tcp_ipv4.c

/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Implementation of the Transmission Control Protocol(TCP).
 *
 * Version:	$Id: tcp_ipv4.c,v 1.240 2002/02/01 22:01:04 davem Exp $
 *
 *		IPv4 specific functions
 *
 *
 *		code split from:
 *		linux/ipv4/tcp.c
 *		linux/ipv4/tcp_input.c
 *		linux/ipv4/tcp_output.c
 *
 *		See tcp.c for author information
 *
 *	This program is free software; you can redistribute it and/or
 *      modify it under the terms of the GNU General Public License
 *      as published by the Free Software Foundation; either version
 *      2 of the License, or (at your option) any later version.
 */

/*
 * Changes:
 *		David S. Miller	:	New socket lookup architecture.
 *					This code is dedicated to John Dyson.
 *		David S. Miller :	Change semantics of established hash,
 *					half is devoted to TIME_WAIT sockets
 *					and the rest go in the other half.
 *		Andi Kleen :		Add support for syncookies and fixed
 *					some bugs: ip options weren't passed to
 *					the TCP layer, missed a check for an
 *					ACK bit.
 *		Andi Kleen :		Implemented fast path mtu discovery.
 *	     				Fixed many serious bugs in the
 *					request_sock handling and moved
 *					most of it into the af independent code.
 *					Added tail drop and some other bugfixes.
 *					Added new listen sematics.
 *		Mike McLagan	:	Routing by source
 *	Juan Jose Ciarlante:		ip_dynaddr bits
 *		Andi Kleen:		various fixes.
 *	Vitaly E. Lavrov	:	Transparent proxy revived after year
 *					coma.
 *	Andi Kleen		:	Fix new listen.
 *	Andi Kleen		:	Fix accept error reporting.
 *	YOSHIFUJI Hideaki @USAGI and:	Support IPV6_V6ONLY socket option, which
 *	Alexey Kuznetsov		allow both IPv4 and IPv6 sockets to bind
 *					a single port at the same time.
 */

#include <linux/config.h>

#include <linux/types.h>
#include <linux/fcntl.h>
#include <linux/module.h>
#include <linux/random.h>
#include <linux/cache.h>
#include <linux/jhash.h>
#include <linux/init.h>
#include <linux/times.h>

#include <net/icmp.h>
#include <net/tcp.h>
#include <net/ipv6.h>
#include <net/inet_common.h>
#include <net/xfrm.h>

#include <linux/inet.h>
#include <linux/ipv6.h>
#include <linux/stddef.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>

extern int sysctl_ip_dynaddr;
int sysctl_tcp_tw_reuse;
int sysctl_tcp_low_latency;

/* Check TCP sequence numbers in ICMP packets. */
#define ICMP_MIN_LENGTH 8

/* Socket used for sending RSTs */
static struct socket *tcp_socket;

void tcp_v4_send_check(struct sock *sk, struct tcphdr *th, int len,
		       struct sk_buff *skb);

struct tcp_hashinfo __cacheline_aligned tcp_hashinfo = {
	.__tcp_lhash_lock	=	RW_LOCK_UNLOCKED,
	.__tcp_lhash_users	=	ATOMIC_INIT(0),
	.__tcp_lhash_wait
	  = __WAIT_QUEUE_HEAD_INITIALIZER(tcp_hashinfo.__tcp_lhash_wait),
	.__tcp_portalloc_lock	=	SPIN_LOCK_UNLOCKED
};

/*
 * This array holds the first and last local port number.
 * For high-usage systems, use sysctl to change this to
 * 32768-61000
 */
int sysctl_local_port_range[2] = { 1024, 4999 };
int tcp_port_rover = 1024 - 1;

static __inline__ int tcp_hashfn(__u32 laddr, __u16 lport,
				 __u32 faddr, __u16 fport)
{
	int h = (laddr ^ lport) ^ (faddr ^ fport);
	h ^= h >> 16;
	h ^= h >> 8;
	return h & (tcp_ehash_size - 1);
}

static __inline__ int tcp_sk_hashfn(struct sock *sk)
{
	struct inet_sock *inet = inet_sk(sk);
	__u32 laddr = inet->rcv_saddr;
	__u16 lport = inet->num;
	__u32 faddr = inet->daddr;
	__u16 fport = inet->dport;

