tunnels: Remove encapsulation offloads on decap.

If a packet is either locally encapsulated or processed through GRO
it is marked with the offloads that it requires. However, when it is
decapsulated these tunnel offload indications are not removed. This
means that if we receive an encapsulated TCP packet, aggregate it with
GRO, decapsulate, and retransmit the resulting frame on a NIC that does
not support encapsulation, we won't be able to take advantage of hardware
offloads even though it is just a simple TCP packet at this point.

This fixes the problem by stripping off encapsulation offload indications
when packets are decapsulated.

The performance impacts of this bug are significant. In a test where a
Geneve encapsulated TCP stream is sent to a hypervisor, GRO'ed, decapsulated,
and bridged to a VM performance is improved by 60% (5Gbps->8Gbps) as a
result of avoiding unnecessary segmentation at the VM tap interface.

Reported-by: Ramu Ramamurthy <sramamur@linux.vnet.ibm.com>
Fixes: 68c33163 ("v4 GRE: Add TCP segmentation offload for GRE")
Signed-off-by: Jesse Gross <jesse@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
index c35dda9ec991163f78736ebcd4e61d07683a7e4c..56050f913339e075f1c22c29bdf5c1975900baed 100644 (file)
--- a/include/net/ip_tunnels.h
+++ b/include/net/ip_tunnels.h
@@ -305,6 +305,22 @@ struct metadata_dst *iptunnel_metadata_reply(struct metadata_dst *md,
 
 struct sk_buff *iptunnel_handle_offloads(struct sk_buff *skb, int gso_type_mask);
 
+static inline int iptunnel_pull_offloads(struct sk_buff *skb)
+{
+       if (skb_is_gso(skb)) {
+               int err;
+
+               err = skb_unclone(skb, GFP_ATOMIC);
+               if (unlikely(err))
+                       return err;
+               skb_shinfo(skb)->gso_type &= ~(NETIF_F_GSO_ENCAP_ALL >>
+                                              NETIF_F_GSO_SHIFT);
+       }
+
+       skb->encapsulation = 0;
+       return 0;
+}
+
 static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len)
 {
        if (pkt_len > 0) {

diff --git a/net/ipv4/fou.c b/net/ipv4/fou.c
index 780484243e144006d8ee409b0377ee4eb4db6dd9..a0586b4a197d9eba50e742cede9f826e432ae5c2 100644 (file)
--- a/net/ipv4/fou.c
+++ b/net/ipv4/fou.c
@@ -48,7 +48,7 @@ static inline struct fou *fou_from_sock(struct sock *sk)
        return sk->sk_user_data;
 }
 
-static void fou_recv_pull(struct sk_buff *skb, size_t len)
+static int fou_recv_pull(struct sk_buff *skb, size_t len)
 {
        struct iphdr *iph = ip_hdr(skb);
 
@@ -59,6 +59,7 @@ static void fou_recv_pull(struct sk_buff *skb, size_t len)
        __skb_pull(skb, len);
        skb_postpull_rcsum(skb, udp_hdr(skb), len);
        skb_reset_transport_header(skb);
+       return iptunnel_pull_offloads(skb);
 }
 
 static int fou_udp_recv(struct sock *sk, struct sk_buff *skb)
@@ -68,9 +69,14 @@ static int fou_udp_recv(struct sock *sk, struct sk_buff *skb)
        if (!fou)
                return 1;
 
-       fou_recv_pull(skb, sizeof(struct udphdr));
+       if (fou_recv_pull(skb, sizeof(struct udphdr)))
+               goto drop;
 
        return -fou->protocol;
+
+drop:
+       kfree_skb(skb);
+       return 0;
 }
 
 static struct guehdr *gue_remcsum(struct sk_buff *skb, struct guehdr *guehdr,
@@ -170,6 +176,9 @@ static int gue_udp_recv(struct sock *sk, struct sk_buff *skb)
        __skb_pull(skb, sizeof(struct udphdr) + hdrlen);
        skb_reset_transport_header(skb);
 
+       if (iptunnel_pull_offloads(skb))
+               goto drop;
+
        return -guehdr->proto_ctype;
 
 drop:

diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c
index d27276f6f8ddcd4d359b26b17b869a8ccd2cb929..02dd990af542bcfe78cd21a80507ba8be1d30335 100644 (file)
--- a/net/ipv4/ip_tunnel_core.c
+++ b/net/ipv4/ip_tunnel_core.c
@@ -114,7 +114,8 @@ int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto,
        skb->vlan_tci = 0;
        skb_set_queue_mapping(skb, 0);
        skb_scrub_packet(skb, xnet);
-       return 0;
+
+       return iptunnel_pull_offloads(skb);
 }
 EXPORT_SYMBOL_GPL(iptunnel_pull_header);
 

diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index f45b8ffc2840bbca9070b96ad85811ea8157f829..83384308d032492fff85c04d4dca196c1bb690fc 100644 (file)
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -681,14 +681,16 @@ static int ipip6_rcv(struct sk_buff *skb)
                skb->mac_header = skb->network_header;
                skb_reset_network_header(skb);
                IPCB(skb)->flags = 0;
-               skb->protocol = htons(ETH_P_IPV6);
+               skb->dev = tunnel->dev;
 
                if (packet_is_spoofed(skb, iph, tunnel)) {
                        tunnel->dev->stats.rx_errors++;
                        goto out;
                }
 
-               __skb_tunnel_rx(skb, tunnel->dev, tunnel->net);
+               if (iptunnel_pull_header(skb, 0, htons(ETH_P_IPV6),
+                   !net_eq(tunnel->net, dev_net(tunnel->dev))))
+                       goto out;
 
                err = IP_ECN_decapsulate(iph, skb);
                if (unlikely(err)) {