net/mlx5e: Split between RX/TX tunnel FW support indication
authorAya Levin <ayal@nvidia.com>
Tue, 1 Dec 2020 22:42:07 +0000 (14:42 -0800)
committerSaeed Mahameed <saeedm@nvidia.com>
Tue, 8 Dec 2020 19:28:48 +0000 (11:28 -0800)
Use the new FW caps to advertise for ip-in-ip tunnel support separately
for RX and TX.

Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
drivers/net/ethernet/mellanox/mlx5/core/en_main.c

index dc744702aee4a567a966e2b08ad6e05b8a9fae63..5749557749b0bd90a8372aeaa6d695b8e072fed4 100644 (file)
@@ -287,8 +287,7 @@ void mlx5e_disable_cvlan_filter(struct mlx5e_priv *priv);
 int mlx5e_create_flow_steering(struct mlx5e_priv *priv);
 void mlx5e_destroy_flow_steering(struct mlx5e_priv *priv);
 
-bool mlx5e_tunnel_proto_supported(struct mlx5_core_dev *mdev, u8 proto_type);
-bool mlx5e_any_tunnel_proto_supported(struct mlx5_core_dev *mdev);
+u8 mlx5e_get_proto_by_tunnel_type(enum mlx5e_tunnel_types tt);
 
 #endif /* __MLX5E_FLOW_STEER_H__ */
 
index 1f48f99c0997d98f62415b3dee239b43a8a145b3..fa8149f6eb088ae86e8d67f9527ae6f714c78741 100644 (file)
@@ -772,25 +772,31 @@ static struct mlx5e_etype_proto ttc_tunnel_rules[] = {
 
 };
 
-bool mlx5e_tunnel_proto_supported(struct mlx5_core_dev *mdev, u8 proto_type)
+u8 mlx5e_get_proto_by_tunnel_type(enum mlx5e_tunnel_types tt)
+{
+       return ttc_tunnel_rules[tt].proto;
+}
+
+static bool mlx5e_tunnel_proto_supported_rx(struct mlx5_core_dev *mdev, u8 proto_type)
 {
        switch (proto_type) {
        case IPPROTO_GRE:
                return MLX5_CAP_ETH(mdev, tunnel_stateless_gre);
        case IPPROTO_IPIP:
        case IPPROTO_IPV6:
-               return MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip);
+               return (MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip) ||
+                       MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip_rx));
        default:
                return false;
        }
 }
 
-bool mlx5e_any_tunnel_proto_supported(struct mlx5_core_dev *mdev)
+static bool mlx5e_tunnel_any_rx_proto_supported(struct mlx5_core_dev *mdev)
 {
        int tt;
 
        for (tt = 0; tt < MLX5E_NUM_TUNNEL_TT; tt++) {
-               if (mlx5e_tunnel_proto_supported(mdev, ttc_tunnel_rules[tt].proto))
+               if (mlx5e_tunnel_proto_supported_rx(mdev, ttc_tunnel_rules[tt].proto))
                        return true;
        }
        return false;
@@ -798,7 +804,7 @@ bool mlx5e_any_tunnel_proto_supported(struct mlx5_core_dev *mdev)
 
 bool mlx5e_tunnel_inner_ft_supported(struct mlx5_core_dev *mdev)
 {
-       return (mlx5e_any_tunnel_proto_supported(mdev) &&
+       return (mlx5e_tunnel_any_rx_proto_supported(mdev) &&
                MLX5_CAP_FLOWTABLE_NIC_RX(mdev, ft_field_support.inner_ip_version));
 }
 
@@ -899,8 +905,8 @@ static int mlx5e_generate_ttc_table_rules(struct mlx5e_priv *priv,
        dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
        dest.ft   = params->inner_ttc->ft.t;
        for (tt = 0; tt < MLX5E_NUM_TUNNEL_TT; tt++) {
-               if (!mlx5e_tunnel_proto_supported(priv->mdev,
-                                                 ttc_tunnel_rules[tt].proto))
+               if (!mlx5e_tunnel_proto_supported_rx(priv->mdev,
+                                                    ttc_tunnel_rules[tt].proto))
                        continue;
                trules[tt] = mlx5e_generate_ttc_rule(priv, ft, &dest,
                                                     ttc_tunnel_rules[tt].etype,
index 3275c4e7cbdff0838629b6c1c036bd6a6ac23a62..2490d68cbd3242798fa22c29a1dab147e9186765 100644 (file)
@@ -4280,6 +4280,20 @@ int mlx5e_get_vf_stats(struct net_device *dev,
 }
 #endif
 
