[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Speak now, or...

> We should put in a question to Sun

i'm getting a little dizzy following this barrage of email.
i haven't been on this list for long so i may not be too familiar
with its history.

is the basic problem that there's no way for init to do the pam_close_session
because there's no way to "match the user with the tty?"
doesn't init know the pid of the login process that just exited?
if so, couldn't init then get the appropriate utmp entry based on
that pid?  the user/tty info (used to identify which session just ended)
would be inside the utmp entry.
having access to any session state is another matter, which is
independent of PAM.

> There's another consideration --- I'm not sure it's a good idea for init
> to be randomly executing shared libraries as dictated by /etc/pam.conf.
> Init is a long-lived program --- if a pam mechanism leaks memory, or
> worse yets, cause a core dump, the results on the system will be
> catastrophic.  So just for development's sake, and general robustness
> issues, it's much better to have login call the end_session routines.

this is a good point.

> I'd be very, very, very suprised
> to hear that they were actually calling any pam calls from init.

our version of init (as specified in the docs)
does make the call to pam_close_session.





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index] []