Skip to contentSkip to conversation

net/sfc: fix an Rx queue double release possibility

PATCH In review Andrew Rybchenko 2 participants 3 messages Sep 14, 2018 – Sep 21, 2018 Network drivers

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 net/sfc: fix an Rx queue double release possibility

From: Igor Romanov <[email protected]>

There are two function that call sfc_rx_qfini():
sfc_rx_fini_queues() and sfc_rx_queue_release(). But only sfc_rx_queue_release() sets rx_queues pointer of the device data to NULL. It may lead to the scenario in which a queue is destroyed by sfc_rx_fini_queues() and after the queue is attempted to be destroyed again by sfc_rx_queue_release().

Move NULL assignment to sfc_rx_qfini().

Fixes ce35b05c635e ("net/sfc: implement Rx queue setup release operations")
---
 drivers/net/sfc/sfc_ethdev.c | 2 --
 drivers/net/sfc/sfc_rx.c     | 1 +
 2 files changed, 1 insertion(+), 2 deletions(-)
drivers/net/sfc/sfc_ethdev.c +0 -2
index 435bde67f..23778c9f3 100644
@@ -444,8 +444,6 @@ sfc_rx_queue_release(void *queue)
 
 	sfc_log_init(sa, "RxQ=%u", sw_index);
 
-	sa->eth_dev->data->rx_queues[sw_index] = NULL;
-
 	sfc_rx_qfini(sa, sw_index);
 
 	sfc_adapter_unlock(sa);
drivers/net/sfc/sfc_rx.c +1 -0
index c6321d174..5bd6bb9be 100644
@@ -1104,6 +1104,7 @@ sfc_rx_qfini(struct sfc_adapter *sa, unsigned int sw_index)
 	struct sfc_rxq *rxq;
 
 	SFC_ASSERT(sw_index < sa->rxq_count);
+	sa->eth_dev->data->rx_queues[sw_index] = NULL;
 
 	rxq_info = &sa->rxq_info[sw_index];
 
Signature · 1 line
2.17.1
#1 lore raw

PATCH 2/2 net/sfc: fix a Tx queue double release possibility

From: Igor Romanov <[email protected]>

There are two function that call sfc_tx_qfini():
sfc_tx_fini_queues() and sfc_tx_queue_release(). But only sfc_tx_queue_release() sets tx_queues pointer of the device data to NULL. It may lead to the scenario in which a queue is destroyed by sfc_tx_fini_queues() and after the queue is attempted to be destroyed again by sfc_tx_queue_release().

Move NULL assignment to sfc_tx_qfini().

Fixes b1b7ad933b39 ("net/sfc: set up and release Tx queues")
---
 drivers/net/sfc/sfc_ethdev.c | 3 ---
 drivers/net/sfc/sfc_tx.c     | 2 ++
 2 files changed, 2 insertions(+), 3 deletions(-)
drivers/net/sfc/sfc_ethdev.c +0 -3
index 23778c9f3..93bfecc4e 100644
@@ -498,9 +498,6 @@ sfc_tx_queue_release(void *queue)
 
 	sfc_adapter_lock(sa);
 
-	SFC_ASSERT(sw_index < sa->eth_dev->data->nb_tx_queues);
-	sa->eth_dev->data->tx_queues[sw_index] = NULL;
-
 	sfc_tx_qfini(sa, sw_index);
 
 	sfc_adapter_unlock(sa);
drivers/net/sfc/sfc_tx.c +2 -0
index 8af08b37c..12665d813 100644
@@ -233,6 +233,8 @@ sfc_tx_qfini(struct sfc_adapter *sa, unsigned int sw_index)
 	sfc_log_init(sa, "TxQ = %u", sw_index);
 
 	SFC_ASSERT(sw_index < sa->txq_count);
+	sa->eth_dev->data->tx_queues[sw_index] = NULL;
+
 	txq_info = &sa->txq_info[sw_index];
 
 	txq = txq_info->txq;
Signature · 1 line
2.17.1
#2 lore raw
In reply to Andrew Rybchenko
Quoted 9 lines · Andrew Rybchenko

From: Igor Romanov <[email protected]>

There are two function that call sfc_rx_qfini():
sfc_rx_fini_queues() and sfc_rx_queue_release(). But only sfc_rx_queue_release() sets rx_queues pointer of the device data to NULL. It may lead to the scenario in which a queue is destroyed by sfc_rx_fini_queues() and after the queue is attempted to be destroyed again by sfc_rx_queue_release().

Quoted 7 lines

Move NULL assignment to sfc_rx_qfini().

Fixes ce35b05c635e ("net/sfc: implement Rx queue setup release operations")

Series applied to dpdk-next-net/master, thanks.