Merge tag 'mlx5-updates-2020-07-02' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5-updates-2020-07-02

Rx and Tx devlink health reporters enhancements.

1) Code cleanup
2) devlink output format improvements
3) Print more useful info on devlink health diagnose output
4) TX timeout recovery, on a single SQ recover failure, stop the loop
and reset all rings (re-open netdev).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

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

@@ -852,26 +852,6 @@ bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq, int napi_budget);
 int mlx5e_poll_rx_cq(struct mlx5e_cq *cq, int budget);
 void mlx5e_free_txqsq_descs(struct mlx5e_txqsq *sq);
 
-static inline u32 mlx5e_rqwq_get_size(struct mlx5e_rq *rq)
-{
-	switch (rq->wq_type) {
-	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
-		return mlx5_wq_ll_get_size(&rq->mpwqe.wq);
-	default:
-		return mlx5_wq_cyc_get_size(&rq->wqe.wq);
-	}
-}
-
-static inline u32 mlx5e_rqwq_get_cur_sz(struct mlx5e_rq *rq)
-{
-	switch (rq->wq_type) {
-	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
-		return rq->mpwqe.wq.cur_sz;
-	default:
-		return rq->wqe.wq.cur_sz;
-	}
-}
-
 bool mlx5e_check_fragmented_striding_rq_cap(struct mlx5_core_dev *mdev);
 bool mlx5e_striding_rq_possible(struct mlx5_core_dev *mdev,
 				struct mlx5e_params *params);

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

@@ -5,7 +5,7 @@
 #include "lib/eq.h"
 #include "lib/mlx5.h"
 
-int mlx5e_reporter_named_obj_nest_start(struct devlink_fmsg *fmsg, char *name)
+int mlx5e_health_fmsg_named_obj_nest_start(struct devlink_fmsg *fmsg, char *name)
 {
 	int err;
 
@@ -20,7 +20,7 @@ int mlx5e_reporter_named_obj_nest_start(struct devlink_fmsg *fmsg, char *name)
 	return 0;
 }
 
-int mlx5e_reporter_named_obj_nest_end(struct devlink_fmsg *fmsg)
+int mlx5e_health_fmsg_named_obj_nest_end(struct devlink_fmsg *fmsg)
 {
 	int err;
 
@@ -35,7 +35,7 @@ int mlx5e_reporter_named_obj_nest_end(struct devlink_fmsg *fmsg)
 	return 0;
 }
 
-int mlx5e_reporter_cq_diagnose(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg)
+int mlx5e_health_cq_diag_fmsg(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg)
 {
 	struct mlx5e_priv *priv = cq->channel->priv;
 	u32 out[MLX5_ST_SZ_DW(query_cq_out)] = {};
@@ -50,7 +50,7 @@ int mlx5e_reporter_cq_diagnose(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg)
 	cqc = MLX5_ADDR_OF(query_cq_out, out, cq_context);
 	hw_status = MLX5_GET(cqc, cqc, status);
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "CQ");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "CQ");
 	if (err)
 		return err;
 
@@ -62,14 +62,22 @@ int mlx5e_reporter_cq_diagnose(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg)
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = devlink_fmsg_u32_pair_put(fmsg, "ci", mlx5_cqwq_get_ci(&cq->wq));
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", mlx5_cqwq_get_size(&cq->wq));
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
 	return 0;
 }
 
-int mlx5e_reporter_cq_common_diagnose(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg)
+int mlx5e_health_cq_common_diag_fmsg(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg)
 {
 	u8 cq_log_stride;
 	u32 cq_sz;
@@ -78,7 +86,7 @@ int mlx5e_reporter_cq_common_diagnose(struct mlx5e_cq *cq, struct devlink_fmsg *
 	cq_sz = mlx5_cqwq_get_size(&cq->wq);
 	cq_log_stride = mlx5_cqwq_get_log_stride_size(&cq->wq);
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "CQ");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "CQ");
 	if (err)
 		return err;
 
@@ -90,26 +98,48 @@ int mlx5e_reporter_cq_common_diagnose(struct mlx5e_cq *cq, struct devlink_fmsg *
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
 	return 0;
 }
 
-int mlx5e_health_create_reporters(struct mlx5e_priv *priv)
+int mlx5e_health_eq_diag_fmsg(struct mlx5_eq_comp *eq, struct devlink_fmsg *fmsg)
 {
 	int err;
 
