[INET]: Generalise tcp_tw_bucket, aka TIME_WAIT sockets

This paves the way to generalise the rest of the sock ID lookup
routines and saves some bytes in TCPv4 TIME_WAIT sockets on distro
kernels (where IPv6 is always built as a module):

[root@qemu ~]# grep tw_sock /proc/slabinfo
tw_sock_TCPv6  0  0  128  31  1
tw_sock_TCP    0  0   96  41  1
[root@qemu ~]#

Now if a protocol wants to use the TIME_WAIT generic infrastructure it
only has to set the sk_prot->twsk_obj_size field with the size of its
inet_timewait_sock derived sock and proto_register will create
sk_prot->twsk_slab, for now its only for INET sockets, but we can
introduce timewait_sock later if some non INET transport protocolo
wants to use this stuff.

Next changesets will take advantage of this new infrastructure to
generalise even more TCP code.

[acme@toy net-2.6.14]$ grep built-in /tmp/before.size /tmp/after.size
/tmp/before.size: 188646   11764    5068  205478   322a6 net/ipv4/built-in.o
/tmp/after.size:  188144   11764    5068  204976   320b0 net/ipv4/built-in.o
[acme@toy net-2.6.14]$

Tested with both IPv4 & IPv6 (::1 (localhost) & ::ffff:172.20.0.1
(qemu host)).

Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 134 insertions, 135 deletions

diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 2f4b1a374bb7..f1a708bf7a97 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -271,8 +271,6 @@ int sysctl_tcp_fin_timeout = TCP_FIN_TIMEOUT;
 
 DEFINE_SNMP_STAT(struct tcp_mib, tcp_statistics);
 
-kmem_cache_t *tcp_timewait_cachep;
-
 atomic_t tcp_orphan_count = ATOMIC_INIT(0);
 
 int sysctl_tcp_mem[3];
@@ -2264,13 +2262,6 @@ void __init tcp_init(void)
 	if (!tcp_hashinfo.bind_bucket_cachep)
 		panic("tcp_init: Cannot alloc tcp_bind_bucket cache.");
 
-	tcp_timewait_cachep = kmem_cache_create("tcp_tw_bucket",
-						sizeof(struct tcp_tw_bucket),
-						0, SLAB_HWCACHE_ALIGN,
-						NULL, NULL);
-	if (!tcp_timewait_cachep)
-		panic("tcp_init: Cannot alloc tcp_tw_bucket cache.");
-
 	/* Size and allocate the main established and bind bucket
 	 * hash tables.
 	 *
@@ -2363,4 +2354,3 @@ EXPORT_SYMBOL(tcp_sendpage);
 EXPORT_SYMBOL(tcp_setsockopt);
 EXPORT_SYMBOL(tcp_shutdown);
 EXPORT_SYMBOL(tcp_statistics);
-EXPORT_SYMBOL(tcp_timewait_cachep);
diff --git a/net/ipv4/tcp_diag.c b/net/ipv4/tcp_diag.c
index 1a89a03c449b..6f2d6f2276b9 100644
--- a/net/ipv4/tcp_diag.c
+++ b/net/ipv4/tcp_diag.c
@@ -81,7 +81,7 @@ static int tcpdiag_fill(struct sk_buff *skb, struct sock *sk,
 	r->id.tcpdiag_cookie[1] = (u32)(((unsigned long)sk >> 31) >> 1);
 
 	if (r->tcpdiag_state == TCP_TIME_WAIT) {
-		struct tcp_tw_bucket *tw = (struct tcp_tw_bucket*)sk;
+		const struct inet_timewait_sock *tw = inet_twsk(sk);
 		long tmo = tw->tw_ttd - jiffies;
 		if (tmo < 0)
 			tmo = 0;
@@ -99,10 +99,12 @@ static int tcpdiag_fill(struct sk_buff *skb, struct sock *sk,
 		r->tcpdiag_inode = 0;
 #ifdef CONFIG_IP_TCPDIAG_IPV6
 		if (r->tcpdiag_family == AF_INET6) {
+			const struct tcp6_timewait_sock *tcp6tw = tcp6_twsk(sk);
+
 			ipv6_addr_copy((struct in6_addr *)r->id.tcpdiag_src,
-				       &tw->tw_v6_rcv_saddr);
+				       &tcp6tw->tw_v6_rcv_saddr);
 			ipv6_addr_copy((struct in6_addr *)r->id.tcpdiag_dst,
-				       &tw->tw_v6_daddr);
+				       &tcp6tw->tw_v6_daddr);
 		}
 #endif
 		nlh->nlmsg_len = skb->tail - b;
@@ -239,7 +241,7 @@ static int tcpdiag_get_exact(struct sk_buff *in_skb, const struct nlmsghdr *nlh)
 out:
 	if (sk) {
 		if (sk->sk_state == TCP_TIME_WAIT)
-			tcp_tw_put((struct tcp_tw_bucket*)sk);
+			inet_twsk_put((struct inet_timewait_sock *)sk);
 		else
 			sock_put(sk);
 	}
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index a678709b36f6..ce423e48ebe0 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -106,7 +106,7 @@ int sysctl_local_port_range[2] = { 1024, 4999 };
 
