net/mlx5: Bridge, allow merged eswitch connectivity

Allow connectivity between representors of different eswitch instances that
are attached to same bridge when merged_eswitch capability is enabled. Add
ports of peer eswitch to bridge instance and mark them with
MLX5_ESW_BRIDGE_PORT_FLAG_PEER. Mark FDBs offloaded on peer ports with
MLX5_ESW_BRIDGE_FLAG_PEER flag. Such FDBs can only be aged out on their
local eswitch instance, which then sends SWITCHDEV_FDB_DEL_TO_BRIDGE event.
Listen to the event on mlx5 bridge implementation and delete peer FDBs in
event handler.

Signed-off-by: Vlad Buslov <[email protected]>
Reviewed-by: Roi Dayan <[email protected]>
Reviewed-by: Mark Bloch <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

Author: w1ldptr

drivers/net/ethernet/mellanox/mlx5/core/en/rep/bridge.c

@@ -15,6 +15,7 @@ struct mlx5_bridge_switchdev_fdb_work {
 	struct work_struct work;
 	struct switchdev_notifier_fdb_info fdb_info;
 	struct net_device *dev;
+	struct mlx5_esw_bridge_offloads *br_offloads;
 	bool add;
 };
 
@@ -25,13 +26,28 @@ static bool mlx5_esw_bridge_dev_same_esw(struct net_device *dev, struct mlx5_esw
 	return esw == priv->mdev->priv.eswitch;
 }
 
+static bool mlx5_esw_bridge_dev_same_hw(struct net_device *dev, struct mlx5_eswitch *esw)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev, *esw_mdev;
+	u64 system_guid, esw_system_guid;
+
+	mdev = priv->mdev;
+	esw_mdev = esw->dev;
+
+	system_guid = mlx5_query_nic_system_image_guid(mdev);
+	esw_system_guid = mlx5_query_nic_system_image_guid(esw_mdev);
+
+	return system_guid == esw_system_guid;
+}
+
 static int mlx5_esw_bridge_vport_num_vhca_id_get(struct net_device *dev, struct mlx5_eswitch *esw,
 						 u16 *vport_num, u16 *esw_owner_vhca_id)
 {
 	struct mlx5e_rep_priv *rpriv;
 	struct mlx5e_priv *priv;
 
-	if (!mlx5e_eswitch_rep(dev) || !mlx5_esw_bridge_dev_same_esw(dev, esw))
+	if (!mlx5e_eswitch_rep(dev) || !mlx5_esw_bridge_dev_same_hw(dev, esw))
 		return -ENODEV;
 
 	priv = netdev_priv(dev);
@@ -48,7 +64,7 @@ mlx5_esw_bridge_lower_rep_vport_num_vhca_id_get(struct net_device *dev, struct m
 	struct net_device *lower_dev;
 	struct list_head *iter;
 
-	if (mlx5e_eswitch_rep(dev) && mlx5_esw_bridge_dev_same_esw(dev, esw))
+	if (mlx5e_eswitch_rep(dev))
 		return mlx5_esw_bridge_vport_num_vhca_id_get(dev, esw, vport_num,
 							     esw_owner_vhca_id);
 
@@ -74,6 +90,7 @@ static int mlx5_esw_bridge_port_changeupper(struct notifier_block *nb, void *ptr
 								    netdev_nb);
 	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
 	struct netdev_notifier_changeupper_info *info = ptr;
+	struct mlx5_eswitch *esw = br_offloads->esw;
 	struct net_device *upper = info->upper_dev;
 	u16 vport_num, esw_owner_vhca_id;
 	struct netlink_ext_ack *extack;
@@ -90,11 +107,20 @@ static int mlx5_esw_bridge_port_changeupper(struct notifier_block *nb, void *ptr
 
 	extack = netdev_notifier_info_to_extack(&info->info);
 
