Alle mercoledì 26 marzo 2008, Ola Lundqvist ha scritto:
> > at this time a "ps" should list no sessvc processes and obviously no
> > Xvnc/X11rdp nor the wm.
> > if your situation is like this, there's some problem with signals
> > (read: sessvc dies but sesman can't get his SIGCHLD).
> > what does ps say?
>
> Before logout.
>
> ...
> 14520 pts/17 Sl 0:17 /usr/bin/xrdp
> 14522 pts/17 Sl 0:00 /usr/bin/sesman
> 14625 pts/17 S 0:00 /usr/lib/xrdp/sessvc 14627 14626
> 14626 pts/17 S 0:00 x-session-manager
> 14627 pts/17 S 0:41 Xvnc :15 -geometry 800x600 -depth 24 -rfbauth
> /home/test/.vnc/sesman_passwd -bs -ac 14661 ? Ss 0:00
> /usr/bin/ssh-agent /usr/bin/dbus-launch --exit-with-session
> x-session-manager ...
>
> After logout:
>
> ...
> 14520 pts/17 S 0:18 /usr/bin/xrdp
> 14522 pts/17 Sl 0:00 /usr/bin/sesman
> 16389 pts/17 R+ 0:00 ps xa
>
> So we have a problem with the signals obviously.
I'm puzzled, I can't think a reason for which sessvc should not send
SIGCHLD, but it seems similiar to what happens on solaris.
can you try this patch? (it's against current cvs but should work or be
trivial to adapt)
S.
--
Follow the blind and talk to the deaf
Listen to the dumb and try to teach the dead
Machinery of lies
Bugs, creeps and fucking flies
Machinery of lies
Destruction - Machinery Of Lies
99,103d98
< void sig_child(int sig)
< {
< //does nothing
< }
<
128,129d122
< signal(SIGCHLD, sig_child);
<
158,159d150
< default:
< LOG_DBG(&(g_cfg.log), "sesman received an unhandled signal: %d", recv_signal);