-	err = mlx5e_reporter_tx_create(priv);
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "EQ");
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_rx_create(priv);
+	err = devlink_fmsg_u8_pair_put(fmsg, "eqn", eq->core.eqn);
 	if (err)
 		return err;
 
-	return 0;
+	err = devlink_fmsg_u32_pair_put(fmsg, "irqn", eq->core.irqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "vecidx", eq->core.vecidx);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "ci", eq->core.cons_index);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", eq->core.nent);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+void mlx5e_health_create_reporters(struct mlx5e_priv *priv)
+{
+	mlx5e_reporter_tx_create(priv);
+	mlx5e_reporter_rx_create(priv);
 }
 
 void mlx5e_health_destroy_reporters(struct mlx5e_priv *priv)
@@ -291,7 +321,7 @@ int mlx5e_health_queue_dump(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, lbl);
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, lbl);
 	if (err)
 		return err;
 
@@ -303,7 +333,7 @@ int mlx5e_health_queue_dump(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 

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

@@ -16,23 +16,25 @@ static inline bool cqe_syndrome_needs_recover(u8 syndrome)
 	       syndrome == MLX5_CQE_SYNDROME_WR_FLUSH_ERR;
 }
 
-int mlx5e_reporter_tx_create(struct mlx5e_priv *priv);
+void mlx5e_reporter_tx_create(struct mlx5e_priv *priv);
 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);
 
-int mlx5e_reporter_cq_diagnose(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg);
-int mlx5e_reporter_cq_common_diagnose(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg);
-int mlx5e_reporter_named_obj_nest_start(struct devlink_fmsg *fmsg, char *name);
-int mlx5e_reporter_named_obj_nest_end(struct devlink_fmsg *fmsg);
+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);
+int mlx5e_health_eq_diag_fmsg(struct mlx5_eq_comp *eq, struct devlink_fmsg *fmsg);
+int mlx5e_health_fmsg_named_obj_nest_start(struct devlink_fmsg *fmsg, char *name);
+int mlx5e_health_fmsg_named_obj_nest_end(struct devlink_fmsg *fmsg);
 
-int mlx5e_reporter_rx_create(struct mlx5e_priv *priv);
+void mlx5e_reporter_rx_create(struct mlx5e_priv *priv);
 void mlx5e_reporter_rx_destroy(struct mlx5e_priv *priv);
 void mlx5e_reporter_icosq_cqe_err(struct mlx5e_icosq *icosq);
 void mlx5e_reporter_rq_cqe_err(struct mlx5e_rq *rq);
 void mlx5e_reporter_rx_timeout(struct mlx5e_rq *rq);
 
 #define MLX5E_REPORTER_PER_Q_MAX_LEN 256
+#define MLX5E_REPORTER_FLUSH_TIMEOUT_MSEC 2000
 
 struct mlx5e_err_ctx {
 	int (*recover)(void *ctx);
@@ -46,7 +48,7 @@ int mlx5e_health_recover_channels(struct mlx5e_priv *priv);
 int mlx5e_health_report(struct mlx5e_priv *priv,
 			struct devlink_health_reporter *reporter, char *err_str,
 			struct mlx5e_err_ctx *err_ctx);
-int mlx5e_health_create_reporters(struct mlx5e_priv *priv);
+void mlx5e_health_create_reporters(struct mlx5e_priv *priv);
 void mlx5e_health_destroy_reporters(struct mlx5e_priv *priv);
 void mlx5e_health_channels_update(struct mlx5e_priv *priv);
 int mlx5e_health_rsc_fmsg_dump(struct mlx5e_priv *priv, struct mlx5_rsc_key *key,

