tcp_ipv6.c 54.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *	TCP over IPv6
3
 *	Linux INET6 implementation
L
Linus Torvalds 已提交
4 5
 *
 *	Authors:
6
 *	Pedro Roque		<roque@di.fc.ul.pt>
L
Linus Torvalds 已提交
7
 *
8
 *	Based on:
L
Linus Torvalds 已提交
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
 *	linux/net/ipv4/tcp.c
 *	linux/net/ipv4/tcp_input.c
 *	linux/net/ipv4/tcp_output.c
 *
 *	Fixes:
 *	Hideaki YOSHIFUJI	:	sin6_scope_id support
 *	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.
 *	YOSHIFUJI Hideaki @USAGI:	convert /proc/net/tcp6 to seq_file.
 *
 *	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.
 */

H
Herbert Xu 已提交
26
#include <linux/bottom_half.h>
L
Linus Torvalds 已提交
27 28 29 30 31 32 33 34 35 36 37 38 39 40
#include <linux/module.h>
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/jiffies.h>
#include <linux/in.h>
#include <linux/in6.h>
#include <linux/netdevice.h>
#include <linux/init.h>
#include <linux/jhash.h>
#include <linux/ipsec.h>
#include <linux/times.h>
41
#include <linux/slab.h>
L
Linus Torvalds 已提交
42 43 44 45 46 47 48

#include <linux/ipv6.h>
#include <linux/icmpv6.h>
#include <linux/random.h>

#include <net/tcp.h>
#include <net/ndisc.h>
49
#include <net/inet6_hashtables.h>
50
#include <net/inet6_connection_sock.h>
L
Linus Torvalds 已提交
51 52 53 54 55 56 57 58 59 60
#include <net/ipv6.h>
#include <net/transp_v6.h>
#include <net/addrconf.h>
#include <net/ip6_route.h>
#include <net/ip6_checksum.h>
#include <net/inet_ecn.h>
#include <net/protocol.h>
#include <net/xfrm.h>
#include <net/snmp.h>
#include <net/dsfield.h>
61
#include <net/timewait_sock.h>
62
#include <net/netdma.h>
63
#include <net/inet_common.h>
64
#include <net/secure_seq.h>
G
Glauber Costa 已提交
65
#include <net/tcp_memcontrol.h>
L
Linus Torvalds 已提交
66 67 68 69 70 71

#include <asm/uaccess.h>

#include <linux/proc_fs.h>
#include <linux/seq_file.h>

72 73 74 75
#include <linux/crypto.h>
#include <linux/scatterlist.h>

static void	tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb);
76 77
static void	tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
				      struct request_sock *req);
L
Linus Torvalds 已提交
78 79

static int	tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb);
80
static void	__tcp_v6_send_check(struct sk_buff *skb,
81 82
				    const struct in6_addr *saddr,
				    const struct in6_addr *daddr);
L
Linus Torvalds 已提交
83

84 85
static const struct inet_connection_sock_af_ops ipv6_mapped;
static const struct inet_connection_sock_af_ops ipv6_specific;
86
#ifdef CONFIG_TCP_MD5SIG
87 88
static const struct tcp_sock_af_ops tcp_sock_ipv6_specific;
static const struct tcp_sock_af_ops tcp_sock_ipv6_mapped_specific;
89 90
#else
static struct tcp_md5sig_key *tcp_v6_md5_do_lookup(struct sock *sk,
91
						   const struct in6_addr *addr)
92 93 94
{
	return NULL;
}
95
#endif
L
Linus Torvalds 已提交
96 97 98 99

static void tcp_v6_hash(struct sock *sk)
{
	if (sk->sk_state != TCP_CLOSE) {
100
		if (inet_csk(sk)->icsk_af_ops == &ipv6_mapped) {
L
Linus Torvalds 已提交
101 102 103 104
			tcp_prot.hash(sk);
			return;
		}
		local_bh_disable();
105
		__inet6_hash(sk, NULL);
L
Linus Torvalds 已提交
106 107 108 109
		local_bh_enable();
	}
}

H
Herbert Xu 已提交
110
static __inline__ __sum16 tcp_v6_check(int len,
111 112
				   const struct in6_addr *saddr,
				   const struct in6_addr *daddr,
113
				   __wsum base)
L
Linus Torvalds 已提交
114 115 116 117
{
	return csum_ipv6_magic(saddr, daddr, len, IPPROTO_TCP, base);
}

118
static __u32 tcp_v6_init_sequence(const struct sk_buff *skb)
L
Linus Torvalds 已提交
119
{
120 121
	return secure_tcpv6_sequence_number(ipv6_hdr(skb)->daddr.s6_addr32,
					    ipv6_hdr(skb)->saddr.s6_addr32,
122 123
					    tcp_hdr(skb)->dest,
					    tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
124 125
}

126
static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
L
Linus Torvalds 已提交
127 128 129
			  int addr_len)
{
	struct sockaddr_in6 *usin = (struct sockaddr_in6 *) uaddr;
130
	struct inet_sock *inet = inet_sk(sk);
131
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
132 133
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
134
	struct in6_addr *saddr = NULL, *final_p, final;
135
	struct rt6_info *rt;
136
	struct flowi6 fl6;
L
Linus Torvalds 已提交
137 138 139 140
	struct dst_entry *dst;
	int addr_type;
	int err;

141
	if (addr_len < SIN6_LEN_RFC2133)
L
Linus Torvalds 已提交
142 143
		return -EINVAL;

144
	if (usin->sin6_family != AF_INET6)
E
Eric Dumazet 已提交
145
		return -EAFNOSUPPORT;
L
Linus Torvalds 已提交
146

147
	memset(&fl6, 0, sizeof(fl6));
L
Linus Torvalds 已提交
148 149

	if (np->sndflow) {
150 151 152
		fl6.flowlabel = usin->sin6_flowinfo&IPV6_FLOWINFO_MASK;
		IP6_ECN_flow_init(fl6.flowlabel);
		if (fl6.flowlabel&IPV6_FLOWLABEL_MASK) {
L
Linus Torvalds 已提交
153
			struct ip6_flowlabel *flowlabel;
154
			flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
L
Linus Torvalds 已提交
155 156
			if (flowlabel == NULL)
				return -EINVAL;
A
Alexey Dobriyan 已提交
157
			usin->sin6_addr = flowlabel->dst;
L
Linus Torvalds 已提交
158 159 160 161 162
			fl6_sock_release(flowlabel);
		}
	}

	/*
163 164 165 166 167
	 *	connect() to INADDR_ANY means loopback (BSD'ism).
	 */

	if(ipv6_addr_any(&usin->sin6_addr))
		usin->sin6_addr.s6_addr[15] = 0x1;
L
Linus Torvalds 已提交
168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198

	addr_type = ipv6_addr_type(&usin->sin6_addr);

	if(addr_type & IPV6_ADDR_MULTICAST)
		return -ENETUNREACH;

	if (addr_type&IPV6_ADDR_LINKLOCAL) {
		if (addr_len >= sizeof(struct sockaddr_in6) &&
		    usin->sin6_scope_id) {
			/* If interface is set while binding, indices
			 * must coincide.
			 */
			if (sk->sk_bound_dev_if &&
			    sk->sk_bound_dev_if != usin->sin6_scope_id)
				return -EINVAL;

			sk->sk_bound_dev_if = usin->sin6_scope_id;
		}

		/* Connect to link-local address requires an interface */
		if (!sk->sk_bound_dev_if)
			return -EINVAL;
	}

	if (tp->rx_opt.ts_recent_stamp &&
	    !ipv6_addr_equal(&np->daddr, &usin->sin6_addr)) {
		tp->rx_opt.ts_recent = 0;
		tp->rx_opt.ts_recent_stamp = 0;
		tp->write_seq = 0;
	}

A
Alexey Dobriyan 已提交
199
	np->daddr = usin->sin6_addr;
200
	np->flow_label = fl6.flowlabel;
L
Linus Torvalds 已提交
201 202 203 204 205 206

	/*
	 *	TCP over IPv4
	 */

	if (addr_type == IPV6_ADDR_MAPPED) {
207
		u32 exthdrlen = icsk->icsk_ext_hdr_len;
L
Linus Torvalds 已提交
208 209 210 211 212 213 214 215 216 217 218
		struct sockaddr_in sin;

		SOCK_DEBUG(sk, "connect: ipv4 mapped\n");

		if (__ipv6_only_sock(sk))
			return -ENETUNREACH;

		sin.sin_family = AF_INET;
		sin.sin_port = usin->sin6_port;
		sin.sin_addr.s_addr = usin->sin6_addr.s6_addr32[3];

219
		icsk->icsk_af_ops = &ipv6_mapped;
L
Linus Torvalds 已提交
220
		sk->sk_backlog_rcv = tcp_v4_do_rcv;
221 222 223
#ifdef CONFIG_TCP_MD5SIG
		tp->af_specific = &tcp_sock_ipv6_mapped_specific;
#endif
L
Linus Torvalds 已提交
224 225 226 227

		err = tcp_v4_connect(sk, (struct sockaddr *)&sin, sizeof(sin));

		if (err) {
228 229
			icsk->icsk_ext_hdr_len = exthdrlen;
			icsk->icsk_af_ops = &ipv6_specific;
L
Linus Torvalds 已提交
230
			sk->sk_backlog_rcv = tcp_v6_do_rcv;
231 232 233
#ifdef CONFIG_TCP_MD5SIG
			tp->af_specific = &tcp_sock_ipv6_specific;
#endif
L
Linus Torvalds 已提交
234 235
			goto failure;
		} else {
E
Eric Dumazet 已提交
236 237 238
			ipv6_addr_set_v4mapped(inet->inet_saddr, &np->saddr);
			ipv6_addr_set_v4mapped(inet->inet_rcv_saddr,
					       &np->rcv_saddr);
L
Linus Torvalds 已提交
239 240 241 242 243 244 245 246
		}

		return err;
	}

	if (!ipv6_addr_any(&np->rcv_saddr))
		saddr = &np->rcv_saddr;

247
	fl6.flowi6_proto = IPPROTO_TCP;
A
Alexey Dobriyan 已提交
248 249
	fl6.daddr = np->daddr;
	fl6.saddr = saddr ? *saddr : np->saddr;
250 251
	fl6.flowi6_oif = sk->sk_bound_dev_if;
	fl6.flowi6_mark = sk->sk_mark;
252 253
	fl6.fl6_dport = usin->sin6_port;
	fl6.fl6_sport = inet->inet_sport;
L
Linus Torvalds 已提交
254

255
	final_p = fl6_update_dst(&fl6, np->opt, &final);
L
Linus Torvalds 已提交
256

257
	security_sk_classify_flow(sk, flowi6_to_flowi(&fl6));
V
Venkat Yekkirala 已提交
258

