Since my FC5 installation a few weeks ago I have a curious problem. The
machine boots up nicely, I get the blue log-in screen, I type username and
password, my account comes up, but just when I should be ready to use it,
the screen goes blank (sometimes displaying some messages for 1/2 a
second), and the machine jumps back to the blue X-window log-in screen,
asking me again to log in. Only after the second, third, fourth, of fifth
log-in
try it works, and I then can use my account normally.
This behavior happens for both of my user accounts and started right
with my plain-vanilla FC5 installation. The only non-plain-vanilla
items are the files that were before in my account (which I copied over
from a backup disk) as well as some tweaking of which packages shall be
loaded.
Thinking that perhaps one of these old files gives this problem, I
tried to copy only some files over, but I was not able to identify any
one file that might cause it. It seems to be a gnome problem with
.gconf, since the problem appears only when I use Gnome, but not when I
use KDE. So I deleted files and once thought I had it narrowed down to
.gconf/apps/panel/applets, but then the error reappeared even if I start
with completely fresh directories .gconf and .gnome* (which made it
necessary for me to individualize the appearance). So the problem does
not seem to be a simple old file. Once I got rid of this problem in one
account (and I don't know how). Then I accidentally deleted the .gconf*
and .gnome* files, and now I am again seeing this problem in this
account.
I don't find any error file created in my account. From dmesg I get
the following error messages, but I think they are not related to this
annoying feature:
audit(1148538077.353:19): avc: granted { execmem } for pid=2488
comm="nautilus" scontext=user_u:system_r:unconfined_t:s0
tcontext=user_u:system_r:unconfined_t:s0 tclass=process
audit(1148538077.413:20): avc: granted { execmem } for pid=2488
comm="nautilus" scontext=user_u:system_r:unconfined_t:s0
tcontext=user_u:system_r:unconfined_t:s0 tclass=process
audit(1148538097.070:21): avc: granted { execstack } for pid=2636
comm="metacity" scontext=user_u:system_r:unconfined_t:s0
tcontext=user_u:system_r:unconfined_t:s0 tclass=process
audit(1148538097.070:22): avc: granted { execmem } for pid=2636
comm="metacity" scontext=user_u:system_r:unconfined_t:s0
tcontext=user_u:system_r:unconfined_t:s0 tclass=process
audit(1148538098.955:23): avc: granted { execstack } for pid=2641
comm="nautilus" scontext=user_u:system_r:unconfined_t:s0
tcontext=user_u:system_r:unconfined_t:s0 tclass=process
I tried also logging in with Gnome Failsafe. It sometimes gave me a
warning that the session lasted less than 10s, and the message in
~./xsession-errors was:
/etc/gdm/PreSession/Default: Registering your session with wtmp and
utmp
/etc/gdm/PreSession/Default: running: /usr/bin/sessreg -a -w
/var/log/wtmp -u /var/run/utmp -x "/var/gdm/":0.Xservers" -h "" -l ".0"
"bauer"
(process: 3797): Gtk-WARNING **: This process is currently running
setuid or setgid.
This is not a supported use of GTK+. You must create a helper program
instead. For further details, see:
http://www.gtk.org/setuid.html
Refusing to initialize GTK+.
I was unable to find anything related to this "feature" on the web
(google and this list archive), and hence I appreciate any suggestion,
like how I could also debug this besides slowly varying my setup.
For example, I would expect somewhere a file to tell me what went wrong.
My system is an Athlon XP 2000+, A7V333 motherboard with a simple
video card, an ATI Mobility-M 8MB. I keep FC5 up-to-date with yum.
Thanks!
--- Johannes Bauer