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

@@ -3,6 +3,7 @@
 
 #include "health.h"
 #include "params.h"
+#include "txrx.h"
 
 static int mlx5e_query_rq_state(struct mlx5_core_dev *dev, u32 rqn, u8 *state)
 {
@@ -29,7 +30,8 @@ static int mlx5e_query_rq_state(struct mlx5_core_dev *dev, u32 rqn, u8 *state)
 
 static int mlx5e_wait_for_icosq_flush(struct mlx5e_icosq *icosq)
 {
-	unsigned long exp_time = jiffies + msecs_to_jiffies(2000);
+	unsigned long exp_time = jiffies +
+				 msecs_to_jiffies(MLX5E_REPORTER_FLUSH_TIMEOUT_MSEC);
 
 	while (time_before(jiffies, exp_time)) {
 		if (icosq->cc == icosq->pc)
@@ -123,25 +125,9 @@ static int mlx5e_rq_to_ready(struct mlx5e_rq *rq, int curr_state)
 
 static int mlx5e_rx_reporter_err_rq_cqe_recover(void *ctx)
 {
-	struct mlx5_core_dev *mdev;
-	struct net_device *dev;
-	struct mlx5e_rq *rq;
-	u8 state;
+	struct mlx5e_rq *rq = ctx;
 	int err;
 
-	rq = ctx;
-	mdev = rq->mdev;
-	dev = rq->netdev;
-	err = mlx5e_query_rq_state(mdev, rq->rqn, &state);
-	if (err) {
-		netdev_err(dev, "Failed to query RQ 0x%x state. err = %d\n",
-			   rq->rqn, err);
-		goto out;
-	}
-
-	if (state != MLX5_RQC_STATE_ERR)
-		goto out;
-
 	mlx5e_deactivate_rq(rq);
 	mlx5e_free_rx_descs(rq);
 
@@ -191,19 +177,71 @@ static int mlx5e_rx_reporter_recover(struct devlink_health_reporter *reporter,
 			 mlx5e_health_recover_channels(priv);
 }
 
+static int mlx5e_reporter_icosq_diagnose(struct mlx5e_icosq *icosq, u8 hw_state,
+					 struct devlink_fmsg *fmsg)
+{
+	int err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "ICOSQ");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "sqn", icosq->sqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "HW state", hw_state);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cc", icosq->cc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "pc", icosq->pc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "WQE size",
+					mlx5_wq_cyc_get_size(&icosq->wq));
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "CQ");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cqn", icosq->cq.mcq.cqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cc", icosq->cq.wq.cc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", mlx5_cqwq_get_size(&icosq->cq.wq));
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
 static int mlx5e_rx_reporter_build_diagnose_output(struct mlx5e_rq *rq,
 						   struct devlink_fmsg *fmsg)
 {
 	struct mlx5e_priv *priv = rq->channel->priv;
-	struct mlx5e_params *params;
 	struct mlx5e_icosq *icosq;
 	u8 icosq_hw_state;
+	u16 wqe_counter;
 	int wqes_sz;
 	u8 hw_state;
 	u16 wq_head;
 	int err;
 
-	params = &priv->channels.params;
 	icosq = &rq->channel->icosq;
 	err = mlx5e_query_rq_state(priv->mdev, rq->rqn, &hw_state);
 	if (err)
@@ -214,8 +252,8 @@ static int mlx5e_rx_reporter_build_diagnose_output(struct mlx5e_rq *rq,
 		return err;
 
 	wqes_sz = mlx5e_rqwq_get_cur_sz(rq);
-	wq_head = params->rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ ?
-		  rq->mpwqe.wq.head : mlx5_wq_cyc_get_head(&rq->wqe.wq);
+	wq_head = mlx5e_rqwq_get_head(rq);
+	wqe_counter = mlx5e_rqwq_get_wqe_counter(rq);
 
 	err = devlink_fmsg_obj_nest_start(fmsg);
 	if (err)
@@ -237,6 +275,10 @@ static int mlx5e_rx_reporter_build_diagnose_output(struct mlx5e_rq *rq,
 	if (err)
 		return err;
 
+	err = devlink_fmsg_u32_pair_put(fmsg, "WQE counter", wqe_counter);
+	if (err)
+		return err;
+
 	err = devlink_fmsg_u32_pair_put(fmsg, "posted WQEs", wqes_sz);
 	if (err)
 		return err;
@@ -245,11 +287,15 @@ static int mlx5e_rx_reporter_build_diagnose_output(struct mlx5e_rq *rq,
 	if (err)
 		return err;
 
-	err = devlink_fmsg_u8_pair_put(fmsg, "ICOSQ HW state", icosq_hw_state);
+	err = mlx5e_health_cq_diag_fmsg(&rq->cq, fmsg);
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_cq_diagnose(&rq->cq, fmsg);
+	err = mlx5e_health_eq_diag_fmsg(rq->cq.mcq.eq, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_reporter_icosq_diagnose(icosq, icosq_hw_state, fmsg);
 	if (err)
 		return err;
 
