Re: NOHZ: local_softirq_pending 08

[Dave Jones <davej@redhat.com>]

On Wed, Jun 06, 2012 at 01:15:50AM +0200, Francois Romieu wrote:
 > Marc Dionne <marc.c.dionne@gmail.com> :
 > > If it helps, I've been seeing the same thing here on a system with an
 > > r8169.  The message appears consistently enough on network startup to
 > > be able to track it down to this specific commit:
 > >     98ddf986fca17840e46e070354b7e2cd2169da15: r8169: bh locking redux
 > > and task scheduling.
 > > 
 > > The softirq pending messages always come right after "link down"
 > > messages for the interface.

for me, they're during boot when the nic comes up, and then very occasionally afterwards.

 > Let's see the traces.
 > 
 > I have added some code to trace the timing and values of the irq in
 > rtl8169_interrupt(). Patches against current git are attached.
 
with these patches applied, I get a lot of spew in dmesg like ..

[   98.659314] r8169 0000:01:00.0: eth0: event ring full.
[   98.763513] r8169 0000:01:00.0: eth0: event ring full.
[   99.182103] r8169 0000:01:00.0: eth0: event ring full.
[   99.294426] r8169 0000:01:00.0: eth0: event ring full.
[   99.420255] r8169 0000:01:00.0: eth0: event ring full.
[   99.455811] r8169 0000:01:00.0: eth0: event ring full.
[   99.520541] r8169 0000:01:00.0: eth0: event ring full.
[   99.730789] r8169 0000:01:00.0: eth0: event ring full.
[   99.887643] r8169 0000:01:00.0: eth0: event ring full.
[  103.798861] net_ratelimit: 43 callbacks suppressed

over and over..

 > 'ip link set dev 8168b-lom up' + 'cat /sys/kernel/debug/tracing/trace':

What tracer do I need to configure to get make this work ? irqsoff ?

finally, /sys/kernel/debug/r8169/ is empty.

	Dave