[DCCP]: Provide 10s of microsecond timesource
[safe/jmp/linux-2.6] / net / ipv4 / udp.c
index fc620a7..ef4d901 100644 (file)
@@ -70,6 +70,7 @@
  *     Alexey Kuznetsov:               allow both IPv4 and IPv6 sockets to bind
  *                                     a single port at the same time.
  *     Derek Atkins <derek@ihtfp.com>: Add Encapulation Support
+ *     James Chapman           :       Add L2TP encapsulation type.
  *
  *
  *             This program is free software; you can redistribute it and/or
@@ -97,6 +98,7 @@
 #include <linux/skbuff.h>
 #include <linux/proc_fs.h>
 #include <linux/seq_file.h>
+#include <net/net_namespace.h>
 #include <net/icmp.h>
 #include <net/route.h>
 #include <net/checksum.h>
@@ -112,9 +114,8 @@ DEFINE_SNMP_STAT(struct udp_mib, udp_statistics) __read_mostly;
 struct hlist_head udp_hash[UDP_HTABLE_SIZE];
 DEFINE_RWLOCK(udp_hash_lock);
 
-static int udp_port_rover;
-
-static inline int __udp_lib_lport_inuse(__u16 num, struct hlist_head udptable[])
+static inline int __udp_lib_lport_inuse(__u16 num,
+                                       const struct hlist_head udptable[])
 {
        struct sock *sk;
        struct hlist_node *node;
@@ -131,11 +132,10 @@ static inline int __udp_lib_lport_inuse(__u16 num, struct hlist_head udptable[])
  *  @sk:          socket struct in question
  *  @snum:        port number to look up
  *  @udptable:    hash list table, must be of UDP_HTABLE_SIZE
- *  @port_rover:  pointer to record of last unallocated port
  *  @saddr_comp:  AF-dependent comparison of bound local IP addresses
  */
 int __udp_lib_get_port(struct sock *sk, unsigned short snum,
-                      struct hlist_head udptable[], int *port_rover,
+                      struct hlist_head udptable[],
                       int (*saddr_comp)(const struct sock *sk1,
                                         const struct sock *sk2 )    )
 {
@@ -145,48 +145,56 @@ int __udp_lib_get_port(struct sock *sk, unsigned short snum,
        int    error = 1;
 
        write_lock_bh(&udp_hash_lock);
-       if (snum == 0) {
-               int best_size_so_far, best, result, i;
-
-               if (*port_rover > sysctl_local_port_range[1] ||
-                   *port_rover < sysctl_local_port_range[0])
-                       *port_rover = sysctl_local_port_range[0];
-               best_size_so_far = 32767;
-               best = result = *port_rover;
-               for (i = 0; i < UDP_HTABLE_SIZE; i++, result++) {
-                       int size;
-
-                       head = &udptable[result & (UDP_HTABLE_SIZE - 1)];
-                       if (hlist_empty(head)) {
-                               if (result > sysctl_local_port_range[1])
-                                       result = sysctl_local_port_range[0] +
-                                               ((result - sysctl_local_port_range[0]) &
-                                                (UDP_HTABLE_SIZE - 1));
+
+       if (!snum) {
+               int i;
+               int low = sysctl_local_port_range[0];
+               int high = sysctl_local_port_range[1];
+               unsigned rover, best, best_size_so_far;
+
+               best_size_so_far = UINT_MAX;
+               best = rover = net_random() % (high - low) + low;
+
+               /* 1st pass: look for empty (or shortest) hash chain */
+               for (i = 0; i < UDP_HTABLE_SIZE; i++) {
+                       int size = 0;
+
+                       head = &udptable[rover & (UDP_HTABLE_SIZE - 1)];
+                       if (hlist_empty(head))
                                goto gotit;
-                       }
-                       size = 0;
+
                        sk_for_each(sk2, node, head) {
                                if (++size >= best_size_so_far)
                                        goto next;
                        }
                        best_size_so_far = size;
-                       best = result;
+                       best = rover;
                next:
-                       ;
+                       /* fold back if end of range */
+                       if (++rover > high)
+                               rover = low + ((rover - low)
+                                              & (UDP_HTABLE_SIZE - 1));
+
+
                }
-               result = best;
-               for(i = 0; i < (1 << 16) / UDP_HTABLE_SIZE; i++, result += UDP_HTABLE_SIZE) {
-                       if (result > sysctl_local_port_range[1])
-                               result = sysctl_local_port_range[0]
-                                       + ((result - sysctl_local_port_range[0]) &
-                                          (UDP_HTABLE_SIZE - 1));
-                       if (! __udp_lib_lport_inuse(result, udptable))
-                               break;
+
+               /* 2nd pass: find hole in shortest hash chain */
+               rover = best;
+               for (i = 0; i < (1 << 16) / UDP_HTABLE_SIZE; i++) {
+                       if (! __udp_lib_lport_inuse(rover, udptable))
+                               goto gotit;
+                       rover += UDP_HTABLE_SIZE;
+                       if (rover > high)
+                               rover = low + ((rover - low)
+                                              & (UDP_HTABLE_SIZE - 1));
                }
-               if (i >= (1 << 16) / UDP_HTABLE_SIZE)
-                       goto fail;
+
+
+               /* All ports in use! */
+               goto fail;
+
 gotit:
-               *port_rover = snum = result;
+               snum = rover;
        } else {
                head = &udptable[snum & (UDP_HTABLE_SIZE - 1)];
 
