You are not logged in.
After upgrading to kernel26-2.6.28 I can't boot the system anymore. At the point where is loading the udev uevents it just stops blinking after a second and the whole system does not react to anything. I have to do a hard system reset (pressing the power button 4s).
No "err" messages are logged to the screen or anywhere for udev. When I set the error level to "info" or "debug" in udev.conf I get a lot of messages (several thousand), which don't appear in any log file on disk. This is what I took from last screen shown:
udevd[546]: seq 893 forked, pid [1120], 'add' 'pci', 0 seconds old
udevd-event[1120]: RUN 'socket:@/org/kernel/udev/monitor' /lib/udev/rules.d/95-udev-late.rules:7
udevd-event[1120]: passed -1 bytes to monitor 0x82c0818
udevd-event[1120]: passed -1 bytes to monitor 0x82c0818
udevd-event[1124]: RUN 'socket:/org/freedesktop/hal/udev_event' /etc/udev/rules.d/90-hal.rules:2
udevd[546]: seq 896 forked, pid [1124], 'add' 'pci', 0 seconds old
udevd[546]: seq 893 exit with 0
udevd-event[1124]: RUN 'socket:@/org/kernel/udev/monitor' /lib/udev/rules.d/95-udev-late.rules:7
There are also other (changing) events with other numbers in [] and other seq, but this is the part that appears everytime, though the exact numbers vary and the monitor adress changes. No other seqs are forked in between, just queued.
my (standard) /lib/udev/rules.d/95-udev-late.rules reads at line 2:
RUN+="socket:@/org/kernel/udev/monitor"
my (standard) /etc/udev/rules.d/90-hal.rules reads at line 2:
RUN+="socket:/org/freedesktop/hal/udev_event"
I use two monitors. Detaching one at a time didn't change anything. I don't know what else to try to get some information. I guess this is not enough information to send to some kernel guy.
This issue is still present in 2.6.28.1
I downgraded to 2.6.27.10, together with an the previous nvdia-173xx-173.14.12-2 in order to make things work again.
EDIT:
changed title
Last edited by JonnyJD (2009-02-26 03:13:44)
Offline
I found somebody with a workaround at archlinux.de: https://forum.archlinux.de/?page=Postin … read=11119
So based on this I did the following:
I changed /etc/rc.sysinit in order to see the modules udev is working on:
--- /etc/rc.sysinit 2009-01-22 00:26:09.000000000 +0100
+++ rc.sysinit 2009-01-22 00:25:43.000000000 +0100
@@ -117,6 +117,7 @@
stat_busy "Loading UDev uevents"
udevstart="$(/bin/date +%s%0N)"
/sbin/udevadm trigger
+ /sbin/udevadm monitor &
/sbin/udevadm settle
stat_done
udevend="$(/bin/date +%s%0N)"
and I edited the grub entry at boot time to remove the VGA=773. This last "tweak" made the system print the last module it war working on. Otherwise it crashes before prints the output.
With these changes I could read these last lines on the system:
add /module/iTCO_vendor_support (module)
add /module/iTCO_wdt (module)
add /bus/platform/drivers/iTCO_wdt (drivers)
add /devices/platform/iTCO_wdt (platform)
After that the system crashes like before and I can only do a hard reset.
So my temporary fix is blacklist this module in /etc/rc.conf:
MODULES=(!iTCO_wdt ....<some other modules>)
Having no watchdog timer is no problem for this system.
FYI
I have a motherboard with an Intel i815 chipset.
Offline
Thanks for posting this info, its gotten my system to work again. I also have the same chipset, and udev was hanging it because of that same module, so definitely there is an issue with this.
Offline
This is the corresponding kernel bug:
http://bugzilla.kernel.org/show_bug.cgi?id=12363
I am testing the debug version right now.
Offline
This is fixed now in 2.6.28.7
so iTCO_wdt doesn't have to be blacklisted anymore.
Offline
4ay
Offline