Okay, I managed to reproduce the USBMS reboots. Will investigate...
In the mean time, good news: it's not a crash, it's a soft reboot, so it should be relatively gentle on the device & filesystem.
EDIT: ... And, of course, now I can't reproduce it again. >_<". I did manage to break everything in a most spectacular way when trying a workaround, so there's that ;p.
At least now I can say that you can boot a PW2 in usbnet mode with the cable plugged in! \o/.
TL;DR: mesquite sometimes fails to stop cleanly when switching to USBMS, and the logic of the mesquite upstart job doesn't seem to know how to deal with this, on one hand it considers this 'normal' if mesquite ran for > 10s, but even in that case, it doesn't ask upstart to stop trying to respawn it. So it does tries to respawn it, which utterly fails without the userstore, and once it hits the 5 restarts limit, it triggers a reboot.
There's only a vague interaction with the Fonts hack potentially involved, so I can't really say who's at fault here, other than finding the job logic a bit convoluted and not terribly solid.
Last edited by NiLuJe; 12-08-2013 at 12:08 PM.
|