@@ -199,6 +207,7 @@ gotit:
                            (*saddr_comp)(sk, sk2)                             )
                                goto fail;
        }
+
        inet_sk(sk)->num = snum;
        sk->sk_hash = snum;
        if (sk_unhashed(sk)) {
@@ -212,13 +221,13 @@ fail:
        return error;
 }
 
-__inline__ int udp_get_port(struct sock *sk, unsigned short snum,
+int udp_get_port(struct sock *sk, unsigned short snum,
                        int (*scmp)(const struct sock *, const struct sock *))
 {
-       return  __udp_lib_get_port(sk, snum, udp_hash, &udp_port_rover, scmp);
+       return  __udp_lib_get_port(sk, snum, udp_hash, scmp);
 }
 
-inline int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
+int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
 {
        struct inet_sock *inet1 = inet_sk(sk1), *inet2 = inet_sk(sk2);
 
@@ -270,10 +279,10 @@ static struct sock *__udp4_lib_lookup(__be32 saddr, __be16 sport,
                                        continue;
                                score+=2;
                        }
-                       if(score == 9) {
+                       if (score == 9) {
                                result = sk;
                                break;
-                       } else if(score > badness) {
+                       } else if (score > badness) {
                                result = sk;
                                badness = score;
                        }
@@ -329,8 +338,8 @@ void __udp4_lib_err(struct sk_buff *skb, u32 info, struct hlist_head udptable[])
        struct inet_sock *inet;
        struct iphdr *iph = (struct iphdr*)skb->data;
        struct udphdr *uh = (struct udphdr*)(skb->data+(iph->ihl<<2));
-       int type = skb->h.icmph->type;
-       int code = skb->h.icmph->code;
+       const int type = icmp_hdr(skb)->type;
+       const int code = icmp_hdr(skb)->code;
        struct sock *sk;
        int harderr;
        int err;
@@ -390,7 +399,7 @@ out:
        sock_put(sk);
 }
 
-__inline__ void udp_err(struct sk_buff *skb, u32 info)
+void udp_err(struct sk_buff *skb, u32 info)
 {
        return __udp4_lib_err(skb, info, udp_hash);
 }
@@ -419,13 +428,14 @@ static void udp4_hwcsum_outgoing(struct sock *sk, struct sk_buff *skb,
                                 __be32 src, __be32 dst, int len      )
 {
        unsigned int offset;
-       struct udphdr *uh = skb->h.uh;
+       struct udphdr *uh = udp_hdr(skb);
        __wsum csum = 0;
 
        if (skb_queue_len(&sk->sk_write_queue) == 1) {
                /*
                 * Only one fragment on the socket.
                 */
+               skb->csum_start = skb_transport_header(skb) - skb->head;
                skb->csum_offset = offsetof(struct udphdr, check);
                uh->check = ~csum_tcpudp_magic(src, dst, len, IPPROTO_UDP, 0);
        } else {
@@ -434,7 +444,7 @@ static void udp4_hwcsum_outgoing(struct sock *sk, struct sk_buff *skb,
                 * fragments on the socket so that all csums of sk_buffs
                 * should be together
                 */
-               offset = skb->h.raw - skb->data;
+               offset = skb_transport_offset(skb);
                skb->csum = skb_checksum(skb, offset, skb->len - offset, 0);
 
                skb->ip_summed = CHECKSUM_NONE;
@@ -469,7 +479,7 @@ static int udp_push_pending_frames(struct sock *sk)
        /*
         * Create a UDP header
         */
-       uh = skb->h.uh;
+       uh = udp_hdr(skb);
        uh->source = fl->fl_ip_sport;
        uh->dest = fl->fl_ip_dport;
        uh->len = htons(up->len);
@@ -502,6 +512,8 @@ send:
 out:
        up->len = 0;
        up->pending = 0;
+       if (!err)
+               UDP_INC_STATS_USER(UDP_MIB_OUTDATAGRAMS, up->pcflag);
        return err;
 }
 
@@ -631,8 +643,11 @@ int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
                                                 .dport = dport } } };
                security_sk_classify_flow(sk, &fl);
                err = ip_route_output_flow(&rt, &fl, sk, 1);
-               if (err)
+               if (err) {
+                       if (err == -ENETUNREACH)
+                               IP_INC_STATS_BH(IPSTATS_MIB_OUTNOROUTES);
                        goto out;
+               }
 
                err = -EACCES;
                if ((rt->rt_flags & RTCF_BROADCAST) &&
@@ -687,10 +702,8 @@ out:
        ip_rt_put(rt);
        if (free)
                kfree(ipc.opt);
-       if (!err) {
-               UDP_INC_STATS_USER(UDP_MIB_OUTDATAGRAMS, is_udplite);
+       if (!err)
                return len;
-       }
        /*
         * ENOBUFS = no kernel mem, SOCK_NOSPACE = no sndbuf space.  Reporting
         * ENOBUFS might not be good (it's not tunable per se), but otherwise
@@ -765,38 +778,38 @@ out:
 
 int udp_ioctl(struct sock *sk, int cmd, unsigned long arg)
 {
-       switch(cmd)
+       switch (cmd) {
+       case SIOCOUTQ:
        {
-               case SIOCOUTQ:
-               {
-                       int amount = atomic_read(&sk->sk_wmem_alloc);
-                       return put_user(amount, (int __user *)arg);
-               }
+               int amount = atomic_read(&sk->sk_wmem_alloc);
+               return put_user(amount, (int __user *)arg);
+       }
 
-               case SIOCINQ:
-               {
-                       struct sk_buff *skb;
-                       unsigned long amount;
-
-                       amount = 0;
-                       spin_lock_bh(&sk->sk_receive_queue.lock);
-                       skb = skb_peek(&sk->sk_receive_queue);
-                       if (skb != NULL) {
-                               /*
-                                * We will only return the amount
-                                * of this packet since that is all
-                                * that will be read.
-                                */
-                               amount = skb->len - sizeof(struct udphdr);
-                       }
-                       spin_unlock_bh(&sk->sk_receive_queue.lock);
-                       return put_user(amount, (int __user *)arg);
+       case SIOCINQ:
+       {
+               struct sk_buff *skb;
+               unsigned long amount;
+
+               amount = 0;
+               spin_lock_bh(&sk->sk_receive_queue.lock);
+               skb = skb_peek(&sk->sk_receive_queue);
+               if (skb != NULL) {
+                       /*
+                        * We will only return the amount
+                        * of this packet since that is all
+                        * that will be read.
+                        */
+                       amount = skb->len - sizeof(struct udphdr);
                }
+               spin_unlock_bh(&sk->sk_receive_queue.lock);
+               return put_user(amount, (int __user *)arg);
+       }
 
-               default:
-                       return -ENOIOCTLCMD;
+       default:
+               return -ENOIOCTLCMD;
        }
-       return(0);
+
+       return 0;
 }
 
 /*
@@ -810,7 +823,9 @@ int udp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
        struct inet_sock *inet = inet_sk(sk);
        struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
        struct sk_buff *skb;
-       int copied, err, copy_only, is_udplite = IS_UDPLITE(sk);
+       unsigned int ulen, copied;
+       int err;
+       int is_udplite = IS_UDPLITE(sk);
 
        /*
         *      Check any passed addresses
@@ -826,28 +841,25 @@ try_again:
        if (!skb)
                goto out;
 
-       copied = skb->len - sizeof(struct udphdr);
-       if (copied > len) {
-               copied = len;
+       ulen = skb->len - sizeof(struct udphdr);
+       copied = len;
+       if (copied > ulen)
+               copied = ulen;
+       else if (copied < ulen)
                msg->msg_flags |= MSG_TRUNC;
-       }
 
        /*
-        *      Decide whether to checksum and/or copy data.
-        *
-        *      UDP:      checksum may have been computed in HW,
-        *                (re-)compute it if message is truncated.
-        *      UDP-Lite: always needs to checksum, no HW support.
+        * If checksum is needed at all, try to do it while copying the
+        * data.  If the data is truncated, or if we only want a partial
+        * coverage checksum (UDP-Lite), do it before the copy.
         */
-       copy_only = (skb->ip_summed==CHECKSUM_UNNECESSARY);
 
-       if (is_udplite  ||  (!copy_only  &&  msg->msg_flags&MSG_TRUNC)) {
-               if (__udp_lib_checksum_complete(skb))
+       if (copied < ulen || UDP_SKB_CB(skb)->partial_cov) {
+               if (udp_lib_checksum_complete(skb))
                        goto csum_copy_err;
-               copy_only = 1;
        }
 
-       if (copy_only)
+       if (skb_csum_unnecessary(skb))
                err = skb_copy_datagram_iovec(skb, sizeof(struct udphdr),
                                              msg->msg_iov, copied       );
        else {
@@ -866,8 +878,8 @@ try_again:
        if (sin)
        {
                sin->sin_family = AF_INET;
-               sin->sin_port = skb->h.uh->source;
-               sin->sin_addr.s_addr = skb->nh.iph->saddr;
+               sin->sin_port = udp_hdr(skb)->source;
+               sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
                memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
        }
        if (inet->cmsg_flags)
@@ -875,7 +887,7 @@ try_again:
 
        err = copied;
        if (flags & MSG_TRUNC)
-               err = skb->len - sizeof(struct udphdr);
+               err = ulen;
 
 out_free:
        skb_free_datagram(sk, skb);
@@ -915,103 +927,6 @@ int udp_disconnect(struct sock *sk, int flags)
        return 0;
 }
 
-/* return:
- *     1  if the the UDP system should process it
- *     0  if we should drop this packet
- *     -1 if it should get processed by xfrm4_rcv_encap
- */
-static int udp_encap_rcv(struct sock * sk, struct sk_buff *skb)
-{
-#ifndef CONFIG_XFRM
-       return 1;
-#else
-       struct udp_sock *up = udp_sk(sk);
-       struct udphdr *uh;
-       struct iphdr *iph;
-       int iphlen, len;
-
-       __u8 *udpdata;
-       __be32 *udpdata32;
-       __u16 encap_type = up->encap_type;
-
-       /* if we're overly short, let UDP handle it */
-       len = skb->len - sizeof(struct udphdr);
-       if (len <= 0)
-               return 1;
-
-       /* if this is not encapsulated socket, then just return now */
-       if (!encap_type)
-               return 1;
-
-       /* If this is a paged skb, make sure we pull up
-        * whatever data we need to look at. */
-       if (!pskb_may_pull(skb, sizeof(struct udphdr) + min(len, 8)))
-               return 1;
-
-       /* Now we can get the pointers */
-       uh = skb->h.uh;
-       udpdata = (__u8 *)uh + sizeof(struct udphdr);
-       udpdata32 = (__be32 *)udpdata;
-
-       switch (encap_type) {
-       default:
-       case UDP_ENCAP_ESPINUDP:
-               /* Check if this is a keepalive packet.  If so, eat it. */
-               if (len == 1 && udpdata[0] == 0xff) {
-                       return 0;
-               } else if (len > sizeof(struct ip_esp_hdr) && udpdata32[0] != 0 ) {
-                       /* ESP Packet without Non-ESP header */
-                       len = sizeof(struct udphdr);
-               } else
-                       /* Must be an IKE packet.. pass it through */
-                       return 1;
-               break;
-       case UDP_ENCAP_ESPINUDP_NON_IKE:
-               /* Check if this is a keepalive packet.  If so, eat it. */
-               if (len == 1 && udpdata[0] == 0xff) {
-                       return 0;
-               } else if (len > 2 * sizeof(u32) + sizeof(struct ip_esp_hdr) &&
-                          udpdata32[0] == 0 && udpdata32[1] == 0) {
-
-                       /* ESP Packet with Non-IKE marker */
-                       len = sizeof(struct udphdr) + 2 * sizeof(u32);
-               } else
-                       /* Must be an IKE packet.. pass it through */
-                       return 1;
-               break;
-       }
-
-       /* At this point we are sure that this is an ESPinUDP packet,
-        * so we need to remove 'len' bytes from the packet (the UDP
-        * header and optional ESP marker bytes) and then modify the
-        * protocol to ESP, and then call into the transform receiver.
-        */
-       if (skb_cloned(skb) && pskb_expand_head(skb, 0, 0, GFP_ATOMIC))
-               return 0;
-
-       /* Now we can update and verify the packet length... */
-       iph = skb->nh.iph;
-       iphlen = iph->ihl << 2;
-       iph->tot_len = htons(ntohs(iph->tot_len) - len);
-       if (skb->len < iphlen + len) {
-               /* packet is too small!?! */
-               return 0;
-       }
-
-       /* pull the data buffer up to the ESP header and set the
-        * transport header to point to ESP.  Keep UDP on the stack
-        * for later.
-        */
-       skb->h.raw = skb_pull(skb, len);
-
-       /* modify the protocol (it's ESP!) */
-       iph->protocol = IPPROTO_ESP;
-
-       /* and let the caller know to send this into the ESP processor... */
-       return -1;
-#endif
-}
-
 /* returns:
  *  -1: error
  *   0: success
@@ -1034,28 +949,28 @@ int udp_queue_rcv_skb(struct sock * sk, struct sk_buff *skb)
 
        if (up->encap_type) {
                /*
-                * This is an encapsulation socket, so let's see if this is
-                * an encapsulated packet.
-                * If it's a keepalive packet, then just eat it.
-                * If it's an encapsulateed packet, then pass it to the
-                * IPsec xfrm input and return the response
-                * appropriately.  Otherwise, just fall through and
-                * pass this up the UDP socket.
+                * This is an encapsulation socket so pass the skb to
+                * the socket's udp_encap_rcv() hook. Otherwise, just
+                * fall through and pass this up the UDP socket.
+                * up->encap_rcv() returns the following value:
+                * =0 if skb was successfully passed to the encap
+                *    handler or was discarded by it.
+                * >0 if skb should be passed on to UDP.
+                * <0 if skb should be resubmitted as proto -N
                 */
-               int ret;
 
-               ret = udp_encap_rcv(sk, skb);
-               if (ret == 0) {
-                       /* Eat the packet .. */
-                       kfree_skb(skb);
-                       return 0;
-               }
-               if (ret < 0) {
-                       /* process the ESP packet */
-                       ret = xfrm4_rcv_encap(skb, up->encap_type);
-                       UDP_INC_STATS_BH(UDP_MIB_INDATAGRAMS, up->pcflag);
-                       return -ret;
+               /* if we're overly short, let UDP handle it */
+               if (skb->len > sizeof(struct udphdr) &&
+                   up->encap_rcv != NULL) {
+                       int ret;
+
+                       ret = (*up->encap_rcv)(sk, skb);
+                       if (ret <= 0) {
+                               UDP_INC_STATS_BH(UDP_MIB_INDATAGRAMS, up->pcflag);
+                               return -ret;
+                       }
                }
+
                /* FALLTHROUGH -- it's a UDP Packet */
        }
 
@@ -1095,10 +1010,9 @@ int udp_queue_rcv_skb(struct sock * sk, struct sk_buff *skb)
                }
        }
 
-       if (sk->sk_filter && skb->ip_summed != CHECKSUM_UNNECESSARY) {
-               if (__udp_lib_checksum_complete(skb))
+       if (sk->sk_filter) {
+               if (udp_lib_checksum_complete(skb))
                        goto drop;
-               skb->ip_summed = CHECKSUM_UNNECESSARY;
        }
 
        if ((rc = sock_queue_rcv_skb(sk,skb)) < 0) {
@@ -1143,10 +1057,10 @@ static int __udp4_lib_mcast_deliver(struct sk_buff *skb,
 
                        sknext = udp_v4_mcast_next(sk_next(sk), uh->dest, daddr,
                                                   uh->source, saddr, dif);
-                       if(sknext)
+                       if (sknext)
                                skb1 = skb_clone(skb, GFP_ATOMIC);
 
-                       if(skb1) {
+                       if (skb1) {
                                int ret = udp_queue_rcv_skb(sk, skb1);
                                if (ret > 0)
                                        /* we should probably re-process instead
@@ -1154,7 +1068,7 @@ static int __udp4_lib_mcast_deliver(struct sk_buff *skb,
                                        kfree_skb(skb1);
                        }
                        sk = sknext;
-               } while(sknext);
+               } while (sknext);
        } else
                kfree_skb(skb);
        read_unlock(&udp_hash_lock);
@@ -1166,25 +1080,37 @@ static int __udp4_lib_mcast_deliver(struct sk_buff *skb,
  * Otherwise, csum completion requires chacksumming packet body,
  * including udp header and folding it to skb->csum.
  */
-static inline void udp4_csum_init(struct sk_buff *skb, struct udphdr *uh)
+static inline int udp4_csum_init(struct sk_buff *skb, struct udphdr *uh,
+                                int proto)
 {
+       const struct iphdr *iph;
+       int err;
+
+       UDP_SKB_CB(skb)->partial_cov = 0;
+       UDP_SKB_CB(skb)->cscov = skb->len;
+
+       if (proto == IPPROTO_UDPLITE) {
+               err = udplite_checksum_init(skb, uh);
+               if (err)
+                       return err;
+       }
+
+       iph = ip_hdr(skb);
        if (uh->check == 0) {
                skb->ip_summed = CHECKSUM_UNNECESSARY;
        } else if (skb->ip_summed == CHECKSUM_COMPLETE) {
-              if (!csum_tcpudp_magic(skb->nh.iph->saddr, skb->nh.iph->daddr,
-                                     skb->len, IPPROTO_UDP, skb->csum       ))
+              if (!csum_tcpudp_magic(iph->saddr, iph->daddr, skb->len,
+                                     proto, skb->csum))
                        skb->ip_summed = CHECKSUM_UNNECESSARY;
        }
-       if (skb->ip_summed != CHECKSUM_UNNECESSARY)
-               skb->csum = csum_tcpudp_nofold(skb->nh.iph->saddr,
-                                              skb->nh.iph->daddr,
-                                              skb->len, IPPROTO_UDP, 0);
+       if (!skb_csum_unnecessary(skb))
+               skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
+                                              skb->len, proto, 0);
        /* Probably, we should checksum udp header (it should be in cache
         * in any case) and data in tiny packets (< rx copybreak).
         */
 
-       /* UDP = UDP-Lite with a non-partial checksum coverage */
-       UDP_SKB_CB(skb)->partial_cov = 0;
+       return 0;
 }
 
 /*
@@ -1192,14 +1118,14 @@ static inline void udp4_csum_init(struct sk_buff *skb, struct udphdr *uh)
  */
 
 int __udp4_lib_rcv(struct sk_buff *skb, struct hlist_head udptable[],
-                  int is_udplite)
+                  int proto)
 {
        struct sock *sk;
-       struct udphdr *uh = skb->h.uh;
+       struct udphdr *uh = udp_hdr(skb);
        unsigned short ulen;
        struct rtable *rt = (struct rtable*)skb->dst;
-       __be32 saddr = skb->nh.iph->saddr;
-       __be32 daddr = skb->nh.iph->daddr;
+       __be32 saddr = ip_hdr(skb)->saddr;
+       __be32 daddr = ip_hdr(skb)->daddr;
 
        /*
         *  Validate the packet.
@@ -1211,20 +1137,17 @@ int __udp4_lib_rcv(struct sk_buff *skb, struct hlist_head udptable[],
        if (ulen > skb->len)
                goto short_packet;
 
-       if(! is_udplite ) {             /* UDP validates ulen. */
-
+       if (proto == IPPROTO_UDP) {
+               /* UDP validates ulen. */
                if (ulen < sizeof(*uh) || pskb_trim_rcsum(skb, ulen))
                        goto short_packet;
-               uh = skb->h.uh;
-
-               udp4_csum_init(skb, uh);
-
-       } else  {                       /* UDP-Lite validates cscov. */
-               if (udplite4_csum_init(skb, uh))
-                       goto csum_error;
+               uh = udp_hdr(skb);
        }
 
-       if(rt->rt_flags & (RTCF_BROADCAST|RTCF_MULTICAST))
+       if (udp4_csum_init(skb, uh, proto))
+               goto csum_error;
+
+       if (rt->rt_flags & (RTCF_BROADCAST|RTCF_MULTICAST))
                return __udp4_lib_mcast_deliver(skb, uh, saddr, daddr, udptable);
 
        sk = __udp4_lib_lookup(saddr, uh->source, daddr, uh->dest,
@@ -1250,7 +1173,7 @@ int __udp4_lib_rcv(struct sk_buff *skb, struct hlist_head udptable[],
        if (udp_lib_checksum_complete(skb))
                goto csum_error;
 
-       UDP_INC_STATS_BH(UDP_MIB_NOPORTS, is_udplite);
+       UDP_INC_STATS_BH(UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE);
        icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PORT_UNREACH, 0);
 
        /*
@@ -1258,11 +1181,11 @@ int __udp4_lib_rcv(struct sk_buff *skb, struct hlist_head udptable[],
         * don't wanna listen.  Ignore it.
         */
        kfree_skb(skb);
-       return(0);
+       return 0;
 
 short_packet:
        LIMIT_NETDEBUG(KERN_DEBUG "UDP%s: short packet: From %u.%u.%u.%u:%u %d/%d to %u.%u.%u.%u:%u\n",
-                      is_udplite? "-Lite" : "",
+                      proto == IPPROTO_UDPLITE ? "-Lite" : "",
                       NIPQUAD(saddr),
                       ntohs(uh->source),
                       ulen,
@@ -1277,21 +1200,21 @@ csum_error:
         * the network is concerned, anyway) as per 4.1.3.4 (MUST).
         */
        LIMIT_NETDEBUG(KERN_DEBUG "UDP%s: bad checksum. From %d.%d.%d.%d:%d to %d.%d.%d.%d:%d ulen %d\n",
-                      is_udplite? "-Lite" : "",
+                      proto == IPPROTO_UDPLITE ? "-Lite" : "",
                       NIPQUAD(saddr),
                       ntohs(uh->source),
                       NIPQUAD(daddr),
                       ntohs(uh->dest),
                       ulen);
 drop:
-       UDP_INC_STATS_BH(UDP_MIB_INERRORS, is_udplite);
+       UDP_INC_STATS_BH(UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE);
        kfree_skb(skb);
-       return(0);
+       return 0;
 }
 
-__inline__ int udp_rcv(struct sk_buff *skb)
+int udp_rcv(struct sk_buff *skb)
 {
-       return __udp4_lib_rcv(skb, udp_hash, 0);
+       return __udp4_lib_rcv(skb, udp_hash, IPPROTO_UDP);
 }
 
 int udp_destroy_sock(struct sock *sk)
@@ -1313,13 +1236,13 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
        int val;
        int err = 0;
 
