ipv6: update skb->csum when CE mark is propagated

[ Upstream commit 34ae6a1aa0540f0f781dd265366036355fdc8930 ]

When a tunnel decapsulates the outer header, it has to comply
with RFC 6080 and eventually propagate CE mark into inner header.

It turns out IP6_ECN_set_ce() does not correctly update skb->csum
for CHECKSUM_COMPLETE packets, triggering infamous "hw csum failure"
messages and stack traces.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 3.2:
 - Adjust context
 - Add skb argument to other callers of IP6_ECN_set_ce()]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

5 files changed, 20 insertions, 7 deletions

diff --git a/include/net/inet_ecn.h b/include/net/inet_ecn.h
index 2fa14691869c..18f2d104b6be 100644
--- a/include/net/inet_ecn.h
+++ b/include/net/inet_ecn.h
@@ -109,11 +109,24 @@ static inline void ipv4_copy_dscp(unsigned int dscp, struct iphdr *inner)
 
 struct ipv6hdr;
 
-static inline int IP6_ECN_set_ce(struct ipv6hdr *iph)
+/* Note:
+ * IP_ECN_set_ce() has to tweak IPV4 checksum when setting CE,
+ * meaning both changes have no effect on skb->csum if/when CHECKSUM_COMPLETE
+ * In IPv6 case, no checksum compensates the change in IPv6 header,
+ * so we have to update skb->csum.
+ */
+static inline int IP6_ECN_set_ce(struct sk_buff *skb, struct ipv6hdr *iph)
 {
+	__be32 from, to;
+
 	if (INET_ECN_is_not_ect(ipv6_get_dsfield(iph)))
 		return 0;
-	*(__be32*)iph |= htonl(INET_ECN_CE << 20);
+
+	from = *(__be32 *)iph;
+	to = from | htonl(INET_ECN_CE << 20);
+	*(__be32 *)iph = to;
+	if (skb->ip_summed == CHECKSUM_COMPLETE)
+		skb->csum = csum_add(csum_sub(skb->csum, from), to);
 	return 1;
 }
 
@@ -138,7 +151,7 @@ static inline int INET_ECN_set_ce(struct sk_buff *skb)
 
 	case cpu_to_be16(ETH_P_IPV6):
 		if (skb->network_header + sizeof(struct ipv6hdr) <= skb->tail)
-			return IP6_ECN_set_ce(ipv6_hdr(skb));
+			return IP6_ECN_set_ce(skb, ipv6_hdr(skb));
 		break;
 	}
 
diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 5f28fabffd67..601254e4f28f 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -535,7 +535,7 @@ static inline void ipgre_ecn_decapsulate(const struct iphdr *iph, struct sk_buff
 		if (skb->protocol == htons(ETH_P_IP)) {
 			IP_ECN_set_ce(ip_hdr(skb));
 		} else if (skb->protocol == htons(ETH_P_IPV6)) {
-			IP6_ECN_set_ce(ipv6_hdr(skb));
+			IP6_ECN_set_ce(skb, ipv6_hdr(skb));
 		}
 	}
 }
diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index ae32ff78ab31..1828c27ba7aa 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -689,7 +689,7 @@ static void ip6ip6_dscp_ecn_decapsulate(const struct ip6_tnl *t,
 		ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
 
 	if (INET_ECN_is_ce(ipv6_get_dsfield(ipv6h)))
-		IP6_ECN_set_ce(ipv6_hdr(skb));
+		IP6_ECN_set_ce(skb, ipv6_hdr(skb));
 }
 
 /* called with rcu_read_lock() */
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index 52ce1969d9b2..1dbd1d10ca9d 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -553,7 +553,7 @@ out:
 static inline void ipip6_ecn_decapsulate(const struct iphdr *iph, struct sk_buff *skb)
 {
 	if (INET_ECN_is_ce(iph->tos))
-		IP6_ECN_set_ce(ipv6_hdr(skb));
+		IP6_ECN_set_ce(skb, ipv6_hdr(skb));
 }
 
 static int ipip6_rcv(struct sk_buff *skb)
diff --git a/net/ipv6/xfrm6_mode_tunnel.c b/net/ipv6/xfrm6_mode_tunnel.c
index 23ecd68a5e62..e33c9d7a4f85 100644
--- a/net/ipv6/xfrm6_mode_tunnel.c
+++ b/net/ipv6/xfrm6_mode_tunnel.c
@@ -24,7 +24,7 @@ static inline void ipip6_ecn_decapsulate(struct sk_buff *skb)
 	struct ipv6hdr *inner_iph = ipipv6_hdr(skb);
 
 	if (INET_ECN_is_ce(ipv6_get_dsfield(outer_iph)))
-		IP6_ECN_set_ce(inner_iph);
+		IP6_ECN_set_ce(skb, inner_iph);
 }
 
 /* Add encapsulation header.