tcp: provide more information on the tcp receive_queue bugs
[safe/jmp/linux-2.6] / net / ipv4 / ipip.c
index 149111f..ae40ed1 100644 (file)
@@ -1,8 +1,6 @@
 /*
  *     Linux NET3:     IP/IP protocol decoder.
  *
- *     Version: $Id: ipip.c,v 1.50 2001/10/02 02:22:36 davem Exp $
- *
  *     Authors:
  *             Sam Lantinga (slouken@cs.ucdavis.edu)  02/01/95
  *
@@ -43,7 +41,7 @@
                Made the tunnels use dev->name not tunnel: when error reporting.
                Added tx_dropped stat
 
-               -Alan Cox       (Alan.Cox@linux.org) 21 March 95
+               -Alan Cox       (alan@lxorguk.ukuu.org.uk) 21 March 95
 
        Reworked:
                Changed to tunnel to destination gateway in addition to the
@@ -132,8 +130,8 @@ struct ipip_net {
        struct net_device *fb_tunnel_dev;
 };
 
-static int ipip_fb_tunnel_init(struct net_device *dev);
-static int ipip_tunnel_init(struct net_device *dev);
+static void ipip_fb_tunnel_init(struct net_device *dev);
+static void ipip_tunnel_init(struct net_device *dev);
 static void ipip_tunnel_setup(struct net_device *dev);
 
 static DEFINE_RWLOCK(ipip_lock);
@@ -247,9 +245,10 @@ static struct ip_tunnel * ipip_tunnel_locate(struct net *net,
        }
 
        nt = netdev_priv(dev);
-       dev->init = ipip_tunnel_init;
        nt->parms = *parms;
 
+       ipip_tunnel_init(dev);
+
        if (register_netdevice(dev) < 0)
                goto failed_free;
 
@@ -278,13 +277,12 @@ static void ipip_tunnel_uninit(struct net_device *dev)
 
 static int ipip_err(struct sk_buff *skb, u32 info)
 {
-#ifndef I_WISH_WORLD_WERE_PERFECT
 
-/* It is not :-( All the routers (except for Linux) return only
+/* All the routers (except for Linux) return only
    8 bytes of packet payload. It means, that precise relaying of
    ICMP in the real Internet is absolutely infeasible.
  */
-       struct iphdr *iph = (struct iphdr*)skb->data;
+       struct iphdr *iph = (struct iphdr *)skb->data;
        const int type = icmp_hdr(skb)->type;
        const int code = icmp_hdr(skb)->code;
        struct ip_tunnel *t;
@@ -329,7 +327,7 @@ static int ipip_err(struct sk_buff *skb, u32 info)
        if (t->parms.iph.ttl == 0 && type == ICMP_TIME_EXCEEDED)
                goto out;
 
-       if (jiffies - t->err_time < IPTUNNEL_ERR_TIMEO)
+       if (time_before(jiffies, t->err_time + IPTUNNEL_ERR_TIMEO))
                t->err_count++;
        else
                t->err_count = 1;
@@ -337,133 +335,6 @@ static int ipip_err(struct sk_buff *skb, u32 info)
 out:
        read_unlock(&ipip_lock);
        return err;
-#else
-       struct iphdr *iph = (struct iphdr*)dp;
-       int hlen = iph->ihl<<2;
-       struct iphdr *eiph;
-       const int type = icmp_hdr(skb)->type;
-       const int code = icmp_hdr(skb)->code;
-       int rel_type = 0;
-       int rel_code = 0;
-       __be32 rel_info = 0;
-       __u32 n = 0;
-       struct sk_buff *skb2;
-       struct flowi fl;
-       struct rtable *rt;
-
-       if (len < hlen + sizeof(struct iphdr))
-               return 0;
-       eiph = (struct iphdr*)(dp + hlen);
-
-       switch (type) {
-       default:
-               return 0;
-       case ICMP_PARAMETERPROB:
-               n = ntohl(icmp_hdr(skb)->un.gateway) >> 24;
-               if (n < hlen)
-                       return 0;
-
-               /* So... This guy found something strange INSIDE encapsulated
-                  packet. Well, he is fool, but what can we do ?
-                */
-               rel_type = ICMP_PARAMETERPROB;
-               rel_info = htonl((n - hlen) << 24);
-               break;
-
-       case ICMP_DEST_UNREACH:
-               switch (code) {
-               case ICMP_SR_FAILED:
-               case ICMP_PORT_UNREACH:
-                       /* Impossible event. */
-                       return 0;
-               case ICMP_FRAG_NEEDED:
-                       /* And it is the only really necessary thing :-) */
-                       n = ntohs(icmp_hdr(skb)->un.frag.mtu);
-                       if (n < hlen+68)
-                               return 0;
-                       n -= hlen;
-                       /* BSD 4.2 MORE DOES NOT EXIST IN NATURE. */
-                       if (n > ntohs(eiph->tot_len))
-                               return 0;
-                       rel_info = htonl(n);
-                       break;
-               default:
-                       /* All others are translated to HOST_UNREACH.
-                          rfc2003 contains "deep thoughts" about NET_UNREACH,
-                          I believe, it is just ether pollution. --ANK
-                        */
-                       rel_type = ICMP_DEST_UNREACH;
-                       rel_code = ICMP_HOST_UNREACH;
-                       break;
-               }
-               break;
-       case ICMP_TIME_EXCEEDED:
-               if (code != ICMP_EXC_TTL)
-                       return 0;
-               break;
-       }
-
-       /* Prepare fake skb to feed it to icmp_send */
-       skb2 = skb_clone(skb, GFP_ATOMIC);
-       if (skb2 == NULL)
-               return 0;
-       dst_release(skb2->dst);
-       skb2->dst = NULL;
-       skb_pull(skb2, skb->data - (u8*)eiph);
-       skb_reset_network_header(skb2);
-
-       /* Try to guess incoming interface */
-       memset(&fl, 0, sizeof(fl));
-       fl.fl4_daddr = eiph->saddr;
-       fl.fl4_tos = RT_TOS(eiph->tos);
-       fl.proto = IPPROTO_IPIP;
-       if (ip_route_output_key(dev_net(skb->dev), &rt, &key)) {
-               kfree_skb(skb2);
-               return 0;
-       }
-       skb2->dev = rt->u.dst.dev;
-
-       /* route "incoming" packet */
-       if (rt->rt_flags&RTCF_LOCAL) {
-               ip_rt_put(rt);
-               rt = NULL;
-               fl.fl4_daddr = eiph->daddr;
-               fl.fl4_src = eiph->saddr;
-               fl.fl4_tos = eiph->tos;
-               if (ip_route_output_key(dev_net(skb->dev), &rt, &fl) ||
-                   rt->u.dst.dev->type != ARPHRD_TUNNEL) {
-                       ip_rt_put(rt);
-                       kfree_skb(skb2);
-                       return 0;
-               }
-       } else {
-               ip_rt_put(rt);
-               if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos, skb2->dev) ||
-                   skb2->dst->dev->type != ARPHRD_TUNNEL) {
-                       kfree_skb(skb2);
-                       return 0;
-               }
-       }
-
-       /* change mtu on this route */
-       if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED) {
-               if (n > dst_mtu(skb2->dst)) {
-                       kfree_skb(skb2);
-                       return 0;
-               }
-               skb2->dst->ops->update_pmtu(skb2->dst, n);
-       } else if (type == ICMP_TIME_EXCEEDED) {
-               struct ip_tunnel *t = netdev_priv(skb2->dev);
-               if (t->parms.iph.ttl) {
-                       rel_type = ICMP_DEST_UNREACH;
-                       rel_code = ICMP_HOST_UNREACH;
-               }
-       }
-
-       icmp_send(skb2, rel_type, rel_code, rel_info);
-       kfree_skb(skb2);
-       return 0;
-#endif
 }
 
 static inline void ipip_ecn_decapsulate(const struct iphdr *outer_iph,
@@ -496,11 +367,10 @@ static int ipip_rcv(struct sk_buff *skb)
                skb->protocol = htons(ETH_P_IP);
                skb->pkt_type = PACKET_HOST;
 
-               tunnel->stat.rx_packets++;
-               tunnel->stat.rx_bytes += skb->len;
+               tunnel->dev->stats.rx_packets++;
+               tunnel->dev->stats.rx_bytes += skb->len;
                skb->dev = tunnel->dev;
-               dst_release(skb->dst);
-               skb->dst = NULL;
+               skb_dst_drop(skb);
                nf_reset(skb);
                ipip_ecn_decapsulate(iph, skb);
                netif_rx(skb);
@@ -517,10 +387,10 @@ static int ipip_rcv(struct sk_buff *skb)
  *     and that skb is filled properly by that function.
  */
 
-static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
+static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 {
        struct ip_tunnel *tunnel = netdev_priv(dev);
-       struct net_device_stats *stats = &tunnel->stat;
+       struct net_device_stats *stats = &tunnel->dev->stats;
        struct iphdr  *tiph = &tunnel->parms.iph;
        u8     tos = tunnel->parms.iph.tos;
        __be16 df = tiph->frag_off;
@@ -532,11 +402,6 @@ static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
        __be32 dst = tiph->daddr;
        int    mtu;
 
-       if (tunnel->recursion++) {
-               tunnel->stat.collisions++;
-               goto tx_error;
-       }
-
        if (skb->protocol != htons(ETH_P_IP))
                goto tx_error;
 
@@ -545,8 +410,8 @@ static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 
        if (!dst) {
                /* NBMA tunnel */
-               if ((rt = skb->rtable) == NULL) {
-                       tunnel->stat.tx_fifo_errors++;
+               if ((rt = skb_rtable(skb)) == NULL) {
+                       stats->tx_fifo_errors++;
                        goto tx_error;
                }
                if ((dst = rt->rt_gateway) == 0)
@@ -561,7 +426,7 @@ static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
                                                .tos = RT_TOS(tos) } },
                                    .proto = IPPROTO_IPIP };
                if (ip_route_output_key(dev_net(dev), &rt, &fl)) {
-                       tunnel->stat.tx_carrier_errors++;
+                       stats->tx_carrier_errors++;
                        goto tx_error_icmp;
                }
        }
@@ -569,33 +434,36 @@ static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 
        if (tdev == dev) {
                ip_rt_put(rt);
-               tunnel->stat.collisions++;
+               stats->collisions++;
                goto tx_error;
        }
 
-       if (tiph->frag_off)
+       df |= old_iph->frag_off & htons(IP_DF);
+
+       if (df) {
                mtu = dst_mtu(&rt->u.dst) - sizeof(struct iphdr);
-       else
-               mtu = skb->dst ? dst_mtu(skb->dst) : dev->mtu;
 
-       if (mtu < 68) {
-               tunnel->stat.collisions++;
-               ip_rt_put(rt);
-               goto tx_error;
-       }
-       if (skb->dst)
-               skb->dst->ops->update_pmtu(skb->dst, mtu);
+               if (mtu < 68) {
+                       stats->collisions++;
+                       ip_rt_put(rt);
+                       goto tx_error;
+               }
 
-       df |= (old_iph->frag_off&htons(IP_DF));
+               if (skb_dst(skb))
+                       skb_dst(skb)->ops->update_pmtu(skb_dst(skb), mtu);
 
-       if ((old_iph->frag_off&htons(IP_DF)) && mtu < ntohs(old_iph->tot_len)) {
-               icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, htonl(mtu));
-               ip_rt_put(rt);
-               goto tx_error;
+               if ((old_iph->frag_off & htons(IP_DF)) &&
+                   mtu < ntohs(old_iph->tot_len)) {
+                       icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
+                                 htonl(mtu));
+                       ip_rt_put(rt);
+                       goto tx_error;
+               }
        }
 
        if (tunnel->err_count > 0) {
-               if (jiffies - tunnel->err_time < IPTUNNEL_ERR_TIMEO) {
+               if (time_before(jiffies,
+                               tunnel->err_time + IPTUNNEL_ERR_TIMEO)) {
                        tunnel->err_count--;
                        dst_link_failure(skb);
                } else
@@ -614,8 +482,7 @@ static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
                        ip_rt_put(rt);
                        stats->tx_dropped++;
                        dev_kfree_skb(skb);
-                       tunnel->recursion--;
-                       return 0;
+                       return NETDEV_TX_OK;
                }
                if (skb->sk)
                        skb_set_owner_w(new_skb, skb->sk);
@@ -630,8 +497,8 @@ static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
        memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
        IPCB(skb)->flags &= ~(IPSKB_XFRM_TUNNEL_SIZE | IPSKB_XFRM_TRANSFORMED |
                              IPSKB_REROUTED);
-       dst_release(skb->dst);
-       skb->dst = &rt->u.dst;
+       skb_dst_drop(skb);
+       skb_dst_set(skb, &rt->u.dst);
 
        /*
         *      Push down and install the IPIP header.
@@ -652,16 +519,14 @@ static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
        nf_reset(skb);
 
