tcp_ipv6.c 56.1 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>
L
Linus Torvalds 已提交
64 65 66 67 68 69

#include <asm/uaccess.h>

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

70 71 72 73
#include <linux/crypto.h>
#include <linux/scatterlist.h>

static void	tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb);
74 75
static void	tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
				      struct request_sock *req);
L
Linus Torvalds 已提交
76 77

static int	tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb);
78 79 80
static void	__tcp_v6_send_check(struct sk_buff *skb,
				    struct in6_addr *saddr,
				    struct in6_addr *daddr);
L
Linus Torvalds 已提交
81

82 83
static const struct inet_connection_sock_af_ops ipv6_mapped;
static const struct inet_connection_sock_af_ops ipv6_specific;
84
#ifdef CONFIG_TCP_MD5SIG
85 86
static const struct tcp_sock_af_ops tcp_sock_ipv6_specific;
static const struct tcp_sock_af_ops tcp_sock_ipv6_mapped_specific;
87 88 89 90 91 92
#else
static struct tcp_md5sig_key *tcp_v6_md5_do_lookup(struct sock *sk,
						   struct in6_addr *addr)
{
	return NULL;
}
93
#endif
L
Linus Torvalds 已提交
94 95 96 97

static void tcp_v6_hash(struct sock *sk)
{
	if (sk->sk_state != TCP_CLOSE) {
98
		if (inet_csk(sk)->icsk_af_ops == &ipv6_mapped) {
L
Linus Torvalds 已提交
99 100 101 102
			tcp_prot.hash(sk);
			return;
		}
		local_bh_disable();
103
		__inet6_hash(sk, NULL);
L
Linus Torvalds 已提交
104 105 106 107
		local_bh_enable();
	}
}

H
Herbert Xu 已提交
108
static __inline__ __sum16 tcp_v6_check(int len,
109 110
				   struct in6_addr *saddr,
				   struct in6_addr *daddr,
111
				   __wsum base)
L
Linus Torvalds 已提交
112 113 114 115
{
	return csum_ipv6_magic(saddr, daddr, len, IPPROTO_TCP, base);
}

116
static __u32 tcp_v6_init_sequence(struct sk_buff *skb)
L
Linus Torvalds 已提交
117
{
118 119
	return secure_tcpv6_sequence_number(ipv6_hdr(skb)->daddr.s6_addr32,
					    ipv6_hdr(skb)->saddr.s6_addr32,
120 121
					    tcp_hdr(skb)->dest,
					    tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
122 123
}

124
static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
L
Linus Torvalds 已提交
125 126 127
			  int addr_len)
{
	struct sockaddr_in6 *usin = (struct sockaddr_in6 *) uaddr;
128
	struct inet_sock *inet = inet_sk(sk);
129
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
130 131
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
132
	struct in6_addr *saddr = NULL, *final_p, final;
L
Linus Torvalds 已提交
133 134 135 136 137
	struct flowi fl;
	struct dst_entry *dst;
	int addr_type;
	int err;

138
	if (addr_len < SIN6_LEN_RFC2133)
L
Linus Torvalds 已提交
139 140
		return -EINVAL;

141
	if (usin->sin6_family != AF_INET6)
E
Eric Dumazet 已提交
142
		return -EAFNOSUPPORT;
L
Linus Torvalds 已提交
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159

	memset(&fl, 0, sizeof(fl));

	if (np->sndflow) {
		fl.fl6_flowlabel = usin->sin6_flowinfo&IPV6_FLOWINFO_MASK;
		IP6_ECN_flow_init(fl.fl6_flowlabel);
		if (fl.fl6_flowlabel&IPV6_FLOWLABEL_MASK) {
			struct ip6_flowlabel *flowlabel;
			flowlabel = fl6_sock_lookup(sk, fl.fl6_flowlabel);
			if (flowlabel == NULL)
				return -EINVAL;
			ipv6_addr_copy(&usin->sin6_addr, &flowlabel->dst);
			fl6_sock_release(flowlabel);
		}
	}

	/*
160 161 162 163 164
	 *	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
165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203

	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;
	}

	ipv6_addr_copy(&np->daddr, &usin->sin6_addr);
	np->flow_label = fl.fl6_flowlabel;

	/*
	 *	TCP over IPv4
	 */

	if (addr_type == IPV6_ADDR_MAPPED) {
204
		u32 exthdrlen = icsk->icsk_ext_hdr_len;
L
Linus Torvalds 已提交
205 206 207 208 209 210 211 212 213 214 215
		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];

216
		icsk->icsk_af_ops = &ipv6_mapped;
L
Linus Torvalds 已提交
217
		sk->sk_backlog_rcv = tcp_v4_do_rcv;
218 219 220
#ifdef CONFIG_TCP_MD5SIG
		tp->af_specific = &tcp_sock_ipv6_mapped_specific;
#endif
L
Linus Torvalds 已提交
221 222 223 224

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

		if (err) {
225 226
			icsk->icsk_ext_hdr_len = exthdrlen;
			icsk->icsk_af_ops = &ipv6_specific;
L
Linus Torvalds 已提交
227
			sk->sk_backlog_rcv = tcp_v6_do_rcv;
228 229 230
#ifdef CONFIG_TCP_MD5SIG
			tp->af_specific = &tcp_sock_ipv6_specific;
#endif
L
Linus Torvalds 已提交
231 232
			goto failure;
		} else {
E
Eric Dumazet 已提交
233 234 235
			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
236 237 238 239 240 241 242 243 244 245 246 247 248
		}

		return err;
	}

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

	fl.proto = IPPROTO_TCP;
	ipv6_addr_copy(&fl.fl6_dst, &np->daddr);
	ipv6_addr_copy(&fl.fl6_src,
		       (saddr ? saddr : &np->saddr));
	fl.oif = sk->sk_bound_dev_if;
B
Brian Haley 已提交
249
	fl.mark = sk->sk_mark;
L
Linus Torvalds 已提交
250
	fl.fl_ip_dport = usin->sin6_port;
E
Eric Dumazet 已提交
251
	fl.fl_ip_sport = inet->inet_sport;
L
Linus Torvalds 已提交
252