	return tcp_hashfn(laddr, lport, faddr, fport);
}

/* Allocate and initialize a new TCP local port bind bucket.
 * The bindhash mutex for snum's hash chain must be held here.
 */
struct tcp_bind_bucket *tcp_bucket_create(struct tcp_bind_hashbucket *head,
					  unsigned short snum)
{
	struct tcp_bind_bucket *tb = kmem_cache_alloc(tcp_bucket_cachep,
						      SLAB_ATOMIC);
	if (tb) {
		tb->port = snum;
		tb->fastreuse = 0;
		INIT_HLIST_HEAD(&tb->owners);
		hlist_add_head(&tb->node, &head->chain);
	}
	return tb;
}

/* Caller must hold hashbucket lock for this tb with local BH disabled */
void tcp_bucket_destroy(struct tcp_bind_bucket *tb)
{
	if (hlist_empty(&tb->owners)) {
		__hlist_del(&tb->node);
		kmem_cache_free(tcp_bucket_cachep, tb);
	}
}

/* Caller must disable local BH processing. */
static __inline__ void __tcp_inherit_port(struct sock *sk, struct sock *child)
{
	struct tcp_bind_hashbucket *head =
				&tcp_bhash[tcp_bhashfn(inet_sk(child)->num)];
	struct tcp_bind_bucket *tb;

	spin_lock(&head->lock);
	tb = tcp_sk(sk)->bind_hash;
	sk_add_bind_node(child, &tb->owners);
	tcp_sk(child)->bind_hash = tb;
	spin_unlock(&head->lock);
}

inline void tcp_inherit_port(struct sock *sk, struct sock *child)
{
	local_bh_disable();
	__tcp_inherit_port(sk, child);
	local_bh_enable();
}

void tcp_bind_hash(struct sock *sk, struct tcp_bind_bucket *tb,
		   unsigned short snum)
{
	inet_sk(sk)->num = snum;
	sk_add_bind_node(sk, &tb->owners);
	tcp_sk(sk)->bind_hash = tb;
}

static inline int tcp_bind_conflict(struct sock *sk, struct tcp_bind_bucket *tb)
{
	const u32 sk_rcv_saddr = tcp_v4_rcv_saddr(sk);
	struct sock *sk2;
	struct hlist_node *node;
	int reuse = sk->sk_reuse;

	sk_for_each_bound(sk2, node, &tb->owners) {
		if (sk != sk2 &&
		    !tcp_v6_ipv6only(sk2) &&
		    (!sk->sk_bound_dev_if ||
		     !sk2->sk_bound_dev_if ||
		     sk->sk_bound_dev_if == sk2->sk_bound_dev_if)) {
			if (!reuse || !sk2->sk_reuse ||
			    sk2->sk_state == TCP_LISTEN) {
				const u32 sk2_rcv_saddr = tcp_v4_rcv_saddr(sk2);
				if (!sk2_rcv_saddr || !sk_rcv_saddr ||
				    sk2_rcv_saddr == sk_rcv_saddr)
					break;
			}
		}
	}
	return node != NULL;
}

/* Obtain a reference to a local port for the given sock,
 * if snum is zero it means select any available local port.
 */
static int tcp_v4_get_port(struct sock *sk, unsigned short snum)
{
	struct tcp_bind_hashbucket *head;
	struct hlist_node *node;
	struct tcp_bind_bucket *tb;
	int ret;

	local_bh_disable();
	if (!snum) {
		int low = sysctl_local_port_range[0];
		int high = sysctl_local_port_range[1];
		int remaining = (high - low) + 1;
		int rover;

		spin_lock(&tcp_portalloc_lock);
		if (tcp_port_rover < low)
			rover = low;
		else
			rover = tcp_port_rover;
		do {
			rover++;
			if (rover > high)
				rover = low;
			head = &tcp_bhash[tcp_bhashfn(rover)];
			spin_lock(&head->lock);
			tb_for_each(tb, node, &head->chain)
				if (tb->port == rover)
					goto next;
			break;
		next:
			spin_unlock(&head->lock);
		} while (--remaining > 0);
		tcp_port_rover = rover;
		spin_unlock(&tcp_portalloc_lock);

		/* Exhausted local port range during search?  It is not
		 * possible for us to be holding one of the bind hash
		 * locks if this test triggers, because if 'remaining'
		 * drops to zero, we broke out of the do/while loop at
		 * the top level, not from the 'break;' statement.
		 */
		ret = 1;
		if (unlikely(remaining <= 0))
			goto fail;

		/* OK, here is the one we will use.  HEAD is
		 * non-NULL and we hold it's mutex.
		 */
		snum = rover;
	} else {
		head = &tcp_bhash[tcp_bhashfn(snum)];
		spin_lock(&head->lock);
		tb_for_each(tb, node, &head->chain)
			if (tb->port == snum)
				goto tb_found;
	}
	tb = NULL;
	goto tb_not_found;
tb_found:
	if (!hlist_empty(&tb->owners)) {
		if (sk->sk_reuse > 1)
			goto success;
		if (tb->fastreuse > 0 &&
		    sk->sk_reuse && sk->sk_state != TCP_LISTEN) {
			goto success;
		} else {
			ret = 1;
			if (tcp_bind_conflict(sk, tb))
				goto fail_unlock;
		}
	}
tb_not_found:
	ret = 1;
	if (!tb && (tb = tcp_bucket_create(head, snum)) == NULL)
		goto fail_unlock;
	if (hlist_empty(&tb->owners)) {
		if (sk->sk_reuse && sk->sk_state != TCP_LISTEN)
			tb->fastreuse = 1;
		else
			tb->fastreuse = 0;
	} else if (tb->fastreuse &&
		   (!sk->sk_reuse || sk->sk_state == TCP_LISTEN))
		tb->fastreuse = 0;
success:
	if (!tcp_sk(sk)->bind_hash)
		tcp_bind_hash(sk, tb, snum);
	BUG_TRAP(tcp_sk(sk)->bind_hash == tb);
 	ret = 0;

fail_unlock:
	spin_unlock(&head->lock);
fail:
	local_bh_enable();
	return ret;
}

/* Get rid of any references to a local port held by the
 * given sock.
 */
static void __tcp_put_port(struct sock *sk)
{
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_bind_hashbucket *head = &tcp_bhash[tcp_bhashfn(inet->num)];
	struct tcp_bind_bucket *tb;

	spin_lock(&head->lock);
	tb = tcp_sk(sk)->bind_hash;
	__sk_del_bind_node(sk);
	tcp_sk(sk)->bind_hash = NULL;
	inet->num = 0;
	tcp_bucket_destroy(tb);
	spin_unlock(&head->lock);
}

void tcp_put_port(struct sock *sk)
{
	local_bh_disable();
	__tcp_put_port(sk);
	local_bh_enable();
}

/* This lock without WQ_FLAG_EXCLUSIVE is good on UP and it can be very bad on SMP.
 * Look, when several writers sleep and reader wakes them up, all but one
 * immediately hit write lock and grab all the cpus. Exclusive sleep solves
 * this, _but_ remember, it adds useless work on UP machines (wake up each
 * exclusive lock release). It should be ifdefed really.
 */

void tcp_listen_wlock(void)
{
	write_lock(&tcp_lhash_lock);

	if (atomic_read(&tcp_lhash_users)) {
		DEFINE_WAIT(wait);

		for (;;) {
			prepare_to_wait_exclusive(&tcp_lhash_wait,
						&wait, TASK_UNINTERRUPTIBLE);
			if (!atomic_read(&tcp_lhash_users))
				break;
			write_unlock_bh(&tcp_lhash_lock);
			schedule();
			write_lock_bh(&tcp_lhash_lock);
		}

		finish_wait(&tcp_lhash_wait, &wait);
	}
}

static __inline__ void __tcp_v4_hash(struct sock *sk, const int listen_possible)
{
	struct hlist_head *list;
	rwlock_t *lock;

