ipv6: introduce tcp_v6_iif()

Commit 971f10eca186 ("tcp: better TCP_SKB_CB layout to reduce cache line
misses") added a regression for SO_BINDTODEVICE on IPv6.

This is because we still use inet6_iif() which expects that IP6 control
block is still at the beginning of skb->cb[]

This patch adds tcp_v6_iif() helper and uses it where necessary.

Because __inet6_lookup_skb() is used by TCP and DCCP, we add an iif
parameter to it.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Fixes: 971f10eca186 ("tcp: better TCP_SKB_CB layout to reduce cache line misses")
Acked-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

5 files changed, 30 insertions, 15 deletions

diff --git a/include/net/inet6_hashtables.h b/include/net/inet6_hashtables.h
index ae06135..d1d2728 100644
--- a/include/net/inet6_hashtables.h
+++ b/include/net/inet6_hashtables.h
@@ -80,7 +80,8 @@ static inline struct sock *__inet6_lookup(struct net *net,
 static inline struct sock *__inet6_lookup_skb(struct inet_hashinfo *hashinfo,
 					      struct sk_buff *skb,
 					      const __be16 sport,
-					      const __be16 dport)
+					      const __be16 dport,
+					      int iif)
 {
 	struct sock *sk = skb_steal_sock(skb);
 
@@ -90,7 +91,7 @@ static inline struct sock *__inet6_lookup_skb(struct inet_hashinfo *hashinfo,
 	return __inet6_lookup(dev_net(skb_dst(skb)->dev), hashinfo,
 			      &ipv6_hdr(skb)->saddr, sport,
 			      &ipv6_hdr(skb)->daddr, ntohs(dport),
-			      inet6_iif(skb));
+			      iif);
 }
 
 struct sock *inet6_lookup(struct net *net, struct inet_hashinfo *hashinfo,
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 3a4bbbf..c9766f8 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -729,6 +729,15 @@ struct tcp_skb_cb {
 
 #define TCP_SKB_CB(__skb)	((struct tcp_skb_cb *)&((__skb)->cb[0]))
 
+
+/* This is the variant of inet6_iif() that must be used by TCP,
+ * as TCP moves IP6CB into a different location in skb->cb[]
+ */
+static inline int tcp_v6_iif(const struct sk_buff *skb)
+{
+	return TCP_SKB_CB(skb)->header.h6.iif;
+}
+
 /* Due to TSO, an SKB can be composed of multiple actual
  * packets.  To keep these tracked properly, we use this.
  */
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index ad2acfe..6bcaa33 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -757,7 +757,8 @@ static int dccp_v6_rcv(struct sk_buff *skb)
 	/* Step 2:
 	 *	Look up flow ID in table and get corresponding socket */
 	sk = __inet6_lookup_skb(&dccp_hashinfo, skb,
-			        dh->dccph_sport, dh->dccph_dport);
+			        dh->dccph_sport, dh->dccph_dport,
+				inet6_iif(skb));
 	/*
 	 * Step 2:
 	 *	If no socket ...
diff --git a/net/ipv6/syncookies.c b/net/ipv6/syncookies.c
index 9a2838e..2a86a0f 100644
--- a/net/ipv6/syncookies.c
+++ b/net/ipv6/syncookies.c
@@ -214,7 +214,7 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
 	/* So that link locals have meaning */
 	if (!sk->sk_bound_dev_if &&
 	    ipv6_addr_type(&ireq->ir_v6_rmt_addr) & IPV6_ADDR_LINKLOCAL)
-		ireq->ir_iif = inet6_iif(skb);
+		ireq->ir_iif = tcp_v6_iif(skb);
 
 	ireq->ir_mark = inet_request_mark(sk, skb);
 
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index cf2e45a..8314955 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -424,6 +424,7 @@ static void tcp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
 		if (sock_owned_by_user(sk))
 			goto out;
 
+		/* Note : We use inet6_iif() here, not tcp_v6_iif() */
 		req = inet6_csk_search_req(sk, &prev, th->dest, &hdr->daddr,
 					   &hdr->saddr, inet6_iif(skb));
 		if (!req)
@@ -738,7 +739,7 @@ static void tcp_v6_init_req(struct request_sock *req, struct sock *sk,
 	/* So that link locals have meaning */
 	if (!sk->sk_bound_dev_if &&
 	    ipv6_addr_type(&ireq->ir_v6_rmt_addr) & IPV6_ADDR_LINKLOCAL)
-		ireq->ir_iif = inet6_iif(skb);
+		ireq->ir_iif = tcp_v6_iif(skb);
 
 	if (!TCP_SKB_CB(skb)->tcp_tw_isn &&
 	    (ipv6_opt_accepted(sk, skb, &TCP_SKB_CB(skb)->header.h6) ||
@@ -860,7 +861,7 @@ static void tcp_v6_send_response(struct sk_buff *skb, u32 seq, u32 ack, u32 win,
 
