tcp_ipv6.c 54.1 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 99 100 101 102 103 104 105 106 107 108
static void inet6_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
{
	struct dst_entry *dst = skb_dst(skb);
	const struct rt6_info *rt = (const struct rt6_info *)dst;

	dst_hold(dst);
	sk->sk_rx_dst = dst;
	inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
	if (rt->rt6i_node)
		inet6_sk(sk)->rx_dst_cookie = rt->rt6i_node->fn_sernum;
}

L
Linus Torvalds 已提交
109 110 111
static void tcp_v6_hash(struct sock *sk)
{
	if (sk->sk_state != TCP_CLOSE) {
112
		if (inet_csk(sk)->icsk_af_ops == &ipv6_mapped) {
L
Linus Torvalds 已提交
113 114 115 116
			tcp_prot.hash(sk);
			return;
		}
		local_bh_disable();
117
		__inet6_hash(sk, NULL);
L
Linus Torvalds 已提交
118 119 120 121
		local_bh_enable();
	}
}

H
Herbert Xu 已提交
122
static __inline__ __sum16 tcp_v6_check(int len,
123 124
				   const struct in6_addr *saddr,
				   const struct in6_addr *daddr,
125
				   __wsum base)
L
Linus Torvalds 已提交
126 127 128 129
{
	return csum_ipv6_magic(saddr, daddr, len, IPPROTO_TCP, base);
}

130
static __u32 tcp_v6_init_sequence(const struct sk_buff *skb)
L
Linus Torvalds 已提交
131
{
132 133
	return secure_tcpv6_sequence_number(ipv6_hdr(skb)->daddr.s6_addr32,
					    ipv6_hdr(skb)->saddr.s6_addr32,
134 135
					    tcp_hdr(skb)->dest,
					    tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
136 137
}

138
static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
L
Linus Torvalds 已提交
139 140 141
			  int addr_len)
{
	struct sockaddr_in6 *usin = (struct sockaddr_in6 *) uaddr;
142
	struct inet_sock *inet = inet_sk(sk);
143
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
144 145
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
146
	struct in6_addr *saddr = NULL, *final_p, final;
147
	struct rt6_info *rt;
148
	struct flowi6 fl6;
L
Linus Torvalds 已提交
149 150 151 152
	struct dst_entry *dst;
	int addr_type;
	int err;

153
	if (addr_len < SIN6_LEN_RFC2133)
L
Linus Torvalds 已提交
154 155
		return -EINVAL;

156
	if (usin->sin6_family != AF_INET6)
E
Eric Dumazet 已提交
157
		return -EAFNOSUPPORT;
L
Linus Torvalds 已提交
158

159
	memset(&fl6, 0, sizeof(fl6));
L
Linus Torvalds 已提交
160 161

	if (np->sndflow) {
162 163 164
		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
165
			struct ip6_flowlabel *flowlabel;
166
			flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
L
Linus Torvalds 已提交
167 168
			if (flowlabel == NULL)
				return -EINVAL;
A
Alexey Dobriyan 已提交
169
			usin->sin6_addr = flowlabel->dst;
L
Linus Torvalds 已提交
170 171 172 173 174
			fl6_sock_release(flowlabel);
		}
	}

	/*
175 176 177 178 179
	 *	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
180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210

	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
211
	np->daddr = usin->sin6_addr;
212
	np->flow_label = fl6.flowlabel;
L
Linus Torvalds 已提交
213 214 215 216 217 218

	/*
	 *	TCP over IPv4
	 */

	if (addr_type == IPV6_ADDR_MAPPED) {
219
		u32 exthdrlen = icsk->icsk_ext_hdr_len;
L
Linus Torvalds 已提交
220 221 222 223 224 225 226 227 228 229 230
		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];

231
		icsk->icsk_af_ops = &ipv6_mapped;
L
Linus Torvalds 已提交
232
		sk->sk_backlog_rcv = tcp_v4_do_rcv;
233 234 235
#ifdef CONFIG_TCP_MD5SIG
		tp->af_specific = &tcp_sock_ipv6_mapped_specific;
#endif
L
Linus Torvalds 已提交
236 237 238 239

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

		if (err) {
240 241
			icsk->icsk_ext_hdr_len = exthdrlen;
			icsk->icsk_af_ops = &ipv6_specific;
L
Linus Torvalds 已提交
242
			sk->sk_backlog_rcv = tcp_v6_do_rcv;
243 244 245
#ifdef CONFIG_TCP_MD5SIG
			tp->af_specific = &tcp_sock_ipv6_specific;
#endif
L
Linus Torvalds 已提交
246 247
			goto failure;
		} else {
E
Eric Dumazet 已提交
248 249 250
			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
251 252 253 254 255 256 257 258
		}

		return err;
	}

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

259
	fl6.flowi6_proto = IPPROTO_TCP;
A
Alexey Dobriyan 已提交
260 261
	fl6.daddr = np->daddr;
	fl6.saddr = saddr ? *saddr : np->saddr;
262 263
	fl6.flowi6_oif = sk->sk_bound_dev_if;
	fl6.flowi6_mark = sk->sk_mark;
264 265
	fl6.fl6_dport = usin->sin6_port;
	fl6.fl6_sport = inet->inet_sport;
L
Linus Torvalds 已提交
266

267
	final_p = fl6_update_dst(&fl6, np->opt, &final);
L
Linus Torvalds 已提交
268

269
	security_sk_classify_flow(sk, flowi6_to_flowi(&fl6));
V
Venkat Yekkirala 已提交
270

271
	dst = ip6_dst_lookup_flow(sk, &fl6, final_p, true);
272 273
	if (IS_ERR(dst)) {
		err = PTR_ERR(dst);
L
Linus Torvalds 已提交
274
		goto failure;
275
	}
L
Linus Torvalds 已提交
276 277

	if (saddr == NULL) {
278
		saddr = &fl6.saddr;
A
Alexey Dobriyan 已提交
279
		np->rcv_saddr = *saddr;
L
Linus Torvalds 已提交
280 281 282
	}

	/* set the source address */
A
Alexey Dobriyan 已提交
283
	np->saddr = *saddr;
E
Eric Dumazet 已提交
284
	inet->inet_rcv_saddr = LOOPBACK4_IPV6;
L
Linus Torvalds 已提交
285

H
Herbert Xu 已提交
286
	sk->sk_gso_type = SKB_GSO_TCPV6;
287
	__ip6_dst_store(sk, dst, NULL, NULL);
L
Linus Torvalds 已提交
288

289 290 291
	rt = (struct rt6_info *) dst;
	if (tcp_death_row.sysctl_tw_recycle &&
	    !tp->rx_opt.ts_recent_stamp &&
292 293
	    ipv6_addr_equal(&rt->rt6i_dst.addr, &np->daddr))
		tcp_fetch_timewait_stamp(sk, dst);
294

295
	icsk->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
296
	if (np->opt)
297 298
		icsk->icsk_ext_hdr_len = (np->opt->opt_flen +
					  np->opt->opt_nflen);
L
Linus Torvalds 已提交
299 300 301

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

E
Eric Dumazet 已提交
302
	inet->inet_dport = usin->sin6_port;
L
Linus Torvalds 已提交
303 304

	tcp_set_state(sk, TCP_SYN_SENT);
305
	err = inet6_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
306 307 308 309 310 311
	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
312 313
							     inet->inet_sport,
							     inet->inet_dport);
L
Linus Torvalds 已提交
314 315 316 317 318 319 320 321 322 323 324

	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
325
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
326 327 328 329
	sk->sk_route_caps = 0;
	return err;
}

330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346
static void tcp_v6_mtu_reduced(struct sock *sk)
{
	struct dst_entry *dst;

	if ((1 << sk->sk_state) & (TCPF_LISTEN | TCPF_CLOSE))
		return;

	dst = inet6_csk_update_pmtu(sk, tcp_sk(sk)->mtu_info);
	if (!dst)
		return;

	if (inet_csk(sk)->icsk_pmtu_cookie > dst_mtu(dst)) {
		tcp_sync_mss(sk, dst_mtu(dst));
		tcp_simple_retransmit(sk);
	}
}