	BUG_TRAP(sk_unhashed(sk));
	if (listen_possible && sk->sk_state == TCP_LISTEN) {
		list = &tcp_listening_hash[tcp_sk_listen_hashfn(sk)];
		lock = &tcp_lhash_lock;
		tcp_listen_wlock();
	} else {
		list = &tcp_ehash[(sk->sk_hashent = tcp_sk_hashfn(sk))].chain;
		lock = &tcp_ehash[sk->sk_hashent].lock;
		write_lock(lock);
	}
	__sk_add_node(sk, list);
	sock_prot_inc_use(sk->sk_prot);
	write_unlock(lock);
	if (listen_possible && sk->sk_state == TCP_LISTEN)
		wake_up(&tcp_lhash_wait);
}

static void tcp_v4_hash(struct sock *sk)
{
	if (sk->sk_state != TCP_CLOSE) {
		local_bh_disable();
		__tcp_v4_hash(sk, 1);
		local_bh_enable();
	}
}

void tcp_unhash(struct sock *sk)
{
	rwlock_t *lock;

	if (sk_unhashed(sk))
		goto ende;

	if (sk->sk_state == TCP_LISTEN) {
		local_bh_disable();
		tcp_listen_wlock();
		lock = &tcp_lhash_lock;
	} else {
		struct tcp_ehash_bucket *head = &tcp_ehash[sk->sk_hashent];
		lock = &head->lock;
		write_lock_bh(&head->lock);
	}

	if (__sk_del_node_init(sk))
		sock_prot_dec_use(sk->sk_prot);
	write_unlock_bh(lock);

 ende:
	if (sk->sk_state == TCP_LISTEN)
		wake_up(&tcp_lhash_wait);
}

/* Don't inline this cruft.  Here are some nice properties to
 * exploit here.  The BSD API does not allow a listening TCP
 * to specify the remote port nor the remote address for the
 * connection.  So always assume those are both wildcarded
 * during the search since they can never be otherwise.
 */
static struct sock *__tcp_v4_lookup_listener(struct hlist_head *head, u32 daddr,
					     unsigned short hnum, int dif)
{
	struct sock *result = NULL, *sk;
	struct hlist_node *node;
	int score, hiscore;

	hiscore=-1;
	sk_for_each(sk, node, head) {
		struct inet_sock *inet = inet_sk(sk);

		if (inet->num == hnum && !ipv6_only_sock(sk)) {
			__u32 rcv_saddr = inet->rcv_saddr;

			score = (sk->sk_family == PF_INET ? 1 : 0);
			if (rcv_saddr) {
				if (rcv_saddr != daddr)
					continue;
				score+=2;
			}
			if (sk->sk_bound_dev_if) {
				if (sk->sk_bound_dev_if != dif)
					continue;
				score+=2;
			}
			if (score == 5)
				return sk;
			if (score > hiscore) {
				hiscore = score;
				result = sk;
			}
		}
	}
	return result;
}

/* Optimize the common listener case. */
static inline struct sock *tcp_v4_lookup_listener(u32 daddr,
		unsigned short hnum, int dif)
{
	struct sock *sk = NULL;
	struct hlist_head *head;

	read_lock(&tcp_lhash_lock);
	head = &tcp_listening_hash[tcp_lhashfn(hnum)];
	if (!hlist_empty(head)) {
		struct inet_sock *inet = inet_sk((sk = __sk_head(head)));

		if (inet->num == hnum && !sk->sk_node.next &&
		    (!inet->rcv_saddr || inet->rcv_saddr == daddr) &&
		    (sk->sk_family == PF_INET || !ipv6_only_sock(sk)) &&
		    !sk->sk_bound_dev_if)
			goto sherry_cache;
		sk = __tcp_v4_lookup_listener(head, daddr, hnum, dif);
	}
	if (sk) {
sherry_cache:
		sock_hold(sk);
	}
	read_unlock(&tcp_lhash_lock);
	return sk;
}

/* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so
 * we need not check it for TCP lookups anymore, thanks Alexey. -DaveM
 *
 * Local BH must be disabled here.
 */

static inline struct sock *__tcp_v4_lookup_established(u32 saddr, u16 sport,
						       u32 daddr, u16 hnum,
						       int dif)
{
	struct tcp_ehash_bucket *head;
	TCP_V4_ADDR_COOKIE(acookie, saddr, daddr)
	__u32 ports = TCP_COMBINED_PORTS(sport, hnum);
	struct sock *sk;
	struct hlist_node *node;
	/* Optimize here for direct hit, only listening connections can
	 * have wildcards anyways.
	 */
	int hash = tcp_hashfn(daddr, hnum, saddr, sport);
	head = &tcp_ehash[hash];
	read_lock(&head->lock);
	sk_for_each(sk, node, &head->chain) {
		if (TCP_IPV4_MATCH(sk, acookie, saddr, daddr, ports, dif))
			goto hit; /* You sunk my battleship! */
	}

	/* Must check for a TIME_WAIT'er before going to listener hash. */
	sk_for_each(sk, node, &(head + tcp_ehash_size)->chain) {
		if (TCP_IPV4_TW_MATCH(sk, acookie, saddr, daddr, ports, dif))
			goto hit;
	}
	sk = NULL;
out:
	read_unlock(&head->lock);
	return sk;
hit:
	sock_hold(sk);
	goto out;
}

static inline struct sock *__tcp_v4_lookup(u32 saddr, u16 sport,
					   u32 daddr, u16 hnum, int dif)
{
	struct sock *sk = __tcp_v4_lookup_established(saddr, sport,
						      daddr, hnum, dif);

	return sk ? : tcp_v4_lookup_listener(daddr, hnum, dif);
}

inline struct sock *tcp_v4_lookup(u32 saddr, u16 sport, u32 daddr,
				  u16 dport, int dif)
{
	struct sock *sk;

	local_bh_disable();
	sk = __tcp_v4_lookup(saddr, sport, daddr, ntohs(dport), dif);
	local_bh_enable();

	return sk;
}

EXPORT_SYMBOL_GPL(tcp_v4_lookup);

static inline __u32 tcp_v4_init_sequence(struct sock *sk, struct sk_buff *skb)
{
	return secure_tcp_sequence_number(skb->nh.iph->daddr,
					  skb->nh.iph->saddr,
					  skb->h.th->dest,
					  skb->h.th->source);
}

/* called with local bh disabled */
static int __tcp_v4_check_established(struct sock *sk, __u16 lport,
				      struct tcp_tw_bucket **twp)
{
	struct inet_sock *inet = inet_sk(sk);
	u32 daddr = inet->rcv_saddr;
	u32 saddr = inet->daddr;
	int dif = sk->sk_bound_dev_if;
	TCP_V4_ADDR_COOKIE(acookie, saddr, daddr)
	__u32 ports = TCP_COMBINED_PORTS(inet->dport, lport);
	int hash = tcp_hashfn(daddr, lport, saddr, inet->dport);
	struct tcp_ehash_bucket *head = &tcp_ehash[hash];
	struct sock *sk2;
	struct hlist_node *node;
	struct tcp_tw_bucket *tw;

	write_lock(&head->lock);

	/* Check TIME-WAIT sockets first. */
	sk_for_each(sk2, node, &(head + tcp_ehash_size)->chain) {
		tw = (struct tcp_tw_bucket *)sk2;

		if (TCP_IPV4_TW_MATCH(sk2, acookie, saddr, daddr, ports, dif)) {
			struct tcp_sock *tp = tcp_sk(sk);

			/* With PAWS, it is safe from the viewpoint
			   of data integrity. Even without PAWS it
			   is safe provided sequence spaces do not
			   overlap i.e. at data rates <= 80Mbit/sec.

			   Actually, the idea is close to VJ's one,
			   only timestamp cache is held not per host,
			   but per port pair and TW bucket is used
			   as state holder.

			   If TW bucket has been already destroyed we
			   fall back to VJ's scheme and use initial
			   timestamp retrieved from peer table.
			 */
			if (tw->tw_ts_recent_stamp &&
			    (!twp || (sysctl_tcp_tw_reuse &&
				      xtime.tv_sec -
				      tw->tw_ts_recent_stamp > 1))) {
				if ((tp->write_seq =
						tw->tw_snd_nxt + 65535 + 2) == 0)
					tp->write_seq = 1;
				tp->rx_opt.ts_recent	   = tw->tw_ts_recent;
				tp->rx_opt.ts_recent_stamp = tw->tw_ts_recent_stamp;
				sock_hold(sk2);
				goto unique;
			} else
				goto not_unique;
		}
	}
	tw = NULL;

	/* And established part... */
	sk_for_each(sk2, node, &head->chain) {
		if (TCP_IPV4_MATCH(sk2, acookie, saddr, daddr, ports, dif))
			goto not_unique;
	}

unique:
	/* Must record num and sport now. Otherwise we will see
	 * in hash table socket with a funny identity. */
	inet->num = lport;
	inet->sport = htons(lport);
	sk->sk_hashent = hash;
	BUG_TRAP(sk_unhashed(sk));
	__sk_add_node(sk, &head->chain);
	sock_prot_inc_use(sk->sk_prot);
	write_unlock(&head->lock);

	if (twp) {
		*twp = tw;
		NET_INC_STATS_BH(LINUX_MIB_TIMEWAITRECYCLED);
	} else if (tw) {
		/* Silly. Should hash-dance instead... */
		tcp_tw_deschedule(tw);
		NET_INC_STATS_BH(LINUX_MIB_TIMEWAITRECYCLED);

		tcp_tw_put(tw);
	}

	return 0;

not_unique:
	write_unlock(&head->lock);
	return -EADDRNOTAVAIL;
}

static inline u32 connect_port_offset(const struct sock *sk)
{
	const struct inet_sock *inet = inet_sk(sk);

	return secure_tcp_port_ephemeral(inet->rcv_saddr, inet->daddr, 
					 inet->dport);
}

/*
 * Bind a port for a connect operation and hash it.
 */
static inline int tcp_v4_hash_connect(struct sock *sk)
{
	unsigned short snum = inet_sk(sk)->num;
 	struct tcp_bind_hashbucket *head;
 	struct tcp_bind_bucket *tb;
	int ret;

 	if (!snum) {
 		int low = sysctl_local_port_range[0];
 		int high = sysctl_local_port_range[1];
		int range = high - low;
 		int i;
		int port;
		static u32 hint;
		u32 offset = hint + connect_port_offset(sk);
		struct hlist_node *node;
 		struct tcp_tw_bucket *tw = NULL;

 		local_bh_disable();
		for (i = 1; i <= range; i++) {
			port = low + (i + offset) % range;
 			head = &tcp_bhash[tcp_bhashfn(port)];
 			spin_lock(&head->lock);

 			/* Does not bother with rcv_saddr checks,
 			 * because the established check is already
 			 * unique enough.
 			 */
			tb_for_each(tb, node, &head->chain) {
 				if (tb->port == port) {
 					BUG_TRAP(!hlist_empty(&tb->owners));
 					if (tb->fastreuse >= 0)
 						goto next_port;
 					if (!__tcp_v4_check_established(sk,
									port,
									&tw))
 						goto ok;
 					goto next_port;
 				}
 			}

 			tb = tcp_bucket_create(head, port);
 			if (!tb) {
 				spin_unlock(&head->lock);
 				break;
 			}
 			tb->fastreuse = -1;
 			goto ok;

 		next_port:
 			spin_unlock(&head->lock);
 		}
 		local_bh_enable();

 		return -EADDRNOTAVAIL;

ok:
		hint += i;

 		/* Head lock still held and bh's disabled */
 		tcp_bind_hash(sk, tb, port);
		if (sk_unhashed(sk)) {
 			inet_sk(sk)->sport = htons(port);
 			__tcp_v4_hash(sk, 0);
 		}
 		spin_unlock(&head->lock);

 		if (tw) {
 			tcp_tw_deschedule(tw);
 			tcp_tw_put(tw);
 		}

		ret = 0;
		goto out;
 	}

 	head  = &tcp_bhash[tcp_bhashfn(snum)];
 	tb  = tcp_sk(sk)->bind_hash;
	spin_lock_bh(&head->lock);
	if (sk_head(&tb->owners) == sk && !sk->sk_bind_node.next) {
		__tcp_v4_hash(sk, 0);
		spin_unlock_bh(&head->lock);
		return 0;
	} else {
		spin_unlock(&head->lock);
		/* No definite answer... Walk to established hash table */
		ret = __tcp_v4_check_established(sk, snum, NULL);
out:
		local_bh_enable();
		return ret;
	}
}

