The problem of device shutting down on Nokia logo after reboot, is now top 1 "mystical" Maemo/N900 thing on my personal list.. Below is a short X-Archive summarize:

1. It seems, that people started encountering this UFO around kernel 42, but other get it with 46, 47, 48 - like me - or, 49.

2. Removing KP always fix the problem, but it re-appears after installing KP again.

3. The problem seems to be not 100% reproducible - it happens most of the times, but not always. Some claim, that holding N900 in "tablet" position (just like the backcover "leg" is used) make chances of successful reboot bigger. Personally, I can't imagine how such a thing may happen - accelerometer isn't initiated at such early stage of boot (fixme?) - but, this claim comes from trusted members, like handaxe

4. If device is able to reboot successfully once - for example, by means of voodoo from point 3 - it can be reboot many times in line without problem. Then, if You do some other things on it/leave it turned off for a while/etc, problem returns. So, it seems, that some unspecific condition change trigger problem again.

5. The same happens, when You try to turn device on while plugged to charger and charging. In fact, device while charging is ion "act-dead" mode, so, turning it on, mean - in fact - reboot. This is possibly the most painful appearance of problem (affecting USB port wear&tear), cause to solve this, one need to unplug, turn device on, and plug again, preferably on 5 dots (any time after Nokia logo will do).

6. It seems, that reflashing and reinstall same packages as before, result in solving the problem (at least, until it re-appears one day ). Ho ever, reflashing, and restoring rootfs/optfs backup - via backupmenu - brings the problem along restored backup. So, it's definitely something rootfs/optfs related (I wonder, if restoring only rootfs would help to sandbox the issue? I know it's way for broken packages, but just testing reboot, and reflashing + restoring backup proper way, might give us idea if it's something on rootfs or optfs)

7. It's still unsure, if R&D mode affect this problem. This one may be verified easily - one just need to turn R&D on, and check. Mentalist Traceur's CLI R&D flag changer, or qwerty's GUI R&D changer, may help here, if someone don't want to use flasher method.

Any clues/idea? I think restoring only rootfs may be a good start, ho ever, even if we know if it is caused by something in rootfs or optfs, it's still searching a drop of water in baltic sea.

/Estel