Hi Pim, Pieter, I reproduced the crash on current master, which already includes 46929: a single NS triggers a SIGSEGV in ip6_fib_table_lookup_exact_match.
With 46038 PS5 applied, the reproducer passes without workers and with two workers. The existing TestIPv6.test_ns regression also passes in both configurations, including the checks that an unknown target does not cause neighbor or /128 route learning. I added Code-Review +1: https://gerrit.fd.io/r/c/vpp/+/46038 The two fixes are complementary: 46929 defers neighbor learning; 46038 guards against an absent link-local FIB. Florin, could you review/merge this and help arrange backports to stable/2610 and stable/2606? Thanks, Jerome
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#27247): https://lists.fd.io/g/vpp-dev/message/27247 Mute This Topic: https://lists.fd.io/mt/121671152/21656 Group Owner: [email protected] Unsubscribe: https://lists.fd.io/g/vpp-dev/leave/14379924/21656/631435203/xyzzy [[email protected]] -=-=-=-=-=-=-=-=-=-=-=-