/* This will initiate an outgoing connection. */
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
{
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
	struct sockaddr_in *usin = (struct sockaddr_in *)uaddr;
	struct rtable *rt;
	u32 daddr, nexthop;
	int tmp;
	int err;

	if (addr_len < sizeof(struct sockaddr_in))
		return -EINVAL;

	if (usin->sin_family != AF_INET)
		return -EAFNOSUPPORT;

	nexthop = daddr = usin->sin_addr.s_addr;
	if (inet->opt && inet->opt->srr) {
		if (!daddr)
			return -EINVAL;
		nexthop = inet->opt->faddr;
	}

	tmp = ip_route_connect(&rt, nexthop, inet->saddr,
			       RT_CONN_FLAGS(sk), sk->sk_bound_dev_if,
			       IPPROTO_TCP,
			       inet->sport, usin->sin_port, sk);
	if (tmp < 0)
		return tmp;

	if (rt->rt_flags & (RTCF_MULTICAST | RTCF_BROADCAST)) {
		ip_rt_put(rt);
		return -ENETUNREACH;
	}

	if (!inet->opt || !inet->opt->srr)
		daddr = rt->rt_dst;

	if (!inet->saddr)
		inet->saddr = rt->rt_src;
	inet->rcv_saddr = inet->saddr;

	if (tp->rx_opt.ts_recent_stamp && inet->daddr != daddr) {
		/* Reset inherited state */
		tp->rx_opt.ts_recent	   = 0;
		tp->rx_opt.ts_recent_stamp = 0;
		tp->write_seq		   = 0;
	}

	if (sysctl_tcp_tw_recycle &&
	    !tp->rx_opt.ts_recent_stamp && rt->rt_dst == daddr) {
		struct inet_peer *peer = rt_get_peer(rt);

		/* VJ's idea. We save last timestamp seen from
		 * the destination in peer table, when entering state TIME-WAIT
		 * and initialize rx_opt.ts_recent from it, when trying new connection.
		 */

		if (peer && peer->tcp_ts_stamp + TCP_PAWS_MSL >= xtime.tv_sec) {
			tp->rx_opt.ts_recent_stamp = peer->tcp_ts_stamp;
			tp->rx_opt.ts_recent = peer->tcp_ts;
		}
	}

	inet->dport = usin->sin_port;
	inet->daddr = daddr;

	tp->ext_header_len = 0;
	if (inet->opt)
		tp->ext_header_len = inet->opt->optlen;

	tp->rx_opt.mss_clamp = 536;

	/* Socket identity is still unknown (sport may be zero).
	 * However we set state to SYN-SENT and not releasing socket
	 * lock select source port, enter ourselves into the hash tables and
	 * complete initialization after this.
	 */
	tcp_set_state(sk, TCP_SYN_SENT);
	err = tcp_v4_hash_connect(sk);
	if (err)
		goto failure;

	err = ip_route_newports(&rt, inet->sport, inet->dport, sk);
	if (err)
		goto failure;

	/* OK, now commit destination to socket.  */
	__sk_dst_set(sk, &rt->u.dst);
	tcp_v4_setup_caps(sk, &rt->u.dst);

	if (!tp->write_seq)
		tp->write_seq = secure_tcp_sequence_number(inet->saddr,
							   inet->daddr,
							   inet->sport,
							   usin->sin_port);

	inet->id = tp->write_seq ^ jiffies;

	err = tcp_connect(sk);
	rt = NULL;
	if (err)
		goto failure;

	return 0;

failure:
	/* This unhashes the socket and releases the local port, if necessary. */
	tcp_set_state(sk, TCP_CLOSE);
	ip_rt_put(rt);
	sk->sk_route_caps = 0;
	inet->dport = 0;
	return err;
}

static __inline__ int tcp_v4_iif(struct sk_buff *skb)
{
	return ((struct rtable *)skb->dst)->rt_iif;
}

static __inline__ u32 tcp_v4_synq_hash(u32 raddr, u16 rport, u32 rnd)
{
	return (jhash_2words(raddr, (u32) rport, rnd) & (TCP_SYNQ_HSIZE - 1));
}

static struct request_sock *tcp_v4_search_req(struct tcp_sock *tp,
					      struct request_sock ***prevp,
					      __u16 rport,
					      __u32 raddr, __u32 laddr)
{
	struct listen_sock *lopt = tp->accept_queue.listen_opt;
	struct request_sock *req, **prev;