L
Linus Torvalds 已提交
347
static void tcp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
348
		u8 type, u8 code, int offset, __be32 info)
L
Linus Torvalds 已提交
349
{
350
	const struct ipv6hdr *hdr = (const struct ipv6hdr*)skb->data;
351
	const struct tcphdr *th = (struct tcphdr *)(skb->data+offset);
L
Linus Torvalds 已提交
352 353 354
	struct ipv6_pinfo *np;
	struct sock *sk;
	int err;
355
	struct tcp_sock *tp;
L
Linus Torvalds 已提交
356
	__u32 seq;
357
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
358

359
	sk = inet6_lookup(net, &tcp_hashinfo, &hdr->daddr,
360
			th->dest, &hdr->saddr, th->source, skb->dev->ifindex);
L
Linus Torvalds 已提交
361 362

	if (sk == NULL) {
363 364
		ICMP6_INC_STATS_BH(net, __in6_dev_get(skb->dev),
				   ICMP6_MIB_INERRORS);
L
Linus Torvalds 已提交
365 366 367 368
		return;
	}

	if (sk->sk_state == TCP_TIME_WAIT) {
369
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
370 371 372 373
		return;
	}

	bh_lock_sock(sk);
374
	if (sock_owned_by_user(sk) && type != ICMPV6_PKT_TOOBIG)
375
		NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
L
Linus Torvalds 已提交
376 377 378 379

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

380 381 382 383 384
	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
385
	tp = tcp_sk(sk);
386
	seq = ntohl(th->seq);
L
Linus Torvalds 已提交
387 388
	if (sk->sk_state != TCP_LISTEN &&
	    !between(seq, tp->snd_una, tp->snd_nxt)) {
389
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
390 391 392 393 394
		goto out;
	}

	np = inet6_sk(sk);

395 396 397
	if (type == NDISC_REDIRECT) {
		struct dst_entry *dst = __sk_dst_check(sk, np->dst_cookie);

398
		if (dst)
399
			dst->ops->redirect(dst, sk, skb);
400 401
	}

L
Linus Torvalds 已提交
402
	if (type == ICMPV6_PKT_TOOBIG) {
403 404 405
		tp->mtu_info = ntohl(info);
		if (!sock_owned_by_user(sk))
			tcp_v6_mtu_reduced(sk);
406 407 408
		else if (!test_and_set_bit(TCP_MTU_REDUCED_DEFERRED,
					   &tp->tsq_flags))
			sock_hold(sk);
L
Linus Torvalds 已提交
409 410 411 412 413
		goto out;
	}

	icmpv6_err_convert(type, code, &err);

414
	/* Might be for an request_sock */
L
Linus Torvalds 已提交
415
	switch (sk->sk_state) {
416
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
417 418 419 420
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

421 422
		req = inet6_csk_search_req(sk, &prev, th->dest, &hdr->daddr,
					   &hdr->saddr, inet6_iif(skb));
L
Linus Torvalds 已提交
423 424 425 426 427 428
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		 * an established socket here.
		 */
429
		WARN_ON(req->sk != NULL);
L
Linus Torvalds 已提交
430

431
		if (seq != tcp_rsk(req)->snt_isn) {
432
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
433 434 435
			goto out;
		}

436
		inet_csk_reqsk_queue_drop(sk, req, prev);
L
Linus Torvalds 已提交
437 438 439 440
		goto out;

	case TCP_SYN_SENT:
	case TCP_SYN_RECV:  /* Cannot happen.
441
			       It can, it SYNs are crossed. --ANK */
L
Linus Torvalds 已提交
442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463
		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);
}


464 465
static int tcp_v6_send_synack(struct sock *sk, struct dst_entry *dst,
			      struct flowi6 *fl6,
466
			      struct request_sock *req,
467 468
			      struct request_values *rvp,
			      u16 queue_mapping)
L
Linus Torvalds 已提交
469
{
470
	struct inet6_request_sock *treq = inet6_rsk(req);
L
Linus Torvalds 已提交
471 472
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct sk_buff * skb;
473
	int err = -ENOMEM;
L
Linus Torvalds 已提交
474

475 476
	/* First, grab a route. */
	if (!dst && (dst = inet6_csk_route_req(sk, fl6, req)) == NULL)
477
		goto done;
478

479
	skb = tcp_make_synack(sk, dst, req, rvp, NULL);
480

L
Linus Torvalds 已提交
481
	if (skb) {
482
		__tcp_v6_send_check(skb, &treq->loc_addr, &treq->rmt_addr);
L
Linus Torvalds 已提交
483

484
		fl6->daddr = treq->rmt_addr;
485
		skb_set_queue_mapping(skb, queue_mapping);
486
		err = ip6_xmit(sk, skb, fl6, np->opt, np->tclass);
487
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
488 489 490 491 492 493
	}

done:
	return err;
}

494 495 496
static int tcp_v6_rtx_synack(struct sock *sk, struct request_sock *req,
			     struct request_values *rvp)
{
497
	struct flowi6 fl6;
498
	int res;
499

500 501 502 503
	res = tcp_v6_send_synack(sk, NULL, &fl6, req, rvp, 0);
	if (!res)
		TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
	return res;
504 505
}

506
static void tcp_v6_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
507
{
508
	kfree_skb(inet6_rsk(req)->pktopts);
L
Linus Torvalds 已提交
509 510
}

511 512
#ifdef CONFIG_TCP_MD5SIG
static struct tcp_md5sig_key *tcp_v6_md5_do_lookup(struct sock *sk,
513
						   const struct in6_addr *addr)
514
{
E
Eric Dumazet 已提交
515
	return tcp_md5_do_lookup(sk, (union tcp_md5_addr *)addr, AF_INET6);
516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545
}

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
546
		if (ipv6_addr_v4mapped(&sin6->sin6_addr))
E
Eric Dumazet 已提交
547 548 549 550
			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);
551 552 553 554 555
	}

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

E
Eric Dumazet 已提交
556 557 558
	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);
559

E
Eric Dumazet 已提交
560 561
	return tcp_md5_do_add(sk, (union tcp_md5_addr *)&sin6->sin6_addr,
			      AF_INET6, cmd.tcpm_key, cmd.tcpm_keylen, GFP_KERNEL);
562 563
}

564
static int tcp_v6_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
565 566
					const struct in6_addr *daddr,
					const struct in6_addr *saddr, int nbytes)
567 568
{
	struct tcp6_pseudohdr *bp;
569
	struct scatterlist sg;
570

571 572
	bp = &hp->md5_blk.ip6;
	/* 1. TCP pseudo-header (RFC2460) */
A
Alexey Dobriyan 已提交
573 574
	bp->saddr = *saddr;
	bp->daddr = *daddr;
575
	bp->protocol = cpu_to_be32(IPPROTO_TCP);
A
Adam Langley 已提交
576
	bp->len = cpu_to_be32(nbytes);
577

578 579 580
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}
581

582
static int tcp_v6_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
583
			       const struct in6_addr *daddr, struct in6_addr *saddr,
E
Eric Dumazet 已提交
584
			       const struct tcphdr *th)
585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602
{
	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))
603 604 605 606
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
607

608 609 610 611
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
612
	return 1;
613 614
}

615
static int tcp_v6_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
616 617 618
			       const struct sock *sk,
			       const struct request_sock *req,
			       const struct sk_buff *skb)
619
{
620
	const struct in6_addr *saddr, *daddr;
621 622
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
E
Eric Dumazet 已提交
623
	const struct tcphdr *th = tcp_hdr(skb);
624 625 626 627

	if (sk) {
		saddr = &inet6_sk(sk)->saddr;
		daddr = &inet6_sk(sk)->daddr;
628
	} else if (req) {
629 630
		saddr = &inet6_rsk(req)->loc_addr;
		daddr = &inet6_rsk(req)->rmt_addr;
631
	} else {
632
		const struct ipv6hdr *ip6h = ipv6_hdr(skb);
633 634
		saddr = &ip6h->saddr;
		daddr = &ip6h->daddr;
635
	}
636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663