259
	dst = ip6_dst_lookup_flow(sk, &fl6, final_p, true);
260 261
	if (IS_ERR(dst)) {
		err = PTR_ERR(dst);
L
Linus Torvalds 已提交
262
		goto failure;
263
	}
L
Linus Torvalds 已提交
264 265

	if (saddr == NULL) {
266
		saddr = &fl6.saddr;
A
Alexey Dobriyan 已提交
267
		np->rcv_saddr = *saddr;
L
Linus Torvalds 已提交
268 269 270
	}

	/* set the source address */
A
Alexey Dobriyan 已提交
271
	np->saddr = *saddr;
E
Eric Dumazet 已提交
272
	inet->inet_rcv_saddr = LOOPBACK4_IPV6;
L
Linus Torvalds 已提交
273

H
Herbert Xu 已提交
274
	sk->sk_gso_type = SKB_GSO_TCPV6;
275
	__ip6_dst_store(sk, dst, NULL, NULL);
L
Linus Torvalds 已提交
276

277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296
	rt = (struct rt6_info *) dst;
	if (tcp_death_row.sysctl_tw_recycle &&
	    !tp->rx_opt.ts_recent_stamp &&
	    ipv6_addr_equal(&rt->rt6i_dst.addr, &np->daddr)) {
		struct inet_peer *peer = rt6_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) {
			inet_peer_refcheck(peer);
			if ((u32)get_seconds() - peer->tcp_ts_stamp <= TCP_PAWS_MSL) {
				tp->rx_opt.ts_recent_stamp = peer->tcp_ts_stamp;
				tp->rx_opt.ts_recent = peer->tcp_ts;
			}
		}
	}

297
	icsk->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
298
	if (np->opt)
299 300
		icsk->icsk_ext_hdr_len = (np->opt->opt_flen +
					  np->opt->opt_nflen);
L
Linus Torvalds 已提交
301 302 303

	tp->rx_opt.mss_clamp = IPV6_MIN_MTU - sizeof(struct tcphdr) - sizeof(struct ipv6hdr);

E
Eric Dumazet 已提交
304
	inet->inet_dport = usin->sin6_port;
L
Linus Torvalds 已提交
305 306

	tcp_set_state(sk, TCP_SYN_SENT);
307
	err = inet6_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
308 309 310 311 312 313
	if (err)
		goto late_failure;

	if (!tp->write_seq)
		tp->write_seq = secure_tcpv6_sequence_number(np->saddr.s6_addr32,
							     np->daddr.s6_addr32,
E
Eric Dumazet 已提交
314 315
							     inet->inet_sport,
							     inet->inet_dport);
L
Linus Torvalds 已提交
316 317 318 319 320 321 322 323 324 325 326

	err = tcp_connect(sk);
	if (err)
		goto late_failure;

	return 0;

late_failure:
	tcp_set_state(sk, TCP_CLOSE);
	__sk_dst_reset(sk);
failure:
E
Eric Dumazet 已提交
327
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
328 329 330 331 332
	sk->sk_route_caps = 0;
	return err;
}

static void tcp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
333
		u8 type, u8 code, int offset, __be32 info)
L
Linus Torvalds 已提交
334
{
335
	const struct ipv6hdr *hdr = (const struct ipv6hdr*)skb->data;
336
	const struct tcphdr *th = (struct tcphdr *)(skb->data+offset);
L
Linus Torvalds 已提交
337 338 339
	struct ipv6_pinfo *np;
	struct sock *sk;
	int err;
340
	struct tcp_sock *tp;
L
Linus Torvalds 已提交
341
	__u32 seq;
342
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
343

344
	sk = inet6_lookup(net, &tcp_hashinfo, &hdr->daddr,
345
			th->dest, &hdr->saddr, th->source, skb->dev->ifindex);
L
Linus Torvalds 已提交
346 347

	if (sk == NULL) {
348 349
		ICMP6_INC_STATS_BH(net, __in6_dev_get(skb->dev),
				   ICMP6_MIB_INERRORS);
L
Linus Torvalds 已提交
350 351 352 353
		return;
	}

	if (sk->sk_state == TCP_TIME_WAIT) {
354
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
355 356 357 358 359
		return;
	}

	bh_lock_sock(sk);
	if (sock_owned_by_user(sk))
360
		NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
L
Linus Torvalds 已提交
361 362 363 364

	if (sk->sk_state == TCP_CLOSE)
		goto out;

365 366 367 368 369
	if (ipv6_hdr(skb)->hop_limit < inet6_sk(sk)->min_hopcount) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto out;
	}

L
Linus Torvalds 已提交
370
	tp = tcp_sk(sk);
371
	seq = ntohl(th->seq);
L
Linus Torvalds 已提交
372 373
	if (sk->sk_state != TCP_LISTEN &&
	    !between(seq, tp->snd_una, tp->snd_nxt)) {
374
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
375 376 377 378 379 380
		goto out;
	}

	np = inet6_sk(sk);

	if (type == ICMPV6_PKT_TOOBIG) {
381
		struct dst_entry *dst;
L
Linus Torvalds 已提交
382 383 384 385 386 387 388 389 390 391 392

		if (sock_owned_by_user(sk))
			goto out;
		if ((1 << sk->sk_state) & (TCPF_LISTEN | TCPF_CLOSE))
			goto out;

		/* icmp should have updated the destination cache entry */
		dst = __sk_dst_check(sk, np->dst_cookie);

		if (dst == NULL) {
			struct inet_sock *inet = inet_sk(sk);
393
			struct flowi6 fl6;
L
Linus Torvalds 已提交
394 395 396 397 398

			/* BUGGG_FUTURE: Again, it is not clear how
			   to handle rthdr case. Ignore this complexity
			   for now.
			 */
399 400
			memset(&fl6, 0, sizeof(fl6));
			fl6.flowi6_proto = IPPROTO_TCP;
A
Alexey Dobriyan 已提交
401 402
			fl6.daddr = np->daddr;
			fl6.saddr = np->saddr;
403 404
			fl6.flowi6_oif = sk->sk_bound_dev_if;
			fl6.flowi6_mark = sk->sk_mark;
405 406
			fl6.fl6_dport = inet->inet_dport;
			fl6.fl6_sport = inet->inet_sport;
407 408 409
			security_skb_classify_flow(skb, flowi6_to_flowi(&fl6));

			dst = ip6_dst_lookup_flow(sk, &fl6, NULL, false);
410 411
			if (IS_ERR(dst)) {
				sk->sk_err_soft = -PTR_ERR(dst);
L
Linus Torvalds 已提交
412 413 414 415 416 417
				goto out;
			}

		} else
			dst_hold(dst);

418
		if (inet_csk(sk)->icsk_pmtu_cookie > dst_mtu(dst)) {
L
Linus Torvalds 已提交
419 420 421 422 423 424 425 426 427
			tcp_sync_mss(sk, dst_mtu(dst));
			tcp_simple_retransmit(sk);
		} /* else let the usual retransmit timer handle it */
		dst_release(dst);
		goto out;
	}

	icmpv6_err_convert(type, code, &err);

428
	/* Might be for an request_sock */
L
Linus Torvalds 已提交
429
	switch (sk->sk_state) {
430
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
431 432 433 434
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

435 436
		req = inet6_csk_search_req(sk, &prev, th->dest, &hdr->daddr,
					   &hdr->saddr, inet6_iif(skb));
L
Linus Torvalds 已提交
437 438 439 440 441 442
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		 * an established socket here.
		 */
443
		WARN_ON(req->sk != NULL);
L
Linus Torvalds 已提交
444

445
		if (seq != tcp_rsk(req)->snt_isn) {
446
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
447 448 449
			goto out;
		}

450
		inet_csk_reqsk_queue_drop(sk, req, prev);
L
Linus Torvalds 已提交
451 452 453 454
		goto out;

	case TCP_SYN_SENT:
	case TCP_SYN_RECV:  /* Cannot happen.
455
			       It can, it SYNs are crossed. --ANK */
L
Linus Torvalds 已提交
456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477
		if (!sock_owned_by_user(sk)) {
			sk->sk_err = err;
			sk->sk_error_report(sk);		/* Wake people up to see the error (see connect in sock.c) */

			tcp_done(sk);
		} else
			sk->sk_err_soft = err;
		goto out;
	}

	if (!sock_owned_by_user(sk) && np->recverr) {
		sk->sk_err = err;
		sk->sk_error_report(sk);
	} else
		sk->sk_err_soft = err;

out:
	bh_unlock_sock(sk);
	sock_put(sk);
}


478
static int tcp_v6_send_synack(struct sock *sk, struct request_sock *req,
479 480
			      struct request_values *rvp,
			      u16 queue_mapping)
L
Linus Torvalds 已提交
481
{
482
	struct inet6_request_sock *treq = inet6_rsk(req);
L
Linus Torvalds 已提交
483 484 485
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct sk_buff * skb;
	struct ipv6_txoptions *opt = NULL;
486
	struct in6_addr * final_p, final;
487
	struct flowi6 fl6;
488
	struct dst_entry *dst;
489
	int err;
L
Linus Torvalds 已提交
490

491 492
	memset(&fl6, 0, sizeof(fl6));
	fl6.flowi6_proto = IPPROTO_TCP;
A
Alexey Dobriyan 已提交
493 494
	fl6.daddr = treq->rmt_addr;
	fl6.saddr = treq->loc_addr;
495 496 497
	fl6.flowlabel = 0;
	fl6.flowi6_oif = treq->iif;
	fl6.flowi6_mark = sk->sk_mark;
498 499
	fl6.fl6_dport = inet_rsk(req)->rmt_port;
	fl6.fl6_sport = inet_rsk(req)->loc_port;
500
	security_req_classify_flow(req, flowi6_to_flowi(&fl6));
L
Linus Torvalds 已提交
501

502
	opt = np->opt;
503
	final_p = fl6_update_dst(&fl6, opt, &final);
L
Linus Torvalds 已提交
504

505
	dst = ip6_dst_lookup_flow(sk, &fl6, final_p, false);
506 507
	if (IS_ERR(dst)) {
		err = PTR_ERR(dst);
508
		dst = NULL;
509
		goto done;
510
	}
511
	skb = tcp_make_synack(sk, dst, req, rvp);
512
	err = -ENOMEM;
L
Linus Torvalds 已提交
513
	if (skb) {
514
		__tcp_v6_send_check(skb, &treq->loc_addr, &treq->rmt_addr);
L
Linus Torvalds 已提交
515

A
Alexey Dobriyan 已提交
516
		fl6.daddr = treq->rmt_addr;
517
		skb_set_queue_mapping(skb, queue_mapping);
518
		err = ip6_xmit(sk, skb, &fl6, opt, np->tclass);
519
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
520 521 522
	}

done:
523
	if (opt && opt != np->opt)
L
Linus Torvalds 已提交
524 525 526 527
		sock_kfree_s(sk, opt, opt->tot_len);
	return err;
}

528 529 530 531
static int tcp_v6_rtx_synack(struct sock *sk, struct request_sock *req,
			     struct request_values *rvp)
{
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
532
	return tcp_v6_send_synack(sk, req, rvp, 0);
533 534
}

