Re: [PATCH net-next v4 0/7] bonding: Extend arp_ip_target format to allow for a list of vlan tags.

[Jay Vosburgh <jv@jvosburgh.net>]

David Wilder <wilder@us.ibm.com> wrote:

>From: Hangbin Liu <liuhangbin@gmail.com>
>Sent: Wednesday, August 6, 2025 9:38 PM
>To: David Wilder
>Cc: netdev@vger.kernel.org; jv@jvosburgh.net; pradeeps@linux.vnet.ibm.com; Pradeep Satyanarayana; i.maximets@ovn.org; Adrian Moreno Zapata
>Subject: [EXTERNAL] Re: [PATCH net-next v4 0/7] bonding: Extend arp_ip_target format to allow for a list of vlan tags.
>
>Hi Hangbin
>
>> Hi David,
>> On Tue, Jul 01, 2025 at 07:13:06AM +0000, Hangbin Liu wrote:
>> > From: Hangbin Liu <liuhangbin@gmail.com>
>> > Sent: Monday, June 30, 2025 3:18 AM
>> > To: David Wilder
>> > Cc: netdev@vger.kernel.org; jv@jvosburgh.net; pradeeps@linux.vnet.ibm.com; Pradeep Satyanarayana; i.maximets@ovn.org; Adrian Moreno Zapata; Hangbin Liu
>> > Subject: [EXTERNAL] Re: [PATCH net-next v4 0/7] bonding: Extend arp_ip_target format to allow for a list of vlan tags.
>> >
>> > > On Fri, Jun 27, 2025 at 01:17:13PM -0700, David Wilder wrote:
>> > > I have run into issues with the ns_ip6_target feature.  I am unable to get
>> > > the existing code to function with vlans. Therefor I am unable to support
>> > > A this change for ns_ip6_target.
>> >
>> > > Any reason why this is incompatible with ns_ip6_target?
>> >
>> > Hi Hangbin
>> >
>> > I am unable to get the existing ns_ip6_target code to function when the target
>> > is in a vlan. If the existing code is not working with vlans it makes no
>> > sense to specify the vlan tags.
>> >
>> > This is what I think is happening:
>> >
>> > In ns_send_all() we have this bit of code:
>> >
>> > dst = ip6_route_output(dev_net(bond->dev), NULL, &fl6);
>> > if (dst->error) {
>> >         dst_release(dst);
>> >         /* there's no route to target - try to send arp
>> >          * probe to generate any traffic (arp_validate=0)
>> >          */
>> >         if (bond->params.arp_validate)
>> >                bond_ns_send(slave, &targets[i], &in6addr_any, tags);
>> >                <.......>
>> >                continue;
>> > }
>> >
>> > ip6_route_output() is returning an error as there is no neighbor entry for
>> > the target. A ns is then sent with no vlan header. I found that the
>> > multicast ns (with no vlan header) is not passed to the vlan siblings
>> > with the target address so no reply is sent.
>> >
>> > The ipv4 code is simmiler but the arp is sent as a brodcast. The broadcast arp
>> > will be propagated to the vlan sibling (in the linux vlan code).
>> >
>> > This could be a testing issue,  I am unsure.  Can you help with
>> > a test case with the target in a vlan?
>
>> I looked into this recently, and you are right — ip6_route_output() returns
>> an error dst. The root cause is that we cannot get the destination IPv6
>> address through the bond interface, because the source IPv6 address is
>> configured on other interfaces, the VLAN sub-interface.
>>
>> This is a key difference between IPv6 and IPv4:
>> In IPv4, it's possible to get a destination route via the bond even when the
>> source IP is configured on a different interface. But in IPv6, the routing
>> mechanism is stricter in requiring the source address to be valid on the
>> outgoing interface.
>>
>> I'm not sure how to fix this yet, as it's fundamentally tied to how IPv6
>> routing behaves.
>
>I am thinking that we don't need to do a route lookup as if we are sending
>from the bonding interface.  We only need to find the interface we should
>send the packet through.  As if we ran "ip route get <dest addr>".

	Assuming I'm following correctly, the whole point of the route
lookup is to determine which interface the ARP (or NS for IPv6) should
nominally sent through (based on the destination address).  This serves
two purposes:

	- collecting the VLAN tags,

	- insuring that the ARP / NS won't be sent on a logically
incorrect interface (e.g., its address corresponds to some totally
unrelated interface).

	So, really, I'm agreed that what we're really looking for is
"what is the proper output interface to use to send to destination X,"
which we can then check to see if that interface is logically connected
to the bond (e.g., a VLAN atop the bond).

	Is the solution to call ip6_route_output() with the flowi6_oif
set to zero?  That seems to be what happens for the "ip route get" case
in inet6_rtm_getroute() (he says, looking at the code but not running
actual tests).

	-J

---
	-Jay Vosburgh, jv@jvosburgh.net