Hi I'm having troubles in BIRD configuration, and I'm pretty sure it's just a trivial wrong configure issue but I really tried my best >_<

Let me give you some background.

1. There are 2 hosts connected by wire (so that they are L2 reachable), one with IPv4 10.22.12.88, and another is 10.22.12.44. Here's one the arp record from "ip neigh sh":
10.22.12.88 dev bond0 lladdr f8:6e:ee:8b:cf:90 REACHABLE

2. Both 2 hosts have some KVM virtual machines, with corresponding routes from host to vm, here's one of the routes from "ip r":
10.233.25.178 dev yap35caa1e88854  scope link  src 10.22.12.44
The 10.233.25.178 is the address of a VM, and the dev yap35caa1e88854 is a tap device connecting eth0 inside the VM.

What I want to do is to build routes between these 2 hosts, in order to let VMs on host A be able to connect VMs on host B.

Hope I make myself clear.

So I'm using BIRD 2.0.3 on Ubuntu 16.04, with the config file as follows:

log syslog all;
router id 10.22.12.44;

protocol device {
  debug { states };
  scan time 2;
}

protocol kernel {
        learn;
        persist;
        scan time 10;
        ipv4 {
                import all;
                export all;
        };
}

protocol bgp {
        debug { states };
        local as 65000;
        neighbor 10.22.12.88 as 65000;
        source address 10.22.12.44;
        multihop;
        ipv4 {
                export filter {
                        if ( net ~ 10.233.0.0/16 ) then {
                                accept;
                        }
                        reject;
                };
                import all;
                next hop self on;
                gateway recursive;
                add paths on;
        };
}

But this doesn't work well, the peer BIRD show the info as follows:

bird> show route protocol bgp1
Table master4:
10.233.25.178/32     unicast [bgp1 18:47:32.910 from 10.22.12.44] * (100/?) [i]
	via 10.22.12.1 on enp2s0f0 onlink
bird> show route export kernel1
Table master4:
10.233.25.178/32     unicast [bgp1 18:47:32.910 from 10.22.12.44] * (100/?) [i]
	via 10.22.12.1 on enp2s0f0 onlink

The problem happened on the gateway, I was hoping the routes should have "via 10.22.12.44" instead of "via 10.22.12.1";

10.22.12.1 came from the host default gateway:
default via 10.22.12.1 dev bond0 onlink
But I have no idea how to prevent BIRD from using it.

The other information may be helpful is included below:

bird> show protocols all bgp1
Name       Proto      Table      State  Since         Info
bgp1       BGP        ---        up     18:28:46.888  Established   
  BGP state:          Established
    Neighbor address: 10.22.12.88
    Neighbor AS:      65000
    Neighbor ID:      10.22.12.88
    Local capabilities
      Multiprotocol
        AF announced: ipv4
      Route refresh
      Graceful restart
      4-octet AS numbers
      ADD-PATH
        RX: ipv4
        TX: ipv4
      Enhanced refresh
      Long-lived graceful restart
    Neighbor capabilities
      Multiprotocol
        AF announced: ipv4
      Route refresh
      Graceful restart
      4-octet AS numbers
      Enhanced refresh
      Long-lived graceful restart
    Session:          internal multihop AS4
    Source address:   10.22.12.44
    Hold timer:       117.086/240
    Keepalive timer:  52.520/80
  Channel ipv4
    State:          UP
    Table:          master4
    Preference:     100
    Input filter:   ACCEPT
    Output filter:  (unnamed)
    Routes:         2 imported, 20 exported
    Route change stats:     received   rejected   filtered    ignored   accepted
      Import updates:              2          0          0          0          2
      Import withdraws:            0          0        ---          0          0
      Export updates:             49          6         23        ---         20
      Export withdraws:            0        ---        ---        ---          4
    BGP Next hop:   10.22.12.44
    IGP IPv4 table: master4
bird> show route export bgp1 
Table master4:
10.233.25.178/32     unicast [kernel1 18:04:09.483] (10)
	dev yap35caa1e88854

Please enlighten me with some advice, and I'll appreciate that very much.

Thank you!