On Mon, 23 Feb 2026 13:39:51 -0800, Martin KaFai Lau wrote: ... > Please check if the v5 discussion make sense (fix in test_run.c). It > should be something simpler than v5, e.g. always assign a dst to skb for > is_lwt case. I think always using net->ipv6.ip6_null_entry should work > but it needs to check ipv6 mod is loaded/enabled before using. Checking > 'ipv6_bpf_stub' should do. I don't think we need to complicate things > for v4 only kernel in test_run. I don't think testing skb->protocol > matters also from looking at bpf_lwt_push_ip_encap but please check.
Sure, I'll check, modify, and test it. Thanks.
