Received-SPF: pass (google.com: domain of linux-kernel+bounces-168061-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
To: linux-kernel@vger.kernel.org,
	netdev@vger.kernel.org
Cc: "David S. Miller" <davem@davemloft.net>,
	Boqun Feng <boqun.feng@gmail.com>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Eric Dumazet <edumazet@google.com>,
	Frederic Weisbecker <frederic@kernel.org>,
	Ingo Molnar <mingo@redhat.com>,
	Jakub Kicinski <kuba@kernel.org>,
	Paolo Abeni <pabeni@redhat.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Waiman Long <longman@redhat.com>,
	Will Deacon <will@kernel.org>,
	Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Subject: [PATCH net-next 10/15] dev: Use nested-BH locking for softnet_data.process_queue.
Date: Fri,  3 May 2024 20:25:14 +0200
Message-ID: <20240503182957.1042122-11-bigeasy@linutronix.de>
In-Reply-To: <20240503182957.1042122-1-bigeasy@linutronix.de>
References: <20240503182957.1042122-1-bigeasy@linutronix.de>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable

softnet_data::process_queue is a per-CPU variable and relies on disabled
BH for its locking. Without per-CPU locking in local_bh_disable() on
PREEMPT_RT this data structure requires explicit locking.

softnet_data::input_queue_head can be updated lockless. This is fine
because this value is only update CPU local by the local backlog_napi
thread.

Add a local_lock_t to softnet_data and use local_lock_nested_bh() for locki=
ng
of process_queue. This change adds only lockdep coverage and does not
alter the functional behaviour for !PREEMPT_RT.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 include/linux/netdevice.h |  1 +
 net/core/dev.c            | 12 +++++++++++-
 2 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index c551ec235f9af..9d19e6ace7cb7 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -3199,6 +3199,7 @@ static inline bool dev_has_header(const struct net_de=
vice *dev)
 struct softnet_data {
 	struct list_head	poll_list;
 	struct sk_buff_head	process_queue;
+	local_lock_t		process_queue_bh_lock;
=20
 	/* stats */
 	unsigned int		processed;
diff --git a/net/core/dev.c b/net/core/dev.c
index cf7b452ce0d74..1503883ce15a4 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -450,7 +450,9 @@ static RAW_NOTIFIER_HEAD(netdev_chain);
  *	queue in the local softnet handler.
  */
=20
-DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
+DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data) =3D {
+	.process_queue_bh_lock =3D INIT_LOCAL_LOCK(process_queue_bh_lock),
+};
 EXPORT_PER_CPU_SYMBOL(softnet_data);
=20
 /* Page_pool has a lockless array/stack to alloc/recycle pages.
@@ -5935,6 +5937,7 @@ static void flush_backlog(struct work_struct *work)
 	}
 	backlog_unlock_irq_enable(sd);
=20
+	local_lock_nested_bh(&softnet_data.process_queue_bh_lock);
 	skb_queue_walk_safe(&sd->process_queue, skb, tmp) {
 		if (skb->dev->reg_state =3D=3D NETREG_UNREGISTERING) {
 			__skb_unlink(skb, &sd->process_queue);
@@ -5942,6 +5945,7 @@ static void flush_backlog(struct work_struct *work)
 			rps_input_queue_head_incr(sd);
 		}
 	}
+	local_unlock_nested_bh(&softnet_data.process_queue_bh_lock);
 	local_bh_enable();
 }
=20
@@ -6063,7 +6067,9 @@ static int process_backlog(struct napi_struct *napi, =
int quota)
 	while (again) {
 		struct sk_buff *skb;
=20
+		local_lock_nested_bh(&softnet_data.process_queue_bh_lock);
 		while ((skb =3D __skb_dequeue(&sd->process_queue))) {
+			local_unlock_nested_bh(&softnet_data.process_queue_bh_lock);
 			rcu_read_lock();
 			__netif_receive_skb(skb);
 			rcu_read_unlock();
@@ -6072,7 +6078,9 @@ static int process_backlog(struct napi_struct *napi, =
int quota)
 				return work;
 			}
=20
+			local_lock_nested_bh(&softnet_data.process_queue_bh_lock);
 		}
+		local_unlock_nested_bh(&softnet_data.process_queue_bh_lock);
=20
 		backlog_lock_irq_disable(sd);
 		if (skb_queue_empty(&sd->input_pkt_queue)) {
@@ -6087,8 +6095,10 @@ static int process_backlog(struct napi_struct *napi,=
 int quota)
 			napi->state &=3D NAPIF_STATE_THREADED;
 			again =3D false;
 		} else {
+			local_lock_nested_bh(&softnet_data.process_queue_bh_lock);
 			skb_queue_splice_tail_init(&sd->input_pkt_queue,
 						   &sd->process_queue);
+			local_unlock_nested_bh(&softnet_data.process_queue_bh_lock);
 		}
 		backlog_unlock_irq_enable(sd);
 	}
--=20
2.43.0