253
	final_p = fl6_update_dst(&fl, np->opt, &final);
L
Linus Torvalds 已提交
254

V
Venkat Yekkirala 已提交
255 256
	security_sk_classify_flow(sk, &fl);

L
Linus Torvalds 已提交
257 258 259 260 261 262
	err = ip6_dst_lookup(sk, &dst, &fl);
	if (err)
		goto failure;
	if (final_p)
		ipv6_addr_copy(&fl.fl6_dst, final_p);

A
Alexey Dobriyan 已提交
263 264
	err = __xfrm_lookup(sock_net(sk), &dst, &fl, sk, XFRM_LOOKUP_WAIT);
	if (err < 0) {
265 266 267 268 269
		if (err == -EREMOTE)
			err = ip6_dst_blackhole(sk, &dst, &fl);
		if (err < 0)
			goto failure;
	}
L
Linus Torvalds 已提交
270 271 272 273 274 275 276 277

	if (saddr == NULL) {
		saddr = &fl.fl6_src;
		ipv6_addr_copy(&np->rcv_saddr, saddr);
	}

	/* set the source address */
	ipv6_addr_copy(&np->saddr, saddr);
E
Eric Dumazet 已提交
278
	inet->inet_rcv_saddr = LOOPBACK4_IPV6;
L
Linus Torvalds 已提交
279

H
Herbert Xu 已提交
280
	sk->sk_gso_type = SKB_GSO_TCPV6;
281
	__ip6_dst_store(sk, dst, NULL, NULL);
L
Linus Torvalds 已提交
282

283
	icsk->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
284
	if (np->opt)
285 286
		icsk->icsk_ext_hdr_len = (np->opt->opt_flen +
					  np->opt->opt_nflen);
L
Linus Torvalds 已提交
287 288 289

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

E
Eric Dumazet 已提交
290
	inet->inet_dport = usin->sin6_port;
L
Linus Torvalds 已提交
291 292

	tcp_set_state(sk, TCP_SYN_SENT);
293
	err = inet6_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
294 295 296 297 298 299
	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
300 301
							     inet->inet_sport,
							     inet->inet_dport);
L
Linus Torvalds 已提交
302 303 304 305 306 307 308 309 310 311 312

	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
313
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
314 315 316 317 318
	sk->sk_route_caps = 0;
	return err;
}

static void tcp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
319
		u8 type, u8 code, int offset, __be32 info)
L
Linus Torvalds 已提交
320 321
{
	struct ipv6hdr *hdr = (struct ipv6hdr*)skb->data;
322
	const struct tcphdr *th = (struct tcphdr *)(skb->data+offset);
L
Linus Torvalds 已提交
323 324 325
	struct ipv6_pinfo *np;
	struct sock *sk;
	int err;
326
	struct tcp_sock *tp;
L
Linus Torvalds 已提交
327
	__u32 seq;
328
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
329

330
	sk = inet6_lookup(net, &tcp_hashinfo, &hdr->daddr,
331
			th->dest, &hdr->saddr, th->source, skb->dev->ifindex);
L
Linus Torvalds 已提交
332 333

	if (sk == NULL) {
334 335
		ICMP6_INC_STATS_BH(net, __in6_dev_get(skb->dev),
				   ICMP6_MIB_INERRORS);
L
Linus Torvalds 已提交
336 337 338 339
		return;
	}

	if (sk->sk_state == TCP_TIME_WAIT) {
340
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
341 342 343 344 345
		return;
	}

	bh_lock_sock(sk);
	if (sock_owned_by_user(sk))
346
		NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
L
Linus Torvalds 已提交
347 348 349 350

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

351 352 353 354 355
	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
356
	tp = tcp_sk(sk);
357
	seq = ntohl(th->seq);
L
Linus Torvalds 已提交
358 359
	if (sk->sk_state != TCP_LISTEN &&
	    !between(seq, tp->snd_una, tp->snd_nxt)) {
360
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
		goto out;
	}

	np = inet6_sk(sk);

	if (type == ICMPV6_PKT_TOOBIG) {
		struct dst_entry *dst = NULL;

		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);
			struct flowi fl;

			/* BUGGG_FUTURE: Again, it is not clear how
			   to handle rthdr case. Ignore this complexity
			   for now.
			 */
			memset(&fl, 0, sizeof(fl));
			fl.proto = IPPROTO_TCP;
			ipv6_addr_copy(&fl.fl6_dst, &np->daddr);
			ipv6_addr_copy(&fl.fl6_src, &np->saddr);
			fl.oif = sk->sk_bound_dev_if;
B
Brian Haley 已提交
390
			fl.mark = sk->sk_mark;
E
Eric Dumazet 已提交
391 392
			fl.fl_ip_dport = inet->inet_dport;
			fl.fl_ip_sport = inet->inet_sport;
V
Venkat Yekkirala 已提交
393
			security_skb_classify_flow(skb, &fl);
L
Linus Torvalds 已提交
394 395 396 397 398 399

			if ((err = ip6_dst_lookup(sk, &dst, &fl))) {
				sk->sk_err_soft = -err;
				goto out;
			}

A
Alexey Dobriyan 已提交
400
			if ((err = xfrm_lookup(net, &dst, &fl, sk, 0)) < 0) {
L
Linus Torvalds 已提交
401 402 403 404 405 406 407
				sk->sk_err_soft = -err;
				goto out;
			}

		} else
			dst_hold(dst);

