net/mlx5e: Wait for concurrent flow deletion during neigh/fib events

Function mlx5e_take_tmp_flow() skips flows with zero reference count. This
can cause syndrome 0x179e84 when the called from neigh or route update code
and the skipped flow is not removed from the hardware by the time
underlying encap/decap resource is deleted. Add new completion
'del_hw_done' that is completed when flow is unoffloaded. This is safe to
do because flow with reference count zero needs to be detached from
encap/decap entry before its memory is deallocated, which requires taking
the encap_tbl_lock mutex that is held by the event handlers code.

Fixes: 8914add ("net/mlx5e: Handle FIB events to update tunnel endpoint device")
Signed-off-by: Vlad Buslov <vladbu@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>

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

@@ -102,6 +102,7 @@ struct mlx5e_tc_flow {
 	refcount_t refcnt;
 	struct rcu_head rcu_head;
 	struct completion init_done;
+	struct completion del_hw_done;
 	int tunnel_id; /* the mapped tunnel id of this flow */
 	struct mlx5_flow_attr *attr;
 };

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

@@ -245,8 +245,14 @@ static void mlx5e_take_tmp_flow(struct mlx5e_tc_flow *flow,
 				struct list_head *flow_list,
 				int index)
 {
-	if (IS_ERR(mlx5e_flow_get(flow)))
+	if (IS_ERR(mlx5e_flow_get(flow))) {
+		/* Flow is being deleted concurrently. Wait for it to be
+		 * unoffloaded from hardware, otherwise deleting encap will
+		 * fail.
+		 */
+		wait_for_completion(&flow->del_hw_done);
 		return;
+	}
 	wait_for_completion(&flow->init_done);
 
 	flow->tmp_entry_index = index;

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

@@ -1600,6 +1600,7 @@ static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
 		else
 			mlx5e_tc_unoffload_fdb_rules(esw, flow, attr);
 	}
+	complete_all(&flow->del_hw_done);
 
 	if (mlx5_flow_has_geneve_opt(flow))
 		mlx5_geneve_tlv_option_del(priv->mdev->geneve);
@@ -4465,6 +4466,7 @@ mlx5e_alloc_flow(struct mlx5e_priv *priv, int attr_size,
 	INIT_LIST_HEAD(&flow->l3_to_l2_reformat);
 	refcount_set(&flow->refcnt, 1);
 	init_completion(&flow->init_done);
+	init_completion(&flow->del_hw_done);
 
 	*__flow = flow;
 	*__parse_attr = parse_attr;