-	return info->linking ?
-		mlx5_esw_bridge_vport_link(ifindex, vport_num, esw_owner_vhca_id, br_offloads,
-					   extack) :
-		mlx5_esw_bridge_vport_unlink(ifindex, vport_num, esw_owner_vhca_id, br_offloads,
-					     extack);
+	if (mlx5_esw_bridge_dev_same_esw(dev, esw))
+		err = info->linking ?
+			mlx5_esw_bridge_vport_link(ifindex, vport_num, esw_owner_vhca_id,
+						   br_offloads, extack) :
+			mlx5_esw_bridge_vport_unlink(ifindex, vport_num, esw_owner_vhca_id,
+						     br_offloads, extack);
+	else if (mlx5_esw_bridge_dev_same_hw(dev, esw))
+		err = info->linking ?
+			mlx5_esw_bridge_vport_peer_link(ifindex, vport_num, esw_owner_vhca_id,
+							br_offloads, extack) :
+			mlx5_esw_bridge_vport_peer_unlink(ifindex, vport_num, esw_owner_vhca_id,
+							  br_offloads, extack);
+
+	return err;
 }
 
 static int mlx5_esw_bridge_switchdev_port_event(struct notifier_block *nb,
@@ -253,16 +279,14 @@ static void mlx5_esw_bridge_switchdev_fdb_event_work(struct work_struct *work)
 		container_of(work, struct mlx5_bridge_switchdev_fdb_work, work);
 	struct switchdev_notifier_fdb_info *fdb_info =
 		&fdb_work->fdb_info;
-	struct mlx5_esw_bridge_offloads *br_offloads;
+	struct mlx5_esw_bridge_offloads *br_offloads =
+		fdb_work->br_offloads;
 	struct net_device *dev = fdb_work->dev;
 	u16 vport_num, esw_owner_vhca_id;
-	struct mlx5e_priv *priv;
 	int err;
 
 	rtnl_lock();
 
-	priv = netdev_priv(dev);
-	br_offloads = priv->mdev->priv.eswitch->br_offloads;
 	err = mlx5_esw_bridge_vport_num_vhca_id_get(dev, br_offloads->esw, &vport_num,
 						    &esw_owner_vhca_id);
 	if (err)
@@ -282,7 +306,8 @@ static void mlx5_esw_bridge_switchdev_fdb_event_work(struct work_struct *work)
 
 static struct mlx5_bridge_switchdev_fdb_work *
 mlx5_esw_bridge_init_switchdev_fdb_work(struct net_device *dev, bool add,
-					struct switchdev_notifier_fdb_info *fdb_info)
+					struct switchdev_notifier_fdb_info *fdb_info,
+					struct mlx5_esw_bridge_offloads *br_offloads)
 {
 	struct mlx5_bridge_switchdev_fdb_work *work;
 	u8 *addr;
@@ -304,6 +329,7 @@ mlx5_esw_bridge_init_switchdev_fdb_work(struct net_device *dev, bool add,
 
 	dev_hold(dev);
 	work->dev = dev;
+	work->br_offloads = br_offloads;
 	work->add = add;
 	return work;
 }
@@ -334,10 +360,13 @@ static int mlx5_esw_bridge_switchdev_event(struct notifier_block *nb,
 
 	if (!mlx5e_eswitch_rep(dev))
 		return NOTIFY_DONE;
-	if (!mlx5_esw_bridge_dev_same_esw(dev, br_offloads->esw))
-		return NOTIFY_DONE;
 
 	switch (event) {
+	case SWITCHDEV_FDB_DEL_TO_BRIDGE:
+		/* only handle the event when source is on another eswitch */
+		if (mlx5_esw_bridge_dev_same_esw(dev, br_offloads->esw))
+			break;
+		fallthrough;
 	case SWITCHDEV_FDB_ADD_TO_DEVICE:
 	case SWITCHDEV_FDB_DEL_TO_DEVICE:
 		fdb_info = container_of(info,
@@ -346,7 +375,8 @@ static int mlx5_esw_bridge_switchdev_event(struct notifier_block *nb,
 
 		work = mlx5_esw_bridge_init_switchdev_fdb_work(dev,
 							       event == SWITCHDEV_FDB_ADD_TO_DEVICE,
-							       fdb_info);
+							       fdb_info,
+							       br_offloads);
 		if (IS_ERR(work)) {
 			WARN_ONCE(1, "Failed to init switchdev work, err=%ld",
 				  PTR_ERR(work));

drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.c

@@ -79,7 +79,7 @@ mlx5_esw_bridge_fdb_offload_notify(struct net_device *dev, const unsigned char *
 static void
 mlx5_esw_bridge_fdb_del_notify(struct mlx5_esw_bridge_fdb_entry *entry)
 {
-	if (!(entry->flags & MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER))
+	if (!(entry->flags & (MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER | MLX5_ESW_BRIDGE_FLAG_PEER)))
 		mlx5_esw_bridge_fdb_offload_notify(entry->dev, entry->key.addr,
 						   entry->key.vid,
 						   SWITCHDEV_FDB_DEL_TO_BRIDGE);
@@ -513,7 +513,7 @@ mlx5_esw_bridge_ingress_filter_flow_create(u16 vport_num, const unsigned char *a
 }
 
 static struct mlx5_flow_handle *
-mlx5_esw_bridge_egress_flow_create(u16 vport_num, const unsigned char *addr,
+mlx5_esw_bridge_egress_flow_create(u16 vport_num, u16 esw_owner_vhca_id, const unsigned char *addr,
 				   struct mlx5_esw_bridge_vlan *vlan,
 				   struct mlx5_esw_bridge *bridge)
 {
@@ -558,6 +558,10 @@ mlx5_esw_bridge_egress_flow_create(u16 vport_num, const unsigned char *addr,
 			 vlan->vid);
 	}
 
+	if (MLX5_CAP_ESW(bridge->br_offloads->esw->dev, merged_eswitch)) {
+		dest.vport.flags = MLX5_FLOW_DEST_VPORT_VHCA_ID;
+		dest.vport.vhca_id = esw_owner_vhca_id;
+	}
 	handle = mlx5_add_flow_rules(bridge->egress_ft, rule_spec, &flow_act, &dest, 1);
 
 	kvfree(rule_spec);
@@ -917,7 +921,7 @@ mlx5_esw_bridge_port_vlan_lookup(u16 vid, u16 vport_num, u16 esw_owner_vhca_id,
 
 static struct mlx5_esw_bridge_fdb_entry *
 mlx5_esw_bridge_fdb_entry_init(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
-			       const unsigned char *addr, u16 vid, bool added_by_user,
+			       const unsigned char *addr, u16 vid, bool added_by_user, bool peer,
 			       struct mlx5_eswitch *esw, struct mlx5_esw_bridge *bridge)
 {
 	struct mlx5_esw_bridge_vlan *vlan = NULL;
@@ -945,6 +949,8 @@ mlx5_esw_bridge_fdb_entry_init(struct net_device *dev, u16 vport_num, u16 esw_ow
 	entry->lastuse = jiffies;
 	if (added_by_user)
 		entry->flags |= MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER;
+	if (peer)
+		entry->flags |= MLX5_ESW_BRIDGE_FLAG_PEER;
 
 	counter = mlx5_fc_create(esw->dev, true);
 	if (IS_ERR(counter)) {
@@ -974,7 +980,8 @@ mlx5_esw_bridge_fdb_entry_init(struct net_device *dev, u16 vport_num, u16 esw_ow
 		entry->filter_handle = handle;
 	}
 
-	handle = mlx5_esw_bridge_egress_flow_create(vport_num, addr, vlan, bridge);
+	handle = mlx5_esw_bridge_egress_flow_create(vport_num, esw_owner_vhca_id, addr, vlan,
+						    bridge);
 	if (IS_ERR(handle)) {
 		err = PTR_ERR(handle);
 		esw_warn(esw->dev, "Failed to create egress flow(vport=%u,err=%d)\n",
@@ -1050,7 +1057,7 @@ int mlx5_esw_bridge_vlan_filtering_set(u16 vport_num, u16 esw_owner_vhca_id, boo
 	return 0;
 }
 
