[v3] net/mlx5: fix external Rx and Tx queues access

Message ID 20251030091138.223868-1-getelson@nvidia.com (mailing list archive)
State Awaiting Upstream
Delegated to: Raslan Darawsheh
Headers
Series [v3] net/mlx5: fix external Rx and Tx queues access |

Checks

Context Check Description
ci/checkpatch success coding style OK
ci/loongarch-compilation success Compilation OK
ci/loongarch-unit-testing success Unit Testing PASS
ci/Intel-compilation success Compilation OK
ci/intel-Functional success Functional PASS
ci/github-robot: build success github build: passed
ci/github-robot-post success github post: success
ci/aws-unit-testing success Unit Testing PASS
ci/intel-Testing success Testing PASS

Commit Message

Gregory Etelson Oct. 30, 2025, 9:11 a.m. UTC
mlx5_ext_rxq_get() and mlx5_ext_txq_get() functions did not return
NULL value if query index was not referencing external queue.

As a result, calling functions did not expect the NULL on return.

External Rx queue:

- In mlx5_ext_rxq_get() remove assert and return NULL if a queue index
  does not point to a valid external queue.

- In mlx5_ext_rxq_verify() validate that probed queue index references
  a valid extern queue.

External Tx queue:

-  In mlx5_ext_txq_get() remove assert and return NULL if a queue index
   does not point to a valid external queue.

- In mlx5_ext_txq_verify() validate that probed queue index references
  a valid extern queue.

Fixes: 311b17e669ab ("net/mlx5: support queue/RSS actions for external Rx queue")

Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
---
v2: fixed MinGW compiler warnings.
v3: fix return value.
---
 drivers/net/mlx5/mlx5_devx.c |  5 +++++
 drivers/net/mlx5/mlx5_flow.h |  2 ++
 drivers/net/mlx5/mlx5_rxq.c  | 17 ++++++++++-------
 drivers/net/mlx5/mlx5_txq.c  | 10 +++++-----
 4 files changed, 22 insertions(+), 12 deletions(-)
  

Comments

Raslan Darawsheh Nov. 6, 2025, 7:21 a.m. UTC | #1
Hi,


On 30/10/2025 11:11 AM, Gregory Etelson wrote:
> mlx5_ext_rxq_get() and mlx5_ext_txq_get() functions did not return
> NULL value if query index was not referencing external queue.
> 
> As a result, calling functions did not expect the NULL on return.
> 
> External Rx queue:
> 
> - In mlx5_ext_rxq_get() remove assert and return NULL if a queue index
>    does not point to a valid external queue.
> 
> - In mlx5_ext_rxq_verify() validate that probed queue index references
>    a valid extern queue.
> 
> External Tx queue:
> 
> -  In mlx5_ext_txq_get() remove assert and return NULL if a queue index
>     does not point to a valid external queue.
> 
> - In mlx5_ext_txq_verify() validate that probed queue index references
>    a valid extern queue.
> 
> Fixes: 311b17e669ab ("net/mlx5: support queue/RSS actions for external Rx queue")
> 
> Signed-off-by: Gregory Etelson <getelson@nvidia.com>
> Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
> ---
> v2: fixed MinGW compiler warnings.
> v3: fix return value.

Patch applied to next-net-mlx,

Kindest regards
Raslan Darawsheh
  

Patch

diff --git a/drivers/net/mlx5/mlx5_devx.c b/drivers/net/mlx5/mlx5_devx.c
index 673c9f3902..523b53d713 100644
--- a/drivers/net/mlx5/mlx5_devx.c
+++ b/drivers/net/mlx5/mlx5_devx.c
@@ -761,6 +761,11 @@  mlx5_devx_ind_table_create_rqt_attr(struct rte_eth_dev *dev,
 			struct mlx5_external_q *ext_rxq =
 					mlx5_ext_rxq_get(dev, queues[i]);
 
+			if (ext_rxq == NULL) {
+				rte_errno = EINVAL;
+				mlx5_free(rqt_attr);
+				return NULL;
+			}
 			rqt_attr->rq_list[i] = ext_rxq->hw_id;
 		} else {
 			struct mlx5_rxq_priv *rxq =
diff --git a/drivers/net/mlx5/mlx5_flow.h b/drivers/net/mlx5/mlx5_flow.h
index ff61706054..5b2e17a2a9 100644
--- a/drivers/net/mlx5/mlx5_flow.h
+++ b/drivers/net/mlx5/mlx5_flow.h
@@ -2039,6 +2039,8 @@  flow_hw_get_sqn(struct rte_eth_dev *dev, uint16_t tx_queue, uint32_t *sqn)
 	}
 	if (mlx5_is_external_txq(dev, tx_queue)) {
 		ext_txq = mlx5_ext_txq_get(dev, tx_queue);
+		if (ext_txq == NULL)
+			return -EINVAL;
 		*sqn = ext_txq->hw_id;
 		return 0;
 	}
diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index 1425886a22..677e8491d3 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -2212,7 +2212,8 @@  mlx5_ext_rxq_ref(struct rte_eth_dev *dev, uint16_t idx)
 {
 	struct mlx5_external_q *rxq = mlx5_ext_rxq_get(dev, idx);
 