        IPTUNNEL_XMIT();
-       tunnel->recursion--;
-       return 0;
+       return NETDEV_TX_OK;
 
 tx_error_icmp:
        dst_link_failure(skb);
 tx_error:
        stats->tx_errors++;
        dev_kfree_skb(skb);
-       tunnel->recursion--;
-       return 0;
+       return NETDEV_TX_OK;
 }
 
 static void ipip_tunnel_bind_dev(struct net_device *dev)
@@ -813,11 +678,6 @@ done:
        return err;
 }
 
-static struct net_device_stats *ipip_tunnel_get_stats(struct net_device *dev)
-{
-       return &(((struct ip_tunnel*)netdev_priv(dev))->stat);
-}
-
 static int ipip_tunnel_change_mtu(struct net_device *dev, int new_mtu)
 {
        if (new_mtu < 68 || new_mtu > 0xFFF8 - sizeof(struct iphdr))
@@ -826,13 +686,17 @@ static int ipip_tunnel_change_mtu(struct net_device *dev, int new_mtu)
        return 0;
 }
 
+static const struct net_device_ops ipip_netdev_ops = {
+       .ndo_uninit     = ipip_tunnel_uninit,
+       .ndo_start_xmit = ipip_tunnel_xmit,
+       .ndo_do_ioctl   = ipip_tunnel_ioctl,
+       .ndo_change_mtu = ipip_tunnel_change_mtu,
+
+};
+
 static void ipip_tunnel_setup(struct net_device *dev)
 {
-       dev->uninit             = ipip_tunnel_uninit;
-       dev->hard_start_xmit    = ipip_tunnel_xmit;
-       dev->get_stats          = ipip_tunnel_get_stats;
-       dev->do_ioctl           = ipip_tunnel_ioctl;
-       dev->change_mtu         = ipip_tunnel_change_mtu;
+       dev->netdev_ops         = &ipip_netdev_ops;
        dev->destructor         = free_netdev;
 
        dev->type               = ARPHRD_TUNNEL;
@@ -842,13 +706,12 @@ static void ipip_tunnel_setup(struct net_device *dev)
        dev->iflink             = 0;
        dev->addr_len           = 4;
        dev->features           |= NETIF_F_NETNS_LOCAL;
+       dev->priv_flags         &= ~IFF_XMIT_DST_RELEASE;
 }
 
-static int ipip_tunnel_init(struct net_device *dev)
+static void ipip_tunnel_init(struct net_device *dev)
 {
-       struct ip_tunnel *tunnel;
-
-       tunnel = netdev_priv(dev);
+       struct ip_tunnel *tunnel = netdev_priv(dev);
 
        tunnel->dev = dev;
        strcpy(tunnel->parms.name, dev->name);
@@ -857,11 +720,9 @@ static int ipip_tunnel_init(struct net_device *dev)
        memcpy(dev->broadcast, &tunnel->parms.iph.daddr, 4);
 
        ipip_tunnel_bind_dev(dev);
-
-       return 0;
 }
 
-static int ipip_fb_tunnel_init(struct net_device *dev)
+static void ipip_fb_tunnel_init(struct net_device *dev)
 {
        struct ip_tunnel *tunnel = netdev_priv(dev);
        struct iphdr *iph = &tunnel->parms.iph;
@@ -876,7 +737,6 @@ static int ipip_fb_tunnel_init(struct net_device *dev)
 
        dev_hold(dev);
        ipn->tunnels_wc[0]      = tunnel;
-       return 0;
 }
 
 static struct xfrm_tunnel ipip_handler = {
@@ -885,7 +745,7 @@ static struct xfrm_tunnel ipip_handler = {
        .priority       =       1,
 };
 
-static char banner[] __initdata =
+static const char banner[] __initconst =
        KERN_INFO "IPv4 over IPv4 tunneling driver\n";
 
 static void ipip_destroy_tunnels(struct ipip_net *ipn)
@@ -928,10 +788,10 @@ static int ipip_init_net(struct net *net)
                err = -ENOMEM;
                goto err_alloc_dev;
        }
-
-       ipn->fb_tunnel_dev->init = ipip_fb_tunnel_init;
        dev_net_set(ipn->fb_tunnel_dev, net);
 
+       ipip_fb_tunnel_init(ipn->fb_tunnel_dev);
+
        if ((err = register_netdev(ipn->fb_tunnel_dev)))
                goto err_reg_dev;