net/mlx5e: Enable CQE based moderation on TX CQ

By using CQE based moderation on TX CQ we can reduce the number of TX
interrupt rate. Besides the benefit of less interrupts, this also
allows the kernel to better utilize TSO. Since TSO has some CPU overhead,
it might not aggregate when CPU is under high stress. By reducing the
interrupt rate and the CPU utilization, we can get better aggregation
and better overall throughput.
The feature is enabled by default and has a private flag in ethtool
for control.

Throughput, interrupt rate and TSO utilization improvements:
(ConnectX-4Lx 40GbE, unidirectional, 1/16 TCP streams, 64B packets)
---------------------------------------------------------
Metric   | Streams | CQE Based | EQE Based | improvement
---------------------------------------------------------
BW       |    1    |  2.4Gb/s  | 2.15Gb/s  |  +11.6%
IR       |    1    |  27Kips   | 50.6Kips  |  -46.7%
TSO Util |    1    |  74.6%    | 71%       |  +5%
BW       |    16   |  29Gb/s   | 25.85Gb/s |  +12.2%
IR       |    16   |  482Kips  | 745Kips   |  -35.3%
TSO Util |    16   |  69.1%    | 49%       |  +41.1%

*BW = Bandwidth, IR = Interrupt rate, ips = interrupt per second.
TSO Util = bytes in TSO sessions / all bytes transferred

Signed-off-by: Tal Gilboa <talgi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

1 files changed, 31 insertions, 8 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index 63d1ac695a75..23425f028405 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -1454,29 +1454,36 @@ static int mlx5e_get_module_eeprom(struct net_device *netdev,
 
 typedef int (*mlx5e_pflag_handler)(struct net_device *netdev, bool enable);
 
-static int set_pflag_rx_cqe_based_moder(struct net_device *netdev, bool enable)
+static int set_pflag_cqe_based_moder(struct net_device *netdev, bool enable,
+				     bool is_rx_cq)
 {
 	struct mlx5e_priv *priv = netdev_priv(netdev);
 	struct mlx5_core_dev *mdev = priv->mdev;
 	struct mlx5e_channels new_channels = {};
-	bool rx_mode_changed;
-	u8 rx_cq_period_mode;
+	bool mode_changed;
+	u8 cq_period_mode, current_cq_period_mode;
 	int err = 0;
 
-	rx_cq_period_mode = enable ?
+	cq_period_mode = enable ?
 		MLX5_CQ_PERIOD_MODE_START_FROM_CQE :
 		MLX5_CQ_PERIOD_MODE_START_FROM_EQE;
-	rx_mode_changed = rx_cq_period_mode != priv->channels.params.rx_cq_period_mode;
+	current_cq_period_mode = is_rx_cq ?
+		priv->channels.params.rx_cq_moderation.cq_period_mode :
+		priv->channels.params.tx_cq_moderation.cq_period_mode;
+	mode_changed = cq_period_mode != current_cq_period_mode;
 
-	if (rx_cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE &&
+	if (cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE &&
 	    !MLX5_CAP_GEN(mdev, cq_period_start_from_cqe))
 		return -EOPNOTSUPP;
 
-	if (!rx_mode_changed)
+	if (!mode_changed)
 		return 0;
 
 	new_channels.params = priv->channels.params;
-	mlx5e_set_rx_cq_mode_params(&new_channels.params, rx_cq_period_mode);
+	if (is_rx_cq)
+		mlx5e_set_rx_cq_mode_params(&new_channels.params, cq_period_mode);
+	else
+		mlx5e_set_tx_cq_mode_params(&new_channels.params, cq_period_mode);
 
 	if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
 		priv->channels.params = new_channels.params;
@@ -1491,6 +1498,16 @@ static int set_pflag_rx_cqe_based_moder(struct net_device *netdev, bool enable)
 	return 0;
 }
 
+static int set_pflag_tx_cqe_based_moder(struct net_device *netdev, bool enable)
+{
+	return set_pflag_cqe_based_moder(netdev, enable, false);
+}
+
+static int set_pflag_rx_cqe_based_moder(struct net_device *netdev, bool enable)
+{
+	return set_pflag_cqe_based_moder(netdev, enable, true);
+}
+
 int mlx5e_modify_rx_cqe_compression_locked(struct mlx5e_priv *priv, bool new_val)
 {
 	bool curr_val = MLX5E_GET_PFLAG(&priv->channels.params, MLX5E_PFLAG_RX_CQE_COMPRESS);
@@ -1579,6 +1596,12 @@ static int mlx5e_set_priv_flags(struct net_device *netdev, u32 pflags)
 		goto out;
 
 	err = mlx5e_handle_pflag(netdev, pflags,
+				 MLX5E_PFLAG_TX_CQE_BASED_MODER,
+				 set_pflag_tx_cqe_based_moder);
+	if (err)
+		goto out;
+
+	err = mlx5e_handle_pflag(netdev, pflags,
 				 MLX5E_PFLAG_RX_CQE_COMPRESS,
 				 set_pflag_rx_cqe_compress);