From: David Ahern <d...@cumulusnetworks.com>
Date: Thu, 7 Apr 2016 11:11:14 -0700
> Similar to 3bfd847203c6 ("net: Use passed in table for nexthop lookups")
> for IPv4, if the route spec contains a table id use that to lookup the
> next hop first and fall back to a full lookup if it fails (per the fix
> 4c9bcd117918b ("net: Fix nexthop lookups")).
> @@ -1940,7 +1940,38 @@ static struct rt6_info *ip6_route_info_create(struct
> fib6_config *cfg)
> if (!(gwa_type & IPV6_ADDR_UNICAST))
> goto out;
>
> - grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex,
> 1);
> + if (cfg->fc_table) {
> + struct flowi6 fl6 = {
> + .flowi6_oif = cfg->fc_ifindex,
> + .daddr = *gw_addr,
> + .saddr = cfg->fc_prefsrc,
> + };
> + struct fib6_table *table;
> + int flags = 0;
> +
> + err = -EHOSTUNREACH;
> + table = fib6_get_table(net, cfg->fc_table);
> + if (!table)
> + goto out;
> +
> + if (!ipv6_addr_any(&cfg->fc_prefsrc))
> + flags |= RT6_LOOKUP_F_HAS_SADDR;
> +
> + grt = ip6_pol_route(net, table, cfg->fc_ifindex,
> + &fl6, flags);
> +
> + /* if table lookup failed, fall back
> + * to full lookup
> + */
This is semantically different from the referenced ipv4 change.
Lack of a matching table for cfg->fc_table does not result in a
failure on the ipv4 side. It falls back in that case.
But here in this ipv6 patch, you do fail if fib6_get_table() gives
NULL.
One thing that drives me absolutely crazy is all of the subtle
semantic differences between our ipv4 and ipv6 stack, so I refuse to
knowingly add new such cases.
Therefore, please make the ipv6 behavior match exactly what ipv4
does.
Thanks.
