From: Matan Barak <matanb@mellanox.com>
Date: Thu, 10 Aug 2017 15:29:12 +0300
Subject: net/mlx5: Fix creating a new FTE when an existing but full FTE exists
Patch-mainline: v4.15-rc1
Git-commit: b92af5a72ca982f0aa3df22f57a178aa5b0f4357
References: bsc#1103990 FATE#326006

Currently, when a flow steering rule is added, we look for a FTE with
an identical value. If we find a match, we try to merge the required
destinations with the existing ones. In a case where the existing
destination list is full, the code should return an error to its
consumer. However, the current code just tries to create another FTE.
Fixing that by returning an error in this special scenario.

Fixes: f478be79a22e ("net/mlx5: Add hash table for flow groups in flow table")
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -1449,7 +1449,7 @@ try_add_to_existing_fg(struct mlx5_flow_
 		       int dest_num)
 {
 	struct mlx5_flow_group *g;
-	struct mlx5_flow_handle *rule = ERR_PTR(-ENOENT);
+	struct mlx5_flow_handle *rule;
 	struct rhlist_head *tmp, *list;
 	struct match_list {
 		struct list_head	list;
@@ -1513,6 +1513,8 @@ try_add_to_existing_fg(struct mlx5_flow_
 		unlock_ref_node(&g->node);
 	}
 
+	rule = ERR_PTR(-ENOENT);
+
 free_list:
 	if (!list_empty(&match_head)) {
 		struct match_list *match_tmp;
@@ -1553,7 +1555,7 @@ _mlx5_add_flow_rules(struct mlx5_flow_ta
 
 	nested_lock_ref_node(&ft->node, FS_MUTEX_GRANDPARENT);
 	rule = try_add_to_existing_fg(ft, spec, flow_act, dest, dest_num);
-	if (!IS_ERR(rule))
+	if (!IS_ERR(rule) || PTR_ERR(rule) != -ENOENT)
 		goto unlock;
 
 	g = create_autogroup(ft, spec->match_criteria_enable,