	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;
664 665
}

E
Eric Dumazet 已提交
666
static int tcp_v6_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
667
{
668
	const __u8 *hash_location = NULL;
669
	struct tcp_md5sig_key *hash_expected;
670
	const struct ipv6hdr *ip6h = ipv6_hdr(skb);
E
Eric Dumazet 已提交
671
	const struct tcphdr *th = tcp_hdr(skb);
672 673 674 675
	int genhash;
	u8 newhash[16];

	hash_expected = tcp_v6_md5_do_lookup(sk, &ip6h->saddr);
676
	hash_location = tcp_parse_md5sig_option(th);
677

678 679 680 681 682 683
	/* 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);
684 685 686
		return 1;
	}

687 688
	if (!hash_expected && hash_location) {
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
689 690 691 692
		return 1;
	}

	/* check the signature */
693 694 695 696
	genhash = tcp_v6_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);

697
	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
698 699 700 701
		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));
702 703 704 705 706 707
		return 1;
	}
	return 0;
}
#endif

708
struct request_sock_ops tcp6_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
709
	.family		=	AF_INET6,
710
	.obj_size	=	sizeof(struct tcp6_request_sock),
711
	.rtx_syn_ack	=	tcp_v6_rtx_synack,
712 713
	.send_ack	=	tcp_v6_reqsk_send_ack,
	.destructor	=	tcp_v6_reqsk_destructor,
714 715
	.send_reset	=	tcp_v6_send_reset,
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
716 717
};

718
#ifdef CONFIG_TCP_MD5SIG
719
static const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = {
720
	.md5_lookup	=	tcp_v6_reqsk_md5_lookup,
721
	.calc_md5_hash	=	tcp_v6_md5_hash_skb,
722
};
723
#endif
724

725
static void __tcp_v6_send_check(struct sk_buff *skb,
726
				const struct in6_addr *saddr, const struct in6_addr *daddr)
L
Linus Torvalds 已提交
727
{
728
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
729

730
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
731
		th->check = ~tcp_v6_check(skb->len, saddr, daddr, 0);
732
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
733
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
734
	} else {
735 736 737
		th->check = tcp_v6_check(skb->len, saddr, daddr,
					 csum_partial(th, th->doff << 2,
						      skb->csum));
L
Linus Torvalds 已提交
738 739 740
	}
}

741
static void tcp_v6_send_check(struct sock *sk, struct sk_buff *skb)
742 743 744 745 746 747
{
	struct ipv6_pinfo *np = inet6_sk(sk);

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

748 749
static int tcp_v6_gso_send_check(struct sk_buff *skb)
{
750
	const struct ipv6hdr *ipv6h;
751 752 753 754 755
	struct tcphdr *th;

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

756
	ipv6h = ipv6_hdr(skb);
757
	th = tcp_hdr(skb);
758 759

	th->check = 0;
760
	skb->ip_summed = CHECKSUM_PARTIAL;
761
	__tcp_v6_send_check(skb, &ipv6h->saddr, &ipv6h->daddr);
762 763
	return 0;
}
L
Linus Torvalds 已提交
764

H
Herbert Xu 已提交
765 766
static struct sk_buff **tcp6_gro_receive(struct sk_buff **head,
					 struct sk_buff *skb)
H
Herbert Xu 已提交
767
{
768
	const struct ipv6hdr *iph = skb_gro_network_header(skb);
E
Eric Dumazet 已提交
769 770
	__wsum wsum;
	__sum16 sum;
H
Herbert Xu 已提交
771 772 773

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
774
		if (!tcp_v6_check(skb_gro_len(skb), &iph->saddr, &iph->daddr,
H
Herbert Xu 已提交
775 776 777 778
				  skb->csum)) {
			skb->ip_summed = CHECKSUM_UNNECESSARY;
			break;
		}
E
Eric Dumazet 已提交
779
flush:
H
Herbert Xu 已提交
780 781
		NAPI_GRO_CB(skb)->flush = 1;
		return NULL;
E
Eric Dumazet 已提交
782 783 784 785 786 787 788 789 790 791 792 793 794 795

	case CHECKSUM_NONE:
		wsum = ~csum_unfold(csum_ipv6_magic(&iph->saddr, &iph->daddr,
						    skb_gro_len(skb),
						    IPPROTO_TCP, 0));
		sum = csum_fold(skb_checksum(skb,
					     skb_gro_offset(skb),
					     skb_gro_len(skb),
					     wsum));
		if (sum)
			goto flush;

		skb->ip_summed = CHECKSUM_UNNECESSARY;
		break;
H
Herbert Xu 已提交
796 797 798 799 800
	}

	return tcp_gro_receive(head, skb);
}

H
Herbert Xu 已提交
801
static int tcp6_gro_complete(struct sk_buff *skb)
H
Herbert Xu 已提交
802
{
803
	const struct ipv6hdr *iph = ipv6_hdr(skb);
H
Herbert Xu 已提交
804 805 806 807 808 809 810 811 812
	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);
}

813
static void tcp_v6_send_response(struct sk_buff *skb, u32 seq, u32 ack, u32 win,
814
				 u32 ts, struct tcp_md5sig_key *key, int rst, u8 tclass)
L
Linus Torvalds 已提交
815
{
816 817
	const struct tcphdr *th = tcp_hdr(skb);
	struct tcphdr *t1;
L
Linus Torvalds 已提交
818
	struct sk_buff *buff;
819
	struct flowi6 fl6;
E
Eric Dumazet 已提交
820
	struct net *net = dev_net(skb_dst(skb)->dev);
821
	struct sock *ctl_sk = net->ipv6.tcp_sk;
822
	unsigned int tot_len = sizeof(struct tcphdr);
E
Eric Dumazet 已提交
823
	struct dst_entry *dst;
824
	__be32 *topt;
L
Linus Torvalds 已提交
825

826 827
	if (ts)
		tot_len += TCPOLEN_TSTAMP_ALIGNED;
828 829 830 831 832 833
#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
834
			 GFP_ATOMIC);
835 836
	if (buff == NULL)
		return;
L
Linus Torvalds 已提交
837

838
	skb_reserve(buff, MAX_HEADER + sizeof(struct ipv6hdr) + tot_len);
L
Linus Torvalds 已提交
839

840
	t1 = (struct tcphdr *) skb_push(buff, tot_len);
841
	skb_reset_transport_header(buff);
L
Linus Torvalds 已提交
842 843 844 845 846

	/* Swap the send and the receive. */
	memset(t1, 0, sizeof(*t1));
	t1->dest = th->source;
	t1->source = th->dest;
847
	t1->doff = tot_len / 4;
848 849 850 851 852
	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
853

854 855
	topt = (__be32 *)(t1 + 1);

856 857 858 859 860 861 862
	if (ts) {
		*topt++ = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				(TCPOPT_TIMESTAMP << 8) | TCPOLEN_TIMESTAMP);
		*topt++ = htonl(tcp_time_stamp);
		*topt++ = htonl(ts);
	}

863 864
#ifdef CONFIG_TCP_MD5SIG
	if (key) {
865 866 867
		*topt++ = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				(TCPOPT_MD5SIG << 8) | TCPOLEN_MD5SIG);
		tcp_v6_md5_hash_hdr((__u8 *)topt, key,
868 869
				    &ipv6_hdr(skb)->saddr,
				    &ipv6_hdr(skb)->daddr, t1);
870 871 872
	}
#endif