 static inline int tcp_bind_conflict(struct sock *sk, struct inet_bind_bucket *tb)
 {
-	const u32 sk_rcv_saddr = tcp_v4_rcv_saddr(sk);
+	const u32 sk_rcv_saddr = inet_rcv_saddr(sk);
 	struct sock *sk2;
 	struct hlist_node *node;
 	int reuse = sk->sk_reuse;
@@ -119,7 +119,7 @@ static inline int tcp_bind_conflict(struct sock *sk, struct inet_bind_bucket *tb
 		     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);
+				const u32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
 				if (!sk2_rcv_saddr || !sk_rcv_saddr ||
 				    sk2_rcv_saddr == sk_rcv_saddr)
 					break;
@@ -251,10 +251,10 @@ static inline struct sock *__tcp_v4_lookup_established(const u32 saddr,
 						       const int dif)
 {
 	struct inet_ehash_bucket *head;
-	TCP_V4_ADDR_COOKIE(acookie, saddr, daddr)
-	__u32 ports = TCP_COMBINED_PORTS(sport, hnum);
+	INET_ADDR_COOKIE(acookie, saddr, daddr)
+	const __u32 ports = INET_COMBINED_PORTS(sport, hnum);
 	struct sock *sk;
-	struct hlist_node *node;
+	const struct hlist_node *node;
 	/* Optimize here for direct hit, only listening connections can
 	 * have wildcards anyways.
 	 */
@@ -262,13 +262,13 @@ static inline struct sock *__tcp_v4_lookup_established(const u32 saddr,
 	head = &tcp_hashinfo.ehash[hash];
 	read_lock(&head->lock);
 	sk_for_each(sk, node, &head->chain) {
-		if (TCP_IPV4_MATCH(sk, acookie, saddr, daddr, ports, dif))
+		if (INET_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_hashinfo.ehash_size)->chain) {
-		if (TCP_IPV4_TW_MATCH(sk, acookie, saddr, daddr, ports, dif))
+		if (INET_TW_MATCH(sk, acookie, saddr, daddr, ports, dif))
 			goto hit;
 	}
 	sk = NULL;
@@ -313,27 +313,28 @@ static inline __u32 tcp_v4_init_sequence(struct sock *sk, struct sk_buff *skb)
 
 /* called with local bh disabled */
 static int __tcp_v4_check_established(struct sock *sk, __u16 lport,
-				      struct tcp_tw_bucket **twp)
+				      struct inet_timewait_sock **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);
+	INET_ADDR_COOKIE(acookie, saddr, daddr)
+	const __u32 ports = INET_COMBINED_PORTS(inet->dport, lport);
 	const int hash = inet_ehashfn(daddr, lport, saddr, inet->dport, tcp_hashinfo.ehash_size);
 	struct inet_ehash_bucket *head = &tcp_hashinfo.ehash[hash];
 	struct sock *sk2;
-	struct hlist_node *node;
-	struct tcp_tw_bucket *tw;
+	const struct hlist_node *node;
+	struct inet_timewait_sock *tw;
 
 	write_lock(&head->lock);
 
 	/* Check TIME-WAIT sockets first. */
 	sk_for_each(sk2, node, &(head + tcp_hashinfo.ehash_size)->chain) {
-		tw = (struct tcp_tw_bucket *)sk2;
+		tw = inet_twsk(sk2);
 
-		if (TCP_IPV4_TW_MATCH(sk2, acookie, saddr, daddr, ports, dif)) {
+		if (INET_TW_MATCH(sk2, acookie, saddr, daddr, ports, dif)) {
+			const struct tcp_timewait_sock *tcptw = tcp_twsk(sk2);
 			struct tcp_sock *tp = tcp_sk(sk);
 
