Re: kernel crash

[Patrick McHardy <kaber@trash.net>]

Jan Engelhardt wrote:
> On Jan 29 2008 18:32, fenglg wrote:
>> I use linux-2.6.18, and there is a bridge with eth0 and eth1. The eth0 and
>> eth1 connetcswitchs which use vlan trunk(802.1q). When i run system some
>> hours, the kernel is crash, anyone can help me.
>>
>> BUG: unable to handle kernel NULL pointer dereference at virtual address 00000010
>> EIP is at br_nf_pre_routing_finish+0x1d/0x340
>> eax: c16e7980   ebx: 00000000   ecx: 00000001   edx: c16e7980
>> esi: de4cb020   edi: c0456520   ebp: de8d2000   esp: c03fbce4
>> Code: 10 01 e9 7a ff ff ff 8d b4 26 00 00 00 00 55 57 56 53 81 ec c0 00 00 00 8b
>> 94 24 d4 00 00 00 8b 9a 80 00 00 00 8b 6a 14 8b 72 20 <8b> 43 10 a8 01 74 14 0f
>> b6 42 75 24 f8 0c 03 88 42 75 8b 43 10
> 
> Thanks for the report.
> 
> All signs point to skb->nf_bridge being NULL.
> 
> 	static int br_nf_pre_routing_finish(struct sk_buff *skb)
> 	{
> 	        struct net_device *dev = skb->dev;
> 	        struct iphdr *iph = ip_hdr(skb);
> 	        struct nf_bridge_info *nf_bridge = skb->nf_bridge;
> 	        int err;
> 
> boom->		if (nf_bridge->mask & BRNF_PKT_TYPE) {
> 
> 
> Hm... now what? :)


2.6.18 and 2.6.24 differ significantly in how the bridging stuff
is handled, so the preferred way would be to try to reproduce this
with 2.6.24. Debugging a 18 month old kernel doesn't seem too
useful unless we know the problem is still present.