@@ -279,11 +325,11 @@ static int mlx5e_rx_reporter_diagnose(struct devlink_health_reporter *reporter,
 	rq_sz = mlx5e_rqwq_get_size(generic_rq);
 	rq_stride = BIT(mlx5e_mpwqe_get_log_stride_size(priv->mdev, params, NULL));
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "Common config");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "Common config");
 	if (err)
 		goto unlock;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "RQ");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "RQ");
 	if (err)
 		goto unlock;
 
@@ -299,15 +345,15 @@ static int mlx5e_rx_reporter_diagnose(struct devlink_health_reporter *reporter,
 	if (err)
 		goto unlock;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_cq_common_diag_fmsg(&generic_rq->cq, fmsg);
 	if (err)
 		goto unlock;
 
-	err = mlx5e_reporter_cq_common_diagnose(&generic_rq->cq, fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		goto unlock;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		goto unlock;
 
@@ -340,7 +386,7 @@ static int mlx5e_rx_reporter_dump_icosq(struct mlx5e_priv *priv, struct devlink_
 	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
 		return 0;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "SX Slice");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SX Slice");
 	if (err)
 		return err;
 
@@ -350,15 +396,15 @@ static int mlx5e_rx_reporter_dump_icosq(struct mlx5e_priv *priv, struct devlink_
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "ICOSQ");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "ICOSQ");
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "QPC");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "QPC");
 	if (err)
 		return err;
 
@@ -370,11 +416,11 @@ static int mlx5e_rx_reporter_dump_icosq(struct mlx5e_priv *priv, struct devlink_
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "send_buff");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "send_buff");
 	if (err)
 		return err;
 
@@ -385,11 +431,11 @@ static int mlx5e_rx_reporter_dump_icosq(struct mlx5e_priv *priv, struct devlink_
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	return mlx5e_reporter_named_obj_nest_end(fmsg);
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 }
 
 static int mlx5e_rx_reporter_dump_rq(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
@@ -402,7 +448,7 @@ static int mlx5e_rx_reporter_dump_rq(struct mlx5e_priv *priv, struct devlink_fms
 	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
 		return 0;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "RX Slice");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "RX Slice");
 	if (err)
 		return err;
 
@@ -412,15 +458,15 @@ static int mlx5e_rx_reporter_dump_rq(struct mlx5e_priv *priv, struct devlink_fms
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "RQ");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "RQ");
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "QPC");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "QPC");
 	if (err)
 		return err;
 
@@ -432,11 +478,11 @@ static int mlx5e_rx_reporter_dump_rq(struct mlx5e_priv *priv, struct devlink_fms
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "receive_buff");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "receive_buff");
 	if (err)
 		return err;
 
@@ -446,11 +492,11 @@ static int mlx5e_rx_reporter_dump_rq(struct mlx5e_priv *priv, struct devlink_fms
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	return mlx5e_reporter_named_obj_nest_end(fmsg);
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 }
 
 static int mlx5e_rx_reporter_dump_all_rqs(struct mlx5e_priv *priv,
@@ -462,7 +508,7 @@ static int mlx5e_rx_reporter_dump_all_rqs(struct mlx5e_priv *priv,
 	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
 		return 0;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "RX Slice");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "RX Slice");
 	if (err)
 		return err;
 
@@ -472,7 +518,7 @@ static int mlx5e_rx_reporter_dump_all_rqs(struct mlx5e_priv *priv,
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
@@ -563,7 +609,7 @@ static const struct devlink_health_reporter_ops mlx5_rx_reporter_ops = {
 
 #define MLX5E_REPORTER_RX_GRACEFUL_PERIOD 500
 
-int mlx5e_reporter_rx_create(struct mlx5e_priv *priv)
+void mlx5e_reporter_rx_create(struct mlx5e_priv *priv)
 {
 	struct devlink *devlink = priv_to_devlink(priv->mdev);
 	struct devlink_health_reporter *reporter;
@@ -575,10 +621,9 @@ int mlx5e_reporter_rx_create(struct mlx5e_priv *priv)
 	if (IS_ERR(reporter)) {
 		netdev_warn(priv->netdev, "Failed to create rx reporter, err = %ld\n",
 			    PTR_ERR(reporter));
-		return PTR_ERR(reporter);
+		return;
 	}
 	priv->rx_reporter = reporter;
-	return 0;
 }
 
 void mlx5e_reporter_rx_destroy(struct mlx5e_priv *priv)