 			/* With PAWS, it is safe from the viewpoint
@@ -350,15 +351,15 @@ static int __tcp_v4_check_established(struct sock *sk, __u16 lport,
 			   fall back to VJ's scheme and use initial
 			   timestamp retrieved from peer table.
 			 */
-			if (tw->tw_ts_recent_stamp &&
+			if (tcptw->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)
+				      tcptw->tw_ts_recent_stamp > 1))) {
+				tp->write_seq = tcptw->tw_snd_nxt + 65535 + 2;
+				if (tp->write_seq == 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;
+				tp->rx_opt.ts_recent	   = tcptw->tw_ts_recent;
+				tp->rx_opt.ts_recent_stamp = tcptw->tw_ts_recent_stamp;
 				sock_hold(sk2);
 				goto unique;
 			} else
@@ -369,7 +370,7 @@ static int __tcp_v4_check_established(struct sock *sk, __u16 lport,
 
 	/* And established part... */
 	sk_for_each(sk2, node, &head->chain) {
-		if (TCP_IPV4_MATCH(sk2, acookie, saddr, daddr, ports, dif))
+		if (INET_MATCH(sk2, acookie, saddr, daddr, ports, dif))
 			goto not_unique;
 	}
 
@@ -392,7 +393,7 @@ unique:
 		tcp_tw_deschedule(tw);
 		NET_INC_STATS_BH(LINUX_MIB_TIMEWAITRECYCLED);
 
-		tcp_tw_put(tw);
+		inet_twsk_put(tw);
 	}
 
 	return 0;
@@ -429,7 +430,7 @@ static inline int tcp_v4_hash_connect(struct sock *sk)
 		static u32 hint;
 		u32 offset = hint + connect_port_offset(sk);
 		struct hlist_node *node;
- 		struct tcp_tw_bucket *tw = NULL;
+ 		struct inet_timewait_sock *tw = NULL;
 
  		local_bh_disable();
 		for (i = 1; i <= range; i++) {
@@ -482,7 +483,7 @@ ok:
 
  		if (tw) {
  			tcp_tw_deschedule(tw);
- 			tcp_tw_put(tw);
+ 			inet_twsk_put(tw);
  		}
 
 		ret = 0;
@@ -757,7 +758,7 @@ void tcp_v4_err(struct sk_buff *skb, u32 info)
 		return;
 	}
 	if (sk->sk_state == TCP_TIME_WAIT) {
-		tcp_tw_put((struct tcp_tw_bucket *)sk);
+		inet_twsk_put((struct inet_timewait_sock *)sk);
 		return;
 	}
 
@@ -1002,12 +1003,13 @@ static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
 
 static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
 {
-	struct tcp_tw_bucket *tw = (struct tcp_tw_bucket *)sk;
+	struct inet_timewait_sock *tw = inet_twsk(sk);
+	const struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
 
-	tcp_v4_send_ack(skb, tw->tw_snd_nxt, tw->tw_rcv_nxt,
-			tw->tw_rcv_wnd >> tw->tw_rcv_wscale, tw->tw_ts_recent);
+	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
+			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale, tcptw->tw_ts_recent);
 
-	tcp_tw_put(tw);
+	inet_twsk_put(tw);
 }
 
 static void tcp_v4_reqsk_send_ack(struct sk_buff *skb, struct request_sock *req)
@@ -1368,7 +1370,7 @@ static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
 			bh_lock_sock(nsk);
 			return nsk;
 		}
-		tcp_tw_put((struct tcp_tw_bucket *)nsk);
+		inet_twsk_put((struct inet_timewait_sock *)nsk);
 		return NULL;
 	}
 
@@ -1557,25 +1559,25 @@ discard_and_relse:
 
 do_time_wait:
 	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
-		tcp_tw_put((struct tcp_tw_bucket *) sk);
+		inet_twsk_put((struct inet_timewait_sock *) sk);
 		goto discard_it;
 	}
 
 	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
 		TCP_INC_STATS_BH(TCP_MIB_INERRS);
-		tcp_tw_put((struct tcp_tw_bucket *) sk);
+		inet_twsk_put((struct inet_timewait_sock *) sk);
 		goto discard_it;
 	}
-	switch (tcp_timewait_state_process((struct tcp_tw_bucket *)sk,
-					   skb, th, skb->len)) {
+	switch (tcp_timewait_state_process((struct inet_timewait_sock *)sk,
+					   skb, th)) {
 	case TCP_TW_SYN: {
 		struct sock *sk2 = inet_lookup_listener(&tcp_hashinfo,
 							skb->nh.iph->daddr,
 							ntohs(th->dest),
 							tcp_v4_iif(skb));
 		if (sk2) {
-			tcp_tw_deschedule((struct tcp_tw_bucket *)sk);
-			tcp_tw_put((struct tcp_tw_bucket *)sk);
+			tcp_tw_deschedule((struct inet_timewait_sock *)sk);
+			inet_twsk_put((struct inet_timewait_sock *)sk);
 			sk = sk2;
 			goto process;
 		}
@@ -1639,18 +1641,18 @@ int tcp_v4_remember_stamp(struct sock *sk)
 	return 0;
 }
 
