linux-kernel - [PATCH 5.6 16/23] staging: wfx: annotate nested gc

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: <20200331085315.519262911@linuxfoundation.org>
Date:   Tue, 31 Mar 2020 10:59:28 +0200
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     linux-kernel@...r.kernel.org
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        stable@...r.kernel.org,
        Michał Mirosław <mirq-linux@...e.qmqm.pl>
Subject: [PATCH 5.6 16/23] staging: wfx: annotate nested gc_list vs tx queue locking

From: Michał Mirosław <mirq-linux@...e.qmqm.pl>

commit e2525a95cc0887c7dc0549cb5d0ac3e796e1d54c upstream.

Lockdep is complaining about recursive locking, because it can't make
a difference between locked skb_queues. Annotate nested locks and avoid
double bh_disable/enable.

[...]
insmod/815 is trying to acquire lock:
cb7d6418 (&(&list->lock)->rlock){+...}, at: wfx_tx_queues_clear+0xfc/0x198 [wfx]

but task is already holding lock:
cb7d61f4 (&(&list->lock)->rlock){+...}, at: wfx_tx_queues_clear+0xa0/0x198 [wfx]

[...]
Possible unsafe locking scenario:

      CPU0
      ----
 lock(&(&list->lock)->rlock);
 lock(&(&list->lock)->rlock);

Cc: stable@...r.kernel.org
Fixes: 9bca45f3d692 ("staging: wfx: allow to send 802.11 frames")
Signed-off-by: Michał Mirosław <mirq-linux@...e.qmqm.pl>
Link: https://lore.kernel.org/r/5e30397af95854b4a7deea073b730c00229f42ba.1581416843.git.mirq-linux@rere.qmqm.pl
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>

---
 drivers/staging/wfx/queue.c |   16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

--- a/drivers/staging/wfx/queue.c
+++ b/drivers/staging/wfx/queue.c
@@ -130,12 +130,12 @@ static void wfx_tx_queue_clear(struct wf
 	spin_lock_bh(&queue->queue.lock);
 	while ((item = __skb_dequeue(&queue->queue)) != NULL)
 		skb_queue_head(gc_list, item);
-	spin_lock_bh(&stats->pending.lock);
+	spin_lock_nested(&stats->pending.lock, 1);
 	for (i = 0; i < ARRAY_SIZE(stats->link_map_cache); ++i) {
 		stats->link_map_cache[i] -= queue->link_map_cache[i];
 		queue->link_map_cache[i] = 0;
 	}
-	spin_unlock_bh(&stats->pending.lock);
+	spin_unlock(&stats->pending.lock);
 	spin_unlock_bh(&queue->queue.lock);
 }
 
@@ -207,9 +207,9 @@ void wfx_tx_queue_put(struct wfx_dev *wd
 
 	++queue->link_map_cache[tx_priv->link_id];
 
-	spin_lock_bh(&stats->pending.lock);
+	spin_lock_nested(&stats->pending.lock, 1);
 	++stats->link_map_cache[tx_priv->link_id];
-	spin_unlock_bh(&stats->pending.lock);
+	spin_unlock(&stats->pending.lock);
 	spin_unlock_bh(&queue->queue.lock);
 }
 
@@ -237,11 +237,11 @@ static struct sk_buff *wfx_tx_queue_get(
 		__skb_unlink(skb, &queue->queue);
 		--queue->link_map_cache[tx_priv->link_id];
 
-		spin_lock_bh(&stats->pending.lock);
+		spin_lock_nested(&stats->pending.lock, 1);
 		__skb_queue_tail(&stats->pending, skb);
 		if (!--stats->link_map_cache[tx_priv->link_id])
 			wakeup_stats = true;
-		spin_unlock_bh(&stats->pending.lock);
+		spin_unlock(&stats->pending.lock);
 	}
 	spin_unlock_bh(&queue->queue.lock);
 	if (wakeup_stats)
@@ -259,10 +259,10 @@ int wfx_pending_requeue(struct wfx_dev *
 	spin_lock_bh(&queue->queue.lock);
 	++queue->link_map_cache[tx_priv->link_id];
 
-	spin_lock_bh(&stats->pending.lock);
+	spin_lock_nested(&stats->pending.lock, 1);
 	++stats->link_map_cache[tx_priv->link_id];
 	__skb_unlink(skb, &stats->pending);
-	spin_unlock_bh(&stats->pending.lock);
+	spin_unlock(&stats->pending.lock);
 	__skb_queue_tail(&queue->queue, skb);
 	spin_unlock_bh(&queue->queue.lock);
 	return 0;