873
	memset(&fl6, 0, sizeof(fl6));
A
Alexey Dobriyan 已提交
874 875
	fl6.daddr = ipv6_hdr(skb)->saddr;
	fl6.saddr = ipv6_hdr(skb)->daddr;
L
Linus Torvalds 已提交
876

877 878 879
	buff->ip_summed = CHECKSUM_PARTIAL;
	buff->csum = 0;

880
	__tcp_v6_send_check(buff, &fl6.saddr, &fl6.daddr);
L
Linus Torvalds 已提交
881

882
	fl6.flowi6_proto = IPPROTO_TCP;
A
Alexey Kuznetsov 已提交
883 884
	if (ipv6_addr_type(&fl6.daddr) & IPV6_ADDR_LINKLOCAL)
		fl6.flowi6_oif = inet6_iif(skb);
885 886
	fl6.fl6_dport = t1->dest;
	fl6.fl6_sport = t1->source;
887
	security_skb_classify_flow(skb, flowi6_to_flowi(&fl6));
L
Linus Torvalds 已提交
888

889 890 891 892
	/* Pass a socket to ip6_dst_lookup either it is for RST
	 * Underlying function will use this to retrieve the network
	 * namespace
	 */
893
	dst = ip6_dst_lookup_flow(ctl_sk, &fl6, NULL, false);
894 895
	if (!IS_ERR(dst)) {
		skb_dst_set(buff, dst);
896
		ip6_xmit(ctl_sk, buff, &fl6, NULL, tclass);
897 898 899 900
		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
901 902 903 904 905
	}

	kfree_skb(buff);
}

906
static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
907
{
908
	const struct tcphdr *th = tcp_hdr(skb);
909
	u32 seq = 0, ack_seq = 0;
910
	struct tcp_md5sig_key *key = NULL;
911 912 913 914 915 916 917
#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
918

919
	if (th->rst)
L
Linus Torvalds 已提交
920 921
		return;

922 923
	if (!ipv6_unicast_destination(skb))
		return;
L
Linus Torvalds 已提交
924

925
#ifdef CONFIG_TCP_MD5SIG
926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951
	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;
	}
952 953
#endif

954 955 956 957 958
	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
959

960
	tcp_v6_send_response(skb, seq, ack_seq, 0, 0, key, 1, 0);
961 962 963 964 965 966 967 968

#ifdef CONFIG_TCP_MD5SIG
release_sk1:
	if (sk1) {
		rcu_read_unlock();
		sock_put(sk1);
	}
#endif
969
}
L
Linus Torvalds 已提交
970

971
static void tcp_v6_send_ack(struct sk_buff *skb, u32 seq, u32 ack, u32 win, u32 ts,
972
			    struct tcp_md5sig_key *key, u8 tclass)
973
{
974
	tcp_v6_send_response(skb, seq, ack, win, ts, key, 0, tclass);
L
Linus Torvalds 已提交
975 976 977 978
}

static void tcp_v6_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
979
	struct inet_timewait_sock *tw = inet_twsk(sk);
980
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
981

982
	tcp_v6_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
983
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
984 985
			tcptw->tw_ts_recent, tcp_twsk_md5_key(tcptw),
			tw->tw_tclass);
L
Linus Torvalds 已提交
986

987
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
988 989
}

990 991
static void tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
				  struct request_sock *req)
L
Linus Torvalds 已提交
992
{
993
	tcp_v6_send_ack(skb, tcp_rsk(req)->snt_isn + 1, tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd, req->ts_recent,
994
			tcp_v6_md5_do_lookup(sk, &ipv6_hdr(skb)->daddr), 0);
L
Linus Torvalds 已提交
995 996 997 998 999
}


static struct sock *tcp_v6_hnd_req(struct sock *sk,struct sk_buff *skb)
{
1000
	struct request_sock *req, **prev;
1001
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1002 1003 1004
	struct sock *nsk;

	/* Find possible connection requests. */
1005
	req = inet6_csk_search_req(sk, &prev, th->source,
1006 1007
				   &ipv6_hdr(skb)->saddr,
				   &ipv6_hdr(skb)->daddr, inet6_iif(skb));
L
Linus Torvalds 已提交
1008
	if (req)
1009
		return tcp_check_req(sk, skb, req, prev, false);
L
Linus Torvalds 已提交
1010

1011
	nsk = __inet6_lookup_established(sock_net(sk), &tcp_hashinfo,
1012 1013
			&ipv6_hdr(skb)->saddr, th->source,
			&ipv6_hdr(skb)->daddr, ntohs(th->dest), inet6_iif(skb));
L
Linus Torvalds 已提交
1014 1015 1016 1017 1018 1019

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1020
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1021 1022 1023
		return NULL;
	}

1024
#ifdef CONFIG_SYN_COOKIES
1025
	if (!th->syn)
1026
		sk = cookie_v6_check(sk, skb);
L
Linus Torvalds 已提交
1027 1028 1029 1030 1031 1032 1033 1034 1035
#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)
{
1036
	struct tcp_extend_values tmp_ext;
1037
	struct tcp_options_received tmp_opt;
1038
	const u8 *hash_location;
1039
	struct request_sock *req;
1040
	struct inet6_request_sock *treq;
L
Linus Torvalds 已提交
1041 1042
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
1043
	__u32 isn = TCP_SKB_CB(skb)->when;
1044
	struct dst_entry *dst = NULL;
1045
	struct flowi6 fl6;
E
Eric Dumazet 已提交
1046
	bool want_cookie = false;
L
Linus Torvalds 已提交
1047 1048 1049 1050 1051

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

	if (!ipv6_unicast_destination(skb))
1052
		goto drop;
L
Linus Torvalds 已提交
1053

1054
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
1055 1056 1057
		want_cookie = tcp_syn_flood_action(sk, skb, "TCPv6");
		if (!want_cookie)
			goto drop;
L
Linus Torvalds 已提交
1058 1059
	}

1060
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1061 1062
		goto drop;

1063
	req = inet6_reqsk_alloc(&tcp6_request_sock_ops);
L
Linus Torvalds 已提交
1064 1065 1066
	if (req == NULL)
		goto drop;

1067 1068 1069 1070
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv6_ops;
#endif

L
Linus Torvalds 已提交
1071 1072 1073
	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;
Y
Yuchung Cheng 已提交
1074
	tcp_parse_options(skb, &tmp_opt, &hash_location, 0, NULL);
1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090

	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 */
1091
		d = (__force u32 *)&ipv6_hdr(skb)->daddr.s6_addr32[0];
1092 1093 1094 1095
		*mess++ ^= *d++;
		*mess++ ^= *d++;
		*mess++ ^= *d++;
		*mess++ ^= *d++;
1096
		d = (__force u32 *)&ipv6_hdr(skb)->saddr.s6_addr32[0];
1097 1098 1099 1100 1101 1102 1103 1104 1105
		*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
1106

E
Eric Dumazet 已提交
1107
		want_cookie = false;	/* not our kind of cookie */
1108 1109 1110 1111 1112 1113 1114 1115 1116 1117
		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
1118

1119
	if (want_cookie && !tmp_opt.saw_tstamp)
1120 1121
		tcp_clear_options(&tmp_opt);

L
Linus Torvalds 已提交
1122 1123 1124
	tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;
	tcp_openreq_init(req, &tmp_opt, skb);

1125
	treq = inet6_rsk(req);
A
Alexey Dobriyan 已提交
1126 1127
	treq->rmt_addr = ipv6_hdr(skb)->saddr;
	treq->loc_addr = ipv6_hdr(skb)->daddr;
1128
	if (!want_cookie || tmp_opt.tstamp_ok)
1129
		TCP_ECN_create_request(req, skb);
1130

1131 1132 1133 1134 1135 1136 1137
	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);

