Skip to contentSkip to conversation

net/mlx5e: Expose the VF/SF RX drop counter on the representor

PATCH Applied Tariq Toukan 3 participants 3 messages Apr 10, 2024 – Apr 12, 2024

Subsystems (4)
lore
Share & tools
lore

Tariq Toukan

author maintainer
#0 lore raw

PATCH v3 1/1 net/mlx5e: Expose the VF/SF RX drop counter on the representor

From: Carolina Jubran <[email protected]>

Q counters are device-level counters that track specific events, among which are out_of_buffer events. These events occur when packets are dropped due to a lack of receive buffer in the RX queue.

Expose the total number of out_of_buffer events on the VFs/SFs to their respective representor, using the "ip stats group link" under the name of "rx_missed".

The "rx_missed" equals the sum of all
Q counters out_of_buffer values allocated on the VFs/SFs.

---
 .../net/ethernet/mellanox/mlx5/core/en_rep.c  | 42 ++++++++++++++++++-
 .../ethernet/mellanox/mlx5/core/en_stats.h    |  2 +
 2 files changed, 43 insertions(+), 1 deletion(-)

V3:
Do not use vendor specific counter in ethtool, nor a generic counter. Use a proper counter from the link group.

drivers/net/ethernet/mellanox/mlx5/core/en_rep.c +41 -1
index a74ee698671c..6acecf2e7cf6 100644
@@ -273,6 +273,40 @@ static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(vport_rep)
 	kvfree(out);
 }
 
+static int mlx5e_rep_query_aggr_q_counter(struct mlx5_core_dev *dev, int vport, void *out)
+{
+	u32 in[MLX5_ST_SZ_DW(query_q_counter_in)] = {};
+
+	MLX5_SET(query_q_counter_in, in, opcode, MLX5_CMD_OP_QUERY_Q_COUNTER);
+	MLX5_SET(query_q_counter_in, in, other_vport, 1);
+	MLX5_SET(query_q_counter_in, in, vport_number, vport);
+	MLX5_SET(query_q_counter_in, in, aggregate, 1);
+
+	return mlx5_cmd_exec_inout(dev, query_q_counter, in, out);
+}
+
+static void mlx5e_rep_update_vport_q_counter(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_stats *rep_stats = &priv->stats.rep_stats;
+	u32 out[MLX5_ST_SZ_DW(query_q_counter_out)] = {};
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	int err;
+
+	if (!MLX5_CAP_GEN(priv->mdev, q_counter_other_vport) ||
+	    !MLX5_CAP_GEN(priv->mdev, q_counter_aggregation))
+		return;
+
+	err = mlx5e_rep_query_aggr_q_counter(priv->mdev, rep->vport, out);
+	if (err) {
+		netdev_warn(priv->netdev, "failed reading stats on vport %d, error %d\n",
+			    rep->vport, err);
+		return;
+	}
+
+	rep_stats->rx_vport_out_of_buffer = MLX5_GET(query_q_counter_out, out, out_of_buffer);
+}
+
 static void mlx5e_rep_get_strings(struct net_device *dev,
 				  u32 stringset, u8 *data)
 {
@@ -1229,6 +1263,12 @@ static int mlx5e_update_rep_rx(struct mlx5e_priv *priv)
 	return 0;
 }
 
+static void mlx5e_rep_stats_update_ndo_stats(struct mlx5e_priv *priv)
+{
+	mlx5e_stats_update_ndo_stats(priv);
+	mlx5e_rep_update_vport_q_counter(priv);
+}
+
 static int mlx5e_rep_event_mpesw(struct mlx5e_priv *priv)
 {
 	struct mlx5e_rep_priv *rpriv = priv->ppriv;
@@ -1421,7 +1461,7 @@ static const struct mlx5e_profile mlx5e_rep_profile = {
 	.enable		        = mlx5e_rep_enable,
 	.disable	        = mlx5e_rep_disable,
 	.update_rx		= mlx5e_update_rep_rx,
-	.update_stats           = mlx5e_stats_update_ndo_stats,
+	.update_stats           = mlx5e_rep_stats_update_ndo_stats,
 	.rx_handlers            = &mlx5e_rx_handlers_rep,
 	.max_tc			= 1,
 	.stats_grps		= mlx5e_rep_stats_grps,
drivers/net/ethernet/mellanox/mlx5/core/en_stats.h +2 -0
index 9cee4c9472e9..650732288616 100644
@@ -484,6 +484,7 @@ struct mlx5e_rep_stats {
 	u64 tx_vport_rdma_multicast_bytes;
 	u64 vport_loopback_packets;
 	u64 vport_loopback_bytes;
+	u64 rx_vport_out_of_buffer;
 };
 
 struct mlx5e_stats {
@@ -504,6 +505,7 @@ static inline void mlx5e_stats_copy_rep_stats(struct rtnl_link_stats64 *vf_vport
 	vf_vport->tx_packets = rep_stats->vport_tx_packets;
 	vf_vport->rx_bytes = rep_stats->vport_rx_bytes;
 	vf_vport->tx_bytes = rep_stats->vport_tx_bytes;
+	vf_vport->rx_missed_errors = rep_stats->rx_vport_out_of_buffer;
 }
 
 extern mlx5e_stats_grp_t mlx5e_nic_stats_grps[];
Signature · 1 line
2.31.1
Reviewed-by on v3 1/1
#1 lore raw
Quoted 11 lines · Tariq Toukan

From: Carolina Jubran <[email protected]>

Q counters are device-level counters that track specific events, among which are out_of_buffer events. These events occur when packets are dropped due to a lack of receive buffer in the RX queue.

Expose the total number of out_of_buffer events on the VFs/SFs to their respective representor, using the "ip stats group link" under the name of "rx_missed".

Quoted 8 lines

The "rx_missed" equals the sum of all
Q counters out_of_buffer values allocated on the VFs/SFs.

#2 lore raw
In reply to Tariq Toukan
This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <[email protected]>:

Quoted 8 lines · On Thu, 11 Apr 2024 00:41:54 +0300 you wrote:

From: Carolina Jubran <[email protected]>

Q counters are device-level counters that track specific events, among which are out_of_buffer events. These events occur when packets are dropped due to a lack of receive buffer in the RX queue.

[...]

Here is the summary with links:
- [net-next,V3] net/mlx5e: Expose the VF/SF RX drop counter on the representor

    https://git.kernel.org/netdev/net-next/c/919b38a916b4

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html