-int tcp_v4_tw_remember_stamp(struct tcp_tw_bucket *tw)
+int tcp_v4_tw_remember_stamp(struct inet_timewait_sock *tw)
 {
-	struct inet_peer *peer = NULL;
-
-	peer = inet_getpeer(tw->tw_daddr, 1);
+	struct inet_peer *peer = inet_getpeer(tw->tw_daddr, 1);
 
 	if (peer) {
-		if ((s32)(peer->tcp_ts - tw->tw_ts_recent) <= 0 ||
+		const struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
+
+		if ((s32)(peer->tcp_ts - tcptw->tw_ts_recent) <= 0 ||
 		    (peer->tcp_ts_stamp + TCP_PAWS_MSL < xtime.tv_sec &&
-		     peer->tcp_ts_stamp <= tw->tw_ts_recent_stamp)) {
-			peer->tcp_ts_stamp = tw->tw_ts_recent_stamp;
-			peer->tcp_ts = tw->tw_ts_recent;
+		     peer->tcp_ts_stamp <= tcptw->tw_ts_recent_stamp)) {
+			peer->tcp_ts_stamp = tcptw->tw_ts_recent_stamp;
+			peer->tcp_ts	   = tcptw->tw_ts_recent;
 		}
 		inet_putpeer(peer);
 		return 1;
@@ -1758,13 +1760,13 @@ EXPORT_SYMBOL(tcp_v4_destroy_sock);
 #ifdef CONFIG_PROC_FS
 /* Proc filesystem TCP sock list dumping. */
 
-static inline struct tcp_tw_bucket *tw_head(struct hlist_head *head)
+static inline struct inet_timewait_sock *tw_head(struct hlist_head *head)
 {
 	return hlist_empty(head) ? NULL :
-		list_entry(head->first, struct tcp_tw_bucket, tw_node);
+		list_entry(head->first, struct inet_timewait_sock, tw_node);
 }
 
-static inline struct tcp_tw_bucket *tw_next(struct tcp_tw_bucket *tw)
+static inline struct inet_timewait_sock *tw_next(struct inet_timewait_sock *tw)
 {
 	return tw->tw_node.next ?
 		hlist_entry(tw->tw_node.next, typeof(*tw), tw_node) : NULL;
@@ -1860,7 +1862,7 @@ static void *established_get_first(struct seq_file *seq)
 	for (st->bucket = 0; st->bucket < tcp_hashinfo.ehash_size; ++st->bucket) {
 		struct sock *sk;
 		struct hlist_node *node;
-		struct tcp_tw_bucket *tw;
+		struct inet_timewait_sock *tw;
 
 		/* We can reschedule _before_ having picked the target: */
 		cond_resched_softirq();
@@ -1874,8 +1876,8 @@ static void *established_get_first(struct seq_file *seq)
 			goto out;
 		}
 		st->state = TCP_SEQ_STATE_TIME_WAIT;
-		tw_for_each(tw, node,
-			    &tcp_hashinfo.ehash[st->bucket + tcp_hashinfo.ehash_size].chain) {
+		inet_twsk_for_each(tw, node,
+				   &tcp_hashinfo.ehash[st->bucket + tcp_hashinfo.ehash_size].chain) {
 			if (tw->tw_family != st->family) {
 				continue;
 			}
@@ -1892,7 +1894,7 @@ out:
 static void *established_get_next(struct seq_file *seq, void *cur)
 {
 	struct sock *sk = cur;
-	struct tcp_tw_bucket *tw;
+	struct inet_timewait_sock *tw;
 	struct hlist_node *node;
 	struct tcp_iter_state* st = seq->private;
 
@@ -2159,7 +2161,7 @@ static void get_tcp4_sock(struct sock *sp, char *tmpbuf, int i)
 		tp->snd_ssthresh >= 0xFFFF ? -1 : tp->snd_ssthresh);
 }
 
-static void get_timewait4_sock(struct tcp_tw_bucket *tw, char *tmpbuf, int i)
+static void get_timewait4_sock(struct inet_timewait_sock *tw, char *tmpbuf, int i)
 {
 	unsigned int dest, src;
 	__u16 destp, srcp;
@@ -2261,6 +2263,7 @@ struct proto tcp_prot = {
 	.sysctl_rmem		= sysctl_tcp_rmem,
 	.max_header		= MAX_TCP_HEADER,
 	.obj_size		= sizeof(struct tcp_sock),
+	.twsk_obj_size		= sizeof(struct tcp_timewait_sock),
 	.rsk_prot		= &tcp_request_sock_ops,
 };
 