-	rte_atomic_fetch_add_explicit(&rxq->refcnt, 1, rte_memory_order_relaxed);
+	if (rxq != NULL)
+		rte_atomic_fetch_add_explicit(&rxq->refcnt, 1, rte_memory_order_relaxed);
 	return rxq;
 }
 
@@ -2232,7 +2233,9 @@  mlx5_ext_rxq_deref(struct rte_eth_dev *dev, uint16_t idx)
 {
 	struct mlx5_external_q *rxq = mlx5_ext_rxq_get(dev, idx);
 
-	return rte_atomic_fetch_sub_explicit(&rxq->refcnt, 1, rte_memory_order_relaxed) - 1;
+	return rxq != NULL ?
+		rte_atomic_fetch_sub_explicit(&rxq->refcnt, 1, rte_memory_order_relaxed) - 1 :
+		UINT32_MAX;
 }
 
 /**
@@ -2251,8 +2254,8 @@  mlx5_ext_rxq_get(struct rte_eth_dev *dev, uint16_t idx)
 {
 	struct mlx5_priv *priv = dev->data->dev_private;
 
-	MLX5_ASSERT(mlx5_is_external_rxq(dev, idx));
-	return &priv->ext_rxqs[idx - RTE_PMD_MLX5_EXTERNAL_RX_QUEUE_ID_MIN];
+	return mlx5_is_external_rxq(dev, idx) ?
+		&priv->ext_rxqs[idx -  RTE_PMD_MLX5_EXTERNAL_RX_QUEUE_ID_MIN] : NULL;
 }
 
 /**
@@ -2415,7 +2418,6 @@  int
 mlx5_ext_rxq_verify(struct rte_eth_dev *dev)
 {
 	struct mlx5_priv *priv = dev->data->dev_private;
-	struct mlx5_external_q *rxq;
 	uint32_t i;
 	int ret = 0;
 
@@ -2423,8 +2425,9 @@  mlx5_ext_rxq_verify(struct rte_eth_dev *dev)
 		return 0;
 
 	for (i = RTE_PMD_MLX5_EXTERNAL_RX_QUEUE_ID_MIN; i <= UINT16_MAX ; ++i) {
-		rxq = mlx5_ext_rxq_get(dev, i);
-		if (rxq->refcnt < 2)
+		struct mlx5_external_q *rxq = mlx5_ext_rxq_get(dev, i);
+
+		if (rxq == NULL || rxq->refcnt < 2)
 			continue;
 		DRV_LOG(DEBUG, "Port %u external RxQ %u still referenced.",
 			dev->data->port_id, i);
diff --git a/drivers/net/mlx5/mlx5_txq.c b/drivers/net/mlx5/mlx5_txq.c
index b090d8274d..2aa2475a8a 100644
--- a/drivers/net/mlx5/mlx5_txq.c
+++ b/drivers/net/mlx5/mlx5_txq.c
@@ -1281,8 +1281,8 @@  mlx5_ext_txq_get(struct rte_eth_dev *dev, uint16_t idx)
 {
 	struct mlx5_priv *priv = dev->data->dev_private;
 
-	MLX5_ASSERT(mlx5_is_external_txq(dev, idx));
-	return &priv->ext_txqs[idx - MLX5_EXTERNAL_TX_QUEUE_ID_MIN];
+	return mlx5_is_external_txq(dev, idx) ?
+		&priv->ext_txqs[idx - MLX5_EXTERNAL_TX_QUEUE_ID_MIN] : NULL;
 }
 
 /**
@@ -1298,7 +1298,6 @@  int
 mlx5_ext_txq_verify(struct rte_eth_dev *dev)
 {
 	struct mlx5_priv *priv = dev->data->dev_private;
-	struct mlx5_external_q *txq;
 	uint32_t i;
 	int ret = 0;
 
@@ -1306,8 +1305,9 @@  mlx5_ext_txq_verify(struct rte_eth_dev *dev)
 		return 0;
 
 	for (i = MLX5_EXTERNAL_TX_QUEUE_ID_MIN; i <= UINT16_MAX ; ++i) {
-		txq = mlx5_ext_txq_get(dev, i);
-		if (txq->refcnt < 2)
+		struct mlx5_external_q *txq = mlx5_ext_txq_get(dev, i);
+
+		if (txq == NULL || txq->refcnt < 2)
 			continue;
 		DRV_LOG(DEBUG, "Port %u external TxQ %u still referenced.",
 			dev->data->port_id, i);