Received-SPF: pass (google.com: domain of linux-wireless+bounces-8332-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
From: Baochen Qiang <quic_bqiang@quicinc.com>
To: <ath11k@lists.infradead.org>
CC: <linux-wireless@vger.kernel.org>, <quic_bqiang@quicinc.com>
Subject: [PATCH] wifi: ath11k: fix RCU documentation in ath11k_mac_op_ipv6_changed()
Date: Fri, 31 May 2024 10:24:11 +0800
Message-ID: <20240531022411.6543-1-quic_bqiang@quicinc.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Type: text/plain

Current documentation on RCU in ath11k_mac_op_ipv6_changed() says:

	/* Note: read_lock_bh() calls rcu_read_lock() */
	read_lock_bh(&idev->lock);

This is wrong because without enabling CONFIG_PREEMPT_RT
rcu_read_lock() is not called by read_lock_bh(). The reason
why current code works even in a CONFIG_PREEMPT_RT=n kernel
is because atomic_notifier_call_chain() already does that for
us, see:

	int atomic_notifier_call_chain()
	{
		...
		rcu_read_lock();
		ret = notifier_call_chain(&nh->head, val, v, -1, NULL);
		rcu_read_unlock();
		...
	}

and backtrace:

	ath11k_mac_op_ipv6_changed
	ieee80211_ifa6_changed
	notifier_call_chain
	atomic_notifier_call_chain

So update the comment to make it correct.

This is found during code review, compile tested only.

Fixes: feafe59c8975 ("wifi: ath11k: use RCU when accessing struct inet6_dev::ac_list")
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
---
 drivers/net/wireless/ath/ath11k/mac.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/net/wireless/ath/ath11k/mac.c b/drivers/net/wireless/ath/ath11k/mac.c
index a1800c75d32b..29af015a16f9 100644
--- a/drivers/net/wireless/ath/ath11k/mac.c
+++ b/drivers/net/wireless/ath/ath11k/mac.c
@@ -9020,7 +9020,12 @@ static void ath11k_mac_op_ipv6_changed(struct ieee80211_hw *hw,
 	offload = &arvif->arp_ns_offload;
 	count = 0;
 
-	/* Note: read_lock_bh() calls rcu_read_lock() */
+	/* The _ipv6_changed() is called with RCU lock already held in
+	 * atomic_notifier_call_chain(), so we don't need to call
+	 * rcu_read_lock() again here. But note that with CONFIG_PREEMPT_RT
+	 * enabled, read_lock_bh() also calls rcu_read_lock(). This is OK
+	 * because RCU read critical section is allowed to get nested.
+	 */
 	read_lock_bh(&idev->lock);
 
 	memset(offload->ipv6_addr, 0, sizeof(offload->ipv6_addr));

base-commit: 6e7a5c6d5e38b93f9cc3289d66a597b9a4ca0403
-- 
2.25.1