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index f29e2f6ebe1b..5b5a49335fbb 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -41,7 +41,7 @@ int sysctl_tcp_max_tw_buckets = NR_FILE*2;
 int sysctl_tcp_syncookies = SYNC_INIT; 
 int sysctl_tcp_abort_on_overflow;
 
-static void tcp_tw_schedule(struct tcp_tw_bucket *tw, int timeo);
+static void tcp_tw_schedule(struct inet_timewait_sock *tw, int timeo);
 
 static __inline__ int tcp_in_window(u32 seq, u32 end_seq, u32 s_win, u32 e_win)
 {
@@ -58,7 +58,7 @@ int tcp_tw_count;
 
 
 /* Must be called with locally disabled BHs. */
-static void tcp_timewait_kill(struct tcp_tw_bucket *tw)
+static void tcp_timewait_kill(struct inet_timewait_sock *tw)
 {
 	struct inet_bind_hashbucket *bhead;
 	struct inet_bind_bucket *tb;
@@ -85,11 +85,11 @@ static void tcp_timewait_kill(struct tcp_tw_bucket *tw)
 
 #ifdef SOCK_REFCNT_DEBUG
 	if (atomic_read(&tw->tw_refcnt) != 1) {
-		printk(KERN_DEBUG "tw_bucket %p refcnt=%d\n", tw,
-		       atomic_read(&tw->tw_refcnt));
+		printk(KERN_DEBUG "%s timewait_sock %p refcnt=%d\n",
+		       tw->tw_prot->name, tw, atomic_read(&tw->tw_refcnt));
 	}
 #endif
-	tcp_tw_put(tw);
+	inet_twsk_put(tw);
 }
 
 /* 
@@ -121,19 +121,20 @@ static void tcp_timewait_kill(struct tcp_tw_bucket *tw)
  * to avoid misread sequence numbers, states etc.  --ANK
  */
 enum tcp_tw_status
-tcp_timewait_state_process(struct tcp_tw_bucket *tw, struct sk_buff *skb,
-			   struct tcphdr *th, unsigned len)
+tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb,
+			   const struct tcphdr *th)
 {
+	struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
 	struct tcp_options_received tmp_opt;
 	int paws_reject = 0;
 
 	tmp_opt.saw_tstamp = 0;
-	if (th->doff > (sizeof(struct tcphdr) >> 2) && tw->tw_ts_recent_stamp) {
+	if (th->doff > (sizeof(*th) >> 2) && tcptw->tw_ts_recent_stamp) {
 		tcp_parse_options(skb, &tmp_opt, 0);
 
 		if (tmp_opt.saw_tstamp) {
-			tmp_opt.ts_recent	   = tw->tw_ts_recent;
-			tmp_opt.ts_recent_stamp = tw->tw_ts_recent_stamp;
+			tmp_opt.ts_recent	= tcptw->tw_ts_recent;
+			tmp_opt.ts_recent_stamp	= tcptw->tw_ts_recent_stamp;
 			paws_reject = tcp_paws_check(&tmp_opt, th->rst);
 		}
 	}
@@ -144,20 +145,20 @@ tcp_timewait_state_process(struct tcp_tw_bucket *tw, struct sk_buff *skb,
 		/* Out of window, send ACK */
 		if (paws_reject ||
 		    !tcp_in_window(TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq,
-				   tw->tw_rcv_nxt,
-				   tw->tw_rcv_nxt + tw->tw_rcv_wnd))
+				   tcptw->tw_rcv_nxt,
+				   tcptw->tw_rcv_nxt + tcptw->tw_rcv_wnd))
 			return TCP_TW_ACK;
 
 		if (th->rst)
 			goto kill;
 
-		if (th->syn && !before(TCP_SKB_CB(skb)->seq, tw->tw_rcv_nxt))
+		if (th->syn && !before(TCP_SKB_CB(skb)->seq, tcptw->tw_rcv_nxt))
 			goto kill_with_rst;
 
 		/* Dup ACK? */
-		if (!after(TCP_SKB_CB(skb)->end_seq, tw->tw_rcv_nxt) ||
+		if (!after(TCP_SKB_CB(skb)->end_seq, tcptw->tw_rcv_nxt) ||
 		    TCP_SKB_CB(skb)->end_seq == TCP_SKB_CB(skb)->seq) {
-			tcp_tw_put(tw);
+			inet_twsk_put(tw);
 			return TCP_TW_SUCCESS;
 		}
 