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

@@ -5,7 +5,8 @@
 
 static int mlx5e_wait_for_sq_flush(struct mlx5e_txqsq *sq)
 {
-	unsigned long exp_time = jiffies + msecs_to_jiffies(2000);
+	unsigned long exp_time = jiffies +
+				 msecs_to_jiffies(MLX5E_REPORTER_FLUSH_TIMEOUT_MSEC);
 
 	while (time_before(jiffies, exp_time)) {
 		if (sq->cc == sq->pc)
@@ -82,17 +83,40 @@ static int mlx5e_tx_reporter_err_cqe_recover(void *ctx)
 	return err;
 }
 
+struct mlx5e_tx_timeout_ctx {
+	struct mlx5e_txqsq *sq;
+	signed int status;
+};
+
 static int mlx5e_tx_reporter_timeout_recover(void *ctx)
 {
+	struct mlx5e_tx_timeout_ctx *to_ctx;
+	struct mlx5e_priv *priv;
 	struct mlx5_eq_comp *eq;
 	struct mlx5e_txqsq *sq;
 	int err;
 
-	sq = ctx;
+	to_ctx = ctx;
+	sq = to_ctx->sq;
 	eq = sq->cq.mcq.eq;
+	priv = sq->channel->priv;
 	err = mlx5e_health_channel_eq_recover(eq, sq->channel);
-	if (err)
+	if (!err) {
+		to_ctx->status = 0; /* this sq recovered */
+		return err;
+	}
+
+	err = mlx5e_safe_reopen_channels(priv);
+	if (!err) {
+		to_ctx->status = 1; /* all channels recovered */
+		return err;
+	}
+
+	to_ctx->status = err;
 	clear_bit(MLX5E_SQ_STATE_ENABLED, &sq->state);
+	netdev_err(priv->netdev,
+		   "mlx5e_safe_reopen_channels failed recovering from a tx_timeout, err(%d).\n",
+		   err);
 
 	return err;
 }
@@ -165,7 +189,11 @@ mlx5e_tx_reporter_build_diagnose_output(struct devlink_fmsg *fmsg,
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_cq_diagnose(&sq->cq, fmsg);
+	err = mlx5e_health_cq_diag_fmsg(&sq->cq, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_eq_diag_fmsg(sq->cq.mcq.eq, fmsg);
 	if (err)
 		return err;
 
@@ -194,11 +222,11 @@ static int mlx5e_tx_reporter_diagnose(struct devlink_health_reporter *reporter,
 	sq_sz = mlx5_wq_cyc_get_size(&generic_sq->wq);
 	sq_stride = MLX5_SEND_WQE_BB;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "Common Config");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "Common Config");
 	if (err)
 		goto unlock;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "SQ");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SQ");
 	if (err)
 		goto unlock;
 
@@ -210,15 +238,15 @@ static int mlx5e_tx_reporter_diagnose(struct devlink_health_reporter *reporter,
 	if (err)
 		goto unlock;
 
-	err = mlx5e_reporter_cq_common_diagnose(&generic_sq->cq, fmsg);
+	err = mlx5e_health_cq_common_diag_fmsg(&generic_sq->cq, fmsg);
 	if (err)
 		goto unlock;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		goto unlock;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		goto unlock;
 
@@ -256,7 +284,7 @@ static int mlx5e_tx_reporter_dump_sq(struct mlx5e_priv *priv, struct devlink_fms
 	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
 		return 0;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "SX Slice");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SX Slice");
 	if (err)
 		return err;
 
@@ -266,15 +294,15 @@ static int mlx5e_tx_reporter_dump_sq(struct mlx5e_priv *priv, struct devlink_fms
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "SQ");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SQ");
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "QPC");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "QPC");
 	if (err)
 		return err;
 
@@ -286,11 +314,11 @@ static int mlx5e_tx_reporter_dump_sq(struct mlx5e_priv *priv, struct devlink_fms
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "send_buff");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "send_buff");
 	if (err)
 		return err;
 