408
		if (inet_csk(sk)->icsk_pmtu_cookie > dst_mtu(dst)) {
L
Linus Torvalds 已提交
409 410 411 412 413 414 415 416 417
			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);

418
	/* Might be for an request_sock */
L
Linus Torvalds 已提交
419
	switch (sk->sk_state) {
420
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
421 422 423 424
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

425 426
		req = inet6_csk_search_req(sk, &prev, th->dest, &hdr->daddr,
					   &hdr->saddr, inet6_iif(skb));
L
Linus Torvalds 已提交
427 428 429 430 431 432
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		 * an established socket here.
		 */
433
		WARN_ON(req->sk != NULL);
L
Linus Torvalds 已提交
434

435
		if (seq != tcp_rsk(req)->snt_isn) {
436
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
437 438 439
			goto out;
		}

440
		inet_csk_reqsk_queue_drop(sk, req, prev);
L
Linus Torvalds 已提交
441 442 443 444
		goto out;

	case TCP_SYN_SENT:
	case TCP_SYN_RECV:  /* Cannot happen.
445
			       It can, it SYNs are crossed. --ANK */
L
Linus Torvalds 已提交
446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
		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);
}


468 469
static int tcp_v6_send_synack(struct sock *sk, struct request_sock *req,
			      struct request_values *rvp)
L
Linus Torvalds 已提交
470
{
471
	struct inet6_request_sock *treq = inet6_rsk(req);
L
Linus Torvalds 已提交
472 473 474
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct sk_buff * skb;
	struct ipv6_txoptions *opt = NULL;
475
	struct in6_addr * final_p, final;
L
Linus Torvalds 已提交
476
	struct flowi fl;
477
	struct dst_entry *dst;
L
Linus Torvalds 已提交
478 479 480 481
	int err = -1;

	memset(&fl, 0, sizeof(fl));
	fl.proto = IPPROTO_TCP;
482 483
	ipv6_addr_copy(&fl.fl6_dst, &treq->rmt_addr);
	ipv6_addr_copy(&fl.fl6_src, &treq->loc_addr);
L
Linus Torvalds 已提交
484
	fl.fl6_flowlabel = 0;
485
	fl.oif = treq->iif;
B
Brian Haley 已提交
486
	fl.mark = sk->sk_mark;
487
	fl.fl_ip_dport = inet_rsk(req)->rmt_port;
488
	fl.fl_ip_sport = inet_rsk(req)->loc_port;
489
	security_req_classify_flow(req, &fl);
L
Linus Torvalds 已提交
490

491
	opt = np->opt;
492
	final_p = fl6_update_dst(&fl, opt, &final);
L
Linus Torvalds 已提交
493

494 495 496 497 498
	err = ip6_dst_lookup(sk, &dst, &fl);
	if (err)
		goto done;
	if (final_p)
		ipv6_addr_copy(&fl.fl6_dst, final_p);
A
Alexey Dobriyan 已提交
499
	if ((err = xfrm_lookup(sock_net(sk), &dst, &fl, sk, 0)) < 0)
500 501
		goto done;

502
	skb = tcp_make_synack(sk, dst, req, rvp);
L
Linus Torvalds 已提交
503
	if (skb) {
504
		__tcp_v6_send_check(skb, &treq->loc_addr, &treq->rmt_addr);
L
Linus Torvalds 已提交
505

506
		ipv6_addr_copy(&fl.fl6_dst, &treq->rmt_addr);
507
		err = ip6_xmit(sk, skb, &fl, opt);
508
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
509 510 511
	}

done:
512
	if (opt && opt != np->opt)
L
Linus Torvalds 已提交
513
		sock_kfree_s(sk, opt, opt->tot_len);
514
	dst_release(dst);
L
Linus Torvalds 已提交
515 516 517
	return err;
}

518 519 520 521 522 523 524
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);
	return tcp_v6_send_synack(sk, req, rvp);
}

525 526 527 528 529 530 531 532 533 534 535 536 537 538
static inline void syn_flood_warning(struct sk_buff *skb)
{
#ifdef CONFIG_SYN_COOKIES
	if (sysctl_tcp_syncookies)
		printk(KERN_INFO
		       "TCPv6: Possible SYN flooding on port %d. "
		       "Sending cookies.\n", ntohs(tcp_hdr(skb)->dest));
	else
#endif
		printk(KERN_INFO
		       "TCPv6: Possible SYN flooding on port %d. "
		       "Dropping request.\n", ntohs(tcp_hdr(skb)->dest));
}

539
static void tcp_v6_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
540
{
541
	kfree_skb(inet6_rsk(req)->pktopts);
L
Linus Torvalds 已提交
542 543
}

544 545 546 547 548 549 550 551 552 553 554 555 556
#ifdef CONFIG_TCP_MD5SIG
static struct tcp_md5sig_key *tcp_v6_md5_do_lookup(struct sock *sk,
						   struct in6_addr *addr)
{
	struct tcp_sock *tp = tcp_sk(sk);
	int i;

	BUG_ON(tp == NULL);

	if (!tp->md5sig_info || !tp->md5sig_info->entries6)
		return NULL;

	for (i = 0; i < tp->md5sig_info->entries6; i++) {
557
		if (ipv6_addr_equal(&tp->md5sig_info->keys6[i].addr, addr))
558
			return &tp->md5sig_info->keys6[i].base;
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578
	}
	return NULL;
}

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_md5_do_add(struct sock *sk, struct in6_addr *peer,
			     char *newkey, u8 newkeylen)
{
	/* Add key to the list */
579
	struct tcp_md5sig_key *key;
580 581 582
	struct tcp_sock *tp = tcp_sk(sk);
	struct tcp6_md5sig_key *keys;

