[PATCH] net/mlx5: break resource release forever loop

[PATCH] net/mlx5: break resource release forever loop

[Alexander Kozyrev]

There is a loop inside the flow_hw_resource_release() function
that tries to free all the template patterns and tables until they
are successfully released. But some of the tables may be still in use
in case of the ungraceful application termination. Which causes the
forever loop in the app on the exit. Don't wait for the tables release
and try them to free only once and proceed with the exit.

Fixes: d1559d66ed ("net/mlx5: add table management")
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
---
 drivers/net/mlx5/mlx5_flow_hw.c | 62 +++++++++++++++++++--------------
 1 file changed, 36 insertions(+), 26 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_flow_hw.c b/drivers/net/mlx5/mlx5_flow_hw.c
index 68c5a36bbb..c68cde14e1 100644
--- a/drivers/net/mlx5/mlx5_flow_hw.c
+++ b/drivers/net/mlx5/mlx5_flow_hw.c
@@ -5048,7 +5048,7 @@ flow_hw_table_destroy(struct rte_eth_dev *dev,
 		return rte_flow_error_set(error, EBUSY,
 				   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
 				   NULL,
-				   "table in use");
+				   "table is in use");
 	}
 	LIST_REMOVE(table, next);
 	for (i = 0; i < table->nb_item_templates; i++)
@@ -7342,7 +7342,7 @@ flow_hw_actions_template_destroy(struct rte_eth_dev *dev,
 		return rte_flow_error_set(error, EBUSY,
 				   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
 				   NULL,
-				   "action template in using");
+				   "action template is in use");
 	}
 	if (template->action_flags & flag)
 		mlx5_free_srh_flex_parser(dev);
@@ -7966,7 +7966,7 @@ flow_hw_pattern_template_destroy(struct rte_eth_dev *dev,
 		return rte_flow_error_set(error, EBUSY,
 				   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
 				   NULL,
-				   "item template in using");
+				   "item template is in use");
 	}
 	if (template->item_flags & (MLX5_FLOW_ITEM_OUTER_IPV6_ROUTING_EXT |
 				    MLX5_FLOW_ITEM_INNER_IPV6_ROUTING_EXT))
@@ -10767,10 +10767,10 @@ void
 flow_hw_resource_release(struct rte_eth_dev *dev)
 {
 	struct mlx5_priv *priv = dev->data->dev_private;
-	struct rte_flow_template_table *tbl;
-	struct rte_flow_pattern_template *it;
-	struct rte_flow_actions_template *at;
-	struct mlx5_flow_group *grp;
+	struct rte_flow_template_table *tbl, *temp_tbl;
+	struct rte_flow_pattern_template *it, *temp_it;
+	struct rte_flow_actions_template *at, *temp_at;
+	struct mlx5_flow_group *grp, *temp_grp;
 	uint32_t i;
 
 	if (!priv->dr_ctx)
@@ -10782,25 +10782,35 @@ flow_hw_resource_release(struct rte_eth_dev *dev)
 	flow_hw_cleanup_tx_repr_tagging(dev);
 	flow_hw_cleanup_ctrl_rx_tables(dev);
 	flow_hw_action_template_drop_release(dev);
-	while (!LIST_EMPTY(&priv->flow_hw_grp)) {
-		grp = LIST_FIRST(&priv->flow_hw_grp);
-		flow_hw_group_unset_miss_group(dev, grp, NULL);
-	}
-	while (!LIST_EMPTY(&priv->flow_hw_tbl_ongo)) {
-		tbl = LIST_FIRST(&priv->flow_hw_tbl_ongo);
-		flow_hw_table_destroy(dev, tbl, NULL);
-	}
-	while (!LIST_EMPTY(&priv->flow_hw_tbl)) {
-		tbl = LIST_FIRST(&priv->flow_hw_tbl);
-		flow_hw_table_destroy(dev, tbl, NULL);
-	}
-	while (!LIST_EMPTY(&priv->flow_hw_itt)) {
-		it = LIST_FIRST(&priv->flow_hw_itt);
-		flow_hw_pattern_template_destroy(dev, it, NULL);
-	}
-	while (!LIST_EMPTY(&priv->flow_hw_at)) {
-		at = LIST_FIRST(&priv->flow_hw_at);
-		flow_hw_actions_template_destroy(dev, at, NULL);
+	grp = LIST_FIRST(&priv->flow_hw_grp);
+	while (grp) {
+		temp_grp = LIST_NEXT(grp, next);
+		claim_zero(flow_hw_group_unset_miss_group(dev, grp, NULL));
+		grp = temp_grp;
+	}
+	tbl = LIST_FIRST(&priv->flow_hw_tbl_ongo);
+	while (tbl) {
+		temp_tbl = LIST_NEXT(tbl, next);
+		claim_zero(flow_hw_table_destroy(dev, tbl, NULL));
+		tbl = temp_tbl;
+	}
+	tbl = LIST_FIRST(&priv->flow_hw_tbl);
+	while (tbl) {
+		temp_tbl = LIST_NEXT(tbl, next);
+		claim_zero(flow_hw_table_destroy(dev, tbl, NULL));
+		tbl = temp_tbl;
+	}
+	it = LIST_FIRST(&priv->flow_hw_itt);
+	while (it) {
+		temp_it = LIST_NEXT(it, next);
+		claim_zero(flow_hw_pattern_template_destroy(dev, it, NULL));
+		it = temp_it;
+	}
+	at = LIST_FIRST(&priv->flow_hw_at);
+	while (at) {
+		temp_at = LIST_NEXT(at, next);
+		claim_zero(flow_hw_actions_template_destroy(dev, at, NULL));
+		at = temp_at;
 	}
 	for (i = 0; i < MLX5_HW_ACTION_FLAG_MAX; i++) {
 		if (priv->hw_drop[i])
-- 
2.18.2

Re: [PATCH] net/mlx5: break resource release forever loop

[Raslan Darawsheh]

Hi,

From: Alexander Kozyrev <akozyrev@nvidia.com>
Sent: Thursday, May 30, 2024 12:46 AM
To: dev@dpdk.org
Cc: stable@dpdk.org; Raslan Darawsheh; Slava Ovsiienko; Matan Azrad; Dariusz Sosnowski; Bing Zhao; Ori Kam; Suanming Mou
Subject: [PATCH] net/mlx5: break resource release forever loop

There is a loop inside the flow_hw_resource_release() function
that tries to free all the template patterns and tables until they
are successfully released. But some of the tables may be still in use
in case of the ungraceful application termination. Which causes the
forever loop in the app on the exit. Don't wait for the tables release
and try them to free only once and proceed with the exit.

Fixes: d1559d66ed ("net/mlx5: add table management")
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>


Patch applied to next-net-mlx,

Kindest regards,
Raslan Darawsheh