Excuse me for (again) hijacking the thread

The problem is apparently that DSME gets confused when shutting down if bq27x00_battery module[*] is loaded. If you unload it before shutdown the N900 turns really off.
[*] I loaded it to be able to see the battery level (as voltage) because I removed the (now useless) standard battery indicator. I have now replaced the hal-device | grep ... command I use in DCE with a script using i2cget, so I don't need bq27x00 anymore.

*** EDIT EDIT EDIT ***
Apparently removing bq27x00 is not enough to make DSME know whether the charger is connected or not.

After browsing through DSME source code I noticed that the initial state is UNKNOWN. Probably BME tells DSME later if it's connected or not. Without BME, somebody else has to tell DSME to chill off.

There comes BATTTEST to the rescue! It's part of DSME, and just by executing:

# batttest --charger 0

DSME is now informed that no charger is connected. Poweroff works as expected!
[ should you consider it necessary, you can imagine me dancing ]
****************************

DSME and HAL will have to wait. OHM is gone and everything seems fine. I will put a SIM card and try to call/receive calls as well as play some music with and without headphones to see if not having OHM has messed up something I haven't noticed.

Once I collect all my hacks I will make a new thread.