net/sched: do not reset queues in graft operations

Following typical script is extremely disruptive,
because each graft operation calls dev_deactivate()
which resets all the queues of the device.

QPARAM="limit 100000 flow_limit 1000 buckets 4096"
TXQS=64
for ETH in eth1
do
 tc qd del dev $ETH root 2>/dev/null
 tc qd add dev $ETH root handle 1: mq
 for i in `seq 1 $TXQS`
 do
   slot=$( printf %x $(( i )) )
   tc qd add dev $ETH parent 1:$slot fq $QPARAM
 done
done

One can add "ip link set dev $ETH down/up" to reduce the disruption time:

QPARAM="limit 100000 flow_limit 1000 buckets 4096"
TXQS=64
for ETH in eth1
do
 ip link set dev $ETH down
 tc qd del dev $ETH root 2>/dev/null
 tc qd add dev $ETH root handle 1: mq
 for i in `seq 1 $TXQS`
 do
   slot=$( printf %x $(( i )) )
   tc qd add dev $ETH parent 1:$slot fq $QPARAM
 done
 ip link set dev $ETH up
done

Or we can add a @reset_needed flag to dev_deactivate() and
dev_deactivate_many().

This flag is set to true at device dismantle or linkwatch_do_dev(),
and to false for graft operations.

In the future, we might only stop one queue instead of the whole
device, ie call dev_deactivate_queue() instead of dev_deactivate().

I think the problem (quadratic behavior) was added in commit
2fb541c862c9 ("net: sch_generic: aviod concurrent reset and enqueue op
for lockless qdisc") but this does not look serious enough to deserve
risky backports.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yunsheng Lin <linyunsheng@huawei.com>
Reviewed-by: Jamal Hadi Salim <jhs@mojatatu.com>
Reviewed-by: Toke Høiland-Jørgensen <toke@redhat.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Link: https://patch.msgid.link/20260307163430.470644-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

8 files changed, 20 insertions, 16 deletions

diff --git a/net/core/dev.c b/net/core/dev.c
index 203dc36aaed5..6fc9350f0be8 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1756,7 +1756,7 @@ static void __dev_close_many(struct list_head *head)
 		smp_mb__after_atomic(); /* Commit netif_running(). */
 	}
 
-	dev_deactivate_many(head);
+	dev_deactivate_many(head, true);
 
 	list_for_each_entry(dev, head, close_list) {
 		const struct net_device_ops *ops = dev->netdev_ops;
diff --git a/net/core/link_watch.c b/net/core/link_watch.c
index 25c455c10a01..ff2c1d4538ef 100644
--- a/net/core/link_watch.c
+++ b/net/core/link_watch.c
@@ -181,7 +181,7 @@ static void linkwatch_do_dev(struct net_device *dev)
 		if (netif_carrier_ok(dev))
 			dev_activate(dev);
 		else
-			dev_deactivate(dev);
+			dev_deactivate(dev, true);
 
 		netif_state_change(dev);
 	}
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index cc43e3f7574f..c0bab092ea80 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -1120,7 +1120,7 @@ static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
 		}
 
 		if (dev->flags & IFF_UP)
-			dev_deactivate(dev);
+			dev_deactivate(dev, false);
 
 		qdisc_offload_graft_root(dev, new, old, extack);
 
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index 556e0d800316..d4fe907c4ad5 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -1370,11 +1370,12 @@ static bool some_qdisc_is_busy(struct net_device *dev)
 /**
  * 	dev_deactivate_many - deactivate transmissions on several devices
  * 	@head: list of devices to deactivate
+ *	@reset_needed: qdisc should be reset if true.
  *
  *	This function returns only when all outstanding transmissions
  *	have completed, unless all devices are in dismantle phase.
  */
-void dev_deactivate_many(struct list_head *head)
+void dev_deactivate_many(struct list_head *head, bool reset_needed)
 {
 	bool sync_needed = false;
 	struct net_device *dev;
@@ -1393,11 +1394,14 @@ void dev_deactivate_many(struct list_head *head)
 	if (sync_needed)
 		synchronize_net();
 
-	list_for_each_entry(dev, head, close_list) {
-		netdev_for_each_tx_queue(dev, dev_reset_queue, NULL);
+	if (reset_needed) {
+		list_for_each_entry(dev, head, close_list) {
+			netdev_for_each_tx_queue(dev, dev_reset_queue, NULL);
 