535
static void tcp_v6_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
536
{
537
	kfree_skb(inet6_rsk(req)->pktopts);
L
Linus Torvalds 已提交
538 539
}

540 541
#ifdef CONFIG_TCP_MD5SIG
static struct tcp_md5sig_key *tcp_v6_md5_do_lookup(struct sock *sk,
542
						   const struct in6_addr *addr)
543
{
E
Eric Dumazet 已提交
544
	return tcp_md5_do_lookup(sk, (union tcp_md5_addr *)addr, AF_INET6);
545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574
}

static struct tcp_md5sig_key *tcp_v6_md5_lookup(struct sock *sk,
						struct sock *addr_sk)
{
	return tcp_v6_md5_do_lookup(sk, &inet6_sk(addr_sk)->daddr);
}

static struct tcp_md5sig_key *tcp_v6_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
{
	return tcp_v6_md5_do_lookup(sk, &inet6_rsk(req)->rmt_addr);
}

static int tcp_v6_parse_md5_keys (struct sock *sk, char __user *optval,
				  int optlen)
{
	struct tcp_md5sig cmd;
	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)&cmd.tcpm_addr;

	if (optlen < sizeof(cmd))
		return -EINVAL;

	if (copy_from_user(&cmd, optval, sizeof(cmd)))
		return -EFAULT;

	if (sin6->sin6_family != AF_INET6)
		return -EINVAL;

	if (!cmd.tcpm_keylen) {
B
Brian Haley 已提交
575
		if (ipv6_addr_v4mapped(&sin6->sin6_addr))
E
Eric Dumazet 已提交
576 577 578 579
			return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin6->sin6_addr.s6_addr32[3],
					      AF_INET);
		return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin6->sin6_addr,
				      AF_INET6);
580 581 582 583 584
	}

	if (cmd.tcpm_keylen > TCP_MD5SIG_MAXKEYLEN)
		return -EINVAL;

E
Eric Dumazet 已提交
585 586 587
	if (ipv6_addr_v4mapped(&sin6->sin6_addr))
		return tcp_md5_do_add(sk, (union tcp_md5_addr *)&sin6->sin6_addr.s6_addr32[3],
				      AF_INET, cmd.tcpm_key, cmd.tcpm_keylen, GFP_KERNEL);
588

E
Eric Dumazet 已提交
589 590
	return tcp_md5_do_add(sk, (union tcp_md5_addr *)&sin6->sin6_addr,
			      AF_INET6, cmd.tcpm_key, cmd.tcpm_keylen, GFP_KERNEL);
591 592
}

593
static int tcp_v6_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
594 595
					const struct in6_addr *daddr,
					const struct in6_addr *saddr, int nbytes)
596 597
{
	struct tcp6_pseudohdr *bp;
598
	struct scatterlist sg;
599

600 601
	bp = &hp->md5_blk.ip6;
	/* 1. TCP pseudo-header (RFC2460) */
A
Alexey Dobriyan 已提交
602 603
	bp->saddr = *saddr;
	bp->daddr = *daddr;
604
	bp->protocol = cpu_to_be32(IPPROTO_TCP);
A
Adam Langley 已提交
605
	bp->len = cpu_to_be32(nbytes);
606

607 608 609
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}
610

611
static int tcp_v6_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
612
			       const struct in6_addr *daddr, struct in6_addr *saddr,
E
Eric Dumazet 已提交
613
			       const struct tcphdr *th)
614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631
{
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;

	hp = tcp_get_md5sig_pool();
	if (!hp)
		goto clear_hash_noput;
	desc = &hp->md5_desc;

	if (crypto_hash_init(desc))
		goto clear_hash;
	if (tcp_v6_md5_hash_pseudoheader(hp, daddr, saddr, th->doff << 2))
		goto clear_hash;
	if (tcp_md5_hash_header(hp, th))
		goto clear_hash;
	if (tcp_md5_hash_key(hp, key))
		goto clear_hash;
	if (crypto_hash_final(desc, md5_hash))
632 633 634 635
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
636

637 638 639 640
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
641
	return 1;
642 643
}

644
static int tcp_v6_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
645 646 647
			       const struct sock *sk,
			       const struct request_sock *req,
			       const struct sk_buff *skb)
648
{
649
	const struct in6_addr *saddr, *daddr;
650 651
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
E
Eric Dumazet 已提交
652
	const struct tcphdr *th = tcp_hdr(skb);
653 654 655 656

	if (sk) {
		saddr = &inet6_sk(sk)->saddr;
		daddr = &inet6_sk(sk)->daddr;
657
	} else if (req) {
658 659
		saddr = &inet6_rsk(req)->loc_addr;
		daddr = &inet6_rsk(req)->rmt_addr;
660
	} else {
661
		const struct ipv6hdr *ip6h = ipv6_hdr(skb);
662 663
		saddr = &ip6h->saddr;
		daddr = &ip6h->daddr;
664
	}
665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692

	hp = tcp_get_md5sig_pool();
	if (!hp)
		goto clear_hash_noput;
	desc = &hp->md5_desc;

	if (crypto_hash_init(desc))
		goto clear_hash;

	if (tcp_v6_md5_hash_pseudoheader(hp, daddr, saddr, skb->len))
		goto clear_hash;
	if (tcp_md5_hash_header(hp, th))
		goto clear_hash;
	if (tcp_md5_hash_skb_data(hp, skb, th->doff << 2))
		goto clear_hash;
	if (tcp_md5_hash_key(hp, key))
		goto clear_hash;
	if (crypto_hash_final(desc, md5_hash))
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;

clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
	return 1;
693 694
}

E
Eric Dumazet 已提交
695
static int tcp_v6_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
696
{
697
	const __u8 *hash_location = NULL;
698
	struct tcp_md5sig_key *hash_expected;
699
	const struct ipv6hdr *ip6h = ipv6_hdr(skb);
E
Eric Dumazet 已提交
700
	const struct tcphdr *th = tcp_hdr(skb);
701 702 703 704
	int genhash;
	u8 newhash[16];

	hash_expected = tcp_v6_md5_do_lookup(sk, &ip6h->saddr);
705
	hash_location = tcp_parse_md5sig_option(th);
706

707 708 709 710 711 712
	/* We've parsed the options - do we have a hash? */
	if (!hash_expected && !hash_location)
		return 0;

	if (hash_expected && !hash_location) {
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
713 714 715
		return 1;
	}

716 717
	if (!hash_expected && hash_location) {
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
718 719 720 721
		return 1;
	}

	/* check the signature */
722 723 724 725
	genhash = tcp_v6_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);

726
	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
727 728 729 730
		net_info_ratelimited("MD5 Hash %s for [%pI6c]:%u->[%pI6c]:%u\n",
				     genhash ? "failed" : "mismatch",
				     &ip6h->saddr, ntohs(th->source),
				     &ip6h->daddr, ntohs(th->dest));
731 732 733 734 735 736
		return 1;
	}
	return 0;
}
#endif

737
struct request_sock_ops tcp6_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
738
	.family		=	AF_INET6,
739
	.obj_size	=	sizeof(struct tcp6_request_sock),
740
	.rtx_syn_ack	=	tcp_v6_rtx_synack,
741 742
	.send_ack	=	tcp_v6_reqsk_send_ack,
	.destructor	=	tcp_v6_reqsk_destructor,
743 744
	.send_reset	=	tcp_v6_send_reset,
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
745 746
};

747
#ifdef CONFIG_TCP_MD5SIG
748
static const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = {
749
	.md5_lookup	=	tcp_v6_reqsk_md5_lookup,
750
	.calc_md5_hash	=	tcp_v6_md5_hash_skb,
751
};
752
#endif
753

754
static void __tcp_v6_send_check(struct sk_buff *skb,
755
				const struct in6_addr *saddr, const struct in6_addr *daddr)
L
Linus Torvalds 已提交
756
{
757
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
758

759
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
760
		th->check = ~tcp_v6_check(skb->len, saddr, daddr, 0);
761
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
762
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
763
	} else {
764 765 766
		th->check = tcp_v6_check(skb->len, saddr, daddr,
					 csum_partial(th, th->doff << 2,
						      skb->csum));
L
Linus Torvalds 已提交
767 768 769
	}
}

770
static void tcp_v6_send_check(struct sock *sk, struct sk_buff *skb)
771 772 773 774 775 776
{
	struct ipv6_pinfo *np = inet6_sk(sk);

	__tcp_v6_send_check(skb, &np->saddr, &np->daddr);
}

777 778
static int tcp_v6_gso_send_check(struct sk_buff *skb)
{
779
	const struct ipv6hdr *ipv6h;
780 781 782 783 784
	struct tcphdr *th;

	if (!pskb_may_pull(skb, sizeof(*th)))
		return -EINVAL;

785
	ipv6h = ipv6_hdr(skb);
786
	th = tcp_hdr(skb);
787 788

	th->check = 0;
789
	skb->ip_summed = CHECKSUM_PARTIAL;
790
	__tcp_v6_send_check(skb, &ipv6h->saddr, &ipv6h->daddr);
791 792
	return 0;
}
L
Linus Torvalds 已提交
793

H
Herbert Xu 已提交
794 795
static struct sk_buff **tcp6_gro_receive(struct sk_buff **head,
					 struct sk_buff *skb)
H
Herbert Xu 已提交
796
{
797
	const struct ipv6hdr *iph = skb_gro_network_header(skb);
H
Herbert Xu 已提交
798 799 800

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
801
		if (!tcp_v6_check(skb_gro_len(skb), &iph->saddr, &iph->daddr,
H
Herbert Xu 已提交
802 803 804 805 806 807 808 809 810 811 812 813 814 815
				  skb->csum)) {
			skb->ip_summed = CHECKSUM_UNNECESSARY;
			break;
		}

		/* fall through */
	case CHECKSUM_NONE:
		NAPI_GRO_CB(skb)->flush = 1;
		return NULL;
	}

	return tcp_gro_receive(head, skb);
}

H
Herbert Xu 已提交
816
static int tcp6_gro_complete(struct sk_buff *skb)
H
Herbert Xu 已提交
817
{
818
	const struct ipv6hdr *iph = ipv6_hdr(skb);
H
Herbert Xu 已提交
819 820 821 822 823 824 825 826 827
	struct tcphdr *th = tcp_hdr(skb);

	th->check = ~tcp_v6_check(skb->len - skb_transport_offset(skb),
				  &iph->saddr, &iph->daddr, 0);
	skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;

	return tcp_gro_complete(skb);
}

828
static void tcp_v6_send_response(struct sk_buff *skb, u32 seq, u32 ack, u32 win,
829
				 u32 ts, struct tcp_md5sig_key *key, int rst, u8 tclass)
L
Linus Torvalds 已提交
830
{
831 832
	const struct tcphdr *th = tcp_hdr(skb);
	struct tcphdr *t1;
L
Linus Torvalds 已提交
833
	struct sk_buff *buff;
834
	struct flowi6 fl6;
E
Eric Dumazet 已提交
835
	struct net *net = dev_net(skb_dst(skb)->dev);
836
	struct sock *ctl_sk = net->ipv6.tcp_sk;
837
	unsigned int tot_len = sizeof(struct tcphdr);
E
Eric Dumazet 已提交
838
	struct dst_entry *dst;
839
	__be32 *topt;
L
Linus Torvalds 已提交
840

841 842
	if (ts)
		tot_len += TCPOLEN_TSTAMP_ALIGNED;
843 844 845 846 847 848
#ifdef CONFIG_TCP_MD5SIG
	if (key)
		tot_len += TCPOLEN_MD5SIG_ALIGNED;
#endif

	buff = alloc_skb(MAX_HEADER + sizeof(struct ipv6hdr) + tot_len,
L
Linus Torvalds 已提交
849
			 GFP_ATOMIC);
850 851
	if (buff == NULL)
		return;
L
Linus Torvalds 已提交
852

853
	skb_reserve(buff, MAX_HEADER + sizeof(struct ipv6hdr) + tot_len);
L
Linus Torvalds 已提交
854

855
	t1 = (struct tcphdr *) skb_push(buff, tot_len);
856
	skb_reset_transport_header(buff);
L
Linus Torvalds 已提交
857 858 859 860 861

	/* Swap the send and the receive. */
	memset(t1, 0, sizeof(*t1));
	t1->dest = th->source;
	t1->source = th->dest;
862
	t1->doff = tot_len / 4;
863 864 865 866 867
	t1->seq = htonl(seq);
	t1->ack_seq = htonl(ack);
	t1->ack = !rst || !th->ack;
	t1->rst = rst;
	t1->window = htons(win);
L
Linus Torvalds 已提交
868

869 870
	topt = (__be32 *)(t1 + 1);

871 872 873 874 875 876 877
	if (ts) {
		*topt++ = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				(TCPOPT_TIMESTAMP << 8) | TCPOLEN_TIMESTAMP);
		*topt++ = htonl(tcp_time_stamp);
		*topt++ = htonl(ts);
	}

878 879
#ifdef CONFIG_TCP_MD5SIG
	if (key) {
880 881 882
		*topt++ = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				(TCPOPT_MD5SIG << 8) | TCPOLEN_MD5SIG);
		tcp_v6_md5_hash_hdr((__u8 *)topt, key,
883 884
				    &ipv6_hdr(skb)->saddr,
				    &ipv6_hdr(skb)->daddr, t1);
885 886 887
	}
#endif

888
	memset(&fl6, 0, sizeof(fl6));
A
Alexey Dobriyan 已提交
889 890
	fl6.daddr = ipv6_hdr(skb)->saddr;
	fl6.saddr = ipv6_hdr(skb)->daddr;
L
Linus Torvalds 已提交
891

892 893 894
	buff->ip_summed = CHECKSUM_PARTIAL;
	buff->csum = 0;

895
	__tcp_v6_send_check(buff, &fl6.saddr, &fl6.daddr);
L
Linus Torvalds 已提交
896

897 898
	fl6.flowi6_proto = IPPROTO_TCP;
	fl6.flowi6_oif = inet6_iif(skb);
899 900
	fl6.fl6_dport = t1->dest;
	fl6.fl6_sport = t1->source;
901
	security_skb_classify_flow(skb, flowi6_to_flowi(&fl6));
L
Linus Torvalds 已提交
902

903 904 905 906
	/* Pass a socket to ip6_dst_lookup either it is for RST
	 * Underlying function will use this to retrieve the network
	 * namespace
	 */
907
	dst = ip6_dst_lookup_flow(ctl_sk, &fl6, NULL, false);
908 909
	if (!IS_ERR(dst)) {
		skb_dst_set(buff, dst);
910
		ip6_xmit(ctl_sk, buff, &fl6, NULL, tclass);
911 912 913 914
		TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
		if (rst)
			TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
		return;
L
Linus Torvalds 已提交
915 916 917 918 919
	}

	kfree_skb(buff);
}

920
static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
921
{
922
	const struct tcphdr *th = tcp_hdr(skb);
923
	u32 seq = 0, ack_seq = 0;
924
	struct tcp_md5sig_key *key = NULL;
925 926 927 928 929 930 931
#ifdef CONFIG_TCP_MD5SIG
	const __u8 *hash_location = NULL;
	struct ipv6hdr *ipv6h = ipv6_hdr(skb);
	unsigned char newhash[16];
	int genhash;
	struct sock *sk1 = NULL;
#endif
L
Linus Torvalds 已提交
932

933
	if (th->rst)
L
Linus Torvalds 已提交
934 935
		return;

936 937
	if (!ipv6_unicast_destination(skb))
		return;
L
Linus Torvalds 已提交
938

939
#ifdef CONFIG_TCP_MD5SIG
940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965
	hash_location = tcp_parse_md5sig_option(th);
	if (!sk && hash_location) {
		/*
		 * active side is lost. Try to find listening socket through
		 * source port, and then find md5 key through listening socket.
		 * we are not loose security here:
		 * Incoming packet is checked with md5 hash with finding key,
		 * no RST generated if md5 hash doesn't match.
		 */
		sk1 = inet6_lookup_listener(dev_net(skb_dst(skb)->dev),
					   &tcp_hashinfo, &ipv6h->daddr,
					   ntohs(th->source), inet6_iif(skb));
		if (!sk1)
			return;

		rcu_read_lock();
		key = tcp_v6_md5_do_lookup(sk1, &ipv6h->saddr);
		if (!key)
			goto release_sk1;

		genhash = tcp_v6_md5_hash_skb(newhash, key, NULL, NULL, skb);
		if (genhash || memcmp(hash_location, newhash, 16) != 0)
			goto release_sk1;
	} else {
		key = sk ? tcp_v6_md5_do_lookup(sk, &ipv6h->saddr) : NULL;
	}
966 967
#endif

968 969 970 971 972
	if (th->ack)
		seq = ntohl(th->ack_seq);
	else
		ack_seq = ntohl(th->seq) + th->syn + th->fin + skb->len -
			  (th->doff << 2);
L
Linus Torvalds 已提交
973

974
	tcp_v6_send_response(skb, seq, ack_seq, 0, 0, key, 1, 0);
975 976 977 978 979 980 981 982

#ifdef CONFIG_TCP_MD5SIG
release_sk1:
	if (sk1) {
		rcu_read_unlock();
		sock_put(sk1);
	}
#endif
983
}
L
Linus Torvalds 已提交
984

985
static void tcp_v6_send_ack(struct sk_buff *skb, u32 seq, u32 ack, u32 win, u32 ts,
986
			    struct tcp_md5sig_key *key, u8 tclass)
987
{
988
	tcp_v6_send_response(skb, seq, ack, win, ts, key, 0, tclass);
L
Linus Torvalds 已提交
989 990 991 992
}

static void tcp_v6_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
993
	struct inet_timewait_sock *tw = inet_twsk(sk);
994
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
995

996
	tcp_v6_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
997
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
998 999
			tcptw->tw_ts_recent, tcp_twsk_md5_key(tcptw),
			tw->tw_tclass);
L
Linus Torvalds 已提交
1000

1001
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
1002 1003
}

1004 1005
static void tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
				  struct request_sock *req)
L
Linus Torvalds 已提交
1006
{
1007
	tcp_v6_send_ack(skb, tcp_rsk(req)->snt_isn + 1, tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd, req->ts_recent,
1008
			tcp_v6_md5_do_lookup(sk, &ipv6_hdr(skb)->daddr), 0);
L
Linus Torvalds 已提交
1009 1010 1011 1012 1013
}


static struct sock *tcp_v6_hnd_req(struct sock *sk,struct sk_buff *skb)
{
1014
	struct request_sock *req, **prev;
1015
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1016 1017 1018
	struct sock *nsk;

	/* Find possible connection requests. */
1019
	req = inet6_csk_search_req(sk, &prev, th->source,
1020 1021
				   &ipv6_hdr(skb)->saddr,
				   &ipv6_hdr(skb)->daddr, inet6_iif(skb));
L
Linus Torvalds 已提交
1022 1023 1024
	if (req)
		return tcp_check_req(sk, skb, req, prev);

1025
	nsk = __inet6_lookup_established(sock_net(sk), &tcp_hashinfo,
1026 1027
			&ipv6_hdr(skb)->saddr, th->source,
			&ipv6_hdr(skb)->daddr, ntohs(th->dest), inet6_iif(skb));
L
Linus Torvalds 已提交
1028 1029 1030 1031 1032 1033

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1034
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1035 1036 1037
		return NULL;
	}

1038
#ifdef CONFIG_SYN_COOKIES
1039
	if (!th->syn)
1040
		sk = cookie_v6_check(sk, skb);
L
Linus Torvalds 已提交
1041 1042 1043 1044 1045 1046 1047 1048 1049
#endif
	return sk;
}

/* FIXME: this is substantially similar to the ipv4 code.
 * Can some kind of merge be done? -- erics
 */
static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
{
1050
	struct tcp_extend_values tmp_ext;
1051
	struct tcp_options_received tmp_opt;
1052
	const u8 *hash_location;
1053
	struct request_sock *req;
1054
	struct inet6_request_sock *treq;
L
Linus Torvalds 已提交
1055 1056
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
1057
	__u32 isn = TCP_SKB_CB(skb)->when;
1058
	struct dst_entry *dst = NULL;
E
Eric Dumazet 已提交
1059
	bool want_cookie = false;
L
Linus Torvalds 已提交
1060 1061 1062 1063 1064

	if (skb->protocol == htons(ETH_P_IP))
		return tcp_v4_conn_request(sk, skb);

	if (!ipv6_unicast_destination(skb))
1065
		goto drop;
L
Linus Torvalds 已提交
1066

1067
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
1068 1069 1070
		want_cookie = tcp_syn_flood_action(sk, skb, "TCPv6");
		if (!want_cookie)
			goto drop;
L
Linus Torvalds 已提交
1071 1072
	}

1073
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1074 1075
		goto drop;

1076
	req = inet6_reqsk_alloc(&tcp6_request_sock_ops);
L
Linus Torvalds 已提交
1077 1078 1079
	if (req == NULL)
		goto drop;

1080 1081 1082 1083
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv6_ops;
#endif

L
Linus Torvalds 已提交
1084 1085 1086
	tcp_clear_options(&tmp_opt);
	tmp_opt.mss_clamp = IPV6_MIN_MTU - sizeof(struct tcphdr) - sizeof(struct ipv6hdr);
	tmp_opt.user_mss = tp->rx_opt.user_mss;
1087
	tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103

	if (tmp_opt.cookie_plus > 0 &&
	    tmp_opt.saw_tstamp &&
	    !tp->rx_opt.cookie_out_never &&
	    (sysctl_tcp_cookie_size > 0 ||
	     (tp->cookie_values != NULL &&
	      tp->cookie_values->cookie_desired > 0))) {
		u8 *c;
		u32 *d;
		u32 *mess = &tmp_ext.cookie_bakery[COOKIE_DIGEST_WORDS];
		int l = tmp_opt.cookie_plus - TCPOLEN_COOKIE_BASE;

		if (tcp_cookie_generator(&tmp_ext.cookie_bakery[0]) != 0)
			goto drop_and_free;

		/* Secret recipe starts with IP addresses */
1104
		d = (__force u32 *)&ipv6_hdr(skb)->daddr.s6_addr32[0];
1105 1106 1107 1108
		*mess++ ^= *d++;
		*mess++ ^= *d++;
		*mess++ ^= *d++;
		*mess++ ^= *d++;
1109
		d = (__force u32 *)&ipv6_hdr(skb)->saddr.s6_addr32[0];
1110 1111 1112 1113 1114 1115 1116 1117 1118
		*mess++ ^= *d++;
		*mess++ ^= *d++;
		*mess++ ^= *d++;
		*mess++ ^= *d++;

		/* plus variable length Initiator Cookie */
		c = (u8 *)mess;
		while (l-- > 0)
			*c++ ^= *hash_location++;
L
Linus Torvalds 已提交
1119

E
Eric Dumazet 已提交
1120
		want_cookie = false;	/* not our kind of cookie */
1121 1122 1123 1124 1125 1126 1127 1128 1129 1130
		tmp_ext.cookie_out_never = 0; /* false */
		tmp_ext.cookie_plus = tmp_opt.cookie_plus;
	} else if (!tp->rx_opt.cookie_in_always) {
		/* redundant indications, but ensure initialization. */
		tmp_ext.cookie_out_never = 1; /* true */
		tmp_ext.cookie_plus = 0;
	} else {
		goto drop_and_free;
	}
	tmp_ext.cookie_in_always = tp->rx_opt.cookie_in_always;
L
Linus Torvalds 已提交
1131

1132
	if (want_cookie && !tmp_opt.saw_tstamp)
1133 1134
		tcp_clear_options(&tmp_opt);

L
Linus Torvalds 已提交
1135 1136 1137
	tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;
	tcp_openreq_init(req, &tmp_opt, skb);

1138
	treq = inet6_rsk(req);
A
Alexey Dobriyan 已提交
1139 1140
	treq->rmt_addr = ipv6_hdr(skb)->saddr;
	treq->loc_addr = ipv6_hdr(skb)->daddr;
1141
	if (!want_cookie || tmp_opt.tstamp_ok)
1142
		TCP_ECN_create_request(req, skb);
1143

1144 1145 1146 1147 1148 1149 1150
	treq->iif = sk->sk_bound_dev_if;

	/* So that link locals have meaning */
	if (!sk->sk_bound_dev_if &&
	    ipv6_addr_type(&treq->rmt_addr) & IPV6_ADDR_LINKLOCAL)
		treq->iif = inet6_iif(skb);

1151
	if (!isn) {
1152 1153
		struct inet_peer *peer = NULL;

1154 1155 1156 1157 1158 1159
		if (ipv6_opt_accepted(sk, skb) ||
		    np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo ||
		    np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim) {
			atomic_inc(&skb->users);
			treq->pktopts = skb;
		}
1160 1161

		if (want_cookie) {
1162 1163
			isn = cookie_v6_init_sequence(sk, skb, &req->mss);
			req->cookie_ts = tmp_opt.tstamp_ok;
1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179
			goto have_isn;
		}

		/* VJ's idea. We save last timestamp seen
		 * from the destination in peer table, when entering
		 * state TIME-WAIT, and check against it before
		 * accepting new connection request.
		 *
		 * If "isn" is not zero, this request hit alive
		 * timewait bucket, so that all the necessary checks
		 * are made in the function processing timewait state.
		 */
		if (tmp_opt.saw_tstamp &&
		    tcp_death_row.sysctl_tw_recycle &&
		    (dst = inet6_csk_route_req(sk, req)) != NULL &&
		    (peer = rt6_get_peer((struct rt6_info *)dst)) != NULL &&
1180
		    ipv6_addr_equal((struct in6_addr *)peer->daddr.addr.a6,
1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
				    &treq->rmt_addr)) {
			inet_peer_refcheck(peer);
			if ((u32)get_seconds() - peer->tcp_ts_stamp < TCP_PAWS_MSL &&
			    (s32)(peer->tcp_ts - req->ts_recent) >
							TCP_PAWS_WINDOW) {
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
				goto drop_and_release;
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
			  (sysctl_max_syn_backlog >> 2)) &&
			 (!peer || !peer->tcp_ts_stamp) &&
			 (!dst || !dst_metric(dst, RTAX_RTT))) {
			/* Without syncookies last quarter of
			 * backlog is filled with destinations,
			 * proven to be alive.
			 * It means that we continue to communicate
			 * to destinations, already remembered
			 * to the moment of synflood.
			 */
			LIMIT_NETDEBUG(KERN_DEBUG "TCP: drop open request from %pI6/%u\n",
				       &treq->rmt_addr, ntohs(tcp_hdr(skb)->source));
			goto drop_and_release;
1206
		}
1207 1208

		isn = tcp_v6_init_sequence(skb);
1209
	}
1210
have_isn:
1211
	tcp_rsk(req)->snt_isn = isn;
1212
	tcp_rsk(req)->snt_synack = tcp_time_stamp;
L
Linus Torvalds 已提交
1213

1214 1215
	security_inet_conn_request(sk, skb, req);

1216
	if (tcp_v6_send_synack(sk, req,
1217 1218
			       (struct request_values *)&tmp_ext,
			       skb_get_queue_mapping(skb)) ||
1219
	    want_cookie)
1220
		goto drop_and_free;
L
Linus Torvalds 已提交
1221

1222 1223
	inet6_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
	return 0;
L
Linus Torvalds 已提交
1224

1225 1226
drop_and_release:
	dst_release(dst);
1227 1228
drop_and_free:
	reqsk_free(req);
L
Linus Torvalds 已提交
1229 1230 1231 1232 1233
drop:
	return 0; /* don't send reset */
}

static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
1234
					  struct request_sock *req,
L
Linus Torvalds 已提交
1235 1236
					  struct dst_entry *dst)
{
1237
	struct inet6_request_sock *treq;
L
Linus Torvalds 已提交
1238 1239 1240 1241 1242 1243
	struct ipv6_pinfo *newnp, *np = inet6_sk(sk);
	struct tcp6_sock *newtcp6sk;
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
	struct ipv6_txoptions *opt;
1244 1245 1246
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
L
Linus Torvalds 已提交
1247 1248 1249 1250 1251 1252 1253 1254

	if (skb->protocol == htons(ETH_P_IP)) {
		/*
		 *	v6 mapped
		 */

		newsk = tcp_v4_syn_recv_sock(sk, skb, req, dst);

1255
		if (newsk == NULL)
L
Linus Torvalds 已提交
1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
			return NULL;

		newtcp6sk = (struct tcp6_sock *)newsk;
		inet_sk(newsk)->pinet6 = &newtcp6sk->inet6;

		newinet = inet_sk(newsk);
		newnp = inet6_sk(newsk);
		newtp = tcp_sk(newsk);

		memcpy(newnp, np, sizeof(struct ipv6_pinfo));

E
Eric Dumazet 已提交
1267
		ipv6_addr_set_v4mapped(newinet->inet_daddr, &newnp->daddr);
L
Linus Torvalds 已提交
1268

E
Eric Dumazet 已提交
1269
		ipv6_addr_set_v4mapped(newinet->inet_saddr, &newnp->saddr);
L
Linus Torvalds 已提交
1270

A
Alexey Dobriyan 已提交
1271
		newnp->rcv_saddr = newnp->saddr;
L
Linus Torvalds 已提交
1272

1273
		inet_csk(newsk)->icsk_af_ops = &ipv6_mapped;
L
Linus Torvalds 已提交
1274
		newsk->sk_backlog_rcv = tcp_v4_do_rcv;
1275 1276 1277 1278
#ifdef CONFIG_TCP_MD5SIG
		newtp->af_specific = &tcp_sock_ipv6_mapped_specific;
#endif

1279 1280
		newnp->ipv6_ac_list = NULL;
		newnp->ipv6_fl_list = NULL;
L
Linus Torvalds 已提交
1281 1282
		newnp->pktoptions  = NULL;
		newnp->opt	   = NULL;
1283
		newnp->mcast_oif   = inet6_iif(skb);
1284
		newnp->mcast_hops  = ipv6_hdr(skb)->hop_limit;
1285
		newnp->rcv_tclass  = ipv6_tclass(ipv6_hdr(skb));
L
Linus Torvalds 已提交
1286

1287 1288 1289 1290
		/*
		 * No need to charge this sock to the relevant IPv6 refcnt debug socks count
		 * here, tcp_create_openreq_child now does this for us, see the comment in
		 * that function for the gory details. -acme
L
Linus Torvalds 已提交
1291 1292 1293
		 */

		/* It is tricky place. Until this moment IPv4 tcp
1294
		   worked with IPv6 icsk.icsk_af_ops.
L
Linus Torvalds 已提交
1295 1296
		   Sync it now.
		 */
1297
		tcp_sync_mss(newsk, inet_csk(newsk)->icsk_pmtu_cookie);
L
Linus Torvalds 已提交
1298 1299 1300 1301

		return newsk;
	}

1302
	treq = inet6_rsk(req);
L
Linus Torvalds 已提交
1303 1304 1305 1306 1307
	opt = np->opt;

	if (sk_acceptq_is_full(sk))
		goto out_overflow;

1308 1309 1310
	if (!dst) {
		dst = inet6_csk_route_req(sk, req);
		if (!dst)
L
Linus Torvalds 已提交
1311
			goto out;
1312
	}
L
Linus Torvalds 已提交
1313 1314 1315

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (newsk == NULL)
1316
		goto out_nonewsk;
L
Linus Torvalds 已提交
1317

1318 1319 1320 1321 1322
	/*
	 * No need to charge this sock to the relevant IPv6 refcnt debug socks
	 * count here, tcp_create_openreq_child now does this for us, see the
	 * comment in that function for the gory details. -acme
	 */
L
Linus Torvalds 已提交
1323

1324
	newsk->sk_gso_type = SKB_GSO_TCPV6;
1325
	__ip6_dst_store(newsk, dst, NULL, NULL);
L
Linus Torvalds 已提交
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335

	newtcp6sk = (struct tcp6_sock *)newsk;
	inet_sk(newsk)->pinet6 = &newtcp6sk->inet6;

	newtp = tcp_sk(newsk);
	newinet = inet_sk(newsk);
	newnp = inet6_sk(newsk);

	memcpy(newnp, np, sizeof(struct ipv6_pinfo));

A
Alexey Dobriyan 已提交
1336 1337 1338
	newnp->daddr = treq->rmt_addr;
	newnp->saddr = treq->loc_addr;
	newnp->rcv_saddr = treq->loc_addr;
1339
	newsk->sk_bound_dev_if = treq->iif;
L
Linus Torvalds 已提交
1340

