Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
On Thu, Jul 14, 2011 at 09:08:33PM +0800, jida...@jidanni.org wrote: seconds the authors have allotted, unless one has expensive Italian hardware. [...] So all along it was that author in his Ferrari responsible for those RACE conditions. Personal abuse does not belong in the BTS. Please stop it. Enrico -- GPG key: 4096R/E7AD5568 2009-05-08 Enrico Zini enr...@enricozini.org signature.asc Description: Digital signature
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
Confirmed!: there must not be any other jobs running at the time nodm starts, else nodm usually cannot set itself up in the precious five seconds the authors have allotted, unless one has expensive Italian hardware. In /etc/rc.local I now do ( sleep 66; batch EOF apt-show-versions|egrep '\*|newer than version in archive' EOF ) to make sure it starts long after nodm has successfully launched. And lo and behold, I can now start nodm from rc.local on all my computers! So all along it was that author in his Ferrari responsible for those RACE conditions. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
Implementing a $ cat /usr/sbin/nodm #!/bin/sh strace $0.real 2 /tmp/ere reveals time(NULL) = 1310476722 send(3, 83Jul 12 21:18:42 nodm[1632]: ..., 72, MSG_NOSIGNAL) = 72 rt_sigprocmask(SIG_SETMASK, ~[KILL STOP RTMIN RT_1], NULL, 8) = 0 waitpid(1633, 0xbfcc4f68, WNOHANG) = 0 time(NULL) = 1310476722 send(3, 85Jul 12 21:18:42 nodm[1632]: ..., 69, MSG_NOSIGNAL) = 69 kill(1633, SIGTERM) = 0 kill(1633, SIGCONT) = 0 from which we see /var/log/auth.log Jul 12 21:18:37 jidanni2 nodm[1632]: starting nodm Jul 12 21:18:42 jidanni2 nodm[1632]: X server did not respond after 5 seconds Jul 12 21:18:42 jidanni2 nodm[1632]: sending X server 1633 the TERM signal from which we conclude maybe my hardware is not as fast as Enrico's. Fortunately there is a way for the average home user to try six seconds instead of five. What is it? Thanks. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
And indeed, here is the corresponding Xorg.0.log showing what was going on in those vital five seconds before the assassination. Indeed apparently the clock started ticking even before X was logging things. You know, I bet the best thing to do would not be to create such RACE CONDITIONS at all, and instead have it be a not default option in /etc/default/nodm for those who want it. Xorg.0.log.old.gz Description: Xorg.0.log However I have been wrong in the past.
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
Yes nodm is working fine now... except I can't call it from
/etc/rc.local anymore. I must now do /etc/init.d/nodm by hand.
Looking at a slice of /var/log/Xorg.0.log for a normal session
[ 1403.567] (**) Option xkb_model pc105
[ 1403.567] (**) Option xkb_layout us
[ 1403.567] (**) Option xkb_options lv3:ralt_switch
[ 26103.653] (II) Power Button: Close
[ 26103.702] (II) UnloadModule: evdev
[ 26103.702] (II) Unloading evdev
and then for a failed session
[42.759] (**) Option xkb_model pc105
[42.759] (**) Option xkb_layout us
[42.760] (**) Option xkb_options lv3:ralt_switch
[42.769] (II) Power Button: Close
[42.769] (II) UnloadModule: evdev
[42.769] (II) Unloading evdev
We see that it only takes nodm 0.009 seconds to die.
(We look at the same spot where we used it for hours in the normal
session.)
Of course there is no other trace in /var/log of what happened.
All I know is it happens every time now I start from boot and I call
nodm from /etc/rc.local:
echo $0:
Will execute \${program=/etc/init.d/nodm start}\ in ${seconds=5} seconds.
Hit some chars and RET to cancel.
Hit RET to start right away.
# Hit ^C to cancel -- doesn't work.
for ((; seconds 0; seconds--))
do
read -p $seconds.. -t 1
case $? in 0) #they hit RET...
case $REPLY in
'') break ;; #...with no other chars
*) exit 44;; #...with other chars
esac;;
esac
done
echo running $program...
$program
P.S.,
# sysv-rc-conf --list nodm
nodm 2:off 3:off 4:off 5:off
In /etc/default/nodm I only changed
NODM_ENABLED=true
NODM_USER=jidanni
And /home/jidanni/.xsession-errors
is not touched during by the .009 second crash.