-		if (dev_ingress_queue(dev))
-			dev_reset_queue(dev, dev_ingress_queue(dev), NULL);
+			if (dev_ingress_queue(dev))
+				dev_reset_queue(dev, dev_ingress_queue(dev),
+						NULL);
+		}
 	}
 
 	/* Wait for outstanding qdisc_run calls. */
@@ -1412,12 +1416,12 @@ void dev_deactivate_many(struct list_head *head)
 	}
 }
 
-void dev_deactivate(struct net_device *dev)
+void dev_deactivate(struct net_device *dev, bool reset_needed)
 {
 	LIST_HEAD(single);
 
 	list_add(&dev->close_list, &single);
-	dev_deactivate_many(&single);
+	dev_deactivate_many(&single, reset_needed);
 	list_del(&single);
 }
 EXPORT_SYMBOL(dev_deactivate);
@@ -1473,7 +1477,7 @@ int dev_qdisc_change_tx_queue_len(struct net_device *dev)
 	int ret = 0;
 
 	if (up)
-		dev_deactivate(dev);
+		dev_deactivate(dev, false);
 
 	for (i = 0; i < dev->num_tx_queues; i++) {
 		ret = qdisc_change_tx_queue_len(dev, &dev->_tx[i]);
diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c
index cf6cd4ccfa20..eb12381795ce 100644
--- a/net/sched/sch_htb.c
+++ b/net/sched/sch_htb.c
@@ -1387,7 +1387,7 @@ htb_graft_helper(struct netdev_queue *dev_queue, struct Qdisc *new_q)
 	struct Qdisc *old_q;
 
 	if (dev->flags & IFF_UP)
-		dev_deactivate(dev);
+		dev_deactivate(dev, false);
 	old_q = dev_graft_qdisc(dev_queue, new_q);
 	if (new_q)
 		new_q->flags |= TCQ_F_ONETXQUEUE | TCQ_F_NOPARENT;
@@ -1421,7 +1421,7 @@ static void htb_offload_move_qdisc(struct Qdisc *sch, struct htb_class *cl_old,
 		struct Qdisc *qdisc;
 
 		if (dev->flags & IFF_UP)
-			dev_deactivate(dev);
+			dev_deactivate(dev, false);
 		qdisc = dev_graft_qdisc(queue_old, NULL);
 		WARN_ON(qdisc != cl_old->leaf.q);
 	}
diff --git a/net/sched/sch_mq.c b/net/sched/sch_mq.c
index 0ed199fa18f0..a0133a7b9d3b 100644
--- a/net/sched/sch_mq.c
+++ b/net/sched/sch_mq.c
@@ -201,7 +201,7 @@ static int mq_graft(struct Qdisc *sch, unsigned long cl, struct Qdisc *new,
 	struct net_device *dev = qdisc_dev(sch);
 
 	if (dev->flags & IFF_UP)
-		dev_deactivate(dev);
+		dev_deactivate(dev, false);
 
 	*old = dev_graft_qdisc(dev_queue, new);
 	if (new)
diff --git a/net/sched/sch_mqprio.c b/net/sched/sch_mqprio.c
index b83276409416..002add5ce9e0 100644
--- a/net/sched/sch_mqprio.c
+++ b/net/sched/sch_mqprio.c
@@ -469,7 +469,7 @@ static int mqprio_graft(struct Qdisc *sch, unsigned long cl, struct Qdisc *new,
 		return -EINVAL;
 
 	if (dev->flags & IFF_UP)
-		dev_deactivate(dev);
+		dev_deactivate(dev, false);
 
 	*old = dev_graft_qdisc(dev_queue, new);
 
diff --git a/net/sched/sch_taprio.c b/net/sched/sch_taprio.c
index f721c03514f6..8e3752811950 100644
--- a/net/sched/sch_taprio.c
+++ b/net/sched/sch_taprio.c
@@ -2184,7 +2184,7 @@ static int taprio_graft(struct Qdisc *sch, unsigned long cl,
 		return -EINVAL;
 
 	if (dev->flags & IFF_UP)
-		dev_deactivate(dev);
+		dev_deactivate(dev, false);
 
 	/* In offload mode, the child Qdisc is directly attached to the netdev
 	 * TX queue, and thus, we need to keep its refcount elevated in order