2.5: daemonize() playing tricks with ttys?

2.5: daemonize() playing tricks with ttys?

[Daniel Egger]

[-- Attachment #1: Type: text/plain, Size: 755 bytes --]

Hija,

I'm trying to spawn new threads from a function called from alloc_uid
using daemonize () as soon as a new user appears on the system. Somehow 
this detaches the original shell from the tty causing an exit not
only of the child but also it's parent.

A diagram of the situation would lock like this:

getty -> (login of root) bash -> (su to another user) bash ->
[new thread is spawned] (whatever) -> exit -> getty

Alternativly, when directly logging in a non-root user:
getty -> (login of foo) motd -> [hang]

How can I daemonize something without disturbing other processes?
I already tried playing with reparent_to_init and some signal stuff
as done by other parts of the kernel but to no avail.

-- 
Servus,
       Daniel

[-- Attachment #2: Dies ist ein digital signierter Nachrichtenteil --]
[-- Type: application/pgp-signature, Size: 189 bytes --]