Comment 75 for bug 1165433

To Jussi...

as the original reporter of this bug, I think it has no hope of being fixed even if you provide all the relevant kernel commits.
Which is bad, because the crash is serious and happens 'late' when may kernel data structures may have been written over, with potential data loss.

I tried to provide pointers to all the relevant LKML entries and the only replies I got was that I should have run apport-collect, which was a bit frustrating. I hope that your list, that is much more organized than my posts, has a better success.

Besides that, the bad news is that even on trusty (that has kernel 3.14), bluetooth dun is broken. At least here the issue is mere lack of functionality, no crashes. This is due to:

- another kernel bug (it is probably http://lists.openwall.net/linux-kernel/2014/02/10/54). This prevents modemmanager from seeing rfcomm0 as a modem. Tried mainline 3.15.6 and works fine.

- a bug in modemmanager/networkmanager. This keeps rfcomm0 open when the connection is dropped and causes modemmanager to keep asking the bluetooth phone about the signal quality. When you try to connect again, the connection fails. This is easily worked around by "service modemmanager restart".

Probably a new bug should be opened on launchpad against the kernel, wrt the issue of parenting rfcomm0. This fix should be easy to backport. But I am a bit discouraged.