net/mlx5e: Allow reporting of checksum unnecessary

Currently we practically never report checksum unnecessary, because
for all IP packets we take the checksum complete path.

Enable non-default runs with reprorting checksum unnecessary, using
an ethtool private flag. This can be useful for performance evals
and other explorations.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

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

@@ -209,6 +209,7 @@ enum mlx5e_priv_flag {
 	MLX5E_PFLAG_TX_CQE_BASED_MODER = (1 << 1),
 	MLX5E_PFLAG_RX_CQE_COMPRESS = (1 << 2),
 	MLX5E_PFLAG_RX_STRIDING_RQ = (1 << 3),
+	MLX5E_PFLAG_RX_NO_CSUM_COMPLETE = (1 << 4),
 };
 
 #define MLX5E_SET_PFLAG(params, pflag, enable)			\
@@ -290,6 +291,7 @@ struct mlx5e_dcbx_dp {
 enum {
 	MLX5E_RQ_STATE_ENABLED,
 	MLX5E_RQ_STATE_AM,
+	MLX5E_RQ_STATE_NO_CSUM_COMPLETE,
 };
 
 struct mlx5e_cq {

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

@@ -140,6 +140,7 @@ static const char mlx5e_priv_flags[][ETH_GSTRING_LEN] = {
 	"tx_cqe_moder",
 	"rx_cqe_compress",
 	"rx_striding_rq",
+	"rx_no_csum_complete",
 };
 
 int mlx5e_ethtool_get_sset_count(struct mlx5e_priv *priv, int sset)
@@ -1531,6 +1532,27 @@ static int set_pflag_rx_striding_rq(struct net_device *netdev, bool enable)
 	return 0;
 }
 
+static int set_pflag_rx_no_csum_complete(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_channels *channels = &priv->channels;
+	struct mlx5e_channel *c;
+	int i;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	for (i = 0; i < channels->num; i++) {
+		c = channels->c[i];
+		if (enable)
+			__set_bit(MLX5E_RQ_STATE_NO_CSUM_COMPLETE, &c->rq.state);
+		else
+			__clear_bit(MLX5E_RQ_STATE_NO_CSUM_COMPLETE, &c->rq.state);
+	}
+
+	return 0;
+}
+
 static int mlx5e_handle_pflag(struct net_device *netdev,
 			      u32 wanted_flags,
 			      enum mlx5e_priv_flag flag,
@@ -1582,6 +1604,12 @@ static int mlx5e_set_priv_flags(struct net_device *netdev, u32 pflags)
 	err = mlx5e_handle_pflag(netdev, pflags,
 				 MLX5E_PFLAG_RX_STRIDING_RQ,
 				 set_pflag_rx_striding_rq);
+	if (err)
+		goto out;
+
+	err = mlx5e_handle_pflag(netdev, pflags,
+				 MLX5E_PFLAG_RX_NO_CSUM_COMPLETE,
+				 set_pflag_rx_no_csum_complete);
 
 out:
 	mutex_unlock(&priv->state_lock);

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

@@ -929,6 +929,9 @@ static int mlx5e_open_rq(struct mlx5e_channel *c,
 	if (params->rx_dim_enabled)
 		__set_bit(MLX5E_RQ_STATE_AM, &c->rq.state);
 
+	if (params->pflags & MLX5E_PFLAG_RX_NO_CSUM_COMPLETE)
+		__set_bit(MLX5E_RQ_STATE_NO_CSUM_COMPLETE, &c->rq.state);
+
 	return 0;
 
 err_destroy_rq:
@@ -4528,6 +4531,7 @@ void mlx5e_build_nic_params(struct mlx5_core_dev *mdev,
 		params->rx_cqe_compress_def = slow_pci_heuristic(mdev);
 
 	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS, params->rx_cqe_compress_def);
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_NO_CSUM_COMPLETE, false);
 
 	/* RQ */
 	mlx5e_build_rq_params(mdev, params);

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

@@ -782,6 +782,9 @@ static inline void mlx5e_handle_csum(struct net_device *netdev,
 		return;
 	}
 
+	if (unlikely(test_bit(MLX5E_RQ_STATE_NO_CSUM_COMPLETE, &rq->state)))
+		goto csum_unnecessary;
+
 	if (likely(is_last_ethertype_ip(skb, &network_depth, &proto))) {
 		if (unlikely(get_ip_proto(skb, proto) == IPPROTO_SCTP))
 			goto csum_unnecessary;