Skip to contentSkip to conversation

net: airoha: Fix skb->priority underflow in airoha_dev_select_queue()

PATCH Superseded Wayen Yan 3 participants 4 messages Jun 16 – Jun 18

Subsystems (4)

v3 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Wayen Yan

author
#0 lore raw

PATCH v3 1/1 net: airoha: Fix skb->priority underflow in airoha_dev_select_queue()

In airoha_dev_select_queue(), the expression:

  queue = (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES;

implicitly converts to unsigned arithmetic: when skb->priority is 0 (the default for unclassified traffic), (0u - 1u) wraps to UINT_MAX, and UINT_MAX % 8 = 7, routing default best-effort packets to the highest-priority QoS queue. This causes QoS inversion where the majority of traffic on a PON gateway starves actual high-priority flows (VoIP, gaming, etc.).

Fix by guarding the subtraction: when priority is 0, map to queue 0 (lowest priority), otherwise apply the original (priority - 1) % 8 mapping.

Fixes 2b288b81560b ("net: airoha: Introduce ndo_select_queue callback")
---
 drivers/net/ethernet/airoha/airoha_eth.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/airoha/airoha_eth.c +1 -1
index 31cdb11cd7..d476ef83c3 100644
@@ -1933,7 +1933,7 @@ static u16 airoha_dev_select_queue(struct net_device *dev, struct sk_buff *skb,
 	 */
 	channel = netdev_uses_dsa(dev) ? skb_get_queue_mapping(skb) : port->id;
 	channel = channel % AIROHA_NUM_QOS_CHANNELS;
-	queue = (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES; /* QoS queue */
+	queue = skb->priority ? (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES : 0;
 	queue = channel * AIROHA_NUM_QOS_QUEUES + queue;
 
 	return queue < dev->num_tx_queues ? queue : 0;
Signature · 1 line
2.51.0
maintainer
#1 lore raw
Quoted 28 lines · On Sun, 14 Jun 2026 07:30:54 +0800 Wayen Yan wrote:

In airoha_dev_select_queue(), the expression:

  queue = (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES;

implicitly converts to unsigned arithmetic: when skb->priority is 0 (the default for unclassified traffic), (0u - 1u) wraps to UINT_MAX, and UINT_MAX % 8 = 7, routing default best-effort packets to the highest-priority QoS queue. This causes QoS inversion where the majority of traffic on a PON gateway starves actual high-priority flows (VoIP, gaming, etc.).

Fix by guarding the subtraction: when priority is 0, map to queue 0 (lowest priority), otherwise apply the original (priority - 1) % 8 mapping.

Fixes 2b288b81560b ("net: airoha: Introduce ndo_select_queue callback")
---
 drivers/net/ethernet/airoha/airoha_eth.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/airoha/airoha_eth.c +0 -0
index 31cdb11cd7..d476ef83c3 100644

drivers/net/ethernet/airoha/airoha_eth.c static u16 airoha_dev_select_queue(struct net_device *dev, struct sk_buff *skb, · line 1936

Quoted 5 lines
drivers/net/ethernet/airoha/airoha_eth.c +1 -1
@@ -1933,7 +1933,7 @@ static u16 airoha_dev_select_queue(struct net_device *dev, struct sk_buff *skb,
 	 */
 	channel = netdev_uses_dsa(dev) ? skb_get_queue_mapping(skb) : port->id;
 	channel = channel % AIROHA_NUM_QOS_CHANNELS;
-	queue = (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES; /* QoS queue */
+	queue = skb->priority ? (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES : 0;

Hi Lorenzo, is there a reason we're subtracting 1 here in the first place? Could be just me, but may be worth adding a comment here.

Intuitively if we are "narrowing" 16 prios to 8 queues it'd make most sense to group the adjacent ones -- divide by two.

Please respin with some sort of an explanation..

Quoted 3 lines
drivers/net/ethernet/airoha/airoha_eth.c +0 -0
 	queue = channel * AIROHA_NUM_QOS_QUEUES + queue;
 
 	return queue < dev->num_tx_queues ? queue : 0;
Signature · 1 line
pw-bot: cr
maintainer Acked-by
#2 lore raw
Quoted 38 lines · On Sun, 14 Jun 2026 07:30:54 +0800 Wayen Yan wrote:
Quoted 28 lines

In airoha_dev_select_queue(), the expression:

  queue = (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES;

implicitly converts to unsigned arithmetic: when skb->priority is 0 (the default for unclassified traffic), (0u - 1u) wraps to UINT_MAX, and UINT_MAX % 8 = 7, routing default best-effort packets to the highest-priority QoS queue. This causes QoS inversion where the majority of traffic on a PON gateway starves actual high-priority flows (VoIP, gaming, etc.).

Fix by guarding the subtraction: when priority is 0, map to queue 0 (lowest priority), otherwise apply the original (priority - 1) % 8 mapping.

Fixes 2b288b81560b ("net: airoha: Introduce ndo_select_queue callback")
---
 drivers/net/ethernet/airoha/airoha_eth.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/airoha/airoha_eth.c +0 -0
index 31cdb11cd7..d476ef83c3 100644

drivers/net/ethernet/airoha/airoha_eth.c static u16 airoha_dev_select_queue(struct net_device *dev, struct sk_buff *skb, · line 1936

Quoted 5 lines
drivers/net/ethernet/airoha/airoha_eth.c +1 -1
@@ -1933,7 +1933,7 @@ static u16 airoha_dev_select_queue(struct net_device *dev, struct sk_buff *skb,
 	 */
 	channel = netdev_uses_dsa(dev) ? skb_get_queue_mapping(skb) : port->id;
 	channel = channel % AIROHA_NUM_QOS_CHANNELS;
-	queue = (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES; /* QoS queue */
+	queue = skb->priority ? (skb->priority - 1) % AIROHA_NUM_QOS_QUEUES : 0;

Hi Lorenzo, is there a reason we're subtracting 1 here in the first place? Could be just me, but may be worth adding a comment here.

Quoted 4 lines

Intuitively if we are "narrowing" 16 prios to 8 queues it'd make most sense to group the adjacent ones -- divide by two.

Please respin with some sort of an explanation..

IIRC this is a leftover of the ETS offload support. I agree it is righ to just do:

	queue = skb->priority % AIROHA_NUM_QOS_QUEUES; /* QoS queue */

@Wayen: can you please respin fixing the issue? Please add even my Acked-by:

Regards,
Lorenzo

Quoted 6 lines
Quoted 4 lines
drivers/net/ethernet/airoha/airoha_eth.c +0 -0
 	queue = channel * AIROHA_NUM_QOS_QUEUES + queue;
 
 	return queue < dev->num_tx_queues ? queue : 0;
Signature · 1 line
pw-bot: cr
author
#3 lore raw
In reply to Jakub Kicinski
Quoted 4 lines · Jakub Kicinski

Hi Lorenzo, is there a reason we're subtracting 1 here in the first place? Could be just me, but may be worth adding a comment here.

Please respin with some sort of an explanation..

Hi Jakub,

The (priority - 1) mapping predates my involvement — I only addressed the underflow bug when skb->priority is 0, where the unsigned subtraction wraps and routes best-effort packets to the highest-priority queue.

Lorenzo, could you clarify the intended priority-to-queue mapping so I can add a proper comment in the respin?

Regards,
Wayen