ipv6: avoid unregistering inet6_dev for loopback

commit 60abc0be96e00ca71bac083215ac91ad2e575096 upstream.

The per netns loopback_dev->ip6_ptr is unregistered and set to
NULL when its mtu is set to smaller than IPV6_MIN_MTU, this
leads to that we could set rt->rt6i_idev NULL after a
rt6_uncached_list_flush_dev() and then crash after another
call.

In this case we should just bring its inet6_dev down, rather
than unregistering it, at least prior to commit 176c39af29
("netns: fix addrconf_ifdown kernel panic") we always
override the case for loopback.

Thanks a lot to Andrey for finding a reliable reproducer.

Fixes: 176c39af29 ("netns: fix addrconf_ifdown kernel panic")
Reported-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Daniel Lezcano <dlezcano@fr.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: David Ahern <dsahern@gmail.com>
Tested-by: Andrey Konovalov <andreyknvl@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 3.2: the NETDEV_CHANGEMTU case used to fall-through to the
 NETDEV_DOWN case here, so replace that with a separate call to addrconf_ifdown()]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
This commit is contained in:
WANG Cong 2017-06-21 14:34:58 -07:00 committed by syphyr
parent cb4e7ced75
commit 7abc0b32da

View file

@ -3087,6 +3087,8 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
* MTU falled under IPV6_MIN_MTU. * MTU falled under IPV6_MIN_MTU.
* Stop IPv6 on this interface. * Stop IPv6 on this interface.
*/ */
addrconf_ifdown(dev, dev != net->loopback_dev);
break;
case NETDEV_DOWN: case NETDEV_DOWN:
case NETDEV_UNREGISTER: case NETDEV_UNREGISTER: