ipvs: allow tunneling with gre encapsulation

windows real servers can handle gre tunnels, this patch allows
gre encapsulation with the tunneling method, thereby letting ipvs
be load balancer for windows-based services

Signed-off-by: Vadim Fedorenko <vfedorenko@yandex-team.ru>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

diff --git a/include/uapi/linux/ip_vs.h b/include/uapi/linux/ip_vs.h
index e4f18061a4fd3e02d03a19c0341bac55cd15e8fd..4102ddcb4e14ec382e77e0a548b3b6f0796533a4 100644 (file)
--- a/include/uapi/linux/ip_vs.h
+++ b/include/uapi/linux/ip_vs.h
@@ -128,6 +128,7 @@
 enum {
        IP_VS_CONN_F_TUNNEL_TYPE_IPIP = 0,      /* IPIP */
        IP_VS_CONN_F_TUNNEL_TYPE_GUE,           /* GUE */
+       IP_VS_CONN_F_TUNNEL_TYPE_GRE,           /* GRE */
        IP_VS_CONN_F_TUNNEL_TYPE_MAX,
 };
 

diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 84384d896e29d07613fbab3842b120fd830b7677..998353bec74f346de731eae50905a584b19b338d 100644 (file)
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -525,6 +525,7 @@ static void ip_vs_rs_hash(struct netns_ipvs *ipvs, struct ip_vs_dest *dest)
                        port = dest->tun_port;
                        break;
                case IP_VS_CONN_F_TUNNEL_TYPE_IPIP:
+               case IP_VS_CONN_F_TUNNEL_TYPE_GRE:
                        port = 0;
                        break;
                default:

diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index 71fc6d63a67f02ef496324ab93fad93534b5a53f..9c464d24beecdf8b90eb4a058528938e73ca7001 100644 (file)
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -29,6 +29,7 @@
 #include <linux/tcp.h>                  /* for tcphdr */
 #include <net/ip.h>
 #include <net/gue.h>
+#include <net/gre.h>
 #include <net/tcp.h>                    /* for csum_tcpudp_magic */
 #include <net/udp.h>
 #include <net/icmp.h>                   /* for icmp_send */
@@ -388,6 +389,12 @@ __ip_vs_get_out_rt(struct netns_ipvs *ipvs, int skb_af, struct sk_buff *skb,
                             IP_VS_TUNNEL_ENCAP_FLAG_REMCSUM) &&
                            skb->ip_summed == CHECKSUM_PARTIAL)
                                mtu -= GUE_PLEN_REMCSUM + GUE_LEN_PRIV;
+               } else if (dest->tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE) {
+                       __be16 tflags = 0;
+
+                       if (dest->tun_flags & IP_VS_TUNNEL_ENCAP_FLAG_CSUM)
+                               tflags |= TUNNEL_CSUM;
+                       mtu -= gre_calc_hlen(tflags);
                }
                if (mtu < 68) {
                        IP_VS_DBG_RL("%s(): mtu less than 68\n", __func__);
@@ -548,6 +555,12 @@ __ip_vs_get_out_rt_v6(struct netns_ipvs *ipvs, int skb_af, struct sk_buff *skb,
                             IP_VS_TUNNEL_ENCAP_FLAG_REMCSUM) &&
                            skb->ip_summed == CHECKSUM_PARTIAL)
                                mtu -= GUE_PLEN_REMCSUM + GUE_LEN_PRIV;
+               } else if (dest->tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE) {
+                       __be16 tflags = 0;
+
+                       if (dest->tun_flags & IP_VS_TUNNEL_ENCAP_FLAG_CSUM)
+                               tflags |= TUNNEL_CSUM;
+                       mtu -= gre_calc_hlen(tflags);
                }
                if (mtu < IPV6_MIN_MTU) {
                        IP_VS_DBG_RL("%s(): mtu less than %d\n", __func__,
@@ -1079,6 +1092,24 @@ ipvs_gue_encap(struct net *net, struct sk_buff *skb,
        return 0;
 }
 