-static int mlx5_esw_bridge_vport_init(u16 vport_num, u16 esw_owner_vhca_id,
+static int mlx5_esw_bridge_vport_init(u16 vport_num, u16 esw_owner_vhca_id, u16 flags,
 				      struct mlx5_esw_bridge_offloads *br_offloads,
 				      struct mlx5_esw_bridge *bridge)
 {
@@ -1065,6 +1072,7 @@ static int mlx5_esw_bridge_vport_init(u16 vport_num, u16 esw_owner_vhca_id,
 	port->vport_num = vport_num;
 	port->esw_owner_vhca_id = esw_owner_vhca_id;
 	port->bridge = bridge;
+	port->flags |= flags;
 	xa_init(&port->vlans);
 	err = mlx5_esw_bridge_port_insert(port, br_offloads);
 	if (err) {
@@ -1101,9 +1109,10 @@ static int mlx5_esw_bridge_vport_cleanup(struct mlx5_esw_bridge_offloads *br_off
 	return 0;
 }
 
-int mlx5_esw_bridge_vport_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
-			       struct mlx5_esw_bridge_offloads *br_offloads,
-			       struct netlink_ext_ack *extack)
+static int mlx5_esw_bridge_vport_link_with_flags(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+						 u16 flags,
+						 struct mlx5_esw_bridge_offloads *br_offloads,
+						 struct netlink_ext_ack *extack)
 {
 	struct mlx5_esw_bridge *bridge;
 	int err;
@@ -1114,7 +1123,7 @@ int mlx5_esw_bridge_vport_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id
 		return PTR_ERR(bridge);
 	}
 
-	err = mlx5_esw_bridge_vport_init(vport_num, esw_owner_vhca_id, br_offloads, bridge);
+	err = mlx5_esw_bridge_vport_init(vport_num, esw_owner_vhca_id, flags, br_offloads, bridge);
 	if (err) {
 		NL_SET_ERR_MSG_MOD(extack, "Error initializing port");
 		goto err_vport;
@@ -1126,6 +1135,14 @@ int mlx5_esw_bridge_vport_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id
 	return err;
 }
 
+int mlx5_esw_bridge_vport_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+			       struct mlx5_esw_bridge_offloads *br_offloads,
+			       struct netlink_ext_ack *extack)
+{
+	return mlx5_esw_bridge_vport_link_with_flags(ifindex, vport_num, esw_owner_vhca_id, 0,
+						     br_offloads, extack);
+}
+
 int mlx5_esw_bridge_vport_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
 				 struct mlx5_esw_bridge_offloads *br_offloads,
 				 struct netlink_ext_ack *extack)
@@ -1149,6 +1166,26 @@ int mlx5_esw_bridge_vport_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_
 	return err;
 }
 
+int mlx5_esw_bridge_vport_peer_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				    struct mlx5_esw_bridge_offloads *br_offloads,
+				    struct netlink_ext_ack *extack)
+{
+	if (!MLX5_CAP_ESW(br_offloads->esw->dev, merged_eswitch))
+		return 0;
+
+	return mlx5_esw_bridge_vport_link_with_flags(ifindex, vport_num, esw_owner_vhca_id,
+						     MLX5_ESW_BRIDGE_PORT_FLAG_PEER,
+						     br_offloads, extack);
+}
+
+int mlx5_esw_bridge_vport_peer_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				      struct mlx5_esw_bridge_offloads *br_offloads,
+				      struct netlink_ext_ack *extack)
+{
+	return mlx5_esw_bridge_vport_unlink(ifindex, vport_num, esw_owner_vhca_id, br_offloads,
+					    extack);
+}
+
 int mlx5_esw_bridge_port_vlan_add(u16 vport_num, u16 esw_owner_vhca_id, u16 vid, u16 flags,
 				  struct mlx5_esw_bridge_offloads *br_offloads,
 				  struct netlink_ext_ack *extack)
@@ -1206,14 +1243,15 @@ void mlx5_esw_bridge_fdb_create(struct net_device *dev, u16 vport_num, u16 esw_o
 	bridge = port->bridge;
 	entry = mlx5_esw_bridge_fdb_entry_init(dev, vport_num, esw_owner_vhca_id, fdb_info->addr,
 					       fdb_info->vid, fdb_info->added_by_user,
+					       port->flags & MLX5_ESW_BRIDGE_PORT_FLAG_PEER,
 					       br_offloads->esw, bridge);
 	if (IS_ERR(entry))
 		return;
 