1138
	if (!isn) {
1139 1140 1141 1142 1143 1144
		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;
		}
1145 1146

		if (want_cookie) {
1147 1148
			isn = cookie_v6_init_sequence(sk, skb, &req->mss);
			req->cookie_ts = tmp_opt.tstamp_ok;
1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162
			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 &&
1163 1164
		    (dst = inet6_csk_route_req(sk, &fl6, req)) != NULL) {
			if (!tcp_peer_is_proven(req, dst, true)) {
1165 1166 1167 1168 1169 1170 1171 1172
				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)) &&
1173
			 !tcp_peer_is_proven(req, dst, false)) {
1174 1175 1176 1177 1178 1179 1180 1181 1182 1183
			/* 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;
1184
		}
1185 1186

		isn = tcp_v6_init_sequence(skb);
1187
	}
1188
have_isn:
1189
	tcp_rsk(req)->snt_isn = isn;
L
Linus Torvalds 已提交
1190

1191 1192
	if (security_inet_conn_request(sk, skb, req))
		goto drop_and_release;
1193

1194
	if (tcp_v6_send_synack(sk, dst, &fl6, req,
1195 1196
			       (struct request_values *)&tmp_ext,
			       skb_get_queue_mapping(skb)) ||
1197
	    want_cookie)
1198
		goto drop_and_free;
L
Linus Torvalds 已提交
1199

1200
	tcp_rsk(req)->snt_synack = tcp_time_stamp;
1201
	tcp_rsk(req)->listener = NULL;
1202 1203
	inet6_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
	return 0;
L
Linus Torvalds 已提交
1204

1205 1206
drop_and_release:
	dst_release(dst);
1207 1208
drop_and_free:
	reqsk_free(req);
L
Linus Torvalds 已提交
1209 1210 1211 1212 1213
drop:
	return 0; /* don't send reset */
}

static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
1214
					  struct request_sock *req,
L
Linus Torvalds 已提交
1215 1216
					  struct dst_entry *dst)
{
1217
	struct inet6_request_sock *treq;
L
Linus Torvalds 已提交
1218 1219 1220 1221 1222
	struct ipv6_pinfo *newnp, *np = inet6_sk(sk);
	struct tcp6_sock *newtcp6sk;
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1223 1224 1225
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
1226
	struct flowi6 fl6;
L
Linus Torvalds 已提交
1227 1228 1229 1230 1231 1232 1233 1234

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

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

1235
		if (newsk == NULL)
L
Linus Torvalds 已提交
1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246
			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
1247
		ipv6_addr_set_v4mapped(newinet->inet_daddr, &newnp->daddr);
L
Linus Torvalds 已提交
1248

E
Eric Dumazet 已提交
1249
		ipv6_addr_set_v4mapped(newinet->inet_saddr, &newnp->saddr);
L
Linus Torvalds 已提交
1250

A
Alexey Dobriyan 已提交
1251
		newnp->rcv_saddr = newnp->saddr;
L
Linus Torvalds 已提交
1252

1253
		inet_csk(newsk)->icsk_af_ops = &ipv6_mapped;
L
Linus Torvalds 已提交
1254
		newsk->sk_backlog_rcv = tcp_v4_do_rcv;
1255 1256 1257 1258
#ifdef CONFIG_TCP_MD5SIG
		newtp->af_specific = &tcp_sock_ipv6_mapped_specific;
#endif

1259 1260
		newnp->ipv6_ac_list = NULL;
		newnp->ipv6_fl_list = NULL;
L
Linus Torvalds 已提交
1261 1262
		newnp->pktoptions  = NULL;
		newnp->opt	   = NULL;
1263
		newnp->mcast_oif   = inet6_iif(skb);
1264
		newnp->mcast_hops  = ipv6_hdr(skb)->hop_limit;
1265
		newnp->rcv_tclass  = ipv6_tclass(ipv6_hdr(skb));
L
Linus Torvalds 已提交
1266

1267 1268 1269 1270
		/*
		 * 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
1271 1272 1273
		 */

		/* It is tricky place. Until this moment IPv4 tcp
1274
		   worked with IPv6 icsk.icsk_af_ops.
L
Linus Torvalds 已提交
1275 1276
		   Sync it now.
		 */
1277
		tcp_sync_mss(newsk, inet_csk(newsk)->icsk_pmtu_cookie);
L
Linus Torvalds 已提交
1278 1279 1280 1281

		return newsk;
	}

1282
	treq = inet6_rsk(req);
L
Linus Torvalds 已提交
1283 1284 1285 1286

	if (sk_acceptq_is_full(sk))
		goto out_overflow;

1287
	if (!dst) {
1288
		dst = inet6_csk_route_req(sk, &fl6, req);
1289
		if (!dst)
L
Linus Torvalds 已提交
1290
			goto out;
1291
	}
L
Linus Torvalds 已提交
1292 1293 1294

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (newsk == NULL)
1295
		goto out_nonewsk;
L
Linus Torvalds 已提交
1296

1297 1298 1299 1300 1301
	/*
	 * 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
1302

1303
	newsk->sk_gso_type = SKB_GSO_TCPV6;
1304
	__ip6_dst_store(newsk, dst, NULL, NULL);
1305
	inet6_sk_rx_dst_set(newsk, skb);
L
Linus Torvalds 已提交
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315

	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
1316 1317 1318
	newnp->daddr = treq->rmt_addr;
	newnp->saddr = treq->loc_addr;
	newnp->rcv_saddr = treq->loc_addr;
1319
	newsk->sk_bound_dev_if = treq->iif;
L
Linus Torvalds 已提交
1320

1321
	/* Now IPv6 options...
L
Linus Torvalds 已提交
1322 1323 1324

	   First: no IPv4 options.
	 */
1325
	newinet->inet_opt = NULL;
1326
	newnp->ipv6_ac_list = NULL;
1327
	newnp->ipv6_fl_list = NULL;
L
Linus Torvalds 已提交
1328 1329 1330 1331 1332 1333

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

	/* Clone pktoptions received with SYN */
	newnp->pktoptions = NULL;
1334
	if (treq->pktopts != NULL) {
1335 1336
		newnp->pktoptions = skb_clone(treq->pktopts,
					      sk_gfp_atomic(sk, GFP_ATOMIC));
1337
		consume_skb(treq->pktopts);
1338
		treq->pktopts = NULL;
L
Linus Torvalds 已提交
1339 1340 1341 1342
		if (newnp->pktoptions)
			skb_set_owner_r(newnp->pktoptions, newsk);
	}
	newnp->opt	  = NULL;
1343
	newnp->mcast_oif  = inet6_iif(skb);
1344
	newnp->mcast_hops = ipv6_hdr(skb)->hop_limit;
1345
	newnp->rcv_tclass = ipv6_tclass(ipv6_hdr(skb));
L
Linus Torvalds 已提交
1346 1347 1348 1349 1350 1351 1352

	/* 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.
	 */
1353 1354
	if (np->opt)
		newnp->opt = ipv6_dup_options(newsk, np->opt);
L
Linus Torvalds 已提交
1355

1356
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
1357
	if (newnp->opt)
1358 1359
		inet_csk(newsk)->icsk_ext_hdr_len = (newnp->opt->opt_nflen +
						     newnp->opt->opt_flen);
L
Linus Torvalds 已提交
1360

J
John Heffner 已提交
1361
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1362
	tcp_sync_mss(newsk, dst_mtu(dst));
1363
	newtp->advmss = dst_metric_advmss(dst);
1364 1365 1366 1367
	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
1368
	tcp_initialize_rcv_mss(newsk);
1369
	tcp_synack_rtt_meas(newsk, req);
1370
	newtp->total_retrans = req->num_retrans;
L
Linus Torvalds 已提交
1371

E
Eric Dumazet 已提交
1372 1373
	newinet->inet_daddr = newinet->inet_saddr = LOOPBACK4_IPV6;
	newinet->inet_rcv_saddr = LOOPBACK4_IPV6;
L
Linus Torvalds 已提交
1374

1375 1376 1377 1378 1379 1380 1381 1382
#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
1383
		tcp_md5_do_add(newsk, (union tcp_md5_addr *)&newnp->daddr,
1384 1385
			       AF_INET6, key->key, key->keylen,
			       sk_gfp_atomic(sk, GFP_ATOMIC));
1386 1387 1388
	}
#endif

1389 1390 1391 1392
	if (__inet_inherit_port(sk, newsk) < 0) {
		sock_put(newsk);
		goto out;
	}
1393
	__inet6_hash(newsk, NULL);
L
Linus Torvalds 已提交
1394 1395 1396 1397

	return newsk;

out_overflow:
1398
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1399
out_nonewsk:
L
Linus Torvalds 已提交
1400
	dst_release(dst);
1401 1402
out:
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1403 1404 1405
	return NULL;
}

1406
static __sum16 tcp_v6_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1407
{
1408
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
H
Herbert Xu 已提交
1409
		if (!tcp_v6_check(skb->len, &ipv6_hdr(skb)->saddr,
1410
				  &ipv6_hdr(skb)->daddr, skb->csum)) {
1411
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1412
			return 0;
1413
		}
L
Linus Torvalds 已提交
1414
	}
1415

H
Herbert Xu 已提交
1416
	skb->csum = ~csum_unfold(tcp_v6_check(skb->len,
1417 1418
					      &ipv6_hdr(skb)->saddr,
					      &ipv6_hdr(skb)->daddr, 0));
1419

L
Linus Torvalds 已提交
1420
	if (skb->len <= 76) {
1421
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
	}
	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);

1451 1452 1453 1454 1455
#ifdef CONFIG_TCP_MD5SIG
	if (tcp_v6_inbound_md5_hash (sk, skb))
		goto discard;
#endif

1456
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474
		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.
1475
					       --ANK (980728)
L
Linus Torvalds 已提交
1476 1477
	 */
	if (np->rxopt.all)
1478
		opt_skb = skb_clone(skb, sk_gfp_atomic(sk, GFP_ATOMIC));
L
Linus Torvalds 已提交
1479 1480

	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
E
Eric Dumazet 已提交
1481 1482
		struct dst_entry *dst = sk->sk_rx_dst;

1483
		sock_rps_save_rxhash(sk, skb);
E
Eric Dumazet 已提交
1484 1485 1486 1487 1488 1489 1490 1491
		if (dst) {
			if (inet_sk(sk)->rx_dst_ifindex != skb->skb_iif ||
			    dst->ops->check(dst, np->rx_dst_cookie) == NULL) {
				dst_release(dst);
				sk->sk_rx_dst = NULL;
			}
		}

1492
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len))
L
Linus Torvalds 已提交
1493 1494 1495 1496 1497 1498
			goto reset;
		if (opt_skb)
			goto ipv6_pktoptions;
		return 0;
	}

1499
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1500 1501
		goto csum_err;

1502
	if (sk->sk_state == TCP_LISTEN) {
L
Linus Torvalds 已提交
1503 1504 1505 1506 1507 1508 1509 1510 1511
		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..
		 */
1512
		if(nsk != sk) {
1513
			sock_rps_save_rxhash(nsk, skb);
L
Linus Torvalds 已提交
1514 1515 1516 1517 1518 1519
			if (tcp_child_process(sk, nsk, skb))
				goto reset;
			if (opt_skb)
				__kfree_skb(opt_skb);
			return 0;
		}
1520
	} else
1521
		sock_rps_save_rxhash(sk, skb);
L
Linus Torvalds 已提交
1522

1523
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len))
L
Linus Torvalds 已提交
1524 1525 1526 1527 1528 1529
		goto reset;
	if (opt_skb)
		goto ipv6_pktoptions;
	return 0;

reset:
1530
	tcp_v6_send_reset(sk, skb);
L
Linus Torvalds 已提交
1531 1532 1533 1534 1535 1536
discard:
	if (opt_skb)
		__kfree_skb(opt_skb);
	kfree_skb(skb);
	return 0;
csum_err:
1537
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
	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))) {
1552
		if (np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo)
1553
			np->mcast_oif = inet6_iif(opt_skb);
1554
		if (np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim)
1555
			np->mcast_hops = ipv6_hdr(opt_skb)->hop_limit;
1556 1557
		if (np->rxopt.bits.rxtclass)
			np->rcv_tclass = ipv6_tclass(ipv6_hdr(skb));
L
Linus Torvalds 已提交
1558 1559 1560 1561 1562 1563 1564 1565 1566
		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);
		}
	}

1567
	kfree_skb(opt_skb);
L
Linus Torvalds 已提交
1568 1569 1570
	return 0;
}

1571
static int tcp_v6_rcv(struct sk_buff *skb)
L
Linus Torvalds 已提交
1572
{
1573
	const struct tcphdr *th;
1574
	const struct ipv6hdr *hdr;
L
Linus Torvalds 已提交
1575 1576
	struct sock *sk;
	int ret;
1577
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1578 1579 1580 1581 1582 1583 1584

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

	/*
	 *	Count it even if it's bad.
	 */
1585
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1586 1587 1588 1589

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

1590
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1591 1592 1593 1594 1595 1596

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

1597
	if (!skb_csum_unnecessary(skb) && tcp_v6_checksum_init(skb))
L
Linus Torvalds 已提交
1598 1599
		goto bad_packet;

1600
	th = tcp_hdr(skb);
1601
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
1602 1603 1604 1605 1606
	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;
1607
	TCP_SKB_CB(skb)->ip_dsfield = ipv6_get_dsfield(hdr);
L
Linus Torvalds 已提交
1608 1609
	TCP_SKB_CB(skb)->sacked = 0;

1610
	sk = __inet6_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616 1617
	if (!sk)
		goto no_tcp_socket;

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

1618 1619 1620 1621 1622
	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
1623 1624 1625
	if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;

1626
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1627 1628 1629 1630
		goto discard_and_relse;

	skb->dev = NULL;

1631
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1632 1633
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1634
#ifdef CONFIG_NET_DMA
1635
		struct tcp_sock *tp = tcp_sk(sk);
1636
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1637
			tp->ucopy.dma_chan = net_dma_find_channel();
1638 1639 1640
		if (tp->ucopy.dma_chan)
			ret = tcp_v6_do_rcv(sk, skb);
		else
C
Chris Leech 已提交
1641 1642 1643 1644 1645
#endif
		{
			if (!tcp_prequeue(sk, skb))
				ret = tcp_v6_do_rcv(sk, skb);
		}
1646 1647
	} else if (unlikely(sk_add_backlog(sk, skb,
					   sk->sk_rcvbuf + sk->sk_sndbuf))) {
Z
Zhu Yi 已提交
1648
		bh_unlock_sock(sk);
1649
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1650 1651
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662
	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:
1663
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1664
	} else {
1665
		tcp_v6_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682
	}

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)) {
1683
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1684 1685 1686 1687
		goto discard_it;
	}

	if (skb->len < (th->doff<<2) || tcp_checksum_complete(skb)) {
1688
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1689
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1690 1691 1692
		goto discard_it;
	}

1693
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1694 1695 1696 1697
	case TCP_TW_SYN:
	{
		struct sock *sk2;

1698
		sk2 = inet6_lookup_listener(dev_net(skb->dev), &tcp_hashinfo,
1699
					    &ipv6_hdr(skb)->daddr,
1700
					    ntohs(th->dest), inet6_iif(skb));
L
Linus Torvalds 已提交
1701
		if (sk2 != NULL) {
1702 1703 1704
			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
1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719
			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;
}