1341
	/* Now IPv6 options...
L
Linus Torvalds 已提交
1342 1343 1344

	   First: no IPv4 options.
	 */
1345
	newinet->inet_opt = NULL;
1346
	newnp->ipv6_ac_list = NULL;
1347
	newnp->ipv6_fl_list = NULL;
L
Linus Torvalds 已提交
1348 1349 1350 1351 1352 1353

	/* Clone RX bits */
	newnp->rxopt.all = np->rxopt.all;

	/* Clone pktoptions received with SYN */
	newnp->pktoptions = NULL;
1354 1355
	if (treq->pktopts != NULL) {
		newnp->pktoptions = skb_clone(treq->pktopts, GFP_ATOMIC);
1356
		consume_skb(treq->pktopts);
1357
		treq->pktopts = NULL;
L
Linus Torvalds 已提交
1358 1359 1360 1361
		if (newnp->pktoptions)
			skb_set_owner_r(newnp->pktoptions, newsk);
	}
	newnp->opt	  = NULL;
1362
	newnp->mcast_oif  = inet6_iif(skb);
1363
	newnp->mcast_hops = ipv6_hdr(skb)->hop_limit;
1364
	newnp->rcv_tclass = ipv6_tclass(ipv6_hdr(skb));
L
Linus Torvalds 已提交
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377

	/* Clone native IPv6 options from listening socket (if any)

	   Yes, keeping reference count would be much more clever,
	   but we make one more one thing there: reattach optmem
	   to newsk.
	 */
	if (opt) {
		newnp->opt = ipv6_dup_options(newsk, opt);
		if (opt != np->opt)
			sock_kfree_s(sk, opt, opt->tot_len);
	}

1378
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
1379
	if (newnp->opt)
1380 1381
		inet_csk(newsk)->icsk_ext_hdr_len = (newnp->opt->opt_nflen +
						     newnp->opt->opt_flen);
L
Linus Torvalds 已提交
1382

J
John Heffner 已提交
1383
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1384
	tcp_sync_mss(newsk, dst_mtu(dst));
1385
	newtp->advmss = dst_metric_advmss(dst);
1386 1387 1388 1389
	if (tcp_sk(sk)->rx_opt.user_mss &&
	    tcp_sk(sk)->rx_opt.user_mss < newtp->advmss)
		newtp->advmss = tcp_sk(sk)->rx_opt.user_mss;

L
Linus Torvalds 已提交
1390
	tcp_initialize_rcv_mss(newsk);
1391 1392 1393 1394
	if (tcp_rsk(req)->snt_synack)
		tcp_valid_rtt_meas(newsk,
		    tcp_time_stamp - tcp_rsk(req)->snt_synack);
	newtp->total_retrans = req->retrans;
L
Linus Torvalds 已提交
1395

E
Eric Dumazet 已提交
1396 1397
	newinet->inet_daddr = newinet->inet_saddr = LOOPBACK4_IPV6;
	newinet->inet_rcv_saddr = LOOPBACK4_IPV6;
L
Linus Torvalds 已提交
1398

1399 1400 1401 1402 1403 1404 1405 1406
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
	if ((key = tcp_v6_md5_do_lookup(sk, &newnp->daddr)) != NULL) {
		/* We're using one, so create a matching key
		 * on the newsk structure. If we fail to get
		 * memory, then we end up not copying the key
		 * across. Shucks.
		 */
E
Eric Dumazet 已提交
1407 1408
		tcp_md5_do_add(newsk, (union tcp_md5_addr *)&newnp->daddr,
			       AF_INET6, key->key, key->keylen, GFP_ATOMIC);
1409 1410 1411
	}
#endif

1412 1413 1414 1415
	if (__inet_inherit_port(sk, newsk) < 0) {
		sock_put(newsk);
		goto out;
	}
1416
	__inet6_hash(newsk, NULL);
L
Linus Torvalds 已提交
1417 1418 1419 1420

	return newsk;

out_overflow:
1421
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1422
out_nonewsk:
L
Linus Torvalds 已提交
1423 1424 1425
	if (opt && opt != np->opt)
		sock_kfree_s(sk, opt, opt->tot_len);
	dst_release(dst);
1426 1427
out:
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1428 1429 1430
	return NULL;
}

1431
static __sum16 tcp_v6_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1432
{
1433
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
H
Herbert Xu 已提交
1434
		if (!tcp_v6_check(skb->len, &ipv6_hdr(skb)->saddr,
1435
				  &ipv6_hdr(skb)->daddr, skb->csum)) {
1436
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1437
			return 0;
1438
		}
L
Linus Torvalds 已提交
1439
	}
1440

H
Herbert Xu 已提交
1441
	skb->csum = ~csum_unfold(tcp_v6_check(skb->len,
1442 1443
					      &ipv6_hdr(skb)->saddr,
					      &ipv6_hdr(skb)->daddr, 0));
1444

L
Linus Torvalds 已提交
1445
	if (skb->len <= 76) {
1446
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475
	}
	return 0;
}

/* The socket must have it's spinlock held when we get
 * here.
 *
 * We have a potential double-lock case here, so even when
 * doing backlog processing we use the BH locking scheme.
 * This is because we cannot sleep with the original spinlock
 * held.
 */
static int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct tcp_sock *tp;
	struct sk_buff *opt_skb = NULL;

	/* Imagine: socket is IPv6. IPv4 packet arrives,
	   goes to IPv4 receive handler and backlogged.
	   From backlog it always goes here. Kerboom...
	   Fortunately, tcp_rcv_established and rcv_established
	   handle them correctly, but it is not case with
	   tcp_v6_hnd_req and tcp_v6_send_reset().   --ANK
	 */

	if (skb->protocol == htons(ETH_P_IP))
		return tcp_v4_do_rcv(sk, skb);

1476 1477 1478 1479 1480
#ifdef CONFIG_TCP_MD5SIG
	if (tcp_v6_inbound_md5_hash (sk, skb))
		goto discard;
#endif

1481
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499
		goto discard;

	/*
	 *	socket locking is here for SMP purposes as backlog rcv
	 *	is currently called with bh processing disabled.
	 */

	/* Do Stevens' IPV6_PKTOPTIONS.

	   Yes, guys, it is the only place in our code, where we
	   may make it not affecting IPv4.
	   The rest of code is protocol independent,
	   and I do not like idea to uglify IPv4.

	   Actually, all the idea behind IPV6_PKTOPTIONS
	   looks not very well thought. For now we latch
	   options, received in the last packet, enqueued
	   by tcp. Feel free to propose better solution.
1500
					       --ANK (980728)
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505
	 */
	if (np->rxopt.all)
		opt_skb = skb_clone(skb, GFP_ATOMIC);

	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1506
		sock_rps_save_rxhash(sk, skb);
1507
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len))
L
Linus Torvalds 已提交
1508 1509 1510 1511 1512 1513
			goto reset;
		if (opt_skb)
			goto ipv6_pktoptions;
		return 0;
	}

1514
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1515 1516
		goto csum_err;

1517
	if (sk->sk_state == TCP_LISTEN) {
L
Linus Torvalds 已提交
1518 1519 1520 1521 1522 1523 1524 1525 1526
		struct sock *nsk = tcp_v6_hnd_req(sk, skb);
		if (!nsk)
			goto discard;

		/*
		 * Queue it on the new socket if the new socket is active,
		 * otherwise we just shortcircuit this and continue with
		 * the new socket..
		 */
1527
		if(nsk != sk) {
1528
			sock_rps_save_rxhash(nsk, skb);
L
Linus Torvalds 已提交
1529 1530 1531 1532 1533 1534
			if (tcp_child_process(sk, nsk, skb))
				goto reset;
			if (opt_skb)
				__kfree_skb(opt_skb);
			return 0;
		}
1535
	} else
1536
		sock_rps_save_rxhash(sk, skb);
L
Linus Torvalds 已提交
1537

1538
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len))
L
Linus Torvalds 已提交
1539 1540 1541 1542 1543 1544
		goto reset;
	if (opt_skb)
		goto ipv6_pktoptions;
	return 0;

reset:
1545
	tcp_v6_send_reset(sk, skb);
L
Linus Torvalds 已提交
1546 1547 1548 1549 1550 1551
discard:
	if (opt_skb)
		__kfree_skb(opt_skb);
	kfree_skb(skb);
	return 0;
csum_err:
1552
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566
	goto discard;


ipv6_pktoptions:
	/* Do you ask, what is it?

	   1. skb was enqueued by tcp.
	   2. skb is added to tail of read queue, rather than out of order.
	   3. socket is not in passive state.
	   4. Finally, it really contains options, which user wants to receive.
	 */
	tp = tcp_sk(sk);
	if (TCP_SKB_CB(opt_skb)->end_seq == tp->rcv_nxt &&
	    !((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN))) {
1567
		if (np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo)
1568
			np->mcast_oif = inet6_iif(opt_skb);
1569
		if (np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim)
1570
			np->mcast_hops = ipv6_hdr(opt_skb)->hop_limit;
1571 1572
		if (np->rxopt.bits.rxtclass)
			np->rcv_tclass = ipv6_tclass(ipv6_hdr(skb));
L
Linus Torvalds 已提交
1573 1574 1575 1576 1577 1578 1579 1580 1581
		if (ipv6_opt_accepted(sk, opt_skb)) {
			skb_set_owner_r(opt_skb, sk);
			opt_skb = xchg(&np->pktoptions, opt_skb);
		} else {
			__kfree_skb(opt_skb);
			opt_skb = xchg(&np->pktoptions, NULL);
		}
	}

1582
	kfree_skb(opt_skb);
L
Linus Torvalds 已提交
1583 1584 1585
	return 0;
}

1586
static int tcp_v6_rcv(struct sk_buff *skb)
L
Linus Torvalds 已提交
1587
{
1588
	const struct tcphdr *th;
1589
	const struct ipv6hdr *hdr;
L
Linus Torvalds 已提交
1590 1591
	struct sock *sk;
	int ret;
1592
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1593 1594 1595 1596 1597 1598 1599

	if (skb->pkt_type != PACKET_HOST)
		goto discard_it;

	/*
	 *	Count it even if it's bad.
	 */
1600
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1601 1602 1603 1604

	if (!pskb_may_pull(skb, sizeof(struct tcphdr)))
		goto discard_it;

1605
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1606 1607 1608 1609 1610 1611

	if (th->doff < sizeof(struct tcphdr)/4)
		goto bad_packet;
	if (!pskb_may_pull(skb, th->doff*4))
		goto discard_it;

1612
	if (!skb_csum_unnecessary(skb) && tcp_v6_checksum_init(skb))
L
Linus Torvalds 已提交
1613 1614
		goto bad_packet;

