Fix SmartSwitch ARP test BGP filter for VLAN-based DPC dataplane

[nnelluri-cisco]

After VLAN-based DPC config moved dataplane IPs to Vlan55, the ARP neighbor MAC test failed to exclude connected routes because it queried per-port addresses. Read VLAN interface IPs from config_db instead.

Description of PR

Summary:
Fixes # (issue)

Type of change

• [X ] Bug fix
• Testbed and Framework(new/improvement)
• New Test case
  • Skipped for non-supported platforms
• Test case improvement

Back port request

• 202205
• 202305
• 202311
• 202405
• 202411
• [X ] 202505
• [ X] 202511

Approach

What is the motivation for this PR?

PR sonic-net#22902 moved SmartSwitch DPC dataplane connectivity to a VLAN-based model: DPC ports (Ethernet224/232/240/248) are untagged members of Vlan55, and the gateway IP (20.0.200.254/24) is assigned to Vlan55 instead of individual DPC ports.

test_neighbor_mac_noptf.py shuts down BGP in setup and waits until no BGP routes remain in ASIC_DB. For SmartSwitch, it excludes local connected routes by filtering IPs returned from _get_back_plane_port_ips(). That function still used ip addr show on each DPC port, which returns empty after the VLAN change. Without those filter IPs, connected routes (20.0.200.0/24 and 20.0.200.254/32) are not excluded and the test times out waiting for BGP shutdown.

How did you do it?

Updated _get_back_plane_port_ips() to:

Read DPC ports from config_facts (PORT table, role == "dpc").
Find VLANs those ports belong to via VLAN_MEMBER.
Extract IPv4 addresses from VLAN_INTERFACE for those VLANs (e.g. 20.0.200.254 on Vlan55).
Keep a fallback to the legacy ip addr show per-port lookup for older configs where DPC ports still have addresses directly assigned.

How did you verify/test it?

Code review against the VLAN dataplane config in ansible/module_utils/smartswitch_utils.py and generate_golden_config_db.py to confirm Vlan55 IP placement matches what the test now reads from config_db.
Verified the filter logic in _get_bgp_routes_asic() uses the returned IPs as route-prefix filters, which correctly excludes both the /24 subnet and /32 host routes when 20.0.200.254 is present.
Pre-commit hooks (flake8, etc.) pass on the changed file.

Any platform specific information?

Affects SmartSwitch platforms only (is_smartswitch == True). The change is scoped to _get_back_plane_port_ips(), which is only called from _check_no_bgp_routes() when the DUT is a SmartSwitch. Non-SmartSwitch platforms are unchanged.

Supported testbed topology if it's a new test case?

Relevant SmartSwitch SKUs: Cisco-8102-* and Mellanox-SN4280-* with DPC role ports.
N/A — this is a bug fix to an existing test case (arp/test_neighbor_mac_noptf.py), not a new test. The test already runs on any topology and applies the SmartSwitch-specific logic only when is_smartswitch is set.

Documentation

Pass Logs:
grep 'testNeighborMacNoPtf' test_neighbor_mac_noptf.log | grep 'SectionStartLogger'
INFO SectionStartLogger:init.py:170 ==================== tests/arp/test_neighbor_mac_noptf.py::TestNeighborMacNoPtf::testNeighborMacNoPtf[4-MtFuji-dut-None] setup ====================
INFO SectionStartLogger:init.py:170 ==================== tests/arp/test_neighbor_mac_noptf.py::TestNeighborMacNoPtf::testNeighborMacNoPtf[4-MtFuji-dut-None] call ====================
INFO SectionStartLogger:init.py:170 ==================== tests/arp/test_neighbor_mac_noptf.py::TestNeighborMacNoPtf::testNeighborMacNoPtf[4-MtFuji-dut-None] teardown ====================
INFO SectionStartLogger:init.py:170 ==================== tests/arp/test_neighbor_mac_noptf.py::TestNeighborMacNoPtf::testNeighborMacNoPtf[6-MtFuji-dut-None] setup ====================
INFO SectionStartLogger:init.py:170 ==================== tests/arp/test_neighbor_mac_noptf.py::TestNeighborMacNoPtf::testNeighborMacNoPtf[6-MtFuji-dut-None] call ====================
INFO SectionStartLogger:init.py:170 ==================== tests/arp/test_neighbor_mac_noptf.py::TestNeighborMacNoPtf::testNeighborMacNoPtf[6-MtFuji-dut-None] teardown ====================

[nnelluri-cisco]

@rameshraghupathi and @prabhataravind

Please review and approve the PR

[rameshraghupathy]

@nnelluri-cisco Do we need to append both iface.ip and iface.network.network_address; otherwise the VLAN gateway /32 connected route may remain unfiltered ?

[nnelluri-cisco]

@rameshraghupathy
Filtered both routes.

grep -E 'custom filter|back plane port IPs' test_neighbor_mac_noptf.log
INFO tests.arp.test_neighbor_mac_noptf:test_neighbor_mac_noptf.py:89 back plane port IPs: ['20.0.200.0', '20.0.200.254']
INFO tests.arp.test_neighbor_mac_noptf:test_neighbor_mac_noptf.py:112 custom filter: {'20.0.200.0', '20.0.200.254'}
INFO tests.arp.test_neighbor_mac_noptf:test_neighbor_mac_noptf.py:89 back plane port IPs: ['20.0.200.0', '20.0.200.254']
INFO tests.arp.test_neighbor_mac_noptf:test_neighbor_mac_noptf.py:112 custom filter: {'20.0.200.0', '20.0.200.254'}

[rameshraghupathy]

@nnelluri-cisco This still does not fully address the /32 gateway/local route.

For example, if the VLAN interface is 20.0.200.254/24, we may need to filter both:

20.0.200.0      # subnet route
20.0.200.254    # local /32 route

Please append both:

back_plane_port_ips.append(str(iface.ip))
back_plane_port_ips.append(str(iface.network.network_address))

Otherwise, 20.0.200.254/32 may remain unfiltered.

Also, please verify the actual VLAN_MEMBER config_facts shape on the DUT.

[nnelluri-cisco]

@rameshraghupathy

The latest commit already filters both the subnet prefix (20.0.200.0) and the local /32 host route (20.0.200.254) by appending both iface.ip and iface.network.network_address to the BGP route filter list.

Verified on DUT sonic-ucs-m6-19 (MtFuji / Cisco-8102): logs show back plane port IPs: ['20.0.200.0', '20.0.200.254'], filtered 2, and BGP shutdown completes within 120s.

                if iface.version == 4:
                    # Redis KEYS glob prefix-matches dest; filter both subnet and /32 host.
                    # e.g. Vlan55 20.0.200.254/24 -> 20.0.200.0 and 20.0.200.254
                    back_plane_port_ips.append(str(iface.ip))
                    back_plane_port_ips.append(str(iface.network.network_address))

test cases already passed in latest run

[prabhataravind]

@nnelluri-cisco could you raise the PR against sonic-net/sonic-mgmt master ?

[nnelluri-cisco]

@prabhataravind
created new PR against to sonic-net/sonic-mgmt master
sonic-net#25676