net/mlx5e: Add recovery flow for tx devlink health reporter for unhealthy PTP SQ

A new check for the tx devlink health reporter is introduced for
determining when the PTP port timestamping SQ is considered unhealthy. If
there are enough CQEs considered never to be delivered, the space that can
be utilized on the SQ decreases significantly, impacting performance and
usability of the SQ. The health reporter is triggered when the number of
likely never delivered port timestamping CQEs that utilize the space of the
PTP SQ is greater than 93.75% of the total capacity of the SQ. A devlink
health reporter recover method is also provided for this specific TX error
context that restarts the PTP SQ.

Signed-off-by: Rahul Rameshbabu <[e-mail]>
Reviewed-by: Tariq Toukan <[e-mail]>
Signed-off-by: Saeed Mahameed <[e-mail]>

Documentation/networking/devlink/mlx5.rst

···
 
 tx reporter
 -----------
-The tx reporter is responsible for reporting and recovering of the following two error scenarios:
+The tx reporter is responsible for reporting and recovering of the following three error scenarios:
 
 - tx timeout
     Report on kernel tx timeout detection.
···
 - tx error completion
     Report on error tx completion.
     Recover by flushing the tx queue and reset it.
+- tx PTP port timestamping CQ unhealthy
+    Report too many CQEs never delivered on port ts CQ.
+    Recover by flushing and re-creating all PTP channels.
 
 tx reporter also support on demand diagnose callback, on which it provides
 real time information of its send queues status.

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

···
 void mlx5e_reporter_tx_destroy(struct mlx5e_priv *priv);
 void mlx5e_reporter_tx_err_cqe(struct mlx5e_txqsq *sq);
 int mlx5e_reporter_tx_timeout(struct mlx5e_txqsq *sq);
+void mlx5e_reporter_tx_ptpsq_unhealthy(struct mlx5e_ptpsq *ptpsq);
 
 int mlx5e_health_cq_diag_fmsg(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg);
 int mlx5e_health_cq_common_diag_fmsg(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg);

drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c

···
 // Copyright (c) 2020 Mellanox Technologies
 
 #include "en/ptp.h"
+#include "en/health.h"
 #include "en/txrx.h"
 #include "en/params.h"
 #include "en/fs_tt_redirect.h"
···
 	return skb;
 }
 
+static bool mlx5e_ptp_metadata_map_unhealthy(struct mlx5e_ptp_metadata_map *map)
+{
+	/* Considered beginning unhealthy state if size * 15 / 2^4 cannot be reclaimed. */
+	return map->undelivered_counter > (map->capacity >> 4) * 15;
+}
+
 static void mlx5e_ptpsq_mark_ts_cqes_undelivered(struct mlx5e_ptpsq *ptpsq,
 						 ktime_t port_tstamp)
 {
···
 out:
 	napi_consume_skb(skb, budget);
 	mlx5e_ptp_metadata_fifo_push(&ptpsq->metadata_freelist, metadata_id);
+	if (unlikely(mlx5e_ptp_metadata_map_unhealthy(&ptpsq->metadata_map)) &&
+	    !test_and_set_bit(MLX5E_SQ_STATE_RECOVERING, &sq->state))
+		queue_work(ptpsq->txqsq.priv->wq, &ptpsq->report_unhealthy_work);
 }
 
 static bool mlx5e_ptp_poll_ts_cq(struct mlx5e_cq *cq, int budget)
···
 	kvfree(ptpsq->ts_cqe_pending_list);
 }
 
+static void mlx5e_ptpsq_unhealthy_work(struct work_struct *work)
+{
+	struct mlx5e_ptpsq *ptpsq =
+		container_of(work, struct mlx5e_ptpsq, report_unhealthy_work);
+
+	mlx5e_reporter_tx_ptpsq_unhealthy(ptpsq);
+}
+
 static int mlx5e_ptp_open_txqsq(struct mlx5e_ptp *c, u32 tisn,
 				int txq_ix, struct mlx5e_ptp_params *cparams,
 				int tc, struct mlx5e_ptpsq *ptpsq)
···
 	if (err)
 		goto err_free_txqsq;
 
+	INIT_WORK(&ptpsq->report_unhealthy_work, mlx5e_ptpsq_unhealthy_work);
+
 	return 0;
 
 err_free_txqsq:
···
 	struct mlx5e_txqsq *sq = &ptpsq->txqsq;
 	struct mlx5_core_dev *mdev = sq->mdev;
 