1615
	th = tcp_hdr(skb);
1616
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
1617 1618 1619 1620 1621
	TCP_SKB_CB(skb)->seq = ntohl(th->seq);
	TCP_SKB_CB(skb)->end_seq = (TCP_SKB_CB(skb)->seq + th->syn + th->fin +
				    skb->len - th->doff*4);
	TCP_SKB_CB(skb)->ack_seq = ntohl(th->ack_seq);
	TCP_SKB_CB(skb)->when = 0;
1622
	TCP_SKB_CB(skb)->ip_dsfield = ipv6_get_dsfield(hdr);
L
Linus Torvalds 已提交
1623 1624
	TCP_SKB_CB(skb)->sacked = 0;

1625
	sk = __inet6_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1626 1627 1628 1629 1630 1631 1632
	if (!sk)
		goto no_tcp_socket;

process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1633 1634 1635 1636 1637
	if (hdr->hop_limit < inet6_sk(sk)->min_hopcount) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto discard_and_relse;
	}

L
Linus Torvalds 已提交
1638 1639 1640
	if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;

1641
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1642 1643 1644 1645
		goto discard_and_relse;

	skb->dev = NULL;

1646
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1647 1648
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1649
#ifdef CONFIG_NET_DMA
1650
		struct tcp_sock *tp = tcp_sk(sk);
1651
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1652
			tp->ucopy.dma_chan = net_dma_find_channel();
1653 1654 1655
		if (tp->ucopy.dma_chan)
			ret = tcp_v6_do_rcv(sk, skb);
		else
C
Chris Leech 已提交
1656 1657 1658 1659 1660
#endif
		{
			if (!tcp_prequeue(sk, skb))
				ret = tcp_v6_do_rcv(sk, skb);
		}
1661 1662
	} else if (unlikely(sk_add_backlog(sk, skb,
					   sk->sk_rcvbuf + sk->sk_sndbuf))) {
Z
Zhu Yi 已提交
1663
		bh_unlock_sock(sk);
1664
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1665 1666
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677
	bh_unlock_sock(sk);

	sock_put(sk);
	return ret ? -1 : 0;

no_tcp_socket:
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
		goto discard_it;

	if (skb->len < (th->doff<<2) || tcp_checksum_complete(skb)) {
bad_packet:
1678
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1679
	} else {
1680
		tcp_v6_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
	}

discard_it:

	/*
	 *	Discard frame
	 */

	kfree_skb(skb);
	return 0;

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1698
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1699 1700 1701 1702
		goto discard_it;
	}

	if (skb->len < (th->doff<<2) || tcp_checksum_complete(skb)) {
1703
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1704
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1705 1706 1707
		goto discard_it;
	}

1708
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1709 1710 1711 1712
	case TCP_TW_SYN:
	{
		struct sock *sk2;

1713
		sk2 = inet6_lookup_listener(dev_net(skb->dev), &tcp_hashinfo,
1714
					    &ipv6_hdr(skb)->daddr,
1715
					    ntohs(th->dest), inet6_iif(skb));
L
Linus Torvalds 已提交
1716
		if (sk2 != NULL) {
1717 1718 1719
			struct inet_timewait_sock *tw = inet_twsk(sk);
			inet_twsk_deschedule(tw, &tcp_death_row);
			inet_twsk_put(tw);
L
Linus Torvalds 已提交
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734
			sk = sk2;
			goto process;
		}
		/* Fall through to ACK */
	}
	case TCP_TW_ACK:
		tcp_v6_timewait_ack(sk, skb);
		break;
	case TCP_TW_RST:
		goto no_tcp_socket;
	case TCP_TW_SUCCESS:;
	}
	goto discard_it;
}

1735 1736
static struct inet_peer *tcp_v6_get_peer(struct sock *sk, bool *release_it)
{
1737 1738
	struct rt6_info *rt = (struct rt6_info *) __sk_dst_get(sk);
	struct ipv6_pinfo *np = inet6_sk(sk);
1739
	struct net *net = sock_net(sk);
1740 1741 1742 1743
	struct inet_peer *peer;

	if (!rt ||
	    !ipv6_addr_equal(&np->daddr, &rt->rt6i_dst.addr)) {
1744
		peer = inet_getpeer_v6(net, &np->daddr, 1);
1745 1746 1747 1748 1749
		*release_it = true;
	} else {
		if (!rt->rt6i_peer)
			rt6_bind_peer(rt, 1);
		peer = rt->rt6i_peer;
1750
		*release_it = false;
1751 1752 1753
	}

	return peer;
1754 1755 1756
}

static void *tcp_v6_tw_get_peer(struct sock *sk)
L
Linus Torvalds 已提交
1757
{
1758 1759
	const struct inet6_timewait_sock *tw6 = inet6_twsk(sk);
	const struct inet_timewait_sock *tw = inet_twsk(sk);
1760
	struct net *net = sock_net(sk);
1761 1762 1763 1764

	if (tw->tw_family == AF_INET)
		return tcp_v4_tw_get_peer(sk);

1765
	return inet_getpeer_v6(net, &tw6->tw_v6_daddr, 1);
L
Linus Torvalds 已提交
1766 1767
}

1768 1769 1770 1771 1772 1773 1774
static struct timewait_sock_ops tcp6_timewait_sock_ops = {
	.twsk_obj_size	= sizeof(struct tcp6_timewait_sock),
	.twsk_unique	= tcp_twsk_unique,
	.twsk_destructor= tcp_twsk_destructor,
	.twsk_getpeer	= tcp_v6_tw_get_peer,
};

1775
static const struct inet_connection_sock_af_ops ipv6_specific = {
1776 1777 1778 1779 1780
	.queue_xmit	   = inet6_csk_xmit,
	.send_check	   = tcp_v6_send_check,
	.rebuild_header	   = inet6_sk_rebuild_header,
	.conn_request	   = tcp_v6_conn_request,
	.syn_recv_sock	   = tcp_v6_syn_recv_sock,
1781
	.get_peer	   = tcp_v6_get_peer,
1782
	.net_header_len	   = sizeof(struct ipv6hdr),
1783
	.net_frag_header_len = sizeof(struct frag_hdr),
1784 1785 1786 1787
	.setsockopt	   = ipv6_setsockopt,
	.getsockopt	   = ipv6_getsockopt,
	.addr2sockaddr	   = inet6_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in6),
1788
	.bind_conflict	   = inet6_csk_bind_conflict,
1789
#ifdef CONFIG_COMPAT
1790 1791
	.compat_setsockopt = compat_ipv6_setsockopt,
	.compat_getsockopt = compat_ipv6_getsockopt,
1792
#endif
L
Linus Torvalds 已提交
1793 1794
};

1795
#ifdef CONFIG_TCP_MD5SIG
1796
static const struct tcp_sock_af_ops tcp_sock_ipv6_specific = {
1797
	.md5_lookup	=	tcp_v6_md5_lookup,
1798
	.calc_md5_hash	=	tcp_v6_md5_hash_skb,
1799 1800
	.md5_parse	=	tcp_v6_parse_md5_keys,
};
1801
#endif
1802

L
Linus Torvalds 已提交
1803 1804 1805 1806
/*
 *	TCP over IPv4 via INET6 API
 */

1807
static const struct inet_connection_sock_af_ops ipv6_mapped = {
1808 1809 1810 1811 1812
	.queue_xmit	   = ip_queue_xmit,
	.send_check	   = tcp_v4_send_check,
	.rebuild_header	   = inet_sk_rebuild_header,
	.conn_request	   = tcp_v6_conn_request,
	.syn_recv_sock	   = tcp_v6_syn_recv_sock,
1813
	.get_peer	   = tcp_v4_get_peer,
1814 1815 1816 1817 1818
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ipv6_setsockopt,
	.getsockopt	   = ipv6_getsockopt,
	.addr2sockaddr	   = inet6_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in6),
1819
	.bind_conflict	   = inet6_csk_bind_conflict,
1820
#ifdef CONFIG_COMPAT
1821 1822
	.compat_setsockopt = compat_ipv6_setsockopt,
	.compat_getsockopt = compat_ipv6_getsockopt,
1823
#endif
L
Linus Torvalds 已提交
1824 1825
};

1826
#ifdef CONFIG_TCP_MD5SIG
1827
static const struct tcp_sock_af_ops tcp_sock_ipv6_mapped_specific = {
1828
	.md5_lookup	=	tcp_v4_md5_lookup,
1829
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1830 1831
	.md5_parse	=	tcp_v6_parse_md5_keys,
};
1832
#endif
1833

L
Linus Torvalds 已提交
1834 1835 1836 1837 1838
/* NOTE: A lot of things set to zero explicitly by call to
 *       sk_alloc() so need not be done here.
 */
static int tcp_v6_init_sock(struct sock *sk)
{
1839
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1840

1841
	tcp_init_sock(sk);
L
Linus Torvalds 已提交
1842

1843
	icsk->icsk_af_ops = &ipv6_specific;
L
Linus Torvalds 已提交
1844

1845
#ifdef CONFIG_TCP_MD5SIG
1846
	tcp_sk(sk)->af_specific = &tcp_sock_ipv6_specific;
1847 1848
#endif

L
Linus Torvalds 已提交
1849 1850 1851
	return 0;
}

1852
static void tcp_v6_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1853 1854
{
	tcp_v4_destroy_sock(sk);
1855
	inet6_destroy_sock(sk);
L
Linus Torvalds 已提交
1856 1857
}

1858
#ifdef CONFIG_PROC_FS
L
Linus Torvalds 已提交
1859
/* Proc filesystem TCPv6 sock list dumping. */
1860
static void get_openreq6(struct seq_file *seq,
1861
			 const struct sock *sk, struct request_sock *req, int i, int uid)
L
Linus Torvalds 已提交
1862 1863
{
	int ttd = req->expires - jiffies;
1864 1865
	const struct in6_addr *src = &inet6_rsk(req)->loc_addr;
	const struct in6_addr *dest = &inet6_rsk(req)->rmt_addr;
L
Linus Torvalds 已提交
1866 1867 1868 1869 1870 1871

	if (ttd < 0)
		ttd = 0;

	seq_printf(seq,
		   "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X "
D
Dan Rosenberg 已提交
1872
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK\n",
L
Linus Torvalds 已提交
1873 1874 1875
		   i,
		   src->s6_addr32[0], src->s6_addr32[1],
		   src->s6_addr32[2], src->s6_addr32[3],
1876
		   ntohs(inet_rsk(req)->loc_port),
L
Linus Torvalds 已提交
1877 1878
		   dest->s6_addr32[0], dest->s6_addr32[1],
		   dest->s6_addr32[2], dest->s6_addr32[3],
1879
		   ntohs(inet_rsk(req)->rmt_port),
L
Linus Torvalds 已提交
1880 1881
		   TCP_SYN_RECV,
		   0,0, /* could print option size, but that is af dependent. */
1882 1883
		   1,   /* timers active (only the expire timer) */
		   jiffies_to_clock_t(ttd),
L
Linus Torvalds 已提交
1884 1885
		   req->retrans,
		   uid,
1886
		   0,  /* non standard timer */
L
Linus Torvalds 已提交
1887 1888 1889 1890 1891 1892
		   0, /* open_requests have no inode */
		   0, req);
}

static void get_tcp6_sock(struct seq_file *seq, struct sock *sp, int i)
{
1893
	const struct in6_addr *dest, *src;
L
Linus Torvalds 已提交
1894 1895 1896
	__u16 destp, srcp;
	int timer_active;
	unsigned long timer_expires;
1897 1898
	const struct inet_sock *inet = inet_sk(sp);
	const struct tcp_sock *tp = tcp_sk(sp);
1899
	const struct inet_connection_sock *icsk = inet_csk(sp);
1900
	const struct ipv6_pinfo *np = inet6_sk(sp);
L
Linus Torvalds 已提交
1901 1902 1903

	dest  = &np->daddr;
	src   = &np->rcv_saddr;
E
Eric Dumazet 已提交
1904 1905
	destp = ntohs(inet->inet_dport);
	srcp  = ntohs(inet->inet_sport);
1906 1907

