mlxsw: spectrum_router: After underlay moves, demote conflicting tunnels

When a GRE tunnel is bound to an underlay netdevice and that netdevice is
moved to a different VRF, that could cause two tunnels to have the same
underlay local address in the same VRF. Linux in this situation dispatches
the traffic according to the tunnel key (or lack thereof), but that cannot
be offloaded to Spectrum devices.

Detect this situation and unoffload the two impacted tunnels when it
happens.

Signed-off-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

1 files changed, 38 insertions, 1 deletions

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 86e25824fcd8..4c4d99ab15a0 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -1614,8 +1614,25 @@ static int
 mlxsw_sp_netdevice_ipip_ul_vrf_event(struct mlxsw_sp *mlxsw_sp,
 				     struct mlxsw_sp_ipip_entry *ipip_entry,
 				     struct net_device *ul_dev,
+				     bool *demote_this,
 				     struct netlink_ext_ack *extack)
 {
+	u32 ul_tb_id = l3mdev_fib_table(ul_dev) ? : RT_TABLE_MAIN;
+	enum mlxsw_sp_l3proto ul_proto;
+	union mlxsw_sp_l3addr saddr;
+
+	/* Moving underlay to a different VRF might cause local address
+	 * conflict, and the conflicting tunnels need to be demoted.
+	 */
+	ul_proto = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt]->ul_proto;
+	saddr = mlxsw_sp_ipip_netdev_saddr(ul_proto, ipip_entry->ol_dev);
+	if (mlxsw_sp_ipip_demote_tunnel_by_saddr(mlxsw_sp, ul_proto,
+						 saddr, ul_tb_id,
+						 ipip_entry)) {
+		*demote_this = true;
+		return 0;
+	}
+
 	return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
 						   true, true, false, extack);
 }
@@ -1766,6 +1783,7 @@ static int
 __mlxsw_sp_netdevice_ipip_ul_event(struct mlxsw_sp *mlxsw_sp,
 				   struct mlxsw_sp_ipip_entry *ipip_entry,
 				   struct net_device *ul_dev,
+				   bool *demote_this,
 				   unsigned long event,
 				   struct netdev_notifier_info *info)
 {
@@ -1780,6 +1798,7 @@ __mlxsw_sp_netdevice_ipip_ul_event(struct mlxsw_sp *mlxsw_sp,
 			return mlxsw_sp_netdevice_ipip_ul_vrf_event(mlxsw_sp,
 								    ipip_entry,
 								    ul_dev,
+								    demote_this,
 								    extack);
 		break;
 
@@ -1806,13 +1825,31 @@ mlxsw_sp_netdevice_ipip_ul_event(struct mlxsw_sp *mlxsw_sp,
 	while ((ipip_entry = mlxsw_sp_ipip_entry_find_by_ul_dev(mlxsw_sp,
 								ul_dev,
 								ipip_entry))) {
+		struct mlxsw_sp_ipip_entry *prev;
+		bool demote_this = false;
+
 		err = __mlxsw_sp_netdevice_ipip_ul_event(mlxsw_sp, ipip_entry,
-							 ul_dev, event, info);
+							 ul_dev, &demote_this,
+							 event, info);
 		if (err) {
 			mlxsw_sp_ipip_demote_tunnel_by_ul_netdev(mlxsw_sp,
 								 ul_dev);
 			return err;
 		}
+
+		if (demote_this) {
+			if (list_is_first(&ipip_entry->ipip_list_node,
+					  &mlxsw_sp->router->ipip_list))
+				prev = NULL;
+			else
+				/* This can't be cached from previous iteration,
+				 * because that entry could be gone now.
+				 */
+				prev = list_prev_entry(ipip_entry,
+						       ipip_list_node);
+			mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
+			ipip_entry = prev;
+		}
 	}
 
 	return 0;