net/mlx5: Lag, Add single RDMA device in multiport mode

In MultiPort E-Switch mode a single RDMA is created. This device has multiple
RDMA ports that represent the uplink ports that are connected to the E-Switch.
Account for this when creating the RDMA device so it has an additional port for
the non native uplink.

As a side effect of this patch, use shared fdb in multiport eswitch mode.
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>

drivers/infiniband/hw/mlx5/ib_rep.c

@@ -37,6 +37,7 @@ mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 	const struct mlx5_ib_profile *profile;
 	struct mlx5_core_dev *peer_dev;
 	struct mlx5_ib_dev *ibdev;
+	int second_uplink = false;
 	u32 peer_num_ports;
 	int vport_index;
 	int ret;
@@ -47,17 +48,24 @@ mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 		peer_dev = mlx5_lag_get_peer_mdev(dev);
 		peer_num_ports = mlx5_eswitch_get_total_vports(peer_dev);
 		if (mlx5_lag_is_master(dev)) {
-			/* Only 1 ib port is the representor for both uplinks */
-			num_ports += peer_num_ports - 1;
+			if (mlx5_lag_is_mpesw(dev))
+				num_ports += peer_num_ports;
+			else
+				num_ports += peer_num_ports - 1;
+
 		} else {
-			if (rep->vport == MLX5_VPORT_UPLINK)
-				return 0;
+			if (rep->vport == MLX5_VPORT_UPLINK) {
+				if (!mlx5_lag_is_mpesw(dev))
+					return 0;
+				second_uplink = true;
+			}
+
 			vport_index += peer_num_ports;
 			dev = peer_dev;
 		}
 	}
 
-	if (rep->vport == MLX5_VPORT_UPLINK)
+	if (rep->vport == MLX5_VPORT_UPLINK && !second_uplink)
 		profile = &raw_eth_profile;
 	else
 		return mlx5_ib_set_vport_rep(dev, rep, vport_index);

drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c

@@ -644,7 +644,7 @@ int mlx5_activate_lag(struct mlx5_lag *ldev,
 	return 0;
 }
 
-static int mlx5_deactivate_lag(struct mlx5_lag *ldev)
+int mlx5_deactivate_lag(struct mlx5_lag *ldev)
 {
 	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
 	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
@@ -721,7 +721,7 @@ bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
 	return true;
 }
 
-static void mlx5_lag_add_devices(struct mlx5_lag *ldev)
+void mlx5_lag_add_devices(struct mlx5_lag *ldev)
 {
 	int i;
 
@@ -738,7 +738,7 @@ static void mlx5_lag_add_devices(struct mlx5_lag *ldev)
 	}
 }
 
-static void mlx5_lag_remove_devices(struct mlx5_lag *ldev)
+void mlx5_lag_remove_devices(struct mlx5_lag *ldev)
 {
 	int i;
 

drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h

@@ -120,5 +120,8 @@ void mlx5_infer_tx_enabled(struct lag_tracker *tracker, u8 num_ports,
 void mlx5_ldev_add_debugfs(struct mlx5_core_dev *dev);
 void mlx5_ldev_remove_debugfs(struct dentry *dbg);
 void mlx5_disable_lag(struct mlx5_lag *ldev);
+void mlx5_lag_remove_devices(struct mlx5_lag *ldev);
+int mlx5_deactivate_lag(struct mlx5_lag *ldev);
+void mlx5_lag_add_devices(struct mlx5_lag *ldev);
 
 #endif /* __MLX5_LAG_H__ */

drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c

@@ -67,15 +67,16 @@ static int mlx5_mpesw_metadata_set(struct mlx5_lag *ldev)
 
 static int enable_mpesw(struct mlx5_lag *ldev)
 {
-	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
 	int err;
 
 	if (ldev->mode != MLX5_LAG_MODE_NONE)
 		return -EINVAL;
 
