net/mlx5e: IPsec: Add TX steering rule per IPsec state

Add new FTE in TX IPsec FT per IPsec state. It has the
same matching criteria as the RX steering rule.

The IPsec FT is created/destroyed when the first/last rule
is added/deleted respectively.

Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Reviewed-by: Boris Pismenny <borisp@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>

3 files changed, 181 insertions, 5 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.h
index 0fc8b4d4f4a3..6164c7f59efb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.h
@@ -76,6 +76,7 @@ struct mlx5e_ipsec_stats {
 };
 
 struct mlx5e_accel_fs_esp;
+struct mlx5e_ipsec_tx;
 
 struct mlx5e_ipsec {
 	struct mlx5e_priv *en_priv;
@@ -87,6 +88,7 @@ struct mlx5e_ipsec {
 	struct mlx5e_ipsec_stats stats;
 	struct workqueue_struct *wq;
 	struct mlx5e_accel_fs_esp *rx_fs;
+	struct mlx5e_ipsec_tx *tx_fs;
 };
 
 struct mlx5e_ipsec_esn_state {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c
index b974f3cd1005..0e45590662a8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c
@@ -34,6 +34,12 @@ struct mlx5e_accel_fs_esp {
 	struct mlx5e_accel_fs_esp_prot fs_prot[ACCEL_FS_ESP_NUM_TYPES];
 };
 
+struct mlx5e_ipsec_tx {
+	struct mlx5_flow_table *ft;
+	struct mutex mutex; /* Protect IPsec TX steering */
+	u32 refcnt;
+};
+
 /* IPsec RX flow steering */
 static enum mlx5e_traffic_types fs_esp2tt(enum accel_fs_esp_type i)
 {
@@ -323,6 +329,77 @@ out:
 	mutex_unlock(&fs_prot->prot_mutex);
 }
 
+/* IPsec TX flow steering */
+static int tx_create(struct mlx5e_priv *priv)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5e_ipsec *ipsec = priv->ipsec;
+	struct mlx5_flow_table *ft;
+	int err;
+
+	priv->fs.egress_ns =
+		mlx5_get_flow_namespace(priv->mdev,
+					MLX5_FLOW_NAMESPACE_EGRESS_KERNEL);
+	if (!priv->fs.egress_ns)
+		return -EOPNOTSUPP;
+
+	ft_attr.max_fte = NUM_IPSEC_FTE;
+	ft_attr.autogroup.max_num_groups = 1;
+	ft = mlx5_create_auto_grouped_flow_table(priv->fs.egress_ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		netdev_err(priv->netdev, "fail to create ipsec tx ft err=%d\n", err);
+		return err;
+	}
+	ipsec->tx_fs->ft = ft;
+	return 0;
+}
+
+static void tx_destroy(struct mlx5e_priv *priv)
+{
+	struct mlx5e_ipsec *ipsec = priv->ipsec;
+
+	if (IS_ERR_OR_NULL(ipsec->tx_fs->ft))
+		return;
+
+	mlx5_destroy_flow_table(ipsec->tx_fs->ft);
+	ipsec->tx_fs->ft = NULL;
+}
+
+static int tx_ft_get(struct mlx5e_priv *priv)
+{
+	struct mlx5e_ipsec_tx *tx_fs = priv->ipsec->tx_fs;
+	int err = 0;
+
+	mutex_lock(&tx_fs->mutex);
+	if (tx_fs->refcnt++)
+		goto out;
+
+	err = tx_create(priv);
+	if (err) {
+		tx_fs->refcnt--;
+		goto out;
+	}
+
+out:
+	mutex_unlock(&tx_fs->mutex);
+	return err;
+}
+
+static void tx_ft_put(struct mlx5e_priv *priv)
+{
+	struct mlx5e_ipsec_tx *tx_fs = priv->ipsec->tx_fs;
+
+	mutex_lock(&tx_fs->mutex);
+	if (--tx_fs->refcnt)
+		goto out;
+
+	tx_destroy(priv);
+
+out:
+	mutex_unlock(&tx_fs->mutex);
+}
+
 static void setup_fte_common(struct mlx5_accel_esp_xfrm_attrs *attrs,
 			     u32 ipsec_obj_id,
 			     struct mlx5_flow_spec *spec,
@@ -457,6 +534,54 @@ out:
 	return err;
 }
 
+static int tx_add_rule(struct mlx5e_priv *priv,
+		       struct mlx5_accel_esp_xfrm_attrs *attrs,
+		       u32 ipsec_obj_id,
+		       struct mlx5e_ipsec_rule *ipsec_rule)
+{
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	err = tx_ft_get(priv);
+	if (err)
+		return err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	setup_fte_common(attrs, ipsec_obj_id, spec, &flow_act);
+
+	/* Add IPsec indicator in metadata_reg_a */
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
+	MLX5_SET(fte_match_param, spec->match_criteria, misc_parameters_2.metadata_reg_a,
+		 MLX5_ETH_WQE_FT_META_IPSEC);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_a,
+		 MLX5_ETH_WQE_FT_META_IPSEC);
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW |
+			  MLX5_FLOW_CONTEXT_ACTION_IPSEC_ENCRYPT;
+	rule = mlx5_add_flow_rules(priv->ipsec->tx_fs->ft, spec, &flow_act, NULL, 0);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(priv->netdev, "fail to add ipsec rule attrs->action=0x%x, err=%d\n",
+			   attrs->action, err);
+		goto out;
+	}
+
+	ipsec_rule->rule = rule;
+
+out:
+	kvfree(spec);
+	if (err)
+		tx_ft_put(priv);
+	return err;
+}
+
 static void rx_del_rule(struct mlx5e_priv *priv,
 			struct mlx5_accel_esp_xfrm_attrs *attrs,
 			struct mlx5e_ipsec_rule *ipsec_rule)
@@ -470,15 +595,27 @@ static void rx_del_rule(struct mlx5e_priv *priv,
 	rx_ft_put(priv, attrs->is_ipv6 ? ACCEL_FS_ESP6 : ACCEL_FS_ESP4);
 }
 
