[openib-general] [RFC] [PATCH v2] IB/ipoib: Add bonding support to IPoIB

[Michael S. Tsirkin]

> 
> During my tests I found that when running 
> 
> 	1. modprobe -r ib_mthca (to delete IPoIB interfaces)
> 	2. ping somewhere on the subnet of bond0
> 
> I get this stack dump (which ends with kernel death)
> 	 [<ffffffff8037ff32>] skb_under_panic+0x5c/0x60
> 	 [<ffffffff882e00c2>] :ib_ipoib:ipoib_hard_header+0xa6/0xc0
> 	 [<ffffffff803c3c98>] arp_create+0x120/0x226
> 	 [<ffffffff803c3dc3>] arp_send+0x25/0x3b
> 	 [<ffffffff803c466a>] arp_solicit+0x186/0x195
> 	 [<ffffffff8038c0ac>] neigh_timer_handler+0x2b5/0x309
> 	 [<ffffffff8038bdf7>] neigh_timer_handler+0x0/0x309
> 	 [<ffffffff80239599>] run_timer_softirq+0x130/0x19e
> 	 [<ffffffff80235fcc>] __do_softirq+0x55/0xc3
> 	 [<ffffffff8020acac>] call_softirq+0x1c/0x28
> 	 [<ffffffff8020c02b>] do_softirq+0x2c/0x7d
> 	 [<ffffffff8021864a>] smp_apic_timer_interrupt+0x57/0x6a
> 	 [<ffffffff80208e19>] mwait_idle+0x0/0x45
> 	 [<ffffffff8020a756>] apic_timer_interrupt+0x66/0x70
> 	 <EOI>  [<ffffffff80208e5b>] mwait_idle+0x42/0x45
> 	 [<ffffffff80208db1>] cpu_idle+0x8b/0xae
> 	 [<ffffffff80217d60>] start_secondary+0x47f/0x48f
> 
> The only way I found to avoid this (for now) is to check skb headroom in
> ipoib_hard_header. I guess that this safety check doesn't harm regular IPoIB 
> operation and it seems to solve my problem. However, I would be happy to hear what
> others think of this last issue.

This seems to mean that hard_header_len is not copied from slave to master
device. Right? Maybe that's what needs to be fixed.

-- 
MST