$ /usr/sbin/sysv-rc-conf --list
acpid2:on 3:on4:on5:on
alsa-utils 0:off 6:off S:on
anacron 2:off 3:off 4:off 5:off
apache2 0:off 1:off 2:off 3:off 4:off 5:off 6:off
atd 0:off 1:off 2:on3:on4:on5:on6:off
bootlogd S:on
bootlogs 1:on 2:on3:on4:on5:on
console-setu S:on
cron 1:off 2:on3:on4:on5:on
dbus 2:on 3:on4:on5:on
dictd1:off 2:on3:on4:on5:on
exim40:off 1:off 2:on3:on4:on5:on6:off
halt 0:off
ifplugd 0:off 1:off 2:off 3:off 4:off 5:off 6:off
ifupdown S:on
ifupdown-cle S:on
kbd S:on
keyboard-set S:on
killprocs1:on
module-init- S:on
mountoverflo S:on
mysql0:off 1:off 2:off 3:off 4:off 5:off 6:off
networking 0:off 6:off S:on
nodm 2:off 3:off 4:off 5:off
ntp 0:off 1:off 2:off 3:off 4:off 5:off 6:off
pdnsd0:off 1:off 2:on3:on4:on5:on6:off S:off
pppd-dns S:on
procps S:on
rc.local 2:on 3:on4:on5:on
reboot 6:off
resolvconf 0:off 6:off S:on
rmnologin2:on 3:on4:on5:on
rsync2:on 3:on4:on5:on
rsyslog 0:off 1:off 2:on3:on4:on5:on6:off
screen-clean S:on
sendsigs 0:off 6:off
single 1:on
smartd
smartmontool 1:off 2:on3:on4:on5:on
ssh 2:off 3:off 4:off 5:off
stop-bootlog 2:on 3:on4:on5:on
stop-bootlog S:on
udev S:on
udev-mtabS:on
umountfs 0:off 6:off
umountroot 0:off 6:off
urandom 0:off 6:off S:on
wicd 0:off 1:off 2:off 3:off 4:off 5:off 6:off
wwwoffle 0:off 1:off 2:on3:on4:on5:on6:off
x11-common S:on
xfstt0:off 1:off 2:off 3:off 4:off 5:off 6:off
-- System Information:
Debian Release: wheezy/sid
APT prefers unstable
APT policy: (500, 'unstable'), (1, 'experimental') [Ah, so that's how
I've set it up]
Architecture: i386 (i686)
Kernel: Linux 2.6.39-2-486
Locale: LANG=zh_TW.UTF-8, LC_CTYPE=zh_TW.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash
Versions of packages nodm depends on:
ii debconf [debconf-2.0] 1.5.40 Debian configuration management sy
ii libc6 2.13-10Embedded GNU C Library: Shared lib
ii libpam0g 1.1.3-2Pluggable Authentication Modules l
ii libx11-6 2:1.4.3-2 X11 client-side library
ii x11-common1:7.6+7X Window System (X.Org) infrastruc
ii x11-xserver-utils 7.6+3 X server utilities
nodm recommends no packages.
nodm suggests no packages.
-- debconf information:
* nodm/xinit: /usr/bin/xinit
* nodm/x_options: vt7 -nolisten tcp
* nodm/min_session_time: 60
* nodm/enabled: true
* nodm/xsession: /etc/X11/Xsession
* nodm/first_vt: 7
* nodm/user: jidanni
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
On Fri, Jul 08, 2011 at 08:40:48PM +0800, jida...@jidanni.org wrote: EZ Did you set NODM_XSESSION to something different than /etc/X11/Xsession? No I never touched that stuff. I assume there is some kind of race condition going on. Well have to wait for other users reports. For now it's aptitude forbid-version for me. I suspect the resources were forgotten during the many X resets that happened with 0.8, and that the fix to X resets in 0.8 also fixed this bug. Can you please try nodm 0.9 and let me know if this has been fixed? Ciao, Enrico -- GPG key: 4096R/E7AD5568 2009-05-08 Enrico Zini enr...@enricozini.org signature.asc Description: Digital signature
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
EZ Can you please try nodm 0.9 and let me know if this has been fixed? Yes, thank you. It is much improved now, and now has only refused to start once out of 10 times. So maybe the problem is 90% fixed. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
Package: nodm Version: 0.8-1 Starting today you no longer read the user's .Xresources . Also we now see several flashes of an empty screen with an X in the middle, whereas in the past we were spared of such a display. I am working on finding out what else is now broken. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
On Fri, Jul 08, 2011 at 05:48:07PM +0800, jida...@jidanni.org wrote: Starting today you no longer read the user's .Xresources . Did you set NODM_XSESSION to something different than /etc/X11/Xsession? nodm just runs /etc/X11/Xsession after starting X, and the standard /etc/X11/Xsession, at least in my system, loads X resources and does all the other things found in /etc/X11/Xsession.d Ciao, Enrico P.S. Would you mind keeping your emotions to yourself and sticking to what is useful for the bug report? -- GPG key: 4096R/E7AD5568 2009-05-08 Enrico Zini enr...@enricozini.org signature.asc Description: Digital signature
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
EZ == Enrico Zini enr...@enricozini.org writes: EZ Did you set NODM_XSESSION to something different than /etc/X11/Xsession? No I never touched that stuff. I assume there is some kind of race condition going on. Well have to wait for other users reports. For now it's aptitude forbid-version for me. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#633089: doesn't read ~/.Xresources anymore etc. etc.
On Fri, Jul 08, 2011 at 08:40:48PM +0800, jida...@jidanni.org wrote: EZ == Enrico Zini enr...@enricozini.org writes: EZ Did you set NODM_XSESSION to something different than /etc/X11/Xsession? No I never touched that stuff. I assume there is some kind of race condition going on. Is there anything specific that makes you think it's a race condition? Ciao, Enrico -- GPG key: 4096R/E7AD5568 2009-05-08 Enrico Zini enr...@enricozini.org signature.asc Description: Digital signature
