Hello,
On Wed, 14 Nov 2018, Xin Long wrote:
> ip_vs_dst_event is supposed to clean up all dst used in ipvs'
> destinations when a net dev is going down. But it works only
> when the dst's dev is the same as the dev from the event.
>
> Now with the same priority but late registration,
> ip_vs_dst_notifier is always called after ipv6_dev_notf where
> the dst's dev is set to lo for NETDEV_DOWN event.
>
> As the dst's dev lo is not the same as the dev from the event
> in ip_vs_dst_event(), ipv6_dev_notf can actually never work.
> Also as these dst have to wait for dest_trash_timer to clean
> them up. It would cause some non-permanent kernel warnings:
>
> unregister_netdevice: waiting for br0 to become free. Usage count = 3
>
> To fix it, call ip_vs_dst_notifier before ipv6_dev_notf by
> increasing its priority to ADDRCONF_NOTIFY_PRIORITY + 5.
>
> Fixes: 7a4f0761fce3 ("IPVS: init and cleanup restructuring")
> Reported-by: Li Shuang <shuali@xxxxxxxxxx>
> Signed-off-by: Xin Long <lucien.xin@xxxxxxxxx>
OK,
Acked-by: Julian Anastasov <ja@xxxxxx>
But this fix may hide another problem...
Isn't that racy to change dst.dev during NETDEV_DOWN?
IPv4 calls rt_flush_dev() only on NETDEV_UNREGISTER, while
rt6_uncached_list_flush_dev() is called in rt6_disable_ip(),
even on NETDEV_DOWN. How fatal is to call it only on
NETDEV_UNREGISTER? By this way, we rely on synchronize_net()
before the NETDEV_UNREGISTER event to avoid any dst.dev access.
> ---
> net/netfilter/ipvs/ip_vs_ctl.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
> index 83395bf6..aded477 100644
> --- a/net/netfilter/ipvs/ip_vs_ctl.c
> +++ b/net/netfilter/ipvs/ip_vs_ctl.c
> @@ -3980,6 +3980,7 @@ static void __net_exit ip_vs_control_net_cleanup_sysctl(struct netns_ipvs *ipvs)
>
> static struct notifier_block ip_vs_dst_notifier = {
> .notifier_call = ip_vs_dst_event,
> + .priority = ADDRCONF_NOTIFY_PRIORITY + 5,
> };
>
> int __net_init ip_vs_control_net_init(struct netns_ipvs *ipvs)
> --
> 2.1.0
Regards
--
Julian Anastasov <ja@xxxxxx>
