events/0 high cpu usage

_dunmer · 2010-01-14 16:43:41

Hello

From yesterday i'm experiencing strange problem. Process "events/0" uses approximately 66% of CPU. I didnt do any major system changes or upgrade.
It's anoying becouse writing text is really delayed and if i'm writing a bit faster i "loose" some chars or even whole words.

Can anybody tell me what could be the problem or what to investigate because i have'nt any idea.

tlawson · 2010-02-02 13:42:15

I'm experiencing the same problem as the above poster (although the cpu being under load switches from cpu0 to cpu1 periodically). This is on a Thinkpad x61, running Arch i686. The system pauses roughly every 2 seconds for about a half-second, making typing a bit of an ordeal. Standard browsing doesn't seem to be significantly affected because the mouse is still operational while this is going on.

Browsing my logfiles, dmesg, etc doesn't seem to be showing anything unusual. The problem is still present after dropping out of X, killing numerous daemons, disabling some standard culprits (wireless card), and finally dropping to single-user mode, so I suspect there is some driver I can't track down at fault. However, I'm at the limit of my bugtracking ability and the difficulty in typing text is making it quite difficult to progress further.

What strikes me as strange about this is that it has been several days since a major pacman - in particular, my pacman.log shows that my last updating was about 3 days ago, when I was fiddling around trying to fix a texlive issue, and this pausing definitely wasn't present then.

Stunix · 2010-02-02 14:03:20

Hello tlawson,

I had the exact same issue you are experiencing only a couple of days ago. I am also running Arch on a Thinkpad X61 with HDAPS.

I found that after playing around for a couple of hours I had made an error while following the slim daemon configuration. I had basically set my Openbox session to run with the following in my /home/<user>/.xinitrc

 exec ck-launch-session openbox-session

After resorting to my normal this problem seems to be resolved:

 exec openbox-session

Perhaps this might fix your issue.

If not could you tell me if you are running hdaps on your thinkpad as I originally thought it may be related to that?

Hope this helps & best of luck.

Last edited by Stunix (2010-02-02 14:03:52)

tlawson · 2010-02-02 15:57:57

I did indeed have ck-launch-lession in my .xinitrc. However, removing that doesn't seem to have made a difference. As I mentioned, the problem seems to be present even outside X.

I don't have hdaps running - I don't even have the module loaded or the relevant package installed.

tlawson · 2010-02-03 03:17:26

After further investigation it seems that only certain input devices are subject to this freezing, and only in certain situations. The stylus for my tablet freezes regularly, but the thinkpad mouse pointer does not freeze at all. Currently it seems that input from the keyboard freezes regularly in most programs (xterm, etc) but not in Namoroka/Firefox.

Stunix · 2010-02-04 09:38:06

tlawson,

Do you happen to use wicd for your networking setup?

I spent some time playing last night with my machine trying to replicate the issue you are tackling and found that the wicd package was to blame. I noticed that if I enable the daemon within the rc.conf the machine exhibits the problem you are seeing.

daemons( hal alsa wicd ....... )

Once I run an

/etc/rc.d/wicd stop

the issue persists. The only way I could find to counter the problem was disabling wicd and rebooting the laptop.

Again this could be the cause of your issue (although I have no idea how to fix it, sorry).

tlawson · 2010-02-04 12:45:28

Stunix wrote:

Once I run an
/etc/rc.d/wicd stop
the issue persists. The only way I could find to counter the problem was disabling wicd and rebooting the laptop.
Again this could be the cause of your issue (although I have no idea how to fix it, sorry).

Your suggestion worked. Removing wicd from rc.conf made the problem go away. Restarting the wicd daemon makes the problem start up again (though not immediately) and then, just like in your case, the only way I can get it to stop is a full reboot.

Starting up the networking manually using wpa_supplicant doesn't create the same problem, and so that's what I'm doing as a stopgap measure while I investigate further. Thanks.

Last edited by tlawson (2010-02-04 12:45:59)

Proofrific · 2010-03-06 00:50:21

I'm having the same problem with wicd running up events/0 and making typing very slow. Removing it from /etc/rc.conf stopped the problem, but now I need a new wireless management program. Although there's a bunch listed at http://wiki.archlinux.org/index.php/Wir … management, do you all have recommendations? I use xfce, so I liked wicd for its fewer dependencies.

Cyrusm · 2010-03-06 02:55:07

I've always been happy with netcfg for wireless management, if it's dependencies you care about I think it's definitely the way to go.
It is command line based and text file based, so using it is significantly different from wicd. however, once you get used to it, it works
like a champ imo.

Proofrific · 2010-03-10 08:38:57

For the past few days, I've been using netcfg instead of wicd for wireless. And, I'm again having problems with high CPU usage. This time, it's X. Any ideas?

bk201 · 2010-09-16 22:19:22

I just ran into this exact problem on Gentoo. No recent updates or provocation, it just suddenly broke. event/0 has high cpu usage and I get keyboard stutter. wicd is definitely the culprit as well.

codycarey · 2010-09-16 22:27:27

events/0 is the worker thread used by the kernel to process work queues (bottom halves) for CPU 0. wicd must be hammering your networking drivers and causing a lot of interrupts to be generated. I would guess that means it's an upstream issue for the wicd developers.

EDIT: Check if https://bugs.launchpad.net/wicd/+bug/635607 is related to your problem.

I use wicd-curses without adding the wicd daemon, you might try something similar.

Last edited by codycarey (2010-09-16 22:37:18)

bk201 · 2010-09-16 22:43:57

The problem persists after stopping wicd and even after unloading the iwlagn kernel module for wifi. It doesn't seem like wicd is actively causing the issue but maybe just breaking something else when its started?

codycarey · 2010-09-17 01:11:02

bk201 wrote:

The problem persists after stopping wicd and even after unloading the iwlagn kernel module for wifi. It doesn't seem like wicd is actively causing the issue but maybe just breaking something else when its started?

Well iwlagn is only part of the wireless subsystem. If the problem persists after 'rmmod'ing iwlagn you can try removing iwlcore, mac80211, and cfg80211 in that order until something changes. After you're done you can just 'sudo modprobe iwlagn' and it'll bring them all back.

If you're feeling really brave you can 'strace' the wicd process to see what exactly it's doing.

strace -f -F -r -T -p <wicd PID>

If you want to save that output to a file you can use the '-o <filename>' argument as well.

Last edited by codycarey (2010-09-17 01:12:42)

bk201 · 2010-09-17 02:01:52

Tracing /usr/share/wicd/daemon/wicd-daemon.py and /usr/share/wicd/daemon/monitor.py both show a huge number of system calls, though I really don't know how to interpret them. A common theme seems to be the phrase "Resource temporarily unavailable". Killing both these processes, however, does nothing to alleviate the problem.

btw, I do use wicd-curses. I don't even have X running while I debug this (screen ). wicd still has a daemon process though.

I also just upgraded to 2.6.35 from 2.6.34 but it has not changed the situation.

bk201 · 2010-09-17 02:32:03

ok, so doing

# ifconfig eth0 down

fixes the issue.

if wicd is still running, it fixes the issue for a few seconds until wicd rescans interfaces and does its magic.

EDIT: Actually, it seems that eth0 is now not working at all, with wicd or otherwise. The two ethernet lights are also stuck turned on (the amber and green ones in the port) and I don't recall these ever having been on before when the port wasn't in use. So... how did my ethernet suddenly break and cause all this mayhem?

Last edited by bk201 (2010-09-17 02:58:04)

Proofrific · 2011-03-29 17:46:39

bk201 wrote:

ok, so doing
# ifconfig eth0 down
fixes the issue.
if wicd is still running, it fixes the issue for a few seconds until wicd rescans interfaces and does its magic.
EDIT: Actually, it seems that eth0 is now not working at all, with wicd or otherwise. The two ethernet lights are also stuck turned on (the amber and green ones in the port) and I don't recall these ever having been on before when the port wasn't in use. So... how did my ethernet suddenly break and cause all this mayhem?

My ethernet adapter uses the e1000e module. Removing it via `modprobe -r e1000e` also fixes the problem. I blacklisted it in /etc/rc.conf since I use wireless.

lemagrag · 2011-05-15 09:11:57

I noticed that the problem occurs mostly on lenovo 61's. I have T61p and recently had the same problem: 1 processor virtually frozen every 2 seconds for half a second with 100% load of events/0 process. Keyboard not responding at that time, eth0 not working, `ifconfig eth0 down` works out the CPU load.

I did one thing -- unplugged the power cord, removed the battery waited a while and inserted everything back again.
Everything works fine now. (Simple restart or even computer shut down didn't work).

It seems that the lenovo ethernet card experiences some kind of hardware deadlock such that the drivers can't solve by simple software-resetting the device.
I think that battery removal is necessary in this case, because the ethernet card should have power supply connected all the time no matter if the computer is shut down (it has a wake-on-LAN feature after all).

Hope I helped,
Piotr.

Arch Linux

#1 2010-01-14 16:43:41

events/0 high cpu usage

#2 2010-02-02 13:42:15

Re: events/0 high cpu usage

#3 2010-02-02 14:03:20

Re: events/0 high cpu usage

#4 2010-02-02 15:57:57

Re: events/0 high cpu usage

#5 2010-02-03 03:17:26

Re: events/0 high cpu usage

#6 2010-02-04 09:38:06

Re: events/0 high cpu usage

#7 2010-02-04 12:45:28

Re: events/0 high cpu usage

#8 2010-03-06 00:50:21

Re: events/0 high cpu usage

#9 2010-03-06 02:55:07

Re: events/0 high cpu usage

#10 2010-03-10 08:38:57

Re: events/0 high cpu usage

#11 2010-09-16 22:19:22

Re: events/0 high cpu usage

#12 2010-09-16 22:27:27

Re: events/0 high cpu usage

#13 2010-09-16 22:43:57

Re: events/0 high cpu usage

#14 2010-09-17 01:11:02

Re: events/0 high cpu usage

#15 2010-09-17 02:01:52

Re: events/0 high cpu usage

#16 2010-09-17 02:32:03

Re: events/0 high cpu usage

#17 2011-03-29 17:46:39

Re: events/0 high cpu usage

#18 2011-05-15 09:11:57

Re: events/0 high cpu usage

Board footer