E
Eric Dumazet 已提交
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746
static void tcp_v6_early_demux(struct sk_buff *skb)
{
	const struct ipv6hdr *hdr;
	const struct tcphdr *th;
	struct sock *sk;

	if (skb->pkt_type != PACKET_HOST)
		return;

	if (!pskb_may_pull(skb, skb_transport_offset(skb) + sizeof(struct tcphdr)))
		return;

	hdr = ipv6_hdr(skb);
	th = tcp_hdr(skb);

	if (th->doff < sizeof(struct tcphdr) / 4)
		return;

	sk = __inet6_lookup_established(dev_net(skb->dev), &tcp_hashinfo,
					&hdr->saddr, th->source,
					&hdr->daddr, ntohs(th->dest),
					inet6_iif(skb));
	if (sk) {
		skb->sk = sk;
		skb->destructor = sock_edemux;
		if (sk->sk_state != TCP_TIME_WAIT) {
			struct dst_entry *dst = sk->sk_rx_dst;
1747

E
Eric Dumazet 已提交
1748
			if (dst)
E
Eric Dumazet 已提交
1749
				dst = dst_check(dst, inet6_sk(sk)->rx_dst_cookie);
E
Eric Dumazet 已提交
1750
			if (dst &&
1751
			    inet_sk(sk)->rx_dst_ifindex == skb->skb_iif)
E
Eric Dumazet 已提交
1752 1753 1754 1755 1756
				skb_dst_set_noref(skb, dst);
		}
	}
}

1757 1758 1759 1760 1761 1762
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,
};

1763
static const struct inet_connection_sock_af_ops ipv6_specific = {
1764 1765 1766
	.queue_xmit	   = inet6_csk_xmit,
	.send_check	   = tcp_v6_send_check,
	.rebuild_header	   = inet6_sk_rebuild_header,
E
Eric Dumazet 已提交
1767
	.sk_rx_dst_set	   = inet6_sk_rx_dst_set,
1768 1769 1770
	.conn_request	   = tcp_v6_conn_request,
	.syn_recv_sock	   = tcp_v6_syn_recv_sock,
	.net_header_len	   = sizeof(struct ipv6hdr),
1771
	.net_frag_header_len = sizeof(struct frag_hdr),
1772 1773 1774 1775
	.setsockopt	   = ipv6_setsockopt,
	.getsockopt	   = ipv6_getsockopt,
	.addr2sockaddr	   = inet6_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in6),
1776
	.bind_conflict	   = inet6_csk_bind_conflict,
1777
#ifdef CONFIG_COMPAT
1778 1779
	.compat_setsockopt = compat_ipv6_setsockopt,
	.compat_getsockopt = compat_ipv6_getsockopt,
1780
#endif
L
Linus Torvalds 已提交
1781 1782
};

1783
#ifdef CONFIG_TCP_MD5SIG
1784
static const struct tcp_sock_af_ops tcp_sock_ipv6_specific = {
1785
	.md5_lookup	=	tcp_v6_md5_lookup,
1786
	.calc_md5_hash	=	tcp_v6_md5_hash_skb,
1787 1788
	.md5_parse	=	tcp_v6_parse_md5_keys,
};
1789
#endif
1790

L
Linus Torvalds 已提交
1791 1792 1793 1794
/*
 *	TCP over IPv4 via INET6 API
 */

1795
static const struct inet_connection_sock_af_ops ipv6_mapped = {
1796 1797 1798
	.queue_xmit	   = ip_queue_xmit,
	.send_check	   = tcp_v4_send_check,
	.rebuild_header	   = inet_sk_rebuild_header,
1799
	.sk_rx_dst_set	   = inet_sk_rx_dst_set,
1800 1801 1802 1803 1804 1805 1806
	.conn_request	   = tcp_v6_conn_request,
	.syn_recv_sock	   = tcp_v6_syn_recv_sock,
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ipv6_setsockopt,
	.getsockopt	   = ipv6_getsockopt,
	.addr2sockaddr	   = inet6_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in6),
1807
	.bind_conflict	   = inet6_csk_bind_conflict,
1808
#ifdef CONFIG_COMPAT
1809 1810
	.compat_setsockopt = compat_ipv6_setsockopt,
	.compat_getsockopt = compat_ipv6_getsockopt,
1811
#endif
L
Linus Torvalds 已提交
1812 1813
};

1814
#ifdef CONFIG_TCP_MD5SIG
1815
static const struct tcp_sock_af_ops tcp_sock_ipv6_mapped_specific = {
1816
	.md5_lookup	=	tcp_v4_md5_lookup,
1817
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1818 1819
	.md5_parse	=	tcp_v6_parse_md5_keys,
};
1820
#endif
1821

L
Linus Torvalds 已提交
1822 1823 1824 1825 1826
/* 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)
{
1827
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1828

1829
	tcp_init_sock(sk);
L
Linus Torvalds 已提交
1830

1831
	icsk->icsk_af_ops = &ipv6_specific;
L
Linus Torvalds 已提交
1832

1833
#ifdef CONFIG_TCP_MD5SIG
1834
	tcp_sk(sk)->af_specific = &tcp_sock_ipv6_specific;
1835 1836
#endif

L
Linus Torvalds 已提交
1837 1838 1839
	return 0;
}

1840
static void tcp_v6_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1841 1842
{
	tcp_v4_destroy_sock(sk);
1843
	inet6_destroy_sock(sk);
L
Linus Torvalds 已提交
1844 1845
}

1846
#ifdef CONFIG_PROC_FS
L
Linus Torvalds 已提交
1847
/* Proc filesystem TCPv6 sock list dumping. */
1848
static void get_openreq6(struct seq_file *seq,
1849
			 const struct sock *sk, struct request_sock *req, int i, kuid_t uid)
L
Linus Torvalds 已提交
1850 1851
{
	int ttd = req->expires - jiffies;
1852 1853
	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
1854 1855 1856 1857 1858 1859

	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
1860
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK\n",
L
Linus Torvalds 已提交
1861 1862 1863
		   i,
		   src->s6_addr32[0], src->s6_addr32[1],
		   src->s6_addr32[2], src->s6_addr32[3],
1864
		   ntohs(inet_rsk(req)->loc_port),
L
Linus Torvalds 已提交
1865 1866
		   dest->s6_addr32[0], dest->s6_addr32[1],
		   dest->s6_addr32[2], dest->s6_addr32[3],
1867
		   ntohs(inet_rsk(req)->rmt_port),
L
Linus Torvalds 已提交
1868 1869
		   TCP_SYN_RECV,
		   0,0, /* could print option size, but that is af dependent. */
1870 1871
		   1,   /* timers active (only the expire timer) */
		   jiffies_to_clock_t(ttd),
1872
		   req->num_timeout,
1873
		   from_kuid_munged(seq_user_ns(seq), uid),
1874
		   0,  /* non standard timer */
L
Linus Torvalds 已提交
1875 1876 1877 1878 1879 1880
		   0, /* open_requests have no inode */
		   0, req);
}

static void get_tcp6_sock(struct seq_file *seq, struct sock *sp, int i)
{
1881
	const struct in6_addr *dest, *src;
L
Linus Torvalds 已提交
1882 1883 1884
	__u16 destp, srcp;
	int timer_active;
	unsigned long timer_expires;
1885 1886
	const struct inet_sock *inet = inet_sk(sp);
	const struct tcp_sock *tp = tcp_sk(sp);
1887
	const struct inet_connection_sock *icsk = inet_csk(sp);
1888
	const struct ipv6_pinfo *np = inet6_sk(sp);
L
Linus Torvalds 已提交
1889 1890 1891

	dest  = &np->daddr;
	src   = &np->rcv_saddr;
E
Eric Dumazet 已提交
1892 1893
	destp = ntohs(inet->inet_dport);
	srcp  = ntohs(inet->inet_sport);
1894 1895