583
	key = tcp_v6_md5_do_lookup(sk, peer);
584 585
	if (key) {
		/* modify existing entry - just update that one */
586 587 588
		kfree(key->key);
		key->key = newkey;
		key->keylen = newkeylen;
589 590 591 592 593 594 595 596
	} else {
		/* reallocate new list if current one is full. */
		if (!tp->md5sig_info) {
			tp->md5sig_info = kzalloc(sizeof(*tp->md5sig_info), GFP_ATOMIC);
			if (!tp->md5sig_info) {
				kfree(newkey);
				return -ENOMEM;
			}
E
Eric Dumazet 已提交
597
			sk_nocaps_add(sk, NETIF_F_GSO_MASK);
598
		}
599
		if (tcp_alloc_md5sig_pool(sk) == NULL) {
600 601 602
			kfree(newkey);
			return -ENOMEM;
		}
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624
		if (tp->md5sig_info->alloced6 == tp->md5sig_info->entries6) {
			keys = kmalloc((sizeof (tp->md5sig_info->keys6[0]) *
				       (tp->md5sig_info->entries6 + 1)), GFP_ATOMIC);

			if (!keys) {
				tcp_free_md5sig_pool();
				kfree(newkey);
				return -ENOMEM;
			}

			if (tp->md5sig_info->entries6)
				memmove(keys, tp->md5sig_info->keys6,
					(sizeof (tp->md5sig_info->keys6[0]) *
					 tp->md5sig_info->entries6));

			kfree(tp->md5sig_info->keys6);
			tp->md5sig_info->keys6 = keys;
			tp->md5sig_info->alloced6++;
		}

		ipv6_addr_copy(&tp->md5sig_info->keys6[tp->md5sig_info->entries6].addr,
			       peer);
625 626
		tp->md5sig_info->keys6[tp->md5sig_info->entries6].base.key = newkey;
		tp->md5sig_info->keys6[tp->md5sig_info->entries6].base.keylen = newkeylen;
627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645

		tp->md5sig_info->entries6++;
	}
	return 0;
}

static int tcp_v6_md5_add_func(struct sock *sk, struct sock *addr_sk,
			       u8 *newkey, __u8 newkeylen)
{
	return tcp_v6_md5_do_add(sk, &inet6_sk(addr_sk)->daddr,
				 newkey, newkeylen);
}

static int tcp_v6_md5_do_del(struct sock *sk, struct in6_addr *peer)
{
	struct tcp_sock *tp = tcp_sk(sk);
	int i;

	for (i = 0; i < tp->md5sig_info->entries6; i++) {
646
		if (ipv6_addr_equal(&tp->md5sig_info->keys6[i].addr, peer)) {
647
			/* Free the key */
648
			kfree(tp->md5sig_info->keys6[i].base.key);
649 650 651 652 653
			tp->md5sig_info->entries6--;

			if (tp->md5sig_info->entries6 == 0) {
				kfree(tp->md5sig_info->keys6);
				tp->md5sig_info->keys6 = NULL;
654
				tp->md5sig_info->alloced6 = 0;
655 656 657 658 659 660 661 662
			} else {
				/* shrink the database */
				if (tp->md5sig_info->entries6 != i)
					memmove(&tp->md5sig_info->keys6[i],
						&tp->md5sig_info->keys6[i+1],
						(tp->md5sig_info->entries6 - i)
						* sizeof (tp->md5sig_info->keys6[0]));
			}
663 664
			tcp_free_md5sig_pool();
			return 0;
665 666 667 668 669 670 671 672 673 674 675 676
		}
	}
	return -ENOENT;
}

static void tcp_v6_clear_md5_list (struct sock *sk)
{
	struct tcp_sock *tp = tcp_sk(sk);
	int i;

	if (tp->md5sig_info->entries6) {
		for (i = 0; i < tp->md5sig_info->entries6; i++)
677
			kfree(tp->md5sig_info->keys6[i].base.key);
678 679 680 681 682 683 684 685 686 687
		tp->md5sig_info->entries6 = 0;
		tcp_free_md5sig_pool();
	}

	kfree(tp->md5sig_info->keys6);
	tp->md5sig_info->keys6 = NULL;
	tp->md5sig_info->alloced6 = 0;

	if (tp->md5sig_info->entries4) {
		for (i = 0; i < tp->md5sig_info->entries4; i++)
688
			kfree(tp->md5sig_info->keys4[i].base.key);
689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716
		tp->md5sig_info->entries4 = 0;
		tcp_free_md5sig_pool();
	}

	kfree(tp->md5sig_info->keys4);
	tp->md5sig_info->keys4 = NULL;
	tp->md5sig_info->alloced4 = 0;
}

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;
	u8 *newkey;

	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) {
		if (!tcp_sk(sk)->md5sig_info)
			return -ENOENT;
B
Brian Haley 已提交
717
		if (ipv6_addr_v4mapped(&sin6->sin6_addr))
718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733
			return tcp_v4_md5_do_del(sk, sin6->sin6_addr.s6_addr32[3]);
		return tcp_v6_md5_do_del(sk, &sin6->sin6_addr);
	}

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

	if (!tcp_sk(sk)->md5sig_info) {
		struct tcp_sock *tp = tcp_sk(sk);
		struct tcp_md5sig_info *p;

		p = kzalloc(sizeof(struct tcp_md5sig_info), GFP_KERNEL);
		if (!p)
			return -ENOMEM;

		tp->md5sig_info = p;
E
Eric Dumazet 已提交
734
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
735 736
	}

A
Arnaldo Carvalho de Melo 已提交
737
	newkey = kmemdup(cmd.tcpm_key, cmd.tcpm_keylen, GFP_KERNEL);
738 739
	if (!newkey)
		return -ENOMEM;
B
Brian Haley 已提交
740
	if (ipv6_addr_v4mapped(&sin6->sin6_addr)) {
741 742 743 744 745 746
		return tcp_v4_md5_do_add(sk, sin6->sin6_addr.s6_addr32[3],
					 newkey, cmd.tcpm_keylen);
	}
	return tcp_v6_md5_do_add(sk, &sin6->sin6_addr, newkey, cmd.tcpm_keylen);
}

