ipv6: Notify multipath route if should be offloaded

In a similar fashion to previous patches, only notify the new multipath
route if it is the first route in the node or if it was appended to such
route.

The type of the notification (replace vs. append) is determined based on
the number of routes added ('nhn') and the number of sibling routes. If
the two do not match, then an append notification should be sent.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index b59940416cb5776c7a77910fef029aef8e9491e7..c0809f52f9eff9486ff955cdf173fcf273fd2a91 100644 (file)
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -5017,6 +5017,32 @@ static void ip6_route_mpath_notify(struct fib6_info *rt,
                inet6_rt_notify(RTM_NEWROUTE, rt, info, nlflags);
 }
 
+static bool ip6_route_mpath_should_notify(const struct fib6_info *rt)
+{
+       bool rt_can_ecmp = rt6_qualify_for_ecmp(rt);
+       bool should_notify = false;
+       struct fib6_info *leaf;
+       struct fib6_node *fn;
+
+       rcu_read_lock();
+       fn = rcu_dereference(rt->fib6_node);
+       if (!fn)
+               goto out;
+
+       leaf = rcu_dereference(fn->leaf);
+       if (!leaf)
+               goto out;
+
+       if (rt == leaf ||
+           (rt_can_ecmp && rt->fib6_metric == leaf->fib6_metric &&
+            rt6_qualify_for_ecmp(leaf)))
+               should_notify = true;
+out:
+       rcu_read_unlock();
+
+       return should_notify;
+}
+
 static int ip6_route_multipath_add(struct fib6_config *cfg,
                                   struct netlink_ext_ack *extack)
 {
@@ -5147,6 +5173,28 @@ static int ip6_route_multipath_add(struct fib6_config *cfg,
                nhn++;
        }
 
+       /* An in-kernel notification should only be sent in case the new
+        * multipath route is added as the first route in the node, or if
+        * it was appended to it. We pass 'rt_notif' since it is the first
+        * sibling and might allow us to skip some checks in the replace case.
+        */
+       if (ip6_route_mpath_should_notify(rt_notif)) {
+               enum fib_event_type fib_event;
+
+               if (rt_notif->fib6_nsiblings != nhn - 1)
+                       fib_event = FIB_EVENT_ENTRY_APPEND;
+               else
+                       fib_event = FIB_EVENT_ENTRY_REPLACE_TMP;
+
+               err = call_fib6_multipath_entry_notifiers(info->nl_net,
+                                                         fib_event, rt_notif,
+                                                         nhn - 1, extack);
+               if (err) {
+                       /* Delete all the siblings that were just added */
+                       err_nh = NULL;
+                       goto add_errout;
+               }
+       }
        event_type = replace ? FIB_EVENT_ENTRY_REPLACE : FIB_EVENT_ENTRY_ADD;
        err = call_fib6_multipath_entry_notifiers(info->nl_net, event_type,
                                                  rt_notif, nhn - 1, extack);