	for (prev = &lopt->syn_table[tcp_v4_synq_hash(raddr, rport, lopt->hash_rnd)];
	     (req = *prev) != NULL;
	     prev = &req->dl_next) {
		const struct inet_request_sock *ireq = inet_rsk(req);

		if (ireq->rmt_port == rport &&
		    ireq->rmt_addr == raddr &&
		    ireq->loc_addr == laddr &&
		    TCP_INET_FAMILY(req->rsk_ops->family)) {
			BUG_TRAP(!req->sk);
			*prevp = prev;
			break;
		}
	}

	return req;
}

static void tcp_v4_synq_add(struct sock *sk, struct request_sock *req)
{
	struct tcp_sock *tp = tcp_sk(sk);
	struct listen_sock *lopt = tp->accept_queue.listen_opt;
	u32 h = tcp_v4_synq_hash(inet_rsk(req)->rmt_addr, inet_rsk(req)->rmt_port, lopt->hash_rnd);

	reqsk_queue_hash_req(&tp->accept_queue, h, req, TCP_TIMEOUT_INIT);
	tcp_synq_added(sk);
}


/*
 * This routine does path mtu discovery as defined in RFC1191.
 */
static inline void do_pmtu_discovery(struct sock *sk, struct iphdr *iph,
				     u32 mtu)
{
	struct dst_entry *dst;
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);

	/* We are not interested in TCP_LISTEN and open_requests (SYN-ACKs
	 * send out by Linux are always <576bytes so they should go through
	 * unfragmented).
	 */
	if (sk->sk_state == TCP_LISTEN)
		return;

	/* We don't check in the destentry if pmtu discovery is forbidden
	 * on this route. We just assume that no packet_to_big packets
	 * are send back when pmtu discovery is not active.
     	 * There is a small race when the user changes this flag in the
	 * route, but I think that's acceptable.
	 */
	if ((dst = __sk_dst_check(sk, 0)) == NULL)
		return;

	dst->ops->update_pmtu(dst, mtu);

	/* Something is about to be wrong... Remember soft error
	 * for the case, if this connection will not able to recover.
	 */
	if (mtu < dst_mtu(dst) && ip_dont_fragment(sk, dst))
		sk->sk_err_soft = EMSGSIZE;

	mtu = dst_mtu(dst);

	if (inet->pmtudisc != IP_PMTUDISC_DONT &&
	    tp->pmtu_cookie > mtu) {
		tcp_sync_mss(sk, mtu);

		/* Resend the TCP packet because it's
		 * clear that the old packet has been
		 * dropped. This is the new "fast" path mtu
		 * discovery.
		 */
		tcp_simple_retransmit(sk);
	} /* else let the usual retransmit timer handle it */
}

/*
 * This routine is called by the ICMP module when it gets some
 * sort of error condition.  If err < 0 then the socket should
 * be closed and the error returned to the user.  If err > 0
 * it's just the icmp type << 8 | icmp code.  After adjustment
 * header points to the first 8 bytes of the tcp header.  We need
 * to find the appropriate port.
 *
 * The locking strategy used here is very "optimistic". When
 * someone else accesses the socket the ICMP is just dropped
 * and for some paths there is no check at all.
 * A more general error queue to queue errors for later handling
 * is probably better.
 *
 */

void tcp_v4_err(struct sk_buff *skb, u32 info)
{
	struct iphdr *iph = (struct iphdr *)skb->data;
	struct tcphdr *th = (struct tcphdr *)(skb->data + (iph->ihl << 2));
	struct tcp_sock *tp;
	struct inet_sock *inet;
	int type = skb->h.icmph->type;
	int code = skb->h.icmph->code;
	struct sock *sk;
	__u32 seq;
	int err;

	if (skb->len < (iph->ihl << 2) + 8) {
		ICMP_INC_STATS_BH(ICMP_MIB_INERRORS);
		return;
	}

	sk = tcp_v4_lookup(iph->daddr, th->dest, iph->saddr,
			   th->source, tcp_v4_iif(skb));
	if (!sk) {
		ICMP_INC_STATS_BH(ICMP_MIB_INERRORS);
		return;