bgpd: Fix BGP to update correct v6 next-hop globally

[routingrocks]

Issue: In a dual-stack IPv4/IPv6 BGP session, after the interface is deactivated (ifdown) and then reloaded (ifreload), the global next-hop within BGP is incorrectly updated. Currently, only unnumbered neighbors are handled. We also need to handle an IPv4 neighbor when IPv6 AFI-SAFI is enabled.

Fix: When the primary global v6 unicast address is received from Zebra, BGP updates the peer's global next-hop with the correct address and send update to its negibors.

Testing: UT, and TestEbgpMultihop tests

UT logs: with fix: root@:mgmt:/tmp# ifdown swp1s0; ifreload -a root@:mgmt:/tmp# vtysh -c "show ip bgp vrf all neighbors" | grep Nexthop Nexthop: 20.1.2.101 Nexthop global: 2001:10:1:2::101 Nexthop local: fe80::7efe:90ff:fefa:e158

without fix after ifreload: Nexthop: 20.1.2.101 Nexthop global: :: Nexthop local: fe80::7efe:90ff:fefa:e158

Ticket: # Signed-off-by: Rajesh Varatharaj [email protected] and Pooja Doijode [email protected]

[ton31337]

@Mergifyio backport stable/10.0

[mergify[bot]]

backport stable/10.0

🟠 Waiting for conditions to match

• [ ] merged [📌 backport requirement]

[ton31337]

@routingrocks could you rebase?

[github-actions[bot]]

This pull request has conflicts, please resolve those before we can evaluate the pull request.

[louis-6wind]

Normally, the issue is fixed by https://github.com/FRRouting/frr/pull/15614, which is now merged Could you test your scenario with the latest master branch?

[riw777]

@frrbot autoclose in 1 month