How force a route with ip6_output

[Kajetan Staszkiewicz]

Hello, group






dst = (struct sockaddr_in *)&ro->ro_dst;
…
gw = (const struct sockaddr *)dst;

and use those without calling fib4_lookup()


But I don't see the same logic in ip6_output. There's this check:

if (ro->ro_nh != NULL && fwd_tag == NULL &&
    ro->ro_dst.sin6_family == AF_INET6 &&
    IN6_ARE_ADDR_EQUAL(&ro->ro_dst.sin6_addr, &ip6->ip6_dst)) {
        /* Nexthop is valid and contains valid ifp */
        nh = ro->ro_nh;
}





Am I holding it wrong?

--
| pozdrawiam / greetings | Powered by macOS, Debian and FreeBSD |
|  Kajetan Staszkiewicz  |  www: http://vegeta.tuxpowered.net   |
`------------------------^--------------------------------------'

I'm experimenting with another approach to pf route-to issues mentioned in D41517 and I can't get ip6_output do what I want. How does one force ip6_output to use the provided next hop? ip_output does not seem to have problems with that. As long as ro and ro->nh are not NULL it will just do Which fails beause the IP address in ro_dst is different than in ip6_dst - I am routing traffic to some host over a gateway different than that host. The next block of code deals with fwd_tag, which I have not used, so it restores the destination for lookup to back to ip6->ip6_dst and that prevents ip6_output from using the supplied route.

OpenPGP_0x80AAD77FA6F30053.asc
Description: OpenPGP public key

OpenPGP_signature
Description: OpenPGP digital signature