ipv4: add a sock pointer to ip_queue_xmit()

ip_queue_xmit() assumes the skb it has to transmit is attached to an
inet socket. Commit 31c70d5 ("l2tp: keep original skb ownership")
changed l2tp to not change skb ownership and thus broke this assumption.

One fix is to add a new 'struct sock *sk' parameter to ip_queue_xmit(),
so that we do not assume skb->sk points to the socket used by l2tp
tunnel.

Fixes: 31c70d5 ("l2tp: keep original skb ownership")
Reported-by: Zhan Jianyu <[email protected]>
Tested-by: Zhan Jianyu <[email protected]>
Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Author: edumazet

include/net/inet6_connection_sock.h

@@ -40,7 +40,7 @@ void inet6_csk_reqsk_queue_hash_add(struct sock *sk, struct request_sock *req,
 
 void inet6_csk_addr2sockaddr(struct sock *sk, struct sockaddr *uaddr);
 
-int inet6_csk_xmit(struct sk_buff *skb, struct flowi *fl);
+int inet6_csk_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
 
 struct dst_entry *inet6_csk_update_pmtu(struct sock *sk, u32 mtu);
 #endif /* _INET6_CONNECTION_SOCK_H */

include/net/inet_connection_sock.h

@@ -36,7 +36,7 @@ struct tcp_congestion_ops;
  * (i.e. things that depend on the address family)
  */
 struct inet_connection_sock_af_ops {
-	int	    (*queue_xmit)(struct sk_buff *skb, struct flowi *fl);
+	int	    (*queue_xmit)(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
 	void	    (*send_check)(struct sock *sk, struct sk_buff *skb);
 	int	    (*rebuild_header)(struct sock *sk);
 	void	    (*sk_rx_dst_set)(struct sock *sk, const struct sk_buff *skb);

include/net/ip.h

@@ -111,7 +111,7 @@ int ip_do_nat(struct sk_buff *skb);
 void ip_send_check(struct iphdr *ip);
 int __ip_local_out(struct sk_buff *skb);
 int ip_local_out(struct sk_buff *skb);
-int ip_queue_xmit(struct sk_buff *skb, struct flowi *fl);
+int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
 void ip_init(void);
 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
 		   int getfrag(void *from, char *to, int offset, int len,

net/dccp/output.c

@@ -138,7 +138,7 @@ static int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb)
 
 		DCCP_INC_STATS(DCCP_MIB_OUTSEGS);
 
-		err = icsk->icsk_af_ops->queue_xmit(skb, &inet->cork.fl);
+		err = icsk->icsk_af_ops->queue_xmit(sk, skb, &inet->cork.fl);
 		return net_xmit_eval(err);
 	}
 	return -ENOBUFS;

net/ipv4/ip_output.c

@@ -315,9 +315,9 @@ static void ip_copy_addrs(struct iphdr *iph, const struct flowi4 *fl4)
 	       sizeof(fl4->saddr) + sizeof(fl4->daddr));
 }
 
-int ip_queue_xmit(struct sk_buff *skb, struct flowi *fl)
+/* Note: skb->sk can be different from sk, in case of tunnels */
+int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl)
 {
-	struct sock *sk = skb->sk;
 	struct inet_sock *inet = inet_sk(sk);
 	struct ip_options_rcu *inet_opt;
 	struct flowi4 *fl4;
@@ -389,6 +389,7 @@ int ip_queue_xmit(struct sk_buff *skb, struct flowi *fl)
 	ip_select_ident_more(skb, &rt->dst, sk,
 			     (skb_shinfo(skb)->gso_segs ?: 1) - 1);
 
+	/* TODO : should we use skb->sk here instead of sk ? */
 	skb->priority = sk->sk_priority;
 	skb->mark = sk->sk_mark;
 

net/ipv4/tcp_output.c

@@ -981,7 +981,7 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
 		TCP_ADD_STATS(sock_net(sk), TCP_MIB_OUTSEGS,
 			      tcp_skb_pcount(skb));
 
-	err = icsk->icsk_af_ops->queue_xmit(skb, &inet->cork.fl);
+	err = icsk->icsk_af_ops->queue_xmit(sk, skb, &inet->cork.fl);
 	if (likely(err <= 0))
 		return err;
 

net/ipv6/inet6_connection_sock.c

@@ -224,9 +224,8 @@ static struct dst_entry *inet6_csk_route_socket(struct sock *sk,
 	return dst;
 }
 
-int inet6_csk_xmit(struct sk_buff *skb, struct flowi *fl_unused)
+int inet6_csk_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl_unused)
 {
-	struct sock *sk = skb->sk;
 	struct ipv6_pinfo *np = inet6_sk(sk);
 	struct flowi6 fl6;
 	struct dst_entry *dst;

net/l2tp/l2tp_core.c

@@ -1131,10 +1131,10 @@ static int l2tp_xmit_core(struct l2tp_session *session, struct sk_buff *skb,
 	skb->local_df = 1;
 #if IS_ENABLED(CONFIG_IPV6)
 	if (tunnel->sock->sk_family == PF_INET6 && !tunnel->v4mapped)
-		error = inet6_csk_xmit(skb, NULL);
+		error = inet6_csk_xmit(tunnel->sock, skb, NULL);
 	else
 #endif
-		error = ip_queue_xmit(skb, fl);
+		error = ip_queue_xmit(tunnel->sock, skb, fl);
 
 	/* Update stats */
 	if (error >= 0) {

net/l2tp/l2tp_ip.c

@@ -487,7 +487,7 @@ static int l2tp_ip_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *m
 
 xmit:
 	/* Queue the packet to IP for output */
-	rc = ip_queue_xmit(skb, &inet->cork.fl);
+	rc = ip_queue_xmit(sk, skb, &inet->cork.fl);
 	rcu_read_unlock();
 
 error:

net/sctp/protocol.c

@@ -957,7 +957,7 @@ static inline int sctp_v4_xmit(struct sk_buff *skb,
 
 	SCTP_INC_STATS(sock_net(&inet->sk), SCTP_MIB_OUTSCTPPACKS);
 
-	return ip_queue_xmit(skb, &transport->fl);
+	return ip_queue_xmit(&inet->sk, skb, &transport->fl);
 }
 
 static struct sctp_af sctp_af_inet;