+static void
+ipvs_gre_encap(struct net *net, struct sk_buff *skb,
+              struct ip_vs_conn *cp, __u8 *next_protocol)
+{
+       __be16 proto = *next_protocol == IPPROTO_IPIP ?
+                               htons(ETH_P_IP) : htons(ETH_P_IPV6);
+       __be16 tflags = 0;
+       size_t hdrlen;
+
+       if (cp->dest->tun_flags & IP_VS_TUNNEL_ENCAP_FLAG_CSUM)
+               tflags |= TUNNEL_CSUM;
+
+       hdrlen = gre_calc_hlen(tflags);
+       gre_build_header(skb, hdrlen, tflags, proto, 0, 0);
+
+       *next_protocol = IPPROTO_GRE;
+}
+
 /*
  *   IP Tunneling transmitter
  *
@@ -1151,6 +1182,15 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
                gue_hdrlen = sizeof(struct guehdr) + gue_optlen;
 
                max_headroom += sizeof(struct udphdr) + gue_hdrlen;
+       } else if (tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE) {
+               size_t gre_hdrlen;
+               __be16 tflags = 0;
+
+               if (tun_flags & IP_VS_TUNNEL_ENCAP_FLAG_CSUM)
+                       tflags |= TUNNEL_CSUM;
+               gre_hdrlen = gre_calc_hlen(tflags);
+
+               max_headroom += gre_hdrlen;
        }
 
        /* We only care about the df field if sysctl_pmtu_disc(ipvs) is set */
@@ -1172,6 +1212,11 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
                    skb->ip_summed == CHECKSUM_PARTIAL) {
                        gso_type |= SKB_GSO_TUNNEL_REMCSUM;
                }
+       } else if (tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE) {
+               if (tun_flags & IP_VS_TUNNEL_ENCAP_FLAG_CSUM)
+                       gso_type |= SKB_GSO_GRE_CSUM;
+               else
+                       gso_type |= SKB_GSO_GRE;
        }
 
        if (iptunnel_handle_offloads(skb, gso_type))
@@ -1192,8 +1237,8 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
                        check = true;
 
                udp_set_csum(!check, skb, saddr, cp->daddr.ip, skb->len);
-       }
-
+       } else if (tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE)
+               ipvs_gre_encap(net, skb, cp, &next_protocol);
 
        skb_push(skb, sizeof(struct iphdr));
        skb_reset_network_header(skb);
@@ -1287,6 +1332,15 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
                gue_hdrlen = sizeof(struct guehdr) + gue_optlen;
 
                max_headroom += sizeof(struct udphdr) + gue_hdrlen;
+       } else if (tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE) {
+               size_t gre_hdrlen;
+               __be16 tflags = 0;
+
+               if (tun_flags & IP_VS_TUNNEL_ENCAP_FLAG_CSUM)
+                       tflags |= TUNNEL_CSUM;
+               gre_hdrlen = gre_calc_hlen(tflags);
+
+               max_headroom += gre_hdrlen;
        }
 
        skb = ip_vs_prepare_tunneled_skb(skb, cp->af, max_headroom,
@@ -1306,6 +1360,11 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
                    skb->ip_summed == CHECKSUM_PARTIAL) {
                        gso_type |= SKB_GSO_TUNNEL_REMCSUM;
                }
+       } else if (tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE) {
+               if (tun_flags & IP_VS_TUNNEL_ENCAP_FLAG_CSUM)
+                       gso_type |= SKB_GSO_GRE_CSUM;
+               else
+                       gso_type |= SKB_GSO_GRE;
        }
 
        if (iptunnel_handle_offloads(skb, gso_type))
@@ -1326,7 +1385,8 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
                        check = true;
 
                udp6_set_csum(!check, skb, &saddr, &cp->daddr.in6, skb->len);
-       }
+       } else if (tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE)
+               ipvs_gre_encap(net, skb, cp, &next_protocol);
 
        skb_push(skb, sizeof(struct ipv6hdr));
        skb_reset_network_header(skb);