netdev - [PATCH net-next 4/9] ipv6: Notify multipath route if should be offloaded

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20191223132820.888247-5-idosch@idosch.org>
Date:   Mon, 23 Dec 2019 15:28:15 +0200
From:   Ido Schimmel <idosch@...sch.org>
To:     netdev@...r.kernel.org
Cc:     davem@...emloft.net, dsahern@...il.com, roopa@...ulusnetworks.com,
        jakub.kicinski@...ronome.com, jiri@...lanox.com,
        Ido Schimmel <idosch@...lanox.com>
Subject: [PATCH net-next 4/9] ipv6: Notify multipath route if should be offloaded

From: Ido Schimmel <idosch@...lanox.com>

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@...lanox.com>
Reviewed-by: Jiri Pirko <jiri@...lanox.com>
---
 net/ipv6/route.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 48 insertions(+)

diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index b59940416cb5..c0809f52f9ef 100644
--- 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);
-- 
2.24.1