Re: Strange Panic (Deadlock)

[Jarek Poplawski <jarkao2@gmail.com>]

On Tue, Dec 25, 2007 at 12:11:50PM +0300, Badalian Vyacheslav wrote:
...
> I have 4 machine. All platforms different.   All machine do 1 time in
> hour rebuild TC and IPTABLES rules.
> After it do
> echo START >> log.txt
> iptables-restore < xxx.txt
> tc qdisc del dev eth0 root
> tc qdisc del dev eth1 root
> tc -b new_rules.txt
> echo END >> log.txt
> 
> and its all that its doing.
> Bug always be between START and END
> All machines have above 300mbs traffic.
> I try turn off rebuilding rules on 1 PC and it work 3 week without reboot!

Hi Slavon,

After some looking around net schedulers' timers I think you could try
these 3 patches (2 in next messages). They should change 3 suspicious
(maybe only to me) places, but it's only guessing, to eliminate some
most nearby possibilities. These patches are independent, but of course
trying all at once should be quicker.
 
Thanks,
Jarek P.

[PATCH 1/3]
---

diff -Nurp linux-2.6.23.12-/net/sched/sch_generic.c linux-2.6.23.12+/net/sched/sch_generic.c
--- linux-2.6.23.12-/net/sched/sch_generic.c	2007-12-21 22:26:15.000000000 +0100
+++ linux-2.6.23.12+/net/sched/sch_generic.c	2007-12-26 18:39:20.000000000 +0100
@@ -251,10 +251,8 @@ static void dev_watchdog_up(struct net_d
 
 static void dev_watchdog_down(struct net_device *dev)
 {
-	netif_tx_lock_bh(dev);
-	if (del_timer(&dev->watchdog_timer))
+	if (del_timer_sync(&dev->watchdog_timer))
 		dev_put(dev);
-	netif_tx_unlock_bh(dev);
 }
 
 void netif_carrier_on(struct net_device *dev)
@@ -560,6 +558,8 @@ void dev_deactivate(struct net_device *d
 	struct Qdisc *qdisc;
 	struct sk_buff *skb;
 
+	dev_watchdog_down(dev);
+
 	spin_lock_bh(&dev->queue_lock);
 	qdisc = dev->qdisc;
 	dev->qdisc = &noop_qdisc;
@@ -572,8 +572,6 @@ void dev_deactivate(struct net_device *d
 
 	kfree_skb(skb);
 
-	dev_watchdog_down(dev);
-
 	/* Wait for outstanding dev_queue_xmit calls. */
 	synchronize_rcu();