Might also fix #9748, depending on whether I simply never noticed in two years (not impossible, we're talking about power off after all) or whether it's actually a fairly new issue.

@Frenzie: Nah, that's an earlier fix of mine ;).

After a quick look, I would bet that #9748 should be fixed, too. (Tthe code cited in #9748 (comment) should not be reached as we are doing a os.execute("sleep 1 && poweroff -f &") in device.lua

@zwim: The sleep and backgrounding are especially there so that we do reach the exit codepaths in the startup script after an UIManager:quit ;).

There seem to be a few seconds involved though. I don't think a poweroff -f is that slow? I'll check soon.

It is surprisingly... weird. I originally caught it because the previous brokenness was actually attempting to restart Nickel during the poweroff, instead of exiting; and that's pretty obvious because of the boot animation (especially when it's KFMon's).

Confirmed that the behavior remains the same. Wrt my statement that it was a few seconds, I was probably counting thirds of a second. Seconds are surprisingly long really. :-)