On Tue, Sep 03, 2019 at 11:37:01AM +0200, Martin Fuzzey wrote: > On 03/09/2019 08:18, Biwen Li wrote: > > diff --git a/Documentation/devicetree/bindings/rtc/pcf85363.txt b/Documentation/devicetree/bindings/rtc/pcf85363.txt > > index 94adc1cf93d9..588f688b30d1 100644 > > --- a/Documentation/devicetree/bindings/rtc/pcf85363.txt > > +++ b/Documentation/devicetree/bindings/rtc/pcf85363.txt > > @@ -8,10 +8,39 @@ Required properties: > > Optional properties: > > - interrupts: IRQ line for the RTC (not implemented). > > +- interrupt-output-pin: The interrupt output pin must be > > + "INTA" or "INTB", default value is "INTA" > > + > > > The hardware has 2 interrupt pins which can be mapped to various interrupt > sources (alarm1, alarm2, periodic, ...) > > Currently the driver only supports alarm1. > > It is even possible to use both pins for the same interrupt (eg if INTA were > wired to the SoC, INTB to a PMIC and both used for alarm...) > > > So maybe it would be better to have > > alarm1-interrupt-output-pin: The interrupt output pin used for the alarm > function. Must be "INTA", "INTB" or "BOTH" That's a property per source. 2 properties possible sources (either a mask or list) would be my preference. Also, whatever you end up with needs a vendor prefix. > > Then, if and when other types of interrupts are supported by the driver new > properties could be added for them. > > > > > +- quartz-load-femtofarads: The internal capacitor to select for the quartz: > > + PCF85263_QUARTZCAP_7pF [0] > > + PCF85263_QUARTZCAP_6pF [1] > > + PCF85263_QUARTZCAP_12p5pF [2] DEFAULT > > + > > > The standard DT property "quartz-load-femtofarads" takes the real physical > value in femto Farads ie values should be 7000, 6000, 12500 without defines. I believe I said this on the last version. > > > > +- nxp,quartz-drive-strength: Drive strength for the quartz: > > + PCF85263_QUARTZDRIVE_100ko [0] DEFAULT > > + PCF85263_QUARTZDRIVE_60ko [1] > > + PCF85263_QUARTZDRIVE_500ko [2] > > + > > > Not sure about this. > > Wouldn't it be better to either use a real impedence value in ohms (like > load property above, even though it is a vendor specific value) rather than > a define, or defines for "Low, Medium, High"? > > > Martin > >