+static void tx_del_rule(struct mlx5e_priv *priv,
+			struct mlx5e_ipsec_rule *ipsec_rule)
+{
+	mlx5_del_flow_rules(ipsec_rule->rule);
+	ipsec_rule->rule = NULL;
+
+	tx_ft_put(priv);
+}
+
 int mlx5e_accel_ipsec_fs_add_rule(struct mlx5e_priv *priv,
 				  struct mlx5_accel_esp_xfrm_attrs *attrs,
 				  u32 ipsec_obj_id,
 				  struct mlx5e_ipsec_rule *ipsec_rule)
 {
-	if (!priv->ipsec->rx_fs || attrs->action != MLX5_ACCEL_ESP_ACTION_DECRYPT)
+	if (!priv->ipsec->rx_fs)
 		return -EOPNOTSUPP;
 
-	return rx_add_rule(priv, attrs, ipsec_obj_id, ipsec_rule);
+	if (attrs->action == MLX5_ACCEL_ESP_ACTION_DECRYPT)
+		return rx_add_rule(priv, attrs, ipsec_obj_id, ipsec_rule);
+	else
+		return tx_add_rule(priv, attrs, ipsec_obj_id, ipsec_rule);
 }
 
 void mlx5e_accel_ipsec_fs_del_rule(struct mlx5e_priv *priv,
@@ -488,7 +625,18 @@ void mlx5e_accel_ipsec_fs_del_rule(struct mlx5e_priv *priv,
 	if (!priv->ipsec->rx_fs)
 		return;
 
-	rx_del_rule(priv, attrs, ipsec_rule);
+	if (attrs->action == MLX5_ACCEL_ESP_ACTION_DECRYPT)
+		rx_del_rule(priv, attrs, ipsec_rule);
+	else
+		tx_del_rule(priv, ipsec_rule);
+}
+
+static void fs_cleanup_tx(struct mlx5e_priv *priv)
+{
+	mutex_destroy(&priv->ipsec->tx_fs->mutex);
+	WARN_ON(priv->ipsec->tx_fs->refcnt);
+	kfree(priv->ipsec->tx_fs);
+	priv->ipsec->tx_fs = NULL;
 }
 
 static void fs_cleanup_rx(struct mlx5e_priv *priv)
@@ -507,6 +655,17 @@ static void fs_cleanup_rx(struct mlx5e_priv *priv)
 	priv->ipsec->rx_fs = NULL;
 }
 
+static int fs_init_tx(struct mlx5e_priv *priv)
+{
+	priv->ipsec->tx_fs =
+		kzalloc(sizeof(struct mlx5e_ipsec_tx), GFP_KERNEL);
+	if (!priv->ipsec->tx_fs)
+		return -ENOMEM;
+
+	mutex_init(&priv->ipsec->tx_fs->mutex);
+	return 0;
+}
+
 static int fs_init_rx(struct mlx5e_priv *priv)
 {
 	struct mlx5e_accel_fs_esp_prot *fs_prot;
@@ -532,13 +691,24 @@ void mlx5e_accel_ipsec_fs_cleanup(struct mlx5e_priv *priv)
 	if (!priv->ipsec->rx_fs)
 		return;
 
+	fs_cleanup_tx(priv);
 	fs_cleanup_rx(priv);
 }
 
 int mlx5e_accel_ipsec_fs_init(struct mlx5e_priv *priv)
 {
+	int err;
+
 	if (!mlx5_is_ipsec_device(priv->mdev) || !priv->ipsec)
 		return -EOPNOTSUPP;
 
-	return fs_init_rx(priv);
+	err = fs_init_tx(priv);
+	if (err)
+		return err;
+
+	err = fs_init_rx(priv);
+	if (err)
+		fs_cleanup_tx(priv);
+
+	return err;
 }
diff --git a/include/linux/mlx5/qp.h b/include/linux/mlx5/qp.h
index 36492a1342cf..d75ef8aa8fac 100644
--- a/include/linux/mlx5/qp.h
+++ b/include/linux/mlx5/qp.h
@@ -245,6 +245,10 @@ enum {
 	MLX5_ETH_WQE_SWP_OUTER_L4_UDP   = 1 << 5,
 };
 
+enum {
+	MLX5_ETH_WQE_FT_META_IPSEC = BIT(0),
+};
+
 struct mlx5_wqe_eth_seg {
 	u8              swp_outer_l4_offset;
 	u8              swp_outer_l3_offset;
@@ -253,7 +257,7 @@ struct mlx5_wqe_eth_seg {
 	u8              cs_flags;
 	u8              swp_flags;
 	__be16          mss;
-	__be32          rsvd2;
+	__be32          flow_table_metadata;
 	union {
 		struct {
 			__be16 sz;