+static bool mlx5e_tunnel_proto_supported_tx(struct mlx5_core_dev *mdev, u8 proto_type)
+{
+       switch (proto_type) {
+       case IPPROTO_GRE:
+               return MLX5_CAP_ETH(mdev, tunnel_stateless_gre);
+       case IPPROTO_IPIP:
+       case IPPROTO_IPV6:
+               return (MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip) ||
+                       MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip_tx));
+       default:
+               return false;
+       }
+}
+
 static bool mlx5e_gre_tunnel_inner_proto_offload_supported(struct mlx5_core_dev *mdev,
                                                           struct sk_buff *skb)
 {
@@ -4322,7 +4336,7 @@ static netdev_features_t mlx5e_tunnel_features_check(struct mlx5e_priv *priv,
                break;
        case IPPROTO_IPIP:
        case IPPROTO_IPV6:
-               if (mlx5e_tunnel_proto_supported(priv->mdev, IPPROTO_IPIP))
+               if (mlx5e_tunnel_proto_supported_tx(priv->mdev, IPPROTO_IPIP))
                        return features;
                break;
        case IPPROTO_UDP:
@@ -4882,6 +4896,17 @@ void mlx5e_vxlan_set_netdev_info(struct mlx5e_priv *priv)
        priv->netdev->udp_tunnel_nic_info = &priv->nic_info;
 }
 
+static bool mlx5e_tunnel_any_tx_proto_supported(struct mlx5_core_dev *mdev)
+{
+       int tt;
+
+       for (tt = 0; tt < MLX5E_NUM_TUNNEL_TT; tt++) {
+               if (mlx5e_tunnel_proto_supported_tx(mdev, mlx5e_get_proto_by_tunnel_type(tt)))
+                       return true;
+       }
+       return (mlx5_vxlan_allowed(mdev->vxlan) || mlx5_geneve_tx_allowed(mdev));
+}
+
 static void mlx5e_build_nic_netdev(struct net_device *netdev)
 {
        struct mlx5e_priv *priv = netdev_priv(netdev);
@@ -4927,8 +4952,7 @@ static void mlx5e_build_nic_netdev(struct net_device *netdev)
 
        mlx5e_vxlan_set_netdev_info(priv);
 
-       if (mlx5_vxlan_allowed(mdev->vxlan) || mlx5_geneve_tx_allowed(mdev) ||
-           mlx5e_any_tunnel_proto_supported(mdev)) {
+       if (mlx5e_tunnel_any_tx_proto_supported(mdev)) {
                netdev->hw_enc_features |= NETIF_F_HW_CSUM;
                netdev->hw_enc_features |= NETIF_F_TSO;
                netdev->hw_enc_features |= NETIF_F_TSO6;
@@ -4945,7 +4969,7 @@ static void mlx5e_build_nic_netdev(struct net_device *netdev)
                                         NETIF_F_GSO_UDP_TUNNEL_CSUM;
        }
 
-       if (mlx5e_tunnel_proto_supported(mdev, IPPROTO_GRE)) {
+       if (mlx5e_tunnel_proto_supported_tx(mdev, IPPROTO_GRE)) {
                netdev->hw_features     |= NETIF_F_GSO_GRE |
                                           NETIF_F_GSO_GRE_CSUM;
                netdev->hw_enc_features |= NETIF_F_GSO_GRE |
@@ -4954,7 +4978,7 @@ static void mlx5e_build_nic_netdev(struct net_device *netdev)
                                                NETIF_F_GSO_GRE_CSUM;
        }
 
-       if (mlx5e_tunnel_proto_supported(mdev, IPPROTO_IPIP)) {
+       if (mlx5e_tunnel_proto_supported_tx(mdev, IPPROTO_IPIP)) {
                netdev->hw_features |= NETIF_F_GSO_IPXIP4 |
                                       NETIF_F_GSO_IPXIP6;
                netdev->hw_enc_features |= NETIF_F_GSO_IPXIP4 |