Hi,
> From: Jay Cliburn <[email protected]>
>
> On Sat, 17 Jan 2009 18:49:18 +0000 (UTC)
> Sitsofe Wheeler wrote:
>
> > On an ever so slightly modified 2.6.28. I think this only happens
> > when the upstream router/switch goes out to lunch and packets are
> > being sent. It is reproducible but only appears on a home network...
> >
> > [ 204.704065] ------------[ cut here ]------------
> > [ 204.704074] WARNING: at net/sched/sch_generic.c:226 dev_watchdog
> > +0x22b/0x240()
> > [ 204.704080] NETDEV WATCHDOG: eth0 (atl2): transmit timed out
>
> Please provide your complete dmesg output.
Unfortunately I only saved that snippet of the log (I can provide you a full dmesg but it won't be
from a run where this problem occured).
> Can you briefly describe the device immediately upstream of the system producing
> the warning?
I believe it is a USRobotics ADSL modem (nmap reckons its a SureConnect 9105 but I have no means of checking as I don't have physical access to the box where it is contained). Every now and again this thing seems to buckle under network load and goes out to lunch. When this happens ethtool reports that the line has dropped and sometimes this warning appears...
On Sun, 18 Jan 2009 02:18:33 -0800 (PST)
Sitsofe Wheeler <[email protected]> wrote:
> Hi,
>
>
> > From: Jay Cliburn <[email protected]>
> >
> > On Sat, 17 Jan 2009 18:49:18 +0000 (UTC)
> > Sitsofe Wheeler wrote:
> >
> > > On an ever so slightly modified 2.6.28. I think this only happens
> > > when the upstream router/switch goes out to lunch and packets are
> > > being sent. It is reproducible but only appears on a home
> > > network...
> > >
> > > [ 204.704065] ------------[ cut here ]------------
> > > [ 204.704074] WARNING: at net/sched/sch_generic.c:226
> > > dev_watchdog +0x22b/0x240()
> > > [ 204.704080] NETDEV WATCHDOG: eth0 (atl2): transmit timed out
> >
> > Please provide your complete dmesg output.
>
> Unfortunately I only saved that snippet of the log (I can provide you
> a full dmesg but it won't be from a run where this problem occured).
I'd like to see the full dmesg that includes the warning, if you can
obtain it next time it occurs. Meanwhile, please see if this patch
helps.
diff --git a/drivers/net/atlx/atl2.c b/drivers/net/atlx/atl2.c
index 8571e8c..376226c 100644
--- a/drivers/net/atlx/atl2.c
+++ b/drivers/net/atlx/atl2.c
@@ -555,14 +555,18 @@ static void atl2_check_for_link(struct atl2_adapter *adapter)
atl2_read_phy_reg(&adapter->hw, MII_BMSR, &phy_data);
spin_unlock(&adapter->stats_lock);
- /* notify upper layer link down ASAP */
if (!(phy_data & BMSR_LSTATUS)) { /* Link Down */
- if (netif_carrier_ok(netdev)) { /* old link state: Up */
- printk(KERN_INFO "%s: %s NIC Link is Down\n",
- atl2_driver_name, netdev->name);
- adapter->link_speed = SPEED_0;
- netif_carrier_off(netdev);
- netif_stop_queue(netdev);
+ if (netif_carrier_ok(netdev)) {
+ /*
+ * Notify the upper layer and restart the netdev
+ * watchdog timer while we try to recover the link.
+ */
+ netif_carrier_off(netdev);
+ netif_stop_queue(netdev);
+ netdev->trans_start = jiffies;
+ printk(KERN_INFO "%s: %s NIC Link is Down\n",
+ atl2_driver_name, netdev->name);
+ adapter->link_speed = SPEED_0;
}
}
schedule_work(&adapter->link_chg_task);