@@ -300,11 +328,11 @@ static int mlx5e_tx_reporter_dump_sq(struct mlx5e_priv *priv, struct devlink_fms
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
-	return mlx5e_reporter_named_obj_nest_end(fmsg);
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 }
 
 static int mlx5e_tx_reporter_dump_all_sqs(struct mlx5e_priv *priv,
@@ -316,7 +344,7 @@ static int mlx5e_tx_reporter_dump_all_sqs(struct mlx5e_priv *priv,
 	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
 		return 0;
 
-	err = mlx5e_reporter_named_obj_nest_start(fmsg, "SX Slice");
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SX Slice");
 	if (err)
 		return err;
 
@@ -326,7 +354,7 @@ static int mlx5e_tx_reporter_dump_all_sqs(struct mlx5e_priv *priv,
 	if (err)
 		return err;
 
-	err = mlx5e_reporter_named_obj_nest_end(fmsg);
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
 	if (err)
 		return err;
 
@@ -384,9 +412,11 @@ int mlx5e_reporter_tx_timeout(struct mlx5e_txqsq *sq)
 {
 	struct mlx5e_priv *priv = sq->channel->priv;
 	char err_str[MLX5E_REPORTER_PER_Q_MAX_LEN];
+	struct mlx5e_tx_timeout_ctx to_ctx = {};
 	struct mlx5e_err_ctx err_ctx = {};
 
-	err_ctx.ctx = sq;
+	to_ctx.sq = sq;
+	err_ctx.ctx = &to_ctx;
 	err_ctx.recover = mlx5e_tx_reporter_timeout_recover;
 	err_ctx.dump = mlx5e_tx_reporter_dump_sq;
 	snprintf(err_str, sizeof(err_str),
@@ -394,7 +424,8 @@ int mlx5e_reporter_tx_timeout(struct mlx5e_txqsq *sq)
 		 sq->channel->ix, sq->sqn, sq->cq.mcq.cqn, sq->cc, sq->pc,
 		 jiffies_to_usecs(jiffies - sq->txq->trans_start));
 
-	return mlx5e_health_report(priv, priv->tx_reporter, err_str, &err_ctx);
+	mlx5e_health_report(priv, priv->tx_reporter, err_str, &err_ctx);
+	return to_ctx.status;
 }
 
 static const struct devlink_health_reporter_ops mlx5_tx_reporter_ops = {
@@ -406,7 +437,7 @@ static const struct devlink_health_reporter_ops mlx5_tx_reporter_ops = {
 
 #define MLX5_REPORTER_TX_GRACEFUL_PERIOD 500
 
-int mlx5e_reporter_tx_create(struct mlx5e_priv *priv)
+void mlx5e_reporter_tx_create(struct mlx5e_priv *priv)
 {
 	struct devlink_health_reporter *reporter;
 	struct mlx5_core_dev *mdev = priv->mdev;
@@ -421,10 +452,9 @@ int mlx5e_reporter_tx_create(struct mlx5e_priv *priv)
 		netdev_warn(priv->netdev,
 			    "Failed to create tx reporter, err = %ld\n",
 			    PTR_ERR(reporter));
-		return PTR_ERR(reporter);
+		return;
 	}
 	priv->tx_reporter = reporter;
-	return 0;
 }
 
 void mlx5e_reporter_tx_destroy(struct mlx5e_priv *priv)

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

@@ -268,7 +268,7 @@ static inline void mlx5e_rqwq_reset(struct mlx5e_rq *rq)
 	}
 }
 
-static inline void mlx5e_dump_error_cqe(struct mlx5e_cq *cq, u32 sqn,
+static inline void mlx5e_dump_error_cqe(struct mlx5e_cq *cq, u32 qn,
 					struct mlx5_err_cqe *err_cqe)
 {
 	struct mlx5_cqwq *wq = &cq->wq;
@@ -277,13 +277,53 @@ static inline void mlx5e_dump_error_cqe(struct mlx5e_cq *cq, u32 sqn,
 	ci = mlx5_cqwq_ctr2ix(wq, wq->cc - 1);
 
 	netdev_err(cq->channel->netdev,
-		   "Error cqe on cqn 0x%x, ci 0x%x, sqn 0x%x, opcode 0x%x, syndrome 0x%x, vendor syndrome 0x%x\n",
-		   cq->mcq.cqn, ci, sqn,
+		   "Error cqe on cqn 0x%x, ci 0x%x, qn 0x%x, opcode 0x%x, syndrome 0x%x, vendor syndrome 0x%x\n",
+		   cq->mcq.cqn, ci, qn,
 		   get_cqe_opcode((struct mlx5_cqe64 *)err_cqe),
 		   err_cqe->syndrome, err_cqe->vendor_err_synd);
 	mlx5_dump_err_cqe(cq->mdev, err_cqe);
 }
 
