Hi Rob, > On Mon, Apr 8, 2019 at 2:22 AM Flavio Suligoi <f.suligoi@xxxxxxx> wrote: > > > > HI, > > > > > On 06/04/2019 01:07:13-0500, Rob Herring wrote: > > > > On Wed, Apr 03, 2019 at 04:52:44PM +0200, Flavio Suligoi wrote: > > > > > Some RTC devices have a battery-low automatic detection circuit. > > > > > The battery-low event is usually reported with: > > > > > > > > > > - a bit change in a RTC status register > > > > > - a hw signaling (generally using an interrupt generation), > changing > > > > > the hw level of a specific pin; > > > > > > > > > > The new property "battery-low-hw-alarm" enable the RTC to generate > the > > > > > hw signaling in case of battery-low event. > > > > > > > > > > Signed-off-by: Flavio Suligoi <f.suligoi@xxxxxxx> > > > > > --- > > > > > Documentation/devicetree/bindings/rtc/rtc.txt | 3 +++ > > > > > 1 file changed, 3 insertions(+) > > > > > > > > > > diff --git a/Documentation/devicetree/bindings/rtc/rtc.txt > > > b/Documentation/devicetree/bindings/rtc/rtc.txt > > > > > index a97fc6a..f93a44d 100644 > > > > > --- a/Documentation/devicetree/bindings/rtc/rtc.txt > > > > > +++ b/Documentation/devicetree/bindings/rtc/rtc.txt > > > > > @@ -31,6 +31,9 @@ below. > > > > > expressed in femto Farad (fF). > > > > > The default value shall be listed (if > > > optional), > > > > > and likewise all valid values. > > > > > +- battery-low-hw-alarm : Enable the "battery-low" output pin. > This > > > function > > > > > + is available on the following > devices: > > > > > + - pcf2127 - pin used for alarm: INTn > > > > > > > > Boolean? If there's cases where which pin is selectable, then we'd > need > > > > this to take a value. Not sure how likely that is? > > > > > > > > > > Indeed, there is at least the pcf85363 that has two possible pins for > > > that interrupt. How would you select the pin then? a zero based index? > a > > > string? > > I prefer an index. Ok, so we can call this property as: low-voltage-alarm and we can select the pin using a zero-base index, also for future developments. > > > I think the string could be clearer for the final user and would give > > more freedom for future changes. > > For example, we can call this property, instead of "battery-low-alarm" > or > > "low-voltage-alarm", simply: "alarm-pin_1" and then the string argument > > can describe the function used; for example: > > > > alarm-pin_1 = "backup-supply-low-voltage-alarm"; > > alarm-pin_2 = "......"; > > How many pins and functions then? And how does this relate to any > interrupts? If we use index, we don't use strings any more.
