On Sat, Jan 12, 2019 at 2:06 AM Matthias Kaehlcke <mka@xxxxxxxxxxxx> wrote:
>
> Another concern about adding trip points later could be the node
> name. We currently have:
>
>
> trips {
> cpu0_alert0: trip0 {
> ...
> };
>
> cpu0_crit: trip1 {
> ...
> };
> };
>
> If we keep increasing enumeration with the node name this would become:
>
> trips {
> cpu0_alert0: trip0 {
> ...
> };
>
> cpu0_alert1: trip1 {
> ...
> };
>
> cpu0_crit: trip2 {
> ...
> };
> };
>
> i.e. the node name of the critical trip-point changes, which might be
> a concern for dtsi's that override a value, though they should
> probably use the phandle &cpu0_crit anyway. If this is a concern we
> could change the node names to 'alert0' and 'crit'.
>
> I looked around a bit and actually I kinda like the naming scheme used
> by hisilicon/hi6220.dtsi, mediatek/mt8173.dtsi and rockchip/rk3328.dtsi
> (with minor variations):
>
> trips {
> threshold: trip-point@0 {
> temperature = <68000>;
> hysteresis = <2000>;
> type = "passive";
> };
>
> target: trip-point@1 {
> temperature = <85000>;
> hysteresis = <2000>;
> type = "passive";
> };
>
> cpu_crit: cpu_crit@0 {
> temperature = <115000>;
> hysteresis = <2000>;
> type = "critical";
> };
> };
>
> If we were to use this we'd have to adapt it slightly since we have
> multiple thermal zones. In line with the other scheme this could be
> cpuN_threshold, cpuN_target and cpuN_crit.
>
I like this scheme enough that I adopted it for v2.
