Messages in this thread | | | Date | Fri, 08 Jun 2012 19:46:24 +0300 | From | Nikos Chantziaras <> | Subject | Re: NOHZ: local_softirq_pending 08 |
| |
On 31/05/12 17:04, Dave Jones wrote: > On Thu, May 31, 2012 at 09:56:25AM +0200, Eric Dumazet wrote: > > On Thu, 2012-05-31 at 09:52 +0200, Thomas Gleixner wrote: > > > Dave, > > > > > > On Wed, 30 May 2012, Dave Jones wrote: > > > > > > > May 30 17:43:04 [ 1095.335659] NOHZ: local_softirq_pending 08 > > > > > > > > According to git, you ratelimited this back in 2007 with the message > > > > "until we found the root cause of that problem." > > > > > > > > Is this an old problem that has returned ? > > > > > > Looks like. > > > > > > > Maybe interesting note: Both machines I see this on regularly are Atom cpus, > > > > and it's always '08'. > > > > > > So it goes idle with NET_RX softirq pending. That really should not > > > happen. I fear you need to fire up the tracer :) > > > > It might be a bug in the network driver, a race with NAPI completion. > > > > Dave, which net drivers run on these machines ? > > r8169. > > I got an email from someone else asking if I used Realtek drivers, which I thought > was an uncanny guess.
Same thing happens here (kernel 3.4.1, also happened with 3.4.0):
r8169 0000:06:00.0: eth0: link down r8169 0000:06:00.0: eth0: link down NOHZ: local_softirq_pending 08 NOHZ: local_softirq_pending 08 NOHZ: local_softirq_pending 08 r8169 0000:06:00.0: eth0: link up
| |