747 748 749
static int tcp_v6_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					struct in6_addr *daddr,
					struct in6_addr *saddr, int nbytes)
750 751
{
	struct tcp6_pseudohdr *bp;
752
	struct scatterlist sg;
753

754 755 756 757
	bp = &hp->md5_blk.ip6;
	/* 1. TCP pseudo-header (RFC2460) */
	ipv6_addr_copy(&bp->saddr, saddr);
	ipv6_addr_copy(&bp->daddr, daddr);
758
	bp->protocol = cpu_to_be32(IPPROTO_TCP);
A
Adam Langley 已提交
759
	bp->len = cpu_to_be32(nbytes);
760

761 762 763
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}
764

765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
static int tcp_v6_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
			       struct in6_addr *daddr, struct in6_addr *saddr,
			       struct tcphdr *th)
{
	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))
786 787 788 789
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
790

791 792 793 794
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
795
	return 1;
796 797
}

798 799 800
static int tcp_v6_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
			       struct sock *sk, struct request_sock *req,
			       struct sk_buff *skb)
801 802
{
	struct in6_addr *saddr, *daddr;
803 804 805
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
	struct tcphdr *th = tcp_hdr(skb);
806 807 808 809

	if (sk) {
		saddr = &inet6_sk(sk)->saddr;
		daddr = &inet6_sk(sk)->daddr;
810
	} else if (req) {
811 812
		saddr = &inet6_rsk(req)->loc_addr;
		daddr = &inet6_rsk(req)->rmt_addr;
813 814 815 816
	} else {
		struct ipv6hdr *ip6h = ipv6_hdr(skb);
		saddr = &ip6h->saddr;
		daddr = &ip6h->daddr;
817
	}
818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845

	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;
846 847 848 849 850 851
}

static int tcp_v6_inbound_md5_hash (struct sock *sk, struct sk_buff *skb)
{
	__u8 *hash_location = NULL;
	struct tcp_md5sig_key *hash_expected;
852
	struct ipv6hdr *ip6h = ipv6_hdr(skb);
853
	struct tcphdr *th = tcp_hdr(skb);
854 855 856 857
	int genhash;
	u8 newhash[16];

	hash_expected = tcp_v6_md5_do_lookup(sk, &ip6h->saddr);
858
	hash_location = tcp_parse_md5sig_option(th);
859

860 861 862 863 864 865
	/* 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);
866 867 868
		return 1;
	}

869 870
	if (!hash_expected && hash_location) {
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
871 872 873 874
		return 1;
	}

	/* check the signature */
875 876 877 878
	genhash = tcp_v6_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);

879 880
	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
		if (net_ratelimit()) {
881
			printk(KERN_INFO "MD5 Hash %s for [%pI6c]:%u->[%pI6c]:%u\n",
882
			       genhash ? "failed" : "mismatch",
883 884
			       &ip6h->saddr, ntohs(th->source),
			       &ip6h->daddr, ntohs(th->dest));
885 886 887 888 889 890 891
		}
		return 1;
	}
	return 0;
}
#endif

892
struct request_sock_ops tcp6_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
893
	.family		=	AF_INET6,
894
	.obj_size	=	sizeof(struct tcp6_request_sock),
895
	.rtx_syn_ack	=	tcp_v6_rtx_synack,
896 897
	.send_ack	=	tcp_v6_reqsk_send_ack,
	.destructor	=	tcp_v6_reqsk_destructor,
898 899
	.send_reset	=	tcp_v6_send_reset,
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
900 901
};

902
#ifdef CONFIG_TCP_MD5SIG
903
static const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = {
904
	.md5_lookup	=	tcp_v6_reqsk_md5_lookup,
905
	.calc_md5_hash	=	tcp_v6_md5_hash_skb,
906
};
907
#endif
908

909 910
static void __tcp_v6_send_check(struct sk_buff *skb,
				struct in6_addr *saddr, struct in6_addr *daddr)
L
Linus Torvalds 已提交
911
{
912
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
913

914
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
915
		th->check = ~tcp_v6_check(skb->len, saddr, daddr, 0);
916
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
917
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
918
	} else {
919 920 921
		th->check = tcp_v6_check(skb->len, saddr, daddr,
					 csum_partial(th, th->doff << 2,
						      skb->csum));
L
Linus Torvalds 已提交
922 923 924
	}
}

925
static void tcp_v6_send_check(struct sock *sk, struct sk_buff *skb)
926 927 928 929 930 931
{
	struct ipv6_pinfo *np = inet6_sk(sk);

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

932 933 934 935 936 937 938 939
static int tcp_v6_gso_send_check(struct sk_buff *skb)
{
	struct ipv6hdr *ipv6h;
	struct tcphdr *th;

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

940
	ipv6h = ipv6_hdr(skb);
941
	th = tcp_hdr(skb);
942 943

	th->check = 0;
944
	skb->ip_summed = CHECKSUM_PARTIAL;
945
	__tcp_v6_send_check(skb, &ipv6h->saddr, &ipv6h->daddr);
946 947
	return 0;
}
L
Linus Torvalds 已提交
948

H
Herbert Xu 已提交
949 950
static struct sk_buff **tcp6_gro_receive(struct sk_buff **head,
					 struct sk_buff *skb)
H
Herbert Xu 已提交
951
{
H
Herbert Xu 已提交
952
	struct ipv6hdr *iph = skb_gro_network_header(skb);
H
Herbert Xu 已提交
953 954 955

