net/mlx5e: Update neighbour 'used' state using HW flow rules counters

When IP tunnel encapsulation rules are offloaded, the kernel can't see
the traffic of the offloaded flow. The neighbour for the IP tunnel
destination of the offloaded flow can mistakenly become STALE and
deleted by the kernel since its 'used' value wasn't changed.

To make sure that a neighbour which is used by the HW won't become
STALE, we proactively update the neighbour 'used' value every
DELAY_PROBE_TIME period, when packets were matched and counted by the HW
for one of the tunnel encap flows related to this neighbour.

The periodic task that updates the used neighbours is scheduled when a
tunnel encap rule is successfully offloaded into HW and keeps re-scheduling
itself as long as the representor's neighbours list isn't empty.

Add, remove, lookup and status change operations done over the
representor's neighbours list or the neighbour hash entry encaps list
are all serialized by RTNL lock.

Signed-off-by: Hadar Hen Zion <hadarh@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

1 files changed, 52 insertions, 0 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
index 730de6b7e46e..af61b10b85bf 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
@@ -41,6 +41,7 @@
 #include "en.h"
 #include "en_rep.h"
 #include "en_tc.h"
+#include "fs_core.h"
 
 static const char mlx5e_rep_driver_name[] = "mlx5e_rep";
 
@@ -226,6 +227,51 @@ void mlx5e_remove_sqs_fwd_rules(struct mlx5e_priv *priv)
 	mlx5_eswitch_sqs2vport_stop(esw, rep);
 }
 
+static void mlx5e_rep_neigh_update_init_interval(struct mlx5e_rep_priv *rpriv)
+{
+#if IS_ENABLED(CONFIG_IPV6)
+	unsigned long ipv6_interval = NEIGH_VAR(&ipv6_stub->nd_tbl->parms,
+						DELAY_PROBE_TIME);
+#else
+	unsigned long ipv6_interval = ~0UL;
+#endif
+	unsigned long ipv4_interval = NEIGH_VAR(&arp_tbl.parms,
+						DELAY_PROBE_TIME);
+	struct net_device *netdev = rpriv->rep->netdev;
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	rpriv->neigh_update.min_interval = min_t(unsigned long, ipv6_interval, ipv4_interval);
+	mlx5_fc_update_sampling_interval(priv->mdev, rpriv->neigh_update.min_interval);
+}
+
+void mlx5e_rep_queue_neigh_stats_work(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5e_neigh_update_table *neigh_update = &rpriv->neigh_update;
+
+	mlx5_fc_queue_stats_work(priv->mdev,
+				 &neigh_update->neigh_stats_work,
+				 neigh_update->min_interval);
+}
+
+static void mlx5e_rep_neigh_stats_work(struct work_struct *work)
+{
+	struct mlx5e_rep_priv *rpriv = container_of(work, struct mlx5e_rep_priv,
+						    neigh_update.neigh_stats_work.work);
+	struct net_device *netdev = rpriv->rep->netdev;
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_neigh_hash_entry *nhe;
+
+	rtnl_lock();
+	if (!list_empty(&rpriv->neigh_update.neigh_list))
+		mlx5e_rep_queue_neigh_stats_work(priv);
+
+	list_for_each_entry(nhe, &rpriv->neigh_update.neigh_list, neigh_list)
+		mlx5e_tc_update_neigh_used_value(nhe);
+
+	rtnl_unlock();
+}
+
 static void mlx5e_rep_neigh_entry_hold(struct mlx5e_neigh_hash_entry *nhe)
 {
 	refcount_inc(&nhe->refcnt);
@@ -325,6 +371,7 @@ static int mlx5e_rep_netevent_event(struct notifier_block *nb,
 			return NOTIFY_DONE;
 
 		m_neigh.dev = n->dev;
+		m_neigh.family = n->ops->family;
 		memcpy(&m_neigh.dst_ip, n->primary_key, n->tbl->key_len);
 
 		/* We are in atomic context and can't take RTNL mutex, so use
@@ -378,6 +425,9 @@ static int mlx5e_rep_neigh_init(struct mlx5e_rep_priv *rpriv)
 
 	INIT_LIST_HEAD(&neigh_update->neigh_list);
 	spin_lock_init(&neigh_update->encap_lock);
+	INIT_DELAYED_WORK(&neigh_update->neigh_stats_work,
+			  mlx5e_rep_neigh_stats_work);
+	mlx5e_rep_neigh_update_init_interval(rpriv);
 
 	rpriv->neigh_update.netevent_nb.notifier_call = mlx5e_rep_netevent_event;
 	err = register_netevent_notifier(&rpriv->neigh_update.netevent_nb);
@@ -399,6 +449,8 @@ static void mlx5e_rep_neigh_cleanup(struct mlx5e_rep_priv *rpriv)
 
 	flush_workqueue(priv->wq); /* flush neigh update works */
 
+	cancel_delayed_work_sync(&rpriv->neigh_update.neigh_stats_work);
+
 	rhashtable_destroy(&neigh_update->neigh_ht);
 }