@@ -165,19 +166,19 @@ tcp_timewait_state_process(struct tcp_tw_bucket *tw, struct sk_buff *skb,
 		 * reset.
 		 */
 		if (!th->fin ||
-		    TCP_SKB_CB(skb)->end_seq != tw->tw_rcv_nxt + 1) {
+		    TCP_SKB_CB(skb)->end_seq != tcptw->tw_rcv_nxt + 1) {
 kill_with_rst:
 			tcp_tw_deschedule(tw);
-			tcp_tw_put(tw);
+			inet_twsk_put(tw);
 			return TCP_TW_RST;
 		}
 
 		/* FIN arrived, enter true time-wait state. */
-		tw->tw_substate	= TCP_TIME_WAIT;
-		tw->tw_rcv_nxt	= TCP_SKB_CB(skb)->end_seq;
+		tw->tw_substate	  = TCP_TIME_WAIT;
+		tcptw->tw_rcv_nxt = TCP_SKB_CB(skb)->end_seq;
 		if (tmp_opt.saw_tstamp) {
-			tw->tw_ts_recent_stamp	= xtime.tv_sec;
-			tw->tw_ts_recent	= tmp_opt.rcv_tsval;
+			tcptw->tw_ts_recent_stamp = xtime.tv_sec;
+			tcptw->tw_ts_recent	  = tmp_opt.rcv_tsval;
 		}
 
 		/* I am shamed, but failed to make it more elegant.
@@ -186,7 +187,7 @@ kill_with_rst:
 		 * do not undertsnad recycling in any case, it not
 		 * a big problem in practice. --ANK */
 		if (tw->tw_family == AF_INET &&
-		    sysctl_tcp_tw_recycle && tw->tw_ts_recent_stamp &&
+		    sysctl_tcp_tw_recycle && tcptw->tw_ts_recent_stamp &&
 		    tcp_v4_tw_remember_stamp(tw))
 			tcp_tw_schedule(tw, tw->tw_timeout);
 		else
@@ -212,7 +213,7 @@ kill_with_rst:
 	 */
 
 	if (!paws_reject &&
-	    (TCP_SKB_CB(skb)->seq == tw->tw_rcv_nxt &&
+	    (TCP_SKB_CB(skb)->seq == tcptw->tw_rcv_nxt &&
 	     (TCP_SKB_CB(skb)->seq == TCP_SKB_CB(skb)->end_seq || th->rst))) {
 		/* In window segment, it may be only reset or bare ack. */
 
@@ -224,18 +225,18 @@ kill_with_rst:
 			if (sysctl_tcp_rfc1337 == 0) {
 kill:
 				tcp_tw_deschedule(tw);
-				tcp_tw_put(tw);
+				inet_twsk_put(tw);
 				return TCP_TW_SUCCESS;
 			}
 		}
 		tcp_tw_schedule(tw, TCP_TIMEWAIT_LEN);
 
 		if (tmp_opt.saw_tstamp) {
-			tw->tw_ts_recent	= tmp_opt.rcv_tsval;
-			tw->tw_ts_recent_stamp	= xtime.tv_sec;
+			tcptw->tw_ts_recent	  = tmp_opt.rcv_tsval;
+			tcptw->tw_ts_recent_stamp = xtime.tv_sec;
 		}
 
-		tcp_tw_put(tw);
+		inet_twsk_put(tw);
 		return TCP_TW_SUCCESS;
 	}
 
@@ -257,9 +258,10 @@ kill:
 	 */
 
 	if (th->syn && !th->rst && !th->ack && !paws_reject &&
-	    (after(TCP_SKB_CB(skb)->seq, tw->tw_rcv_nxt) ||
-	     (tmp_opt.saw_tstamp && (s32)(tw->tw_ts_recent - tmp_opt.rcv_tsval) < 0))) {
-		u32 isn = tw->tw_snd_nxt + 65535 + 2;
+	    (after(TCP_SKB_CB(skb)->seq, tcptw->tw_rcv_nxt) ||
+	     (tmp_opt.saw_tstamp &&
+	      (s32)(tcptw->tw_ts_recent - tmp_opt.rcv_tsval) < 0))) {
+		u32 isn = tcptw->tw_snd_nxt + 65535 + 2;
 		if (isn == 0)
 			isn++;
 		TCP_SKB_CB(skb)->when = isn;
@@ -284,7 +286,7 @@ kill:
 		 */
 		return TCP_TW_ACK;
 	}
-	tcp_tw_put(tw);
+	inet_twsk_put(tw);
 	return TCP_TW_SUCCESS;
 }
 