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
956
		if (!tcp_v6_check(skb_gro_len(skb), &iph->saddr, &iph->daddr,
H
Herbert Xu 已提交
957 958 959 960 961 962 963 964 965 966 967 968 969 970
				  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
971
static int tcp6_gro_complete(struct sk_buff *skb)
H
Herbert Xu 已提交
972 973 974 975 976 977 978 979 980 981 982
{
	struct ipv6hdr *iph = ipv6_hdr(skb);
	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);
}

983 984
static void tcp_v6_send_response(struct sk_buff *skb, u32 seq, u32 ack, u32 win,
				 u32 ts, struct tcp_md5sig_key *key, int rst)
L
Linus Torvalds 已提交
985
{
986
	struct tcphdr *th = tcp_hdr(skb), *t1;
L
Linus Torvalds 已提交
987 988
	struct sk_buff *buff;
	struct flowi fl;
E
Eric Dumazet 已提交
989
	struct net *net = dev_net(skb_dst(skb)->dev);
990
	struct sock *ctl_sk = net->ipv6.tcp_sk;
991
	unsigned int tot_len = sizeof(struct tcphdr);
E
Eric Dumazet 已提交
992
	struct dst_entry *dst;
993
	__be32 *topt;
L
Linus Torvalds 已提交
994

995 996
	if (ts)
		tot_len += TCPOLEN_TSTAMP_ALIGNED;
997 998 999 1000 1001 1002
#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
1003
			 GFP_ATOMIC);
1004 1005
	if (buff == NULL)
		return;
L
Linus Torvalds 已提交
1006

1007
	skb_reserve(buff, MAX_HEADER + sizeof(struct ipv6hdr) + tot_len);
L
Linus Torvalds 已提交
1008

1009
	t1 = (struct tcphdr *) skb_push(buff, tot_len);
1010
	skb_reset_transport_header(buff);
L
Linus Torvalds 已提交
1011 1012 1013 1014 1015

	/* Swap the send and the receive. */
	memset(t1, 0, sizeof(*t1));
	t1->dest = th->source;
	t1->source = th->dest;
1016
	t1->doff = tot_len / 4;
1017 1018 1019 1020 1021
	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
1022

1023 1024
	topt = (__be32 *)(t1 + 1);

1025 1026 1027 1028 1029 1030 1031
	if (ts) {
		*topt++ = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				(TCPOPT_TIMESTAMP << 8) | TCPOLEN_TIMESTAMP);
		*topt++ = htonl(tcp_time_stamp);
		*topt++ = htonl(ts);
	}

1032 1033
#ifdef CONFIG_TCP_MD5SIG
	if (key) {
1034 1035 1036
		*topt++ = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				(TCPOPT_MD5SIG << 8) | TCPOLEN_MD5SIG);
		tcp_v6_md5_hash_hdr((__u8 *)topt, key,
1037 1038
				    &ipv6_hdr(skb)->saddr,
				    &ipv6_hdr(skb)->daddr, t1);
1039 1040 1041
	}
#endif

L
Linus Torvalds 已提交
1042
	memset(&fl, 0, sizeof(fl));
1043 1044
	ipv6_addr_copy(&fl.fl6_dst, &ipv6_hdr(skb)->saddr);
	ipv6_addr_copy(&fl.fl6_src, &ipv6_hdr(skb)->daddr);
L
Linus Torvalds 已提交
1045

1046 1047 1048
	buff->ip_summed = CHECKSUM_PARTIAL;
	buff->csum = 0;

1049
	__tcp_v6_send_check(buff, &fl.fl6_src, &fl.fl6_dst);
L
Linus Torvalds 已提交
1050 1051

	fl.proto = IPPROTO_TCP;
1052
	fl.oif = inet6_iif(skb);
L
Linus Torvalds 已提交
1053 1054
	fl.fl_ip_dport = t1->dest;
	fl.fl_ip_sport = t1->source;
V
Venkat Yekkirala 已提交
1055
	security_skb_classify_flow(skb, &fl);
L
Linus Torvalds 已提交
1056

1057 1058 1059 1060
	/* Pass a socket to ip6_dst_lookup either it is for RST
	 * Underlying function will use this to retrieve the network
	 * namespace
	 */
E
Eric Dumazet 已提交
1061 1062 1063
	if (!ip6_dst_lookup(ctl_sk, &dst, &fl)) {
		if (xfrm_lookup(net, &dst, &fl, NULL, 0) >= 0) {
			skb_dst_set(buff, dst);
1064
			ip6_xmit(ctl_sk, buff, &fl, NULL);
1065
			TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
1066 1067
			if (rst)
				TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
L
Linus Torvalds 已提交
1068
			return;
A
Arnaldo Carvalho de Melo 已提交
1069
		}
L
Linus Torvalds 已提交
1070 1071 1072 1073 1074
	}

	kfree_skb(buff);
}

1075
static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
1076
{
1077 1078
	struct tcphdr *th = tcp_hdr(skb);
	u32 seq = 0, ack_seq = 0;
1079
	struct tcp_md5sig_key *key = NULL;
L
Linus Torvalds 已提交
1080

1081
	if (th->rst)
L
Linus Torvalds 已提交
1082 1083
		return;

1084 1085
	if (!ipv6_unicast_destination(skb))
		return;
L
Linus Torvalds 已提交
1086

1087
#ifdef CONFIG_TCP_MD5SIG
1088 1089
	if (sk)
		key = tcp_v6_md5_do_lookup(sk, &ipv6_hdr(skb)->daddr);
1090 1091
#endif

1092 1093 1094 1095 1096
	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
1097

1098 1099
	tcp_v6_send_response(skb, seq, ack_seq, 0, 0, key, 1);
}
L
Linus Torvalds 已提交
1100

