Description of problem:
This bug has discovered with misconfig. 

  NovaPCIPassthrough:
  - devname: "ens1f0"
    physical_network: "sriov1"   <<<<<<
  - devname: "ens1f1"
    physical_network: null

- type: ovs_bridge
  name: br-offload   <<<<<<<<<<<<<<<<<<<Bridge name
  mtu: 9000
  use_dhcp: false
  members:
  - type: linux_bond
    name: bond-pf
    bonding_options: "mode=active-backup miimon=100"
    members:
    - type: sriov_pf
      name: ens1f0    <<<<<<<<<<<<<<<<<<<<<<<<<
      numvfs: 3
      primary: true
      promisc: true
      use_dhcp: false
      defroute: false
      link_mode: switchdev
    - type: sriov_pf
      name: ens1f1
      numvfs: 3
      promisc: true
      use_dhcp: false
      defroute: false
      link_mode: switchdev

  NeutronBridgeMappings: 'datacentre:br-ex,sriov1:br-tenant,dpdk2:br-link2'  <<<<<<
  NeutronNetworkVLANRanges: dpdk1:505:510,dpdk2:505:510,sriov1:505:510,sriov2:505:510
  NeutronFlatNetworks: 'datacentre,dpdk1,dpdk2,sriov1,sriov2'

 In NeutronBridgeMappings, sriov1 is mapped to br-tenant. And this bridge is not at all created on compute node. Rather sriov1(ens1f0) added to br-offload. 
However, packets finds out way to go out from VF(created on ens1f0) via geneve tunnel interface (ens1f1). 

This is ovs-hw-offload feature tested on ml2/ovn. 


Version-Release number of selected component (if applicable):
RHOSP: 16.1.4
RHEL: 8.2 
Kernel: 4.18.0-193.41.1.el8_2.x86_64
OVS: openvswitch2.13-2.13.0-71.el8fdp.x86_64
OVN: ovn2.13-20.09.0-17.el8fdp.x86_64

How reproducible:
Always

Steps to Reproduce:
1. Deploy ml2/ovn with ovs-hw-offload as shown above (Misconfig)
2. Create VM from sriov1 provider network
3. Send packets to outside and observe the dump-flow table

Actual results:
Packets are getting out via geneve tunnel where they are expected to drop. 

Expected results:
Drop the packets.

Additional info:
Such misconfig handled via geneve tunnel fall back mechanism may create issue in network. Packets are floating around unnecessary.