--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Mon, Sep 26, 2016 at 3:20 PM, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
I ask because the 300 seconds uptime rings a bell, maybe not relevant, but suspicious at least. The kernel was (maybe still is) initializing the timer (if I remember correctly) subsystem 5 minutes before wraparound, just so it is easier to catch bugs, by making those wraparound bugs easier triggerable...
ok so it might be a simple matter of putting the right "thing" into the dtb or something. i was using the cubieboard2 dtb (which may not actually be properly up-to-date).
oh... one thing that may be keely relevant: there's no RTC on the EOMA68-A20 board. as in: the normal place where the kernel would get "time" from would be a battery-backed AXP209. in this case, however, it's powered from "cold".
so it's actually quite likely to be a major bug that has simply had insufficient coverage to expose it.
l.