Re: rcu_preempt running flat out on idle desktop.
From: Frederic Weisbecker
Date: Thu Jun 06 2013 - 11:43:23 EST
On Thu, Jun 06, 2013 at 11:31:36AM -0400, Dave Jones wrote:
> On Tue, May 14, 2013 at 03:21:07AM +0200, Frederic Weisbecker wrote:
> > On Thu, May 09, 2013 at 05:10:26PM -0400, Dave Jones wrote:
> > > On Thu, May 09, 2013 at 11:02:08PM +0200, Frederic Weisbecker wrote:
> > >
> > > > > RCU options for this build are..
> > > > >
> > > > I can't reproduce that issue even with the above setting. Could you
> > > > please send me your whole config?
> > >
> > > http://paste.fedoraproject.org/11363/36813375/raw/
> >
> > Sorry it took me some time to get that config booting, I ran into
> > other unrelated crashes.
> >
> > So I'm observing the exact opposite than you: the processes appear to
> > have no load nor to consume cputime. Even if I use some dumb user loop.
> > Top don't even show it.
> >
> > I'm digging further and will tell you when I get more info.
>
> So I'm still seeing this on rc4.
>
> Here's my 'idle' firewall, just routing packets:
>
> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
> 4405 root 20 0 0 0 0 S 400.0 0.0 52336:27 kworker/2:0
> 20968 davej 20 0 115m 1840 1092 R 400.0 0.1 2297:48 top
> 964 named 20 0 383m 32m 2896 S 400.0 1.7 81384:54 named
> 10 root 20 0 0 0 0 S 400.0 0.0 195403:05 rcu_preempt
> 20830 davej 20 0 125m 2172 1024 S 400.0 0.1 939:36.10 sshd
> 11 root 20 0 0 0 0 S 400.0 0.0 42644:33 rcuop/0
> 13 root 20 0 0 0 0 S 400.0 0.0 34729:04 rcuop/2
> 754 lighttpd 20 0 54640 2692 1732 S 400.0 0.1 35286:11 lighttpd
> 756 root 20 0 216m 10m 6116 S 400.0 0.5 34143:41 snmpd
> 35 root 20 0 0 0 0 S 400.0 0.0 12790:56 ksoftirqd/2
> 19893 root 20 0 0 0 0 S 400.0 0.0 12570:49 kworker/0:1
> 20059 root 20 0 0 0 0 S 400.0 0.0 4666:30 kworker/1:1
> 32 root rt 0 0 0 0 S 400.0 0.0 8780:22 watchdog/2
> 792 root 20 0 92220 2516 632 S 400.0 0.1 7821:28 sendmail
> 25 root rt 0 0 0 0 S 400.0 0.0 8780:21 watchdog/0
> 26 root rt 0 0 0 0 S 400.0 0.0 8780:22 watchdog/1
> 38 root rt 0 0 0 0 S 400.0 0.0 8780:22 watchdog/3
> 19167 root 20 0 0 0 0 S 400.0 0.0 7804:04 kworker/3:2
>
>
> Every process is either 400%, or 0%.
>
> Here's my desktop box, just writing this email:
>
> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
> 10 root 20 0 0 0 0 S 200.0 0.0 35134:00 rcu_preempt
> 12 root 20 0 0 0 0 S 200.0 0.0 14059:47 rcuop/1
> 12528 root 20 0 0 0 0 S 200.0 0.0 1012:56 kworker/0:1
> 12999 davej 20 0 115m 1700 956 R 200.0 0.1 1:09.10 top
> 4799 root 20 0 0 0 0 S 200.0 0.0 3510:40 kworker/1:1
> 1 root 20 0 50760 5092 1648 S 0.0 0.2 3:14.96 systemd
> 2 root 20 0 0 0 0 S 0.0 0.0 1:40.55 kthreadd
>
> Every process 200% or 0%.
I see, would you mind testing this branch?
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git
timers/urgent
It might help, I specially think about 45eacc692771bd2b1ea3d384e6345cab3da10861
("vtime: Use consistent clocks among nohz accounting") which fixes an issue
with cputime accounting that resulted in similar symptoms.
Thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/