+static inline u32 mlx5e_rqwq_get_size(struct mlx5e_rq *rq)
+{
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return mlx5_wq_ll_get_size(&rq->mpwqe.wq);
+	default:
+		return mlx5_wq_cyc_get_size(&rq->wqe.wq);
+	}
+}
+
+static inline u32 mlx5e_rqwq_get_cur_sz(struct mlx5e_rq *rq)
+{
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return rq->mpwqe.wq.cur_sz;
+	default:
+		return rq->wqe.wq.cur_sz;
+	}
+}
+
+static inline u16 mlx5e_rqwq_get_head(struct mlx5e_rq *rq)
+{
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return mlx5_wq_ll_get_head(&rq->mpwqe.wq);
+	default:
+		return mlx5_wq_cyc_get_head(&rq->wqe.wq);
+	}
+}
+
+static inline u16 mlx5e_rqwq_get_wqe_counter(struct mlx5e_rq *rq)
+{
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return mlx5_wq_ll_get_counter(&rq->mpwqe.wq);
+	default:
+		return mlx5_wq_cyc_get_counter(&rq->wqe.wq);
+	}
+}
+
 /* SW parser related functions */
 
 struct mlx5e_swp_spec {

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

@@ -4367,8 +4367,6 @@ static void mlx5e_tx_timeout_work(struct work_struct *work)
 {
 	struct mlx5e_priv *priv = container_of(work, struct mlx5e_priv,
 					       tx_timeout_work);
-	bool report_failed = false;
-	int err;
 	int i;
 
 	rtnl_lock();
@@ -4386,18 +4384,10 @@ static void mlx5e_tx_timeout_work(struct work_struct *work)
 			continue;
 
 		if (mlx5e_reporter_tx_timeout(sq))
-			report_failed = true;
+		/* break if tried to reopened channels */
+			break;
 	}
 
-	if (!report_failed)
-		goto unlock;
-
-	err = mlx5e_safe_reopen_channels(priv);
-	if (err)
-		netdev_err(priv->netdev,
-			   "mlx5e_safe_reopen_channels failed recovering from a tx_timeout, err(%d).\n",
-			   err);
-
 unlock:
 	mutex_unlock(&priv->state_lock);
 	rtnl_unlock();

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

@@ -1187,8 +1187,10 @@ static void trigger_report(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
 	struct mlx5_err_cqe *err_cqe = (struct mlx5_err_cqe *)cqe;
 
 	if (cqe_syndrome_needs_recover(err_cqe->syndrome) &&
-	    !test_and_set_bit(MLX5E_RQ_STATE_RECOVERING, &rq->state))
+	    !test_and_set_bit(MLX5E_RQ_STATE_RECOVERING, &rq->state)) {
+		mlx5e_dump_error_cqe(&rq->cq, rq->rqn, err_cqe);
 		queue_work(rq->channel->priv->wq, &rq->recover_work);
+	}
 }
 
 void mlx5e_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)

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

@@ -172,6 +172,11 @@ static inline int mlx5_wq_cyc_cc_bigger(u16 cc1, u16 cc2)
 	return !equal && !smaller;
 }
 
+static inline u16 mlx5_wq_cyc_get_counter(struct mlx5_wq_cyc *wq)
+{
+	return wq->wqe_ctr;
+}
+
 static inline u32 mlx5_cqwq_get_size(struct mlx5_cqwq *wq)
 {
 	return wq->fbc.sz_m1 + 1;
@@ -290,4 +295,14 @@ static inline void mlx5_wq_ll_update_db_record(struct mlx5_wq_ll *wq)
 	*wq->db = cpu_to_be32(wq->wqe_ctr);
 }
 
+static inline u16 mlx5_wq_ll_get_head(struct mlx5_wq_ll *wq)
+{
+	return wq->head;
+}
+
+static inline u16 mlx5_wq_ll_get_counter(struct mlx5_wq_ll *wq)
+{
+	return wq->wqe_ctr;
+}
+
 #endif /* __MLX5_WQ_H__ */