rethink passive cooling w.r.t. charging
As preparation for the summer, I re-tested the passive cooling mechanisms and they work very well. That's the description I'm talking about that defines our policy: https://source.puri.sm/Librem5/linux/-/blob/pureos/byzantium/arch/arm64/boot/dts/freescale/imx8mq.dtsi#L277 which is:
- 50°C: only 1,0 Ghz instead of 1,5 (that doesn't help much)
- 60°C: start the idle_inject tasks and increase their runtime duration depending on a hysteresis (how long the temperature stays the same and does not go lower) up to 60% cpu-time
- 90°C: power off (what NXP does)
- why do we basically stop charging at charger 40 °C (at least that's my observation, but @sebastian.krzyszkowiak you konw more than I do here) when the charger and the gauge have 60 and 70 degrees describes as "max" respectively?
- can we change the above limit in order to allow devices to better charge, especially during summer?
- independently: Should we adjust the cooling via idle-injection to that temperature in order to help charging? Of course the lower we start idle-injection the more often it will be done (and even "maxed-out", see below)
- currently we define the idle-injection maximum cpu-time as 60% in order to ensure a usable system even beyond the "hot" temperature trip point. Should we increase that maximum injection time?