 	fl6.flowi6_proto = IPPROTO_TCP;
 	if (rt6_need_strict(&fl6.daddr) && !oif)
-		fl6.flowi6_oif = inet6_iif(skb);
+		fl6.flowi6_oif = tcp_v6_iif(skb);
 	else
 		fl6.flowi6_oif = oif;
 	fl6.flowi6_mark = IP6_REPLY_MARK(net, skb->mark);
@@ -918,7 +919,7 @@ static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb)
 		sk1 = inet6_lookup_listener(dev_net(skb_dst(skb)->dev),
 					   &tcp_hashinfo, &ipv6h->saddr,
 					   th->source, &ipv6h->daddr,
-					   ntohs(th->source), inet6_iif(skb));
+					   ntohs(th->source), tcp_v6_iif(skb));
 		if (!sk1)
 			return;
 
@@ -1000,13 +1001,14 @@ static struct sock *tcp_v6_hnd_req(struct sock *sk, struct sk_buff *skb)
 	/* Find possible connection requests. */
 	req = inet6_csk_search_req(sk, &prev, th->source,
 				   &ipv6_hdr(skb)->saddr,
-				   &ipv6_hdr(skb)->daddr, inet6_iif(skb));
+				   &ipv6_hdr(skb)->daddr, tcp_v6_iif(skb));
 	if (req)
 		return tcp_check_req(sk, skb, req, prev, false);
 
 	nsk = __inet6_lookup_established(sock_net(sk), &tcp_hashinfo,
-			&ipv6_hdr(skb)->saddr, th->source,
-			&ipv6_hdr(skb)->daddr, ntohs(th->dest), inet6_iif(skb));
+					 &ipv6_hdr(skb)->saddr, th->source,
+					 &ipv6_hdr(skb)->daddr, ntohs(th->dest),
+					 tcp_v6_iif(skb));
 
 	if (nsk) {
 		if (nsk->sk_state != TCP_TIME_WAIT) {
@@ -1090,7 +1092,7 @@ static struct sock *tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 		newnp->ipv6_fl_list = NULL;
 		newnp->pktoptions  = NULL;
 		newnp->opt	   = NULL;
-		newnp->mcast_oif   = inet6_iif(skb);
+		newnp->mcast_oif   = tcp_v6_iif(skb);
 		newnp->mcast_hops  = ipv6_hdr(skb)->hop_limit;
 		newnp->rcv_flowinfo = ip6_flowinfo(ipv6_hdr(skb));
 		if (np->repflow)
@@ -1174,7 +1176,7 @@ static struct sock *tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 			skb_set_owner_r(newnp->pktoptions, newsk);
 	}
 	newnp->opt	  = NULL;
-	newnp->mcast_oif  = inet6_iif(skb);
+	newnp->mcast_oif  = tcp_v6_iif(skb);
 	newnp->mcast_hops = ipv6_hdr(skb)->hop_limit;
 	newnp->rcv_flowinfo = ip6_flowinfo(ipv6_hdr(skb));
 	if (np->repflow)
@@ -1360,7 +1362,7 @@ ipv6_pktoptions:
 	if (TCP_SKB_CB(opt_skb)->end_seq == tp->rcv_nxt &&
 	    !((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN))) {
 		if (np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo)
-			np->mcast_oif = inet6_iif(opt_skb);
+			np->mcast_oif = tcp_v6_iif(opt_skb);
 		if (np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim)
 			np->mcast_hops = ipv6_hdr(opt_skb)->hop_limit;
 		if (np->rxopt.bits.rxflow || np->rxopt.bits.rxtclass)
@@ -1427,7 +1429,8 @@ static int tcp_v6_rcv(struct sk_buff *skb)
 	TCP_SKB_CB(skb)->ip_dsfield = ipv6_get_dsfield(hdr);
 	TCP_SKB_CB(skb)->sacked = 0;
 
-	sk = __inet6_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
+	sk = __inet6_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest,
+				tcp_v6_iif(skb));
 	if (!sk)
 		goto no_tcp_socket;
 
@@ -1514,7 +1517,7 @@ do_time_wait:
 		sk2 = inet6_lookup_listener(dev_net(skb->dev), &tcp_hashinfo,
 					    &ipv6_hdr(skb)->saddr, th->source,
 					    &ipv6_hdr(skb)->daddr,
-					    ntohs(th->dest), inet6_iif(skb));
+					    ntohs(th->dest), tcp_v6_iif(skb));
 		if (sk2 != NULL) {
 			struct inet_timewait_sock *tw = inet_twsk(sk);
 			inet_twsk_deschedule(tw, &tcp_death_row);
@@ -1553,6 +1556,7 @@ static void tcp_v6_early_demux(struct sk_buff *skb)
 	if (th->doff < sizeof(struct tcphdr) / 4)
 		return;
 
+	/* Note : We use inet6_iif() here, not tcp_v6_iif() */
 	sk = __inet6_lookup_established(dev_net(skb->dev), &tcp_hashinfo,
 					&hdr->saddr, th->source,
 					&hdr->daddr, ntohs(th->dest),