1101 1102 1103 1104
static void tcp_v6_send_ack(struct sk_buff *skb, u32 seq, u32 ack, u32 win, u32 ts,
			    struct tcp_md5sig_key *key)
{
	tcp_v6_send_response(skb, seq, ack, win, ts, key, 0);
L
Linus Torvalds 已提交
1105 1106 1107 1108
}

static void tcp_v6_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
1109
	struct inet_timewait_sock *tw = inet_twsk(sk);
1110
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
1111

1112
	tcp_v6_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
1113
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
1114
			tcptw->tw_ts_recent, tcp_twsk_md5_key(tcptw));
L
Linus Torvalds 已提交
1115

1116
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
1117 1118
}

1119 1120
static void tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
				  struct request_sock *req)
L
Linus Torvalds 已提交
1121
{
1122
	tcp_v6_send_ack(skb, tcp_rsk(req)->snt_isn + 1, tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd, req->ts_recent,
1123
			tcp_v6_md5_do_lookup(sk, &ipv6_hdr(skb)->daddr));
L
Linus Torvalds 已提交
1124 1125 1126 1127 1128
}


static struct sock *tcp_v6_hnd_req(struct sock *sk,struct sk_buff *skb)
{
1129
	struct request_sock *req, **prev;
1130
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1131 1132 1133
	struct sock *nsk;

	/* Find possible connection requests. */
1134
	req = inet6_csk_search_req(sk, &prev, th->source,
1135 1136
				   &ipv6_hdr(skb)->saddr,
				   &ipv6_hdr(skb)->daddr, inet6_iif(skb));
L
Linus Torvalds 已提交
1137 1138 1139
	if (req)
		return tcp_check_req(sk, skb, req, prev);

1140
	nsk = __inet6_lookup_established(sock_net(sk), &tcp_hashinfo,
1141 1142
			&ipv6_hdr(skb)->saddr, th->source,
			&ipv6_hdr(skb)->daddr, ntohs(th->dest), inet6_iif(skb));
L
Linus Torvalds 已提交
1143 1144 1145 1146 1147 1148

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1149
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1150 1151 1152
		return NULL;
	}

1153
#ifdef CONFIG_SYN_COOKIES
1154
	if (!th->syn)
1155
		sk = cookie_v6_check(sk, skb);
L
Linus Torvalds 已提交
1156 1157 1158 1159 1160 1161 1162 1163 1164
#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)
{
1165
	struct tcp_extend_values tmp_ext;
1166
	struct tcp_options_received tmp_opt;
1167
	u8 *hash_location;
1168
	struct request_sock *req;
1169
	struct inet6_request_sock *treq;
L
Linus Torvalds 已提交
1170 1171
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
1172
	__u32 isn = TCP_SKB_CB(skb)->when;
1173 1174 1175 1176 1177
#ifdef CONFIG_SYN_COOKIES
	int want_cookie = 0;
#else
#define want_cookie 0
#endif
L
Linus Torvalds 已提交
1178 1179 1180 1181 1182

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

	if (!ipv6_unicast_destination(skb))
1183
		goto drop;
L
Linus Torvalds 已提交
1184

1185
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
L
Linus Torvalds 已提交
1186
		if (net_ratelimit())
1187 1188 1189 1190 1191 1192
			syn_flood_warning(skb);
#ifdef CONFIG_SYN_COOKIES
		if (sysctl_tcp_syncookies)
			want_cookie = 1;
		else
#endif
1193
		goto drop;
L
Linus Torvalds 已提交
1194 1195
	}

1196
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1197 1198
		goto drop;

1199
	req = inet6_reqsk_alloc(&tcp6_request_sock_ops);
L
Linus Torvalds 已提交
1200 1201 1202
	if (req == NULL)
		goto drop;

1203 1204 1205 1206
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv6_ops;
#endif

L
Linus Torvalds 已提交
1207 1208 1209
	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;
1210
	tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226

	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 */
1227
		d = (__force u32 *)&ipv6_hdr(skb)->daddr.s6_addr32[0];
1228 1229 1230 1231
		*mess++ ^= *d++;
		*mess++ ^= *d++;
		*mess++ ^= *d++;
		*mess++ ^= *d++;
1232
		d = (__force u32 *)&ipv6_hdr(skb)->saddr.s6_addr32[0];
1233 1234 1235 1236 1237 1238 1239 1240 1241
		*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
1242

1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255
#ifdef CONFIG_SYN_COOKIES
		want_cookie = 0;	/* not our kind of cookie */
#endif
		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
1256

1257
	if (want_cookie && !tmp_opt.saw_tstamp)
1258 1259
		tcp_clear_options(&tmp_opt);

L
Linus Torvalds 已提交
1260 1261 1262
	tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;
	tcp_openreq_init(req, &tmp_opt, skb);

1263
	treq = inet6_rsk(req);
1264 1265
	ipv6_addr_copy(&treq->rmt_addr, &ipv6_hdr(skb)->saddr);
	ipv6_addr_copy(&treq->loc_addr, &ipv6_hdr(skb)->daddr);
1266
	if (!want_cookie || tmp_opt.tstamp_ok)
1267 1268
		TCP_ECN_create_request(req, tcp_hdr(skb));

1269
	if (!isn) {
1270 1271 1272 1273 1274 1275 1276
		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;
		}
		treq->iif = sk->sk_bound_dev_if;
L
Linus Torvalds 已提交
1277

1278 1279 1280 1281
		/* 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);
1282 1283 1284 1285 1286 1287
		if (!want_cookie) {
			isn = tcp_v6_init_sequence(skb);
		} else {
			isn = cookie_v6_init_sequence(sk, skb, &req->mss);
			req->cookie_ts = tmp_opt.tstamp_ok;
		}
1288
	}
1289
	tcp_rsk(req)->snt_isn = isn;
L
Linus Torvalds 已提交
1290

1291 1292
	security_inet_conn_request(sk, skb, req);

