Hi Geert, On Thu, Oct 12, 2017 at 3:56 PM, Geert Uytterhoeven <[email protected]> wrote: > Hi Marek, > > On Wed, Oct 11, 2017 at 11:39 PM, Marek Vasut <[email protected]> wrote: >> On 10/10/2017 04:58 PM, Geert Uytterhoeven wrote: >>> Add a device node for the ROHM BD9571MWV PMIC, based on the example in >>> the DT binding documentation, but using INTC-EX instead. >>> >>> Signed-off-by: Geert Uytterhoeven <[email protected]> >>> --- >>> Do we need to describe more regulators? >> >> To my knowledge, no. > > OK, thanks! > >>> --- a/arch/arm64/boot/dts/renesas/salvator-common.dtsi >>> +++ b/arch/arm64/boot/dts/renesas/salvator-common.dtsi >>> @@ -353,6 +353,30 @@ >>> >>> &i2c_dvfs { >>> status = "okay"; >>> + >>> + pmic: pmic@30 { >>> + pinctrl-0 = <&irq0_pins>; >>> + pinctrl-names = "default"; >>> + >>> + compatible = "rohm,bd9571mwv"; >>> + reg = <0x30>; >>> + interrupt-parent = <&intc_ex>; >> >> Shouldn't this be gpio2 ? Why intc-ex ? > > Because we now have INTC-EX support ;-) > > Serious: if a pin used for interrupt signalling can be configured for both > GPIO and INTC-EX aka IRQC, we typically configure it for INTC-EX. Probably > because the latter is a simpler block, and thus consumes less power?
I agree with your decision to use INTC-EX over GPIO, however I do think that this discussion smells like software policy... Isn't DT supposed to describe the hardware? =) It's almost like we want to DT to point out the pin, not the function.... Cheers, / magnus
