Re: [BUG] stable v3.10.16+ introduced by "ip6tnl: allow to use rtnl ops on fb tunnel"

[Nicolas Dichtel <nicolas.dichtel@6wind.com>]

Le 11/12/2013 22:53, David Miller a écrit :
> From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> Date: Sun, 8 Dec 2013 16:25:31 -0800
>
>> On Wed, Nov 13, 2013 at 09:14:30PM -0500, Steven Rostedt wrote:
>>> +++ b/net/ipv6/ip6_tunnel.c
>>> @@ -1731,8 +1731,6 @@ static void __net_exit ip6_tnl_destroy_tunnels(struct ip
>>>                  }
>>>          }
>>>
>>> -       t = rtnl_dereference(ip6n->tnls_wc[0]);
>>> -       unregister_netdevice_queue(t->dev, &list);
>>>          unregister_netdevice_many(&list);
>>>   }
>>>
>>>
>>> Which, when applied to 3.10.18, fixes the bug. Was there a reason that
>>> this part of the commit wasn't backported? or was this just an oversight?
>>
>> It looks like it was left out to me as well.
>>
>> David, any objection to me making this fixup in the 3.10-stable tree?
>
> The original patch submitted told me to leave this part of the patch
> out of the backport, explaining that it wasn't necessary in older
> kernels.
Yes, and this was right (in upstream commit, I remove this part because
the fb device is deleted by the loop which check dev->rtnl_ops) , but ...

>
> Can someone please sort this out?
>
> Nicolas please provide some guidance here, thanks.
The original patch left a bug, which was fixed upstream with this commit:
1e9f3d6f1c40 ip6tnl: fix use after free of fb_tnl_dev

The problem is a bit different in 3.10.y, because there is no x-vrf support.
When ip6_tunnel.ko is unloaded, FB device is deleted by rtnl_link_unregister()
and then we try to delete it again in ip6_tnl_destroy_tunnels().
Thus the fix is different and in fact, the above patch is good.

Steven, will you submit this patch properly or should I do this?