Re: [PATCH v2 net] af_unix: Update unix_sk(sk)->oob_skb under sk_receive_queue lock.

[Michal Luczaj <mhal@rbox.co>]

On 5/10/24 11:39, Kuniyuki Iwashima wrote:
> @@ -2655,6 +2661,8 @@ static struct sk_buff *manage_oob(struct sk_buff *skb, struct sock *sk,
>  		consume_skb(skb);
>  		skb = NULL;
>  	} else {
> +		spin_lock(&sk->sk_receive_queue.lock);
> +
>  		if (skb == u->oob_skb) {
>  			if (copied) {
>  				skb = NULL;
> @@ -2666,13 +2674,15 @@ static struct sk_buff *manage_oob(struct sk_buff *skb, struct sock *sk,
>  			} else if (flags & MSG_PEEK) {
>  				skb = NULL;
>  			} else {
> -				skb_unlink(skb, &sk->sk_receive_queue);
> +				__skb_unlink(skb, &sk->sk_receive_queue);
>  				WRITE_ONCE(u->oob_skb, NULL);
>  				if (!WARN_ON_ONCE(skb_unref(skb)))
>  					kfree_skb(skb);
>  				skb = skb_peek(&sk->sk_receive_queue);
>  			}
>  		}
> +
> +		spin_unlock(&sk->sk_receive_queue.lock);
>  	}
>  	return skb;
>  }

Now it is
  
  spin_lock(&sk->sk_receive_queue.lock)
  kfree_skb
    unix_destruct_scm
      unix_notinflight
        spin_lock(&unix_gc_lock)

I.e. sk_receive_queue.lock -> unix_gc_lock, inversion of what unix_gc() does.
But that's benign, right?

thanks,
Michal