@@ -293,7 +295,7 @@ kill:
  * relevant info into it from the SK, and mess with hash chains
  * and list linkage.
  */
-static void __tcp_tw_hashdance(struct sock *sk, struct tcp_tw_bucket *tw)
+static void __tcp_tw_hashdance(struct sock *sk, struct inet_timewait_sock *tw)
 {
 	const struct inet_sock *inet = inet_sk(sk);
 	struct inet_ehash_bucket *ehead = &tcp_hashinfo.ehash[sk->sk_hashent];
@@ -306,7 +308,7 @@ static void __tcp_tw_hashdance(struct sock *sk, struct tcp_tw_bucket *tw)
 	spin_lock(&bhead->lock);
 	tw->tw_tb = inet->bind_hash;
 	BUG_TRAP(inet->bind_hash);
-	tw_add_bind_node(tw, &tw->tw_tb->owners);
+	inet_twsk_add_bind_node(tw, &tw->tw_tb->owners);
 	spin_unlock(&bhead->lock);
 
 	write_lock(&ehead->lock);
@@ -316,7 +318,7 @@ static void __tcp_tw_hashdance(struct sock *sk, struct tcp_tw_bucket *tw)
 		sock_prot_dec_use(sk->sk_prot);
 
 	/* Step 3: Hash TW into TIMEWAIT half of established hash table. */
-	tw_add_node(tw, &(ehead + tcp_hashinfo.ehash_size)->chain);
+	inet_twsk_add_node(tw, &(ehead + tcp_hashinfo.ehash_size)->chain);
 	atomic_inc(&tw->tw_refcnt);
 
 	write_unlock(&ehead->lock);
@@ -327,19 +329,23 @@ static void __tcp_tw_hashdance(struct sock *sk, struct tcp_tw_bucket *tw)
  */ 
 void tcp_time_wait(struct sock *sk, int state, int timeo)
 {
-	struct tcp_tw_bucket *tw = NULL;
-	struct tcp_sock *tp = tcp_sk(sk);
+	struct inet_timewait_sock *tw = NULL;
+	const struct tcp_sock *tp = tcp_sk(sk);
 	int recycle_ok = 0;
 
 	if (sysctl_tcp_tw_recycle && tp->rx_opt.ts_recent_stamp)
 		recycle_ok = tp->af_specific->remember_stamp(sk);
 
 	if (tcp_tw_count < sysctl_tcp_max_tw_buckets)
-		tw = kmem_cache_alloc(tcp_timewait_cachep, SLAB_ATOMIC);
+		tw = kmem_cache_alloc(sk->sk_prot_creator->twsk_slab, SLAB_ATOMIC);
+
+	if (tw != NULL) {
+		struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
+		const struct inet_sock *inet = inet_sk(sk);
+		const int rto = (tp->rto << 2) - (tp->rto >> 1);
 
-	if(tw != NULL) {
-		struct inet_sock *inet = inet_sk(sk);
-		int rto = (tp->rto<<2) - (tp->rto>>1);
+		/* Remember our protocol */
+		tw->tw_prot		= sk->sk_prot_creator;
 
 		/* Give us an identity. */
 		tw->tw_daddr		= inet->daddr;
@@ -356,25 +362,23 @@ void tcp_time_wait(struct sock *sk, int state, int timeo)
 		atomic_set(&tw->tw_refcnt, 1);
 
 		tw->tw_hashent		= sk->sk_hashent;
-		tw->tw_rcv_nxt		= tp->rcv_nxt;
-		tw->tw_snd_nxt		= tp->snd_nxt;
-		tw->tw_rcv_wnd		= tcp_receive_window(tp);
-		tw->tw_ts_recent	= tp->rx_opt.ts_recent;
-		tw->tw_ts_recent_stamp	= tp->rx_opt.ts_recent_stamp;
-		tw_dead_node_init(tw);
+		tcptw->tw_rcv_nxt	= tp->rcv_nxt;
+		tcptw->tw_snd_nxt	= tp->snd_nxt;
+		tcptw->tw_rcv_wnd	= tcp_receive_window(tp);
+		tcptw->tw_ts_recent	= tp->rx_opt.ts_recent;
+		tcptw->tw_ts_recent_stamp = tp->rx_opt.ts_recent_stamp;
+		inet_twsk_dead_node_init(tw);
 
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 		if (tw->tw_family == PF_INET6) {
 			struct ipv6_pinfo *np = inet6_sk(sk);
+			struct tcp6_timewait_sock *tcp6tw = tcp6_twsk((struct sock *)tw);
 
