net/mlx5e: Enhance TX timeout recovery

Upon a TX timeout handle, if the TX reporter was not able to recover
from the error, reopen the channels. If tried to reopen channels, do not
loop over TX queues for timeout.

With that, the reporters state and separation will better
expose the driver's state.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

2 files changed, 33 insertions, 17 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c
index 465c7cc8d909..826584380216 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c
@@ -83,17 +83,40 @@ out:
 	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)
-		clear_bit(MLX5E_SQ_STATE_ENABLED, &sq->state);
+	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;
 }
@@ -389,9 +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),
@@ -399,7 +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 = {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 046cfb0ea180..b04c8572adea 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/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();