	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
1908
		timer_active	= 1;
1909 1910
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
1911
		timer_active	= 4;
1912
		timer_expires	= icsk->icsk_timeout;
L
Linus Torvalds 已提交
1913 1914 1915 1916 1917 1918 1919 1920 1921 1922
	} else if (timer_pending(&sp->sk_timer)) {
		timer_active	= 2;
		timer_expires	= sp->sk_timer.expires;
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

	seq_printf(seq,
		   "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X "
D
Dan Rosenberg 已提交
1923
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %lu %lu %u %u %d\n",
L
Linus Torvalds 已提交
1924 1925 1926 1927 1928
		   i,
		   src->s6_addr32[0], src->s6_addr32[1],
		   src->s6_addr32[2], src->s6_addr32[3], srcp,
		   dest->s6_addr32[0], dest->s6_addr32[1],
		   dest->s6_addr32[2], dest->s6_addr32[3], destp,
1929
		   sp->sk_state,
1930 1931
		   tp->write_seq-tp->snd_una,
		   (sp->sk_state == TCP_LISTEN) ? sp->sk_ack_backlog : (tp->rcv_nxt - tp->copied_seq),
L
Linus Torvalds 已提交
1932 1933
		   timer_active,
		   jiffies_to_clock_t(timer_expires - jiffies),
1934
		   icsk->icsk_retransmits,
L
Linus Torvalds 已提交
1935
		   sock_i_uid(sp),
1936
		   icsk->icsk_probes_out,
L
Linus Torvalds 已提交
1937 1938
		   sock_i_ino(sp),
		   atomic_read(&sp->sk_refcnt), sp,
1939 1940
		   jiffies_to_clock_t(icsk->icsk_rto),
		   jiffies_to_clock_t(icsk->icsk_ack.ato),
1941
		   (icsk->icsk_ack.quick << 1 ) | icsk->icsk_ack.pingpong,
I
Ilpo Järvinen 已提交
1942 1943
		   tp->snd_cwnd,
		   tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh
L
Linus Torvalds 已提交
1944 1945 1946
		   );
}

1947
static void get_timewait6_sock(struct seq_file *seq,
1948
			       struct inet_timewait_sock *tw, int i)
L
Linus Torvalds 已提交
1949
{
1950
	const struct in6_addr *dest, *src;
L
Linus Torvalds 已提交
1951
	__u16 destp, srcp;
1952
	const struct inet6_timewait_sock *tw6 = inet6_twsk((struct sock *)tw);
L
Linus Torvalds 已提交
1953 1954 1955 1956 1957
	int ttd = tw->tw_ttd - jiffies;

	if (ttd < 0)
		ttd = 0;

1958 1959
	dest = &tw6->tw_v6_daddr;
	src  = &tw6->tw_v6_rcv_saddr;
L
Linus Torvalds 已提交
1960 1961 1962 1963 1964
	destp = ntohs(tw->tw_dport);
	srcp  = ntohs(tw->tw_sport);

	seq_printf(seq,
		   "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X "
D
Dan Rosenberg 已提交
1965
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK\n",
L
Linus Torvalds 已提交
1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
		   i,
		   src->s6_addr32[0], src->s6_addr32[1],
		   src->s6_addr32[2], src->s6_addr32[3], srcp,
		   dest->s6_addr32[0], dest->s6_addr32[1],
		   dest->s6_addr32[2], dest->s6_addr32[3], destp,
		   tw->tw_substate, 0, 0,
		   3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
		   atomic_read(&tw->tw_refcnt), tw);
}

static int tcp6_seq_show(struct seq_file *seq, void *v)
{
	struct tcp_iter_state *st;

	if (v == SEQ_START_TOKEN) {
		seq_puts(seq,
			 "  sl  "
			 "local_address                         "
			 "remote_address                        "
			 "st tx_queue rx_queue tr tm->when retrnsmt"
			 "   uid  timeout inode\n");
		goto out;
	}
	st = seq->private;

	switch (st->state) {
	case TCP_SEQ_STATE_LISTENING:
	case TCP_SEQ_STATE_ESTABLISHED:
		get_tcp6_sock(seq, v, st->num);
		break;
	case TCP_SEQ_STATE_OPENREQ:
		get_openreq6(seq, st->syn_wait_sk, v, st->num, st->uid);
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
		get_timewait6_sock(seq, v, st->num);
		break;
	}
out:
	return 0;
}

2007 2008 2009 2010 2011 2012 2013 2014
static const struct file_operations tcp6_afinfo_seq_fops = {
	.owner   = THIS_MODULE,
	.open    = tcp_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
	.release = seq_release_net
};

L
Linus Torvalds 已提交
2015 2016 2017
static struct tcp_seq_afinfo tcp6_seq_afinfo = {
	.name		= "tcp6",
	.family		= AF_INET6,
2018
	.seq_fops	= &tcp6_afinfo_seq_fops,
2019 2020 2021
	.seq_ops	= {
		.show		= tcp6_seq_show,
	},
L
Linus Torvalds 已提交
2022 2023
};

2024
int __net_init tcp6_proc_init(struct net *net)
L
Linus Torvalds 已提交
2025
{
2026
	return tcp_proc_register(net, &tcp6_seq_afinfo);
L
Linus Torvalds 已提交
2027 2028
}

2029
void tcp6_proc_exit(struct net *net)
L
Linus Torvalds 已提交
2030
{
2031
	tcp_proc_unregister(net, &tcp6_seq_afinfo);
L
Linus Torvalds 已提交
2032 2033 2034 2035 2036 2037 2038 2039 2040
}
#endif

struct proto tcpv6_prot = {
	.name			= "TCPv6",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v6_connect,
	.disconnect		= tcp_disconnect,
2041
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2042 2043 2044 2045 2046 2047 2048
	.ioctl			= tcp_ioctl,
	.init			= tcp_v6_init_sock,
	.destroy		= tcp_v6_destroy_sock,
	.shutdown		= tcp_shutdown,
	.setsockopt		= tcp_setsockopt,
	.getsockopt		= tcp_getsockopt,
	.recvmsg		= tcp_recvmsg,
2049 2050
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2051 2052
	.backlog_rcv		= tcp_v6_do_rcv,
	.hash			= tcp_v6_hash,
2053 2054
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2055 2056 2057 2058
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
2059
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2060 2061 2062 2063
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp6_sock),
2064
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2065
	.twsk_prot		= &tcp6_timewait_sock_ops,
2066
	.rsk_prot		= &tcp6_request_sock_ops,
2067
	.h.hashinfo		= &tcp_hashinfo,
2068
	.no_autobind		= true,
2069 2070 2071 2072
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
G
Glauber Costa 已提交
2073 2074 2075
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
	.proto_cgroup		= tcp_proto_cgroup,
#endif
L
Linus Torvalds 已提交
2076 2077
};

2078
static const struct inet6_protocol tcpv6_protocol = {
L
Linus Torvalds 已提交
2079 2080
	.handler	=	tcp_v6_rcv,
	.err_handler	=	tcp_v6_err,
2081
	.gso_send_check	=	tcp_v6_gso_send_check,
H
Herbert Xu 已提交
2082
	.gso_segment	=	tcp_tso_segment,
H
Herbert Xu 已提交
2083 2084
	.gro_receive	=	tcp6_gro_receive,
	.gro_complete	=	tcp6_gro_complete,
L
Linus Torvalds 已提交
2085 2086 2087 2088 2089 2090 2091 2092 2093
	.flags		=	INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
};

static struct inet_protosw tcpv6_protosw = {
	.type		=	SOCK_STREAM,
	.protocol	=	IPPROTO_TCP,
	.prot		=	&tcpv6_prot,
	.ops		=	&inet6_stream_ops,
	.no_check	=	0,
2094 2095
	.flags		=	INET_PROTOSW_PERMANENT |
				INET_PROTOSW_ICSK,
L
Linus Torvalds 已提交
2096 2097
};

2098
static int __net_init tcpv6_net_init(struct net *net)
2099
{
2100 2101
	return inet_ctl_sock_create(&net->ipv6.tcp_sk, PF_INET6,
				    SOCK_RAW, IPPROTO_TCP, net);
2102 2103
}

2104
static void __net_exit tcpv6_net_exit(struct net *net)
2105
{
2106
	inet_ctl_sock_destroy(net->ipv6.tcp_sk);
E
Eric W. Biederman 已提交
2107 2108
}

2109
static void __net_exit tcpv6_net_exit_batch(struct list_head *net_exit_list)
E
Eric W. Biederman 已提交
2110 2111
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET6);
2112 2113 2114
}

static struct pernet_operations tcpv6_net_ops = {
E
Eric W. Biederman 已提交
2115 2116 2117
	.init	    = tcpv6_net_init,
	.exit	    = tcpv6_net_exit,
	.exit_batch = tcpv6_net_exit_batch,
2118 2119
};

2120
int __init tcpv6_init(void)
L
Linus Torvalds 已提交
2121
{
2122 2123 2124 2125 2126 2127
	int ret;

	ret = inet6_add_protocol(&tcpv6_protocol, IPPROTO_TCP);
	if (ret)
		goto out;

L
Linus Torvalds 已提交
2128
	/* register inet6 protocol */
2129 2130 2131 2132
	ret = inet6_register_protosw(&tcpv6_protosw);
	if (ret)
		goto out_tcpv6_protocol;

2133
	ret = register_pernet_subsys(&tcpv6_net_ops);
2134 2135 2136 2137
	if (ret)
		goto out_tcpv6_protosw;
out:
	return ret;
2138

2139 2140 2141 2142 2143 2144 2145
out_tcpv6_protocol:
	inet6_del_protocol(&tcpv6_protocol, IPPROTO_TCP);
out_tcpv6_protosw:
	inet6_unregister_protosw(&tcpv6_protosw);
	goto out;
}

2146
void tcpv6_exit(void)
2147
{
2148
	unregister_pernet_subsys(&tcpv6_net_ops);
2149 2150
	inet6_unregister_protosw(&tcpv6_protosw);
	inet6_del_protocol(&tcpv6_protocol, IPPROTO_TCP);
L
Linus Torvalds 已提交
2151
}