-			ipv6_addr_copy(&tw->tw_v6_daddr, &np->daddr);
-			ipv6_addr_copy(&tw->tw_v6_rcv_saddr, &np->rcv_saddr);
-			tw->tw_v6_ipv6only = np->ipv6only;
-		} else {
-			memset(&tw->tw_v6_daddr, 0, sizeof(tw->tw_v6_daddr));
-			memset(&tw->tw_v6_rcv_saddr, 0, sizeof(tw->tw_v6_rcv_saddr));
-			tw->tw_v6_ipv6only = 0;
-		}
+			ipv6_addr_copy(&tcp6tw->tw_v6_daddr, &np->daddr);
+			ipv6_addr_copy(&tcp6tw->tw_v6_rcv_saddr, &np->rcv_saddr);
+			tw->tw_ipv6only = np->ipv6only;
+		} else
+			tw->tw_ipv6only = 0;
 #endif
 		/* Linkage updates. */
 		__tcp_tw_hashdance(sk, tw);
@@ -392,7 +396,7 @@ void tcp_time_wait(struct sock *sk, int state, int timeo)
 		}
 
 		tcp_tw_schedule(tw, timeo);
-		tcp_tw_put(tw);
+		inet_twsk_put(tw);
 	} else {
 		/* Sorry, if we're out of memory, just CLOSE this
 		 * socket up.  We've got bigger problems than
@@ -427,7 +431,7 @@ static u32 twkill_thread_slots;
 /* Returns non-zero if quota exceeded.  */
 static int tcp_do_twkill_work(int slot, unsigned int quota)
 {
-	struct tcp_tw_bucket *tw;
+	struct inet_timewait_sock *tw;
 	struct hlist_node *node;
 	unsigned int killed;
 	int ret;
@@ -441,11 +445,11 @@ static int tcp_do_twkill_work(int slot, unsigned int quota)
 	killed = 0;
 	ret = 0;
 rescan:
-	tw_for_each_inmate(tw, node, &tcp_tw_death_row[slot]) {
-		__tw_del_dead_node(tw);
+	inet_twsk_for_each_inmate(tw, node, &tcp_tw_death_row[slot]) {
+		__inet_twsk_del_dead_node(tw);
 		spin_unlock(&tw_death_lock);
 		tcp_timewait_kill(tw);
-		tcp_tw_put(tw);
+		inet_twsk_put(tw);
 		killed++;
 		spin_lock(&tw_death_lock);
 		if (killed > quota) {
@@ -531,11 +535,11 @@ static void twkill_work(void *dummy)
  */
 
 /* This is for handling early-kills of TIME_WAIT sockets. */
-void tcp_tw_deschedule(struct tcp_tw_bucket *tw)
+void tcp_tw_deschedule(struct inet_timewait_sock *tw)
 {
 	spin_lock(&tw_death_lock);
-	if (tw_del_dead_node(tw)) {
-		tcp_tw_put(tw);
+	if (inet_twsk_del_dead_node(tw)) {
+		inet_twsk_put(tw);
 		if (--tcp_tw_count == 0)
 			del_timer(&tcp_tw_timer);
 	}
@@ -552,7 +556,7 @@ static struct timer_list tcp_twcal_timer =
 		TIMER_INITIALIZER(tcp_twcal_tick, 0, 0);
 static struct hlist_head tcp_twcal_row[TCP_TW_RECYCLE_SLOTS];
 
-static void tcp_tw_schedule(struct tcp_tw_bucket *tw, int timeo)
+static void tcp_tw_schedule(struct inet_timewait_sock *tw, const int timeo)
 {
 	struct hlist_head *list;
 	int slot;
@@ -586,7 +590,7 @@ static void tcp_tw_schedule(struct tcp_tw_bucket *tw, int timeo)
 	spin_lock(&tw_death_lock);
 
 	/* Unlink it, if it was scheduled */
-	if (tw_del_dead_node(tw))
+	if (inet_twsk_del_dead_node(tw))
 		tcp_tw_count--;
 	else
 		atomic_inc(&tw->tw_refcnt);
@@ -644,13 +648,13 @@ void tcp_twcal_tick(unsigned long dummy)
 	for (n=0; n<TCP_TW_RECYCLE_SLOTS; n++) {
 		if (time_before_eq(j, now)) {
 			struct hlist_node *node, *safe;
-			struct tcp_tw_bucket *tw;
+			struct inet_timewait_sock *tw;
 
-			tw_for_each_inmate_safe(tw, node, safe,
-					   &tcp_twcal_row[slot]) {
-				__tw_del_dead_node(tw);
+			inet_twsk_for_each_inmate_safe(tw, node, safe,
+						       &tcp_twcal_row[slot]) {
+				__inet_twsk_del_dead_node(tw);
 				tcp_timewait_kill(tw);
-				tcp_tw_put(tw);
+				inet_twsk_put(tw);
 				killed++;
 			}
 		} else {