Now, we all have our "of, sh*t!" moments, yep?

Yes, it was it, indeed. the funniest thing is that a) I've thanked post you've mentioned b) I was perfectly, completely, absolutely sure, that I've removed "network" from that file. Unless it was somehow re-inserted (for reasons unknown) during bluetooth-dun package uninstallation (did it at some point, as part for trying to find problem's cause), I must have imagined it. (Update: For the record, one does *not* need to remove bluetooth-dun package for bluetooth PAN to work - editing disabled plugins list and *restarting* device is all that is required to switch from DUN to PAN and other way around).

Now, as it may help someone later:

I tried restarting bluetoothd instead of whole device just for the sake of it, and it *seemed* to work - i.e, pairing was OK, bridge was created and dmesg showed all proper messages about entering forwarding state, but there was *no* network access on other machines. They could talk to each other (I was able to ssh both ways through PAN IP addresses), and bridge was pretending to work (brctl show contained mac of paired interface, mentioned messages in dmesg), nothing would result in proper network access for clients. On top of it, every time i turned bluetooth off and on on N900, it required full re-pairing with clients, setting up PIN, etc.

Now, after rebooting *whole* N900, it all works like charm. I've tested all possible combinations of enabling, disabling, and changing internet access on N90 0during and in-between PAN connection with bridge, and it flawlessly detect changes and propagate them (confirmed both by real-life tests, and by dmesg infos about propagating changed topology)

So, for future readers, restarting bluetoothd doesn't seem to be enough.
---

Thanks a lot for help, and again for making it possible in the first place!

/Estel