IMHO, I think the boot-to-charging mode's a little shakey - If I plug in power when it's off, It will semi-boot and charge. I'll have to unplug the power and wait a minute or so before I can actually boot into maemo.

I suggest simply leaving it on all the time - It uses almost no power in it's sleep mode, and is a lot happier that way.
Just leave it on, plug it in, and in the morning unplug it and off you go.

It happened again today.
Here's the syslog
According to it Feb 14 05:33:53 is the last record before I restarted it at 9:07AM. I had to set the clock and there was no alarm in the morning.

I see nothing bad in there save for the blasted reset. Any new entries in oops? Did last night's boot provide a new entry?

Also, it seems to core right after you stopped using it, is that a rule?

No new entries in oops.

I'm not sure what "to core" means — is that connected with those entries:

csd[759]: ISI_SMS .555575> incoming_cell_broadcast(): Incoming cell broadcast

If so I have no idea...

To core is the Unix equivalent of a Windows application "fatal error" or whatnot. It's not an error, it's a way for whoever wrote the application to get feedback on the state of the app when it crashed, kind of like a picture of the car at the exact moment it died, so mechanics can figure out what it was missing.

In windows, this is done via log dumps, in Unix it's a core dump (saving the memory of the app to disk). An application core dump is the result of an error and, the term got shortened to "to core". As trivia, Windows also dumps core on some components that crash, when the logging mechanism is compromised. E.g., in a kernel crash or a general BSOD and the core is dumped completely (aka all the RAM).

Also, this is probably the most useless post on this thread, since I'm fresh out of ideas. I'm fairly sure it's not the broadcast messages. Mainly, because those flow through all the time and on all phones with no ill effects.

Right now, the only correlation I found is that it cores immediately after plugging it and leaving it alone. If this is repeatable, this might be it. For example, I think phone is locked at a frequency while charging (actually, while in mass storage but it is possible).

Try to get brightness display applet and leave the screen on at low setting (so it won't heat up) overnight when charging. This prevents device sleep - if that fixes it's it's likely sleeping is at fault.

If it still cores, it might be the charger itself - a problem with the SMPSU could fery well put bad, bad juice through the phone and while sane enough to charge battery, might prove too much for an undervolted CPU.

No warning reboots smell like hardware faults.

So:

* does it happen overnight but not in the day?
* does it happen if left on with no auto-TKlock?
* does it happen if left locked but not on charger?

Let's go from here.

Thank you for the explanation

* does it happen overnight but not in the day?

No, not necessarily. But it only happens when my phone is locked - never when the screen is on and as far as I can see it doesn't depend on the charger - it happens both while charging and while not connected via USB

* does it happen if left on with no auto-TKlock?
I think I should try this.
_______________
I used to have Shortcutd installed for some time could be related? I guess I'm out of ideas. Right now I'm going to unistall everything except FBreader and see if it happens again.

When locked, CPU goes to 0. You may have a wakeup issue for the CPU. In which case, installing power kernel and overvolting the CPU might help. If not, Nokia Care?

Will reflashing the device solve the issue?

If it doesn't, it' a hardware issue and you need to return it to Nokia. It's likely, though.

Intall BackupMenu and do a backup first. That way, if it's not solved at least you can recover. Also, the embedded backup app also does good work.

You were right, reflashing did not solve the problem.
Thanks anyway.