Re: [Patch net v2] kcm: close race conditions on sk_receive_queue

[Paolo Abeni <pabeni@redhat.com>]

Hello,
On Thu, 2022-11-03 at 11:46 -0700, Cong Wang wrote:
> @@ -1085,53 +1085,17 @@ static int kcm_sendmsg(struct socket *sock, struct msghdr *msg, size_t len)
>  	return err;
>  }
>  
> -static struct sk_buff *kcm_wait_data(struct sock *sk, int flags,
> -				     long timeo, int *err)
> -{
> -	struct sk_buff *skb;
> -
> -	while (!(skb = skb_peek(&sk->sk_receive_queue))) {
> -		if (sk->sk_err) {
> -			*err = sock_error(sk);
> -			return NULL;
> -		}
> -
> -		if (sock_flag(sk, SOCK_DONE))
> -			return NULL;

It looks like skb_recv_datagram() ignores the SOCK_DONE flag, so this
change could potentially miss some wait_data end coditions. On the flip
side I don't see any place where the SOCK_DONE flag is set for the kcm
socket, so this should be safe, but could you please document this in
the commit message?

[...]

> @@ -1187,11 +1147,7 @@ static ssize_t kcm_splice_read(struct socket *sock, loff_t *ppos,
>  
>  	/* Only support splice for SOCKSEQPACKET */
>  
> -	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
> -
> -	lock_sock(sk);
> -
> -	skb = kcm_wait_data(sk, flags, timeo, &err);
> +	skb = skb_recv_datagram(sk, flags, &err);
>  	if (!skb)
>  		goto err_out;
>  
> @@ -1219,13 +1175,11 @@ static ssize_t kcm_splice_read(struct socket *sock, loff_t *ppos,
>  	 * finish reading the message.
>  	 */
>  
> -	release_sock(sk);
> -
> +	skb_free_datagram(sk, skb);
>  	return copied;
>  
>  err_out:
> -	release_sock(sk);
> -
> +	skb_free_datagram(sk, skb);

We can reach here with skb == NULL and skb_free_datagram() ->
__kfree_skb() -> skb_release_all() does not deal correctly with NULL
skb, you need to check for skb explicitly here (or rearrange the error
paths in a suitable way).

Thanks!

Paolo