-	if (mlx5_eswitch_mode(dev) != MLX5_ESWITCH_OFFLOADS ||
-	    !MLX5_CAP_PORT_SELECTION(dev, port_select_flow_table) ||
-	    !MLX5_CAP_GEN(dev, create_lag_when_not_master_up) ||
+	if (mlx5_eswitch_mode(dev0) != MLX5_ESWITCH_OFFLOADS ||
+	    !MLX5_CAP_PORT_SELECTION(dev0, port_select_flow_table) ||
+	    !MLX5_CAP_GEN(dev0, create_lag_when_not_master_up) ||
 	    !mlx5_lag_check_prereq(ldev))
 		return -EOPNOTSUPP;
 
@@ -83,15 +84,32 @@ static int enable_mpesw(struct mlx5_lag *ldev)
 	if (err)
 		return err;
 
-	err = mlx5_activate_lag(ldev, NULL, MLX5_LAG_MODE_MPESW, false);
+	mlx5_lag_remove_devices(ldev);
+
+	err = mlx5_activate_lag(ldev, NULL, MLX5_LAG_MODE_MPESW, true);
 	if (err) {
-		mlx5_core_warn(dev, "Failed to create LAG in MPESW mode (%d)\n", err);
-		goto out_err;
+		mlx5_core_warn(dev0, "Failed to create LAG in MPESW mode (%d)\n", err);
+		goto err_add_devices;
 	}
 
+	dev0->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+	mlx5_rescan_drivers_locked(dev0);
+	err = mlx5_eswitch_reload_reps(dev0->priv.eswitch);
+	if (!err)
+		err = mlx5_eswitch_reload_reps(dev1->priv.eswitch);
+	if (err)
+		goto err_rescan_drivers;
+
 	return 0;
 
-out_err:
+err_rescan_drivers:
+	dev0->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+	mlx5_rescan_drivers_locked(dev0);
+	mlx5_deactivate_lag(ldev);
+err_add_devices:
+	mlx5_lag_add_devices(ldev);
+	mlx5_eswitch_reload_reps(dev0->priv.eswitch);
+	mlx5_eswitch_reload_reps(dev1->priv.eswitch);
 	mlx5_mpesw_metadata_cleanup(ldev);
 	return err;
 }
@@ -109,6 +127,7 @@ static void mlx5_mpesw_work(struct work_struct *work)
 	struct mlx5_mpesw_work_st *mpesww = container_of(work, struct mlx5_mpesw_work_st, work);
 	struct mlx5_lag *ldev = mpesww->lag;
 
+	mlx5_dev_list_lock();
 	mutex_lock(&ldev->lock);
 	if (ldev->mode_changes_in_progress) {
 		mpesww->result = -EAGAIN;
@@ -121,6 +140,7 @@ static void mlx5_mpesw_work(struct work_struct *work)
 		disable_mpesw(ldev);
 unlock:
 	mutex_unlock(&ldev->lock);
+	mlx5_dev_list_unlock();
 	complete(&mpesww->comp);
 }
 
@@ -187,3 +207,4 @@ bool mlx5_lag_is_mpesw(struct mlx5_core_dev *dev)
 
 	return ldev && ldev->mode == MLX5_LAG_MODE_MPESW;
 }
+EXPORT_SYMBOL(mlx5_lag_is_mpesw);

include/linux/mlx5/driver.h

@@ -1162,6 +1162,7 @@ bool mlx5_lag_is_active(struct mlx5_core_dev *dev);
 bool mlx5_lag_mode_is_hash(struct mlx5_core_dev *dev);
 bool mlx5_lag_is_master(struct mlx5_core_dev *dev);
 bool mlx5_lag_is_shared_fdb(struct mlx5_core_dev *dev);
+bool mlx5_lag_is_mpesw(struct mlx5_core_dev *dev);
 struct net_device *mlx5_lag_get_roce_netdev(struct mlx5_core_dev *dev);
 u8 mlx5_lag_get_slave_port(struct mlx5_core_dev *dev,
 			   struct net_device *slave);