-       if(optlen<sizeof(int))
+       if (optlen<sizeof(int))
                return -EINVAL;
 
        if (get_user(val, (int __user *)optval))
                return -EFAULT;
 
-       switch(optname) {
+       switch (optname) {
        case UDP_CORK:
                if (val != 0) {
                        up->corkflag = 1;
@@ -1336,6 +1259,9 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
                case 0:
                case UDP_ENCAP_ESPINUDP:
                case UDP_ENCAP_ESPINUDP_NON_IKE:
+                       up->encap_rcv = xfrm4_udp_encap_rcv;
+                       /* FALLTHROUGH */
+               case UDP_ENCAP_L2TPINUDP:
                        up->encap_type = val;
                        break;
                default:
@@ -1373,7 +1299,7 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
        default:
                err = -ENOPROTOOPT;
                break;
-       };
+       }
 
        return err;
 }
@@ -1404,15 +1330,15 @@ int udp_lib_getsockopt(struct sock *sk, int level, int optname,
        struct udp_sock *up = udp_sk(sk);
        int val, len;
 
-       if(get_user(len,optlen))
+       if (get_user(len,optlen))
                return -EFAULT;
 
        len = min_t(unsigned int, len, sizeof(int));
 
-       if(len < 0)
+       if (len < 0)
                return -EINVAL;
 
-       switch(optname) {
+       switch (optname) {
        case UDP_CORK:
                val = up->corkflag;
                break;
@@ -1433,11 +1359,11 @@ int udp_lib_getsockopt(struct sock *sk, int level, int optname,
 
        default:
                return -ENOPROTOOPT;
-       };
+       }
 
-       if(put_user(len, optlen))
+       if (put_user(len, optlen))
                return -EFAULT;
-       if(copy_to_user(optval, &val,len))
+       if (copy_to_user(optval, &val,len))
                return -EFAULT;
        return 0;
 }
@@ -1486,15 +1412,11 @@ unsigned int udp_poll(struct file *file, struct socket *sock, poll_table *wait)
                struct sk_buff *skb;
 
                spin_lock_bh(&rcvq->lock);
-               while ((skb = skb_peek(rcvq)) != NULL) {
-                       if (udp_lib_checksum_complete(skb)) {
-                               UDP_INC_STATS_BH(UDP_MIB_INERRORS, is_lite);
-                               __skb_unlink(skb, rcvq);
-                               kfree_skb(skb);
-                       } else {
-                               skb->ip_summed = CHECKSUM_UNNECESSARY;
-                               break;
-                       }
+               while ((skb = skb_peek(rcvq)) != NULL &&
+                      udp_lib_checksum_complete(skb)) {
+                       UDP_INC_STATS_BH(UDP_MIB_INERRORS, is_lite);
+                       __skb_unlink(skb, rcvq);
+                       kfree_skb(skb);
                }
                spin_unlock_bh(&rcvq->lock);
 
@@ -1573,7 +1495,7 @@ static struct sock *udp_get_idx(struct seq_file *seq, loff_t pos)
        struct sock *sk = udp_get_first(seq);
 
        if (sk)
-               while(pos && (sk = udp_get_next(seq, sk)) != NULL)
+               while (pos && (sk = udp_get_next(seq, sk)) != NULL)
                        --pos;
        return pos ? NULL : sk;
 }
@@ -1645,7 +1567,7 @@ int udp_proc_register(struct udp_seq_afinfo *afinfo)
        afinfo->seq_fops->llseek        = seq_lseek;
        afinfo->seq_fops->release       = seq_release_private;
 
-       p = proc_net_fops_create(afinfo->name, S_IRUGO, afinfo->seq_fops);
+       p = proc_net_fops_create(&init_net, afinfo->name, S_IRUGO, afinfo->seq_fops);
        if (p)
                p->data = afinfo;
        else
@@ -1657,7 +1579,7 @@ void udp_proc_unregister(struct udp_seq_afinfo *afinfo)
 {
        if (!afinfo)
                return;
-       proc_net_remove(afinfo->name);
+       proc_net_remove(&init_net, afinfo->name);
        memset(afinfo->seq_fops, 0, sizeof(*afinfo->seq_fops));
 }