Re: [PATCH net v3 1/2] rtnetlink: allow to set iface down before enslaving it

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

Le 05/01/2024 à 12:59, Jiri Pirko a écrit :
> Thu, Jan 04, 2024 at 05:42:59PM CET, nicolas.dichtel@6wind.com wrote:
>> The below commit adds support for:
>>> ip link set dummy0 down
>>> ip link set dummy0 master bond0 up
>>
>> but breaks the opposite:
>>> ip link set dummy0 up
>>> ip link set dummy0 master bond0 down
> 
> It is a bit weird to see these 2 and assume some ordering.
> The first one assumes:
> dummy0 master bond 0, dummy0 up
> The second one assumes:
> dummy0 down, dummy0 master bond 0
> But why?
> 
> What is the practival reason for a4abfa627c38 existence? I mean,
> bond/team bring up the device themselfs when needed. Phil?
> Wouldn't simple revert do better job here?
Yeah, I assumed there was a good reason, but you're right.


> 
> 
>>
>> Let's add a workaround to have both commands working.
>>
>> Cc: stable@vger.kernel.org
>> Fixes: a4abfa627c38 ("net: rtnetlink: Enslave device before bringing it up")
>> Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
>> Acked-by: Phil Sutter <phil@nwl.cc>
>> Reviewed-by: David Ahern <dsahern@kernel.org>
>> ---
>> net/core/rtnetlink.c | 8 ++++++++
>> 1 file changed, 8 insertions(+)
>>
>> diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
>> index e8431c6c8490..dd79693c2d91 100644
>> --- a/net/core/rtnetlink.c
>> +++ b/net/core/rtnetlink.c
>> @@ -2905,6 +2905,14 @@ static int do_setlink(const struct sk_buff *skb,
>> 		call_netdevice_notifiers(NETDEV_CHANGEADDR, dev);
>> 	}
>>
>> +	/* Backward compat: enable to set interface down before enslaving it */
>> +	if (!(ifm->ifi_flags & IFF_UP) && ifm->ifi_change & IFF_UP) {
>> +		err = dev_change_flags(dev, rtnl_dev_combine_flags(dev, ifm),
>> +				       extack);
>> +		if (err < 0)
>> +			goto errout;
>> +	}
>> +
>> 	if (tb[IFLA_MASTER]) {
>> 		err = do_set_master(dev, nla_get_u32(tb[IFLA_MASTER]), extack);
>> 		if (err)
>> -- 
>> 2.39.2
>>
>>