[IPSEC]: make sit use the xfrm4_tunnel_register
Kazunori MIYAZAWA [Tue, 13 Feb 2007 20:55:25 +0000 (12:55 -0800)]
This patch makes sit use xfrm4_tunnel_register instead of
inet_add_protocol. It solves conflict of sit device with
inter address family IPsec tunnel.

Signed-off-by: Kazunori MIYAZAWA <miyazawa@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/ipv6/Kconfig
net/ipv6/sit.c

index deb4101..79682ef 100644 (file)
@@ -156,6 +156,7 @@ config INET6_XFRM_MODE_ROUTEOPTIMIZATION
 config IPV6_SIT
        tristate "IPv6: IPv6-in-IPv4 tunnel (SIT driver)"
        depends on IPV6
+       select INET_TUNNEL
        default y
        ---help---
          Tunneling means encapsulating data of one protocol type within
index 4d3cf30..862ed7c 100644 (file)
@@ -216,7 +216,7 @@ static void ipip6_tunnel_uninit(struct net_device *dev)
 }
 
 
-static void ipip6_err(struct sk_buff *skb, u32 info)
+static int ipip6_err(struct sk_buff *skb, u32 info)
 {
 #ifndef I_WISH_WORLD_WERE_PERFECT
 
@@ -228,21 +228,22 @@ static void ipip6_err(struct sk_buff *skb, u32 info)
        int type = skb->h.icmph->type;
        int code = skb->h.icmph->code;
        struct ip_tunnel *t;
+       int err;
 
        switch (type) {
        default:
        case ICMP_PARAMETERPROB:
-               return;
+               return 0;
 
        case ICMP_DEST_UNREACH:
                switch (code) {
                case ICMP_SR_FAILED:
                case ICMP_PORT_UNREACH:
                        /* Impossible event. */
-                       return;
+                       return 0;
                case ICMP_FRAG_NEEDED:
                        /* Soft state for pmtu is maintained by IP core. */
-                       return;
+                       return 0;
                default:
                        /* All others are translated to HOST_UNREACH.
                           rfc2003 contains "deep thoughts" about NET_UNREACH,
@@ -253,14 +254,18 @@ static void ipip6_err(struct sk_buff *skb, u32 info)
                break;
        case ICMP_TIME_EXCEEDED:
                if (code != ICMP_EXC_TTL)
-                       return;
+                       return 0;
                break;
        }
 
+       err = -ENOENT;
+
        read_lock(&ipip6_lock);
        t = ipip6_tunnel_lookup(iph->daddr, iph->saddr);
        if (t == NULL || t->parms.iph.daddr == 0)
                goto out;
+
+       err = 0;
        if (t->parms.iph.ttl == 0 && type == ICMP_TIME_EXCEEDED)
                goto out;
 
@@ -271,7 +276,7 @@ static void ipip6_err(struct sk_buff *skb, u32 info)
        t->err_time = jiffies;
 out:
        read_unlock(&ipip6_lock);
-       return;
+       return err;
 #else
        struct iphdr *iph = (struct iphdr*)dp;
        int hlen = iph->ihl<<2;
@@ -332,7 +337,7 @@ out:
        /* Prepare fake skb to feed it to icmpv6_send */
        skb2 = skb_clone(skb, GFP_ATOMIC);
        if (skb2 == NULL)
-               return;
+               return 0;
        dst_release(skb2->dst);
        skb2->dst = NULL;
        skb_pull(skb2, skb->data - (u8*)iph6);
@@ -355,7 +360,7 @@ out:
                }
        }
        kfree_skb(skb2);
-       return;
+       return 0;
 #endif
 }
 
@@ -791,9 +796,10 @@ static int __init ipip6_fb_tunnel_init(struct net_device *dev)
        return 0;
 }
 
-static struct net_protocol sit_protocol = {
+static struct xfrm_tunnel sit_handler = {
        .handler        =       ipip6_rcv,
        .err_handler    =       ipip6_err,
+       .priority       =       1,
 };
 
 static void __exit sit_destroy_tunnels(void)
@@ -812,7 +818,7 @@ static void __exit sit_destroy_tunnels(void)
 
 static void __exit sit_cleanup(void)
 {
-       inet_del_protocol(&sit_protocol, IPPROTO_IPV6);
+       xfrm4_tunnel_deregister(&sit_handler, AF_INET6);
 
        rtnl_lock();
        sit_destroy_tunnels();
@@ -826,7 +832,7 @@ static int __init sit_init(void)
 
        printk(KERN_INFO "IPv6 over IPv4 tunneling driver\n");
 
-       if (inet_add_protocol(&sit_protocol, IPPROTO_IPV6) < 0) {
+       if (xfrm4_tunnel_register(&sit_handler, AF_INET6) < 0) {
                printk(KERN_INFO "sit init: Can't add protocol\n");
                return -EAGAIN;
        }
@@ -848,7 +854,7 @@ static int __init sit_init(void)
  err2:
        free_netdev(ipip6_fb_tunnel_dev);
  err1:
-       inet_del_protocol(&sit_protocol, IPPROTO_IPV6);
+       xfrm4_tunnel_deregister(&sit_handler, AF_INET6);
        goto out;
 }