Re: [PATCH] net: ipv6: Make address flushing on ifdown optional - v3

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

Le 13/02/2015 04:47, David Ahern a écrit :
> On 2/12/15 9:57 AM, Nicolas Dichtel wrote:
>> Le 12/02/2015 05:27, David Ahern a écrit :
>>> Currently, all ipv6 addresses are flushed when the interface is
>>> configured
>>> down, even static address:
>>>
>> [snip]
>>>
>>> [root@f20 ~]# echo 0 > /proc/sys/net/ipv6/conf/eth1/flush_addr_on_down
>>> [root@f20 ~]# ip -6 addr add dev eth1 2000:11:1:1::1/64
>>> [root@f20 ~]# ip addr show dev eth1
>>> 3: eth1: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast state DOWN
>>> group default qlen 1000
>>>      link/ether 02:04:11:22:33:01 brd ff:ff:ff:ff:ff:ff
>>>      inet6 2000:11:1:1::1/64 scope global tentative
>>>         valid_lft forever preferred_lft forever
>>> [root@f20 ~]#  ip link set dev eth1 up
>>> [root@f20 ~]#  ip link set dev eth1 down
>>> [root@f20 ~]# ip addr show dev eth1
>>> 3: eth1: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast state DOWN
>>> group default qlen 1000
>>>      link/ether 02:04:11:22:33:01 brd ff:ff:ff:ff:ff:ff
>>>      inet6 2000:11:1:1::1/64 scope global
>>>         valid_lft forever preferred_lft forever
>>>      inet6 fe80::4:11ff:fe22:3301/64 scope link
>>>         valid_lft forever preferred_lft forever
>> Can you show an output of 'ip -6 route list table local' and 'ip -6
>> route' ?
>
> hmmmm.... interesting. once again ipv6 behavior is radically different from ipv4.
>
> This is *after* a configure, up, down cycle:
>
> [root@f21 ~]# ip -6 route list table local
> local ::1 dev lo  proto none  metric 0
> local 2000:11:1:1::1 dev lo  proto none  metric 0
> local fe80::11:22ff:fe33:4402 dev lo  proto none  metric 0
> ff00::/8 dev eth0  metric 256
We also need to check netlink notifications. When you set an interface down, the
kernel doesn't send rtnl notifications for IPv6 routes. Daemons have to take
care of the ifdow notification.
If your patch is going to keep some routes after an ifdown, we need to have a
way in userland to know this.
Advertising the new sysctl (flush_addr_on_down) entry via netconf may be a
solution (Stephen already talks about this feature).

>
> [root@f21 ~]# ip -6 route ls
> unreachable ::/96 dev lo  metric 1024  error -101
> unreachable ::ffff:0.0.0.0/96 dev lo  metric 1024  error -101
> unreachable 2002:a00::/24 dev lo  metric 1024  error -101
> unreachable 2002:7f00::/24 dev lo  metric 1024  error -101
> unreachable 2002:a9fe::/32 dev lo  metric 1024  error -101
> unreachable 2002:ac10::/28 dev lo  metric 1024  error -101
> unreachable 2002:c0a8::/32 dev lo  metric 1024  error -101
> unreachable 2002:e000::/19 dev lo  metric 1024  error -101
> unreachable 3ffe:ffff::/32 dev lo  metric 1024  error -101
> fe80::/64 dev eth0  proto kernel  metric 256
>
> (the ipv6 addresses on lo show up on stock 3.18.3-201.fc21.x86_64; no idea why)
>
> But on a subsequent ifconfig up the route is not inserted:
>
> [root@f21 ~]# ifconfig eth1 up
>
> [root@f21 ~]# ip -6 route ls
> unreachable ::/96 dev lo  metric 1024  error -101
> unreachable ::ffff:0.0.0.0/96 dev lo  metric 1024  error -101
> unreachable 2002:a00::/24 dev lo  metric 1024  error -101
> unreachable 2002:7f00::/24 dev lo  metric 1024  error -101
> unreachable 2002:a9fe::/32 dev lo  metric 1024  error -101
> unreachable 2002:ac10::/28 dev lo  metric 1024  error -101
> unreachable 2002:c0a8::/32 dev lo  metric 1024  error -101
> unreachable 2002:e000::/19 dev lo  metric 1024  error -101
> unreachable 3ffe:ffff::/32 dev lo  metric 1024  error -101
> fe80::/64 dev eth0  proto kernel  metric 256
> fe80::/64 dev eth1  proto kernel  metric 256
>
> So I need to look into why.
Yes. The patch will probably have to play with this connected route.


Regards,
Nicolas