sock: sock_dequeue_err_skb() needs hard irq safety

Non NAPI drivers can call skb_tstamp_tx() and then sock_queue_err_skb()
from hard IRQ context.

Therefore, sock_dequeue_err_skb() needs to block hard irq or
corruptions or hangs can happen.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Fixes: 364a9e9 ("sock: deduplicate errqueue dequeue")
Fixes: cb820f8 ("net: Provide a generic socket error queue delivery method for Tx time stamps.")
Signed-off-by: David S. Miller <davem@davemloft.net>

net/core/skbuff.c

@@ -3621,13 +3621,14 @@ struct sk_buff *sock_dequeue_err_skb(struct sock *sk)
 {
 	struct sk_buff_head *q = &sk->sk_error_queue;
 	struct sk_buff *skb, *skb_next;
+	unsigned long flags;
 	int err = 0;
 
-	spin_lock_bh(&q->lock);
+	spin_lock_irqsave(&q->lock, flags);
 	skb = __skb_dequeue(q);
 	if (skb && (skb_next = skb_peek(q)))
 		err = SKB_EXT_ERR(skb_next)->ee.ee_errno;
-	spin_unlock_bh(&q->lock);
+	spin_unlock_irqrestore(&q->lock, flags);
 
 	sk->sk_err = err;
 	if (err)