[PATCH net-next v2] net/sched: don't use dynamic lockdep keys with clsact/ingress/noqueue

[PATCH net-next v2] net/sched: don't use dynamic lockdep keys with clsact/ingress/noqueue

[Davide Caratti]

Currently we are registering one dynamic lockdep key for each allocated
qdisc, to avoid false deadlock reports when mirred (or TC eBPF) redirects
packets to another device while the root lock is acquired [1].
Since dynamic keys are a limited resource, we can save them at least for
qdiscs that are not meant to acquire the root lock in the traffic path,
or to carry traffic at all, like:

 - clsact
 - ingress
 - noqueue

Don't register dynamic keys for the above schedulers, so that we hit
MAX_LOCKDEP_KEYS later in our tests.

[1] https://github.com/multipath-tcp/mptcp_net-next/issues/451

Changes in v2:
 - change ordering of spin_lock_init() vs. lockdep_register_key()
   (Jakub Kicinski)

Signed-off-by: Davide Caratti <dcaratti@redhat.com>
---
 include/net/pkt_sched.h | 24 ++++++++++++++++++++++++
 net/sched/sch_api.c     |  2 +-
 net/sched/sch_generic.c |  8 +++-----
 3 files changed, 28 insertions(+), 6 deletions(-)

diff --git a/include/net/pkt_sched.h b/include/net/pkt_sched.h
index e703c507d0da..18a419cd9d94 100644
--- a/include/net/pkt_sched.h
+++ b/include/net/pkt_sched.h
@@ -308,4 +308,28 @@ static inline unsigned int qdisc_peek_len(struct Qdisc *sch)
 	return len;
 }
 
+static inline void qdisc_lock_init(struct Qdisc *sch,
+				   const struct Qdisc_ops *ops)
+{
+	spin_lock_init(&sch->q.lock);
+
+	/* Skip dynamic keys if nesting is not possible */
+	if (ops->static_flags & TCQ_F_INGRESS ||
+	    ops == &noqueue_qdisc_ops)
+		return;
+
+	lockdep_register_key(&sch->root_lock_key);
+	lockdep_set_class(&sch->q.lock, &sch->root_lock_key);
+}
+
+static inline void qdisc_lock_uninit(struct Qdisc *sch,
+				     const struct Qdisc_ops *ops)
+{
+	if (ops->static_flags & TCQ_F_INGRESS ||
+	    ops == &noqueue_qdisc_ops)
+		return;
+
+	lockdep_unregister_key(&sch->root_lock_key);
+}
+
 #endif
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index f56b18c8aebf..443c116e8663 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -1353,7 +1353,7 @@ static struct Qdisc *qdisc_create(struct net_device *dev,
 		ops->destroy(sch);
 	qdisc_put_stab(rtnl_dereference(sch->stab));
 err_out3:
-	lockdep_unregister_key(&sch->root_lock_key);
+	qdisc_lock_uninit(sch, ops);
 	netdev_put(dev, &sch->dev_tracker);
 	qdisc_free(sch);
 err_out2:
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index 852e603c1755..98ffe64de51f 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -955,9 +955,7 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 	__skb_queue_head_init(&sch->gso_skb);
 	__skb_queue_head_init(&sch->skb_bad_txq);
 	gnet_stats_basic_sync_init(&sch->bstats);
-	lockdep_register_key(&sch->root_lock_key);
-	spin_lock_init(&sch->q.lock);
-	lockdep_set_class(&sch->q.lock, &sch->root_lock_key);
+	qdisc_lock_init(sch, ops);
 
 	if (ops->static_flags & TCQ_F_CPUSTATS) {
 		sch->cpu_bstats =
@@ -987,7 +985,7 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 
 	return sch;
 errout1:
-	lockdep_unregister_key(&sch->root_lock_key);
+	qdisc_lock_uninit(sch, ops);
 	kfree(sch);
 errout:
 	return ERR_PTR(err);
@@ -1076,7 +1074,7 @@ static void __qdisc_destroy(struct Qdisc *qdisc)
 	if (ops->destroy)
 		ops->destroy(qdisc);
 
-	lockdep_unregister_key(&qdisc->root_lock_key);
+	qdisc_lock_uninit(qdisc, ops);
 	bpf_module_put(ops, ops->owner);
 	netdev_put(dev, &qdisc->dev_tracker);
 
-- 
2.52.0

Re: [PATCH net-next v2] net/sched: don't use dynamic lockdep keys with clsact/ingress/noqueue

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net-next.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Wed,  4 Feb 2026 17:02:31 +0100 you wrote:
> Currently we are registering one dynamic lockdep key for each allocated
> qdisc, to avoid false deadlock reports when mirred (or TC eBPF) redirects
> packets to another device while the root lock is acquired [1].
> Since dynamic keys are a limited resource, we can save them at least for
> qdiscs that are not meant to acquire the root lock in the traffic path,
> or to carry traffic at all, like:
> 
> [...]

Here is the summary with links:
  - [net-next,v2] net/sched: don't use dynamic lockdep keys with clsact/ingress/noqueue
    https://git.kernel.org/netdev/net-next/c/a90f6dcefca6

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html