+	if (current_work() != &ptpsq->report_unhealthy_work)
+		cancel_work_sync(&ptpsq->report_unhealthy_work);
 	mlx5e_ptp_free_traffic_db(ptpsq);
 	cancel_work_sync(&sq->recover_work);
 	mlx5e_ptp_destroy_sq(mdev, sq->sqn);

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

···
 #include <linux/ktime.h>
 #include <linux/ptp_classify.h>
 #include <linux/time64.h>
+#include <linux/workqueue.h>
 
 #define MLX5E_PTP_CHANNEL_IX 0
 #define MLX5E_PTP_MAX_LOG_SQ_SIZE (8U)
···
 	struct mlx5e_ptp_cq_stats *cq_stats;
 	u16                      ts_cqe_ctr_mask;
 
+	struct work_struct                 report_unhealthy_work;
 	struct mlx5e_ptp_port_ts_cqe_list  *ts_cqe_pending_list;
 	struct mlx5e_ptp_metadata_fifo     metadata_freelist;
 	struct mlx5e_ptp_metadata_map      metadata_map;

drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c

···
 	return err;
 }
 
+static int mlx5e_tx_reporter_ptpsq_unhealthy_recover(void *ctx)
+{
+	struct mlx5e_ptpsq *ptpsq = ctx;
+	struct mlx5e_channels *chs;
+	struct net_device *netdev;
+	struct mlx5e_priv *priv;
+	int carrier_ok;
+	int err;
+
+	if (!test_bit(MLX5E_SQ_STATE_RECOVERING, &ptpsq->txqsq.state))
+		return 0;
+
+	priv = ptpsq->txqsq.priv;
+
+	mutex_lock(&priv->state_lock);
+	chs = &priv->channels;
+	netdev = priv->netdev;
+
+	carrier_ok = netif_carrier_ok(netdev);
+	netif_carrier_off(netdev);
+
+	mlx5e_deactivate_priv_channels(priv);
+
+	mlx5e_ptp_close(chs->ptp);
+	err = mlx5e_ptp_open(priv, &chs->params, chs->c[0]->lag_port, &chs->ptp);
+
+	mlx5e_activate_priv_channels(priv);
+
+	/* return carrier back if needed */
+	if (carrier_ok)
+		netif_carrier_on(netdev);
+
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
 /* state lock cannot be grabbed within this function.
  * It can cause a dead lock or a read-after-free.
  */
···
 	return mlx5e_tx_reporter_dump_sq(priv, fmsg, to_ctx->sq);
 }
 
+static int mlx5e_tx_reporter_ptpsq_unhealthy_dump(struct mlx5e_priv *priv,
+						  struct devlink_fmsg *fmsg,
+						  void *ctx)
+{
+	struct mlx5e_ptpsq *ptpsq = ctx;
+
+	return mlx5e_tx_reporter_dump_sq(priv, fmsg, &ptpsq->txqsq);
+}
+
 static int mlx5e_tx_reporter_dump_all_sqs(struct mlx5e_priv *priv,
 					  struct devlink_fmsg *fmsg)
 {
···
 
 	mlx5e_health_report(priv, priv->tx_reporter, err_str, &err_ctx);
 	return to_ctx.status;
+}
+
+void mlx5e_reporter_tx_ptpsq_unhealthy(struct mlx5e_ptpsq *ptpsq)
+{
+	struct mlx5e_ptp_metadata_map *map = &ptpsq->metadata_map;
+	char err_str[MLX5E_REPORTER_PER_Q_MAX_LEN];
+	struct mlx5e_txqsq *txqsq = &ptpsq->txqsq;
+	struct mlx5e_cq *ts_cq = &ptpsq->ts_cq;
+	struct mlx5e_priv *priv = txqsq->priv;
+	struct mlx5e_err_ctx err_ctx = {};
+
+	err_ctx.ctx = ptpsq;
+	err_ctx.recover = mlx5e_tx_reporter_ptpsq_unhealthy_recover;
+	err_ctx.dump = mlx5e_tx_reporter_ptpsq_unhealthy_dump;
+	snprintf(err_str, sizeof(err_str),
+		 "Unhealthy TX port TS queue: %d, SQ: 0x%x, CQ: 0x%x, Undelivered CQEs: %u Map Capacity: %u",
+		 txqsq->ch_ix, txqsq->sqn, ts_cq->mcq.cqn, map->undelivered_counter, map->capacity);
+
+	mlx5e_health_report(priv, priv->tx_reporter, err_str, &err_ctx);
 }
 
 static const struct devlink_health_reporter_ops mlx5_tx_reporter_ops = {