	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
1896
		timer_active	= 1;
1897 1898
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
1899
		timer_active	= 4;
1900
		timer_expires	= icsk->icsk_timeout;
L
Linus Torvalds 已提交
1901 1902 1903 1904 1905 1906 1907 1908 1909 1910
	} 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
1911
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %lu %lu %u %u %d\n",
L
Linus Torvalds 已提交
1912 1913 1914 1915 1916
		   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,
1917
		   sp->sk_state,
1918 1919
		   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
1920
		   timer_active,
1921
		   jiffies_delta_to_clock_t(timer_expires - jiffies),
1922
		   icsk->icsk_retransmits,
1923
		   from_kuid_munged(seq_user_ns(seq), sock_i_uid(sp)),
1924
		   icsk->icsk_probes_out,
L
Linus Torvalds 已提交
1925 1926
		   sock_i_ino(sp),
		   atomic_read(&sp->sk_refcnt), sp,
1927 1928
		   jiffies_to_clock_t(icsk->icsk_rto),
		   jiffies_to_clock_t(icsk->icsk_ack.ato),
1929
		   (icsk->icsk_ack.quick << 1 ) | icsk->icsk_ack.pingpong,
I
Ilpo Järvinen 已提交
1930 1931
		   tp->snd_cwnd,
		   tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh
L
Linus Torvalds 已提交
1932 1933 1934
		   );
}

1935
static void get_timewait6_sock(struct seq_file *seq,
1936
			       struct inet_timewait_sock *tw, int i)
L
Linus Torvalds 已提交
1937
{
1938
	const struct in6_addr *dest, *src;
L
Linus Torvalds 已提交
1939
	__u16 destp, srcp;
1940
	const struct inet6_timewait_sock *tw6 = inet6_twsk((struct sock *)tw);
1941
	long delta = tw->tw_ttd - jiffies;
L
Linus Torvalds 已提交
1942

1943 1944
	dest = &tw6->tw_v6_daddr;
	src  = &tw6->tw_v6_rcv_saddr;
L
Linus Torvalds 已提交
1945 1946 1947 1948 1949
	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
1950
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK\n",
L
Linus Torvalds 已提交
1951 1952 1953 1954 1955 1956
		   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,
1957
		   3, jiffies_delta_to_clock_t(delta), 0, 0, 0, 0,
L
Linus Torvalds 已提交
1958 1959 1960 1961 1962 1963 1964 1965 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
		   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;
}

1992 1993 1994 1995 1996 1997 1998 1999
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
2000 2001 2002
static struct tcp_seq_afinfo tcp6_seq_afinfo = {
	.name		= "tcp6",
	.family		= AF_INET6,
2003
	.seq_fops	= &tcp6_afinfo_seq_fops,
2004 2005 2006
	.seq_ops	= {
		.show		= tcp6_seq_show,
	},
L
Linus Torvalds 已提交
2007 2008
};

2009
int __net_init tcp6_proc_init(struct net *net)
L
Linus Torvalds 已提交
2010
{
2011
	return tcp_proc_register(net, &tcp6_seq_afinfo);
L
Linus Torvalds 已提交
2012 2013
}

2014
void tcp6_proc_exit(struct net *net)
L
Linus Torvalds 已提交
2015
{
2016
	tcp_proc_unregister(net, &tcp6_seq_afinfo);
L
Linus Torvalds 已提交
2017 2018 2019 2020 2021 2022 2023 2024 2025
}
#endif

struct proto tcpv6_prot = {
	.name			= "TCPv6",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v6_connect,
	.disconnect		= tcp_disconnect,
2026
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2027 2028 2029 2030 2031 2032 2033
	.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,
2034 2035
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2036
	.backlog_rcv		= tcp_v6_do_rcv,
E
Eric Dumazet 已提交
2037
	.release_cb		= tcp_release_cb,
2038
	.mtu_reduced		= tcp_v6_mtu_reduced,
L
Linus Torvalds 已提交
2039
	.hash			= tcp_v6_hash,
2040 2041
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2042 2043 2044 2045
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
2046
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2047 2048 2049 2050
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp6_sock),
2051
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2052
	.twsk_prot		= &tcp6_timewait_sock_ops,
2053
	.rsk_prot		= &tcp6_request_sock_ops,
2054
	.h.hashinfo		= &tcp_hashinfo,
2055
	.no_autobind		= true,
2056 2057 2058 2059
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
A
Andrew Morton 已提交
2060
#ifdef CONFIG_MEMCG_KMEM
G
Glauber Costa 已提交
2061 2062
	.proto_cgroup		= tcp_proto_cgroup,
#endif
L
Linus Torvalds 已提交
2063 2064
};

2065
static const struct inet6_protocol tcpv6_protocol = {
E
Eric Dumazet 已提交
2066
	.early_demux	=	tcp_v6_early_demux,
L
Linus Torvalds 已提交
2067 2068
	.handler	=	tcp_v6_rcv,
	.err_handler	=	tcp_v6_err,
2069
	.gso_send_check	=	tcp_v6_gso_send_check,
H
Herbert Xu 已提交
2070
	.gso_segment	=	tcp_tso_segment,
H
Herbert Xu 已提交
2071 2072
	.gro_receive	=	tcp6_gro_receive,
	.gro_complete	=	tcp6_gro_complete,
L
Linus Torvalds 已提交
2073 2074 2075
	.flags		=	INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
};

2076 2077 2078 2079 2080 2081 2082
static const struct net_offload tcpv6_offload = {
	.gso_send_check	=	tcp_v6_gso_send_check,
	.gso_segment	=	tcp_tso_segment,
	.gro_receive	=	tcp6_gro_receive,
	.gro_complete	=	tcp6_gro_complete,
};

L
Linus Torvalds 已提交
2083 2084 2085 2086 2087 2088
static struct inet_protosw tcpv6_protosw = {
	.type		=	SOCK_STREAM,
	.protocol	=	IPPROTO_TCP,
	.prot		=	&tcpv6_prot,
	.ops		=	&inet6_stream_ops,
	.no_check	=	0,
2089 2090
	.flags		=	INET_PROTOSW_PERMANENT |
				INET_PROTOSW_ICSK,
L
Linus Torvalds 已提交
2091 2092
};

2093
static int __net_init tcpv6_net_init(struct net *net)
2094
{
2095 2096
	return inet_ctl_sock_create(&net->ipv6.tcp_sk, PF_INET6,
				    SOCK_RAW, IPPROTO_TCP, net);
2097 2098
}

2099
static void __net_exit tcpv6_net_exit(struct net *net)
2100
{
2101
	inet_ctl_sock_destroy(net->ipv6.tcp_sk);
E
Eric W. Biederman 已提交
2102 2103
}

2104
static void __net_exit tcpv6_net_exit_batch(struct list_head *net_exit_list)
E
Eric W. Biederman 已提交
2105 2106
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET6);
2107 2108 2109
}

static struct pernet_operations tcpv6_net_ops = {
E
Eric W. Biederman 已提交
2110 2111 2112
	.init	    = tcpv6_net_init,
	.exit	    = tcpv6_net_exit,
	.exit_batch = tcpv6_net_exit_batch,
2113 2114
};

2115
int __init tcpv6_init(void)
L
Linus Torvalds 已提交
2116
{
2117 2118 2119 2120 2121 2122
	int ret;

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

L
Linus Torvalds 已提交
2123
	/* register inet6 protocol */
2124 2125 2126 2127
	ret = inet6_register_protosw(&tcpv6_protosw);
	if (ret)
		goto out_tcpv6_protocol;

2128
	ret = register_pernet_subsys(&tcpv6_net_ops);
2129 2130 2131 2132
	if (ret)
		goto out_tcpv6_protosw;
out:
	return ret;
2133

2134 2135 2136 2137 2138 2139 2140
out_tcpv6_protocol:
	inet6_del_protocol(&tcpv6_protocol, IPPROTO_TCP);
out_tcpv6_protosw:
	inet6_unregister_protosw(&tcpv6_protosw);
	goto out;
}

2141
void tcpv6_exit(void)
2142
{
2143
	unregister_pernet_subsys(&tcpv6_net_ops);
2144 2145
	inet6_unregister_protosw(&tcpv6_protosw);
	inet6_del_protocol(&tcpv6_protocol, IPPROTO_TCP);
L
Linus Torvalds 已提交
2146
}