individual properties > or extra regulator and/or clock abstractions in line with: > Documentation/devicetree/bindings/display/panel/ilitek,ili9322.yaml > > Which was originally looking like so: > https://lore.kernel.org/dri-devel/20170813114448.20179-2-linus.walleij@xxxxxxxxxx/ > > To which you replied: > https://lore.kernel.org/dri-devel/20170817204424.e2wdkmyp4vyx2qj3@rob-hp-laptop/ > > "Normally, we the physical panel is described which would imply all these > settings. Are there lots of panels with this controller that would > justify all these settings?" I reserve the right to contradict myself. :) Seriously, it's always a judgement call. > > In that case there was one (1) > > In this case there are two (2) products that I know of. It does not have the > relationship between panel and panel controller products though, but... > it's not very different. > > I don't think this chip was used a lot, I really tried to find other instances. > But they could exist of course. Okay, if you want to leave it like this, I'm fine with that. Rob