linux-kernel - [PATCH net-next 09/24] dev: Use the RPS lock for softnet_data::input_pkt_queue on PREEMPT

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20231215171020.687342-10-bigeasy@linutronix.de>
Date: Fri, 15 Dec 2023 18:07:28 +0100
From: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
To: linux-kernel@...r.kernel.org,
	netdev@...r.kernel.org
Cc: "David S. Miller" <davem@...emloft.net>,
	Boqun Feng <boqun.feng@...il.com>,
	Daniel Borkmann <daniel@...earbox.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Frederic Weisbecker <frederic@...nel.org>,
	Ingo Molnar <mingo@...hat.com>,
	Jakub Kicinski <kuba@...nel.org>,
	Paolo Abeni <pabeni@...hat.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	Waiman Long <longman@...hat.com>,
	Will Deacon <will@...nel.org>,
	Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Subject: [PATCH net-next 09/24] dev: Use the RPS lock for softnet_data::input_pkt_queue on PREEMPT_RT.

There is no additional locking in rps_lock_*() on PREEMPT_RT in the
!CONFIG_RPS case. The reasoning is that local_bh_disable() provides the
required synchronization (everywhere).
All interrupts are threaded and so the required CPU local
synchronization is provided by the per-CPU lock in local_bh_disable()
which is always performed for the forced-threaded interrupts.
Needless to say that the softirq users (NAPI) have BH disabled while
being invoked as a softirq callback.

Without locking in local_bh_disable() on PREEMPT_RT the
softnet_data::input_pkt_queue data structure requires explicit locking
in the !CONFIG_RPS case.

Disabling interrupts for CPU-local locking is undesired because it will
impose restrictions within the locked sections like not allowing to free
a skb.

Utilise the locking which is present in the CONFIG_RPS case for
serialisation on PREEMPT_RT even without CONFIG_RPS enabled.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
---
 net/core/dev.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index c879246be48d8..09232080843ee 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -220,34 +220,34 @@ static inline struct hlist_head *dev_index_hash(struct net *net, int ifindex)
 static inline void rps_lock_irqsave(struct softnet_data *sd,
 				    unsigned long *flags)
 {
-	if (IS_ENABLED(CONFIG_RPS))
+	if (IS_ENABLED(CONFIG_RPS) || IS_ENABLED(CONFIG_PREEMPT_RT))
 		spin_lock_irqsave(&sd->input_pkt_queue.lock, *flags);
-	else if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+	else
 		local_irq_save(*flags);
 }
 
 static inline void rps_lock_irq_disable(struct softnet_data *sd)
 {
-	if (IS_ENABLED(CONFIG_RPS))
+	if (IS_ENABLED(CONFIG_RPS) || IS_ENABLED(CONFIG_PREEMPT_RT))
 		spin_lock_irq(&sd->input_pkt_queue.lock);
-	else if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+	else
 		local_irq_disable();
 }
 
 static inline void rps_unlock_irq_restore(struct softnet_data *sd,
 					  unsigned long *flags)
 {
-	if (IS_ENABLED(CONFIG_RPS))
+	if (IS_ENABLED(CONFIG_RPS) || IS_ENABLED(CONFIG_PREEMPT_RT))
 		spin_unlock_irqrestore(&sd->input_pkt_queue.lock, *flags);
-	else if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+	else
 		local_irq_restore(*flags);
 }
 
 static inline void rps_unlock_irq_enable(struct softnet_data *sd)
 {
-	if (IS_ENABLED(CONFIG_RPS))
+	if (IS_ENABLED(CONFIG_RPS) || IS_ENABLED(CONFIG_PREEMPT_RT))
 		spin_unlock_irq(&sd->input_pkt_queue.lock);
-	else if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+	else
 		local_irq_enable();
 }
 
-- 
2.43.0