 	if (entry->flags & MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER)
 		mlx5_esw_bridge_fdb_offload_notify(dev, entry->key.addr, entry->key.vid,
 						   SWITCHDEV_FDB_OFFLOADED);
-	else
+	else if (!(entry->flags & MLX5_ESW_BRIDGE_FLAG_PEER))
 		/* Take over dynamic entries to prevent kernel bridge from aging them out. */
 		mlx5_esw_bridge_fdb_offload_notify(dev, entry->key.addr, entry->key.vid,
 						   SWITCHDEV_FDB_ADD_TO_BRIDGE);
@@ -1263,7 +1301,8 @@ void mlx5_esw_bridge_update(struct mlx5_esw_bridge_offloads *br_offloads)
 
 			if (time_after(lastuse, entry->lastuse)) {
 				mlx5_esw_bridge_fdb_entry_refresh(lastuse, entry);
-			} else if (time_is_before_jiffies(entry->lastuse + bridge->ageing_time)) {
+			} else if (!(entry->flags & MLX5_ESW_BRIDGE_FLAG_PEER) &&
+				   time_is_before_jiffies(entry->lastuse + bridge->ageing_time)) {
 				mlx5_esw_bridge_fdb_del_notify(entry);
 				mlx5_esw_bridge_fdb_entry_cleanup(entry, bridge);
 			}

drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.h

@@ -40,6 +40,12 @@ int mlx5_esw_bridge_vport_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id
 int mlx5_esw_bridge_vport_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
 				 struct mlx5_esw_bridge_offloads *br_offloads,
 				 struct netlink_ext_ack *extack);
+int mlx5_esw_bridge_vport_peer_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				    struct mlx5_esw_bridge_offloads *br_offloads,
+				    struct netlink_ext_ack *extack);
+int mlx5_esw_bridge_vport_peer_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				      struct mlx5_esw_bridge_offloads *br_offloads,
+				      struct netlink_ext_ack *extack);
 void mlx5_esw_bridge_fdb_create(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
 				struct mlx5_esw_bridge_offloads *br_offloads,
 				struct switchdev_notifier_fdb_info *fdb_info);

drivers/net/ethernet/mellanox/mlx5/core/esw/bridge_priv.h

@@ -19,6 +19,11 @@ struct mlx5_esw_bridge_fdb_key {
 
 enum {
 	MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER = BIT(0),
+	MLX5_ESW_BRIDGE_FLAG_PEER = BIT(1),
+};
+
+enum {
+	MLX5_ESW_BRIDGE_PORT_FLAG_PEER = BIT(0),
 };
 
 struct mlx5_esw_bridge_fdb_entry {
@@ -49,6 +54,7 @@ struct mlx5_esw_bridge_vlan {
 struct mlx5_esw_bridge_port {
 	u16 vport_num;
 	u16 esw_owner_vhca_id;
+	u16 flags;
 	struct mlx5_esw_bridge *bridge;
 	struct xarray vlans;
 };

drivers/net/ethernet/mellanox/mlx5/core/esw/diag/bridge_tracepoint.h

@@ -86,14 +86,17 @@ DECLARE_EVENT_CLASS(mlx5_esw_bridge_port_template,
 		    TP_STRUCT__entry(
 			    __field(u16, vport_num)
 			    __field(u16, esw_owner_vhca_id)
+			    __field(u16, flags)
 			    ),
 		    TP_fast_assign(
 			    __entry->vport_num = port->vport_num;
 			    __entry->esw_owner_vhca_id = port->esw_owner_vhca_id;
+			    __entry->flags = port->flags;
 			    ),
-		    TP_printk("vport_num=%hu esw_owner_vhca_id=%hu",
+		    TP_printk("vport_num=%hu esw_owner_vhca_id=%hu flags=%hx",
 			      __entry->vport_num,
-			      __entry->esw_owner_vhca_id)
+			      __entry->esw_owner_vhca_id,
+			      __entry->flags)
 	);
 
 DEFINE_EVENT(mlx5_esw_bridge_port_template,