1293 1294 1295
	if (tcp_v6_send_synack(sk, req,
			       (struct request_values *)&tmp_ext) ||
	    want_cookie)
1296
		goto drop_and_free;
L
Linus Torvalds 已提交
1297

1298 1299
	inet6_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
	return 0;
L
Linus Torvalds 已提交
1300

1301 1302
drop_and_free:
	reqsk_free(req);
L
Linus Torvalds 已提交
1303 1304 1305 1306 1307
drop:
	return 0; /* don't send reset */
}

static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
1308
					  struct request_sock *req,
L
Linus Torvalds 已提交
1309 1310
					  struct dst_entry *dst)
{
1311
	struct inet6_request_sock *treq;
L
Linus Torvalds 已提交
1312 1313 1314 1315 1316 1317
	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;
1318 1319 1320
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
L
Linus Torvalds 已提交
1321 1322 1323 1324 1325 1326 1327 1328

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

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

1329
		if (newsk == NULL)
L
Linus Torvalds 已提交
1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340
			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
1341
		ipv6_addr_set_v4mapped(newinet->inet_daddr, &newnp->daddr);
L
Linus Torvalds 已提交
1342

E
Eric Dumazet 已提交
1343
		ipv6_addr_set_v4mapped(newinet->inet_saddr, &newnp->saddr);
L
Linus Torvalds 已提交
1344 1345 1346

		ipv6_addr_copy(&newnp->rcv_saddr, &newnp->saddr);

1347
		inet_csk(newsk)->icsk_af_ops = &ipv6_mapped;
L
Linus Torvalds 已提交
1348
		newsk->sk_backlog_rcv = tcp_v4_do_rcv;
1349 1350 1351 1352
#ifdef CONFIG_TCP_MD5SIG
		newtp->af_specific = &tcp_sock_ipv6_mapped_specific;
#endif

L
Linus Torvalds 已提交
1353 1354
		newnp->pktoptions  = NULL;
		newnp->opt	   = NULL;
1355
		newnp->mcast_oif   = inet6_iif(skb);
1356
		newnp->mcast_hops  = ipv6_hdr(skb)->hop_limit;
L
Linus Torvalds 已提交
1357

1358 1359 1360 1361
		/*
		 * 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
1362 1363 1364
		 */

		/* It is tricky place. Until this moment IPv4 tcp
1365
		   worked with IPv6 icsk.icsk_af_ops.
L
Linus Torvalds 已提交
1366 1367
		   Sync it now.
		 */
1368
		tcp_sync_mss(newsk, inet_csk(newsk)->icsk_pmtu_cookie);
L
Linus Torvalds 已提交
1369 1370 1371 1372

		return newsk;
	}

1373
	treq = inet6_rsk(req);
L
Linus Torvalds 已提交
1374 1375 1376 1377 1378 1379
	opt = np->opt;

	if (sk_acceptq_is_full(sk))
		goto out_overflow;

	if (dst == NULL) {
1380
		struct in6_addr *final_p, final;
L
Linus Torvalds 已提交
1381 1382 1383 1384
		struct flowi fl;

		memset(&fl, 0, sizeof(fl));
		fl.proto = IPPROTO_TCP;
1385
		ipv6_addr_copy(&fl.fl6_dst, &treq->rmt_addr);
1386
		final_p = fl6_update_dst(&fl, opt, &final);
1387
		ipv6_addr_copy(&fl.fl6_src, &treq->loc_addr);
L
Linus Torvalds 已提交
1388
		fl.oif = sk->sk_bound_dev_if;
B
Brian Haley 已提交
1389
		fl.mark = sk->sk_mark;
1390
		fl.fl_ip_dport = inet_rsk(req)->rmt_port;
1391
		fl.fl_ip_sport = inet_rsk(req)->loc_port;
1392
		security_req_classify_flow(req, &fl);
L
Linus Torvalds 已提交
1393 1394 1395 1396 1397 1398 1399

		if (ip6_dst_lookup(sk, &dst, &fl))
			goto out;

		if (final_p)
			ipv6_addr_copy(&fl.fl6_dst, final_p);

A
Alexey Dobriyan 已提交
1400
		if ((xfrm_lookup(sock_net(sk), &dst, &fl, sk, 0)) < 0)
L
Linus Torvalds 已提交
1401
			goto out;
1402
	}
L
Linus Torvalds 已提交
1403 1404 1405

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (newsk == NULL)
1406
		goto out_nonewsk;
L
Linus Torvalds 已提交
1407

1408 1409 1410 1411 1412
	/*
	 * 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
1413

1414
	newsk->sk_gso_type = SKB_GSO_TCPV6;
1415
	__ip6_dst_store(newsk, dst, NULL, NULL);
L
Linus Torvalds 已提交
1416 1417 1418 1419 1420 1421 1422 1423 1424 1425

	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));

1426 1427 1428 1429
	ipv6_addr_copy(&newnp->daddr, &treq->rmt_addr);
	ipv6_addr_copy(&newnp->saddr, &treq->loc_addr);
	ipv6_addr_copy(&newnp->rcv_saddr, &treq->loc_addr);
	newsk->sk_bound_dev_if = treq->iif;
L
Linus Torvalds 已提交
1430

1431
	/* Now IPv6 options...
L
Linus Torvalds 已提交
1432 1433 1434 1435

	   First: no IPv4 options.
	 */
	newinet->opt = NULL;
1436
	newnp->ipv6_fl_list = NULL;
L
Linus Torvalds 已提交
1437 1438 1439 1440 1441 1442

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

	/* Clone pktoptions received with SYN */
	newnp->pktoptions = NULL;
1443 1444 1445 1446
	if (treq->pktopts != NULL) {
		newnp->pktoptions = skb_clone(treq->pktopts, GFP_ATOMIC);
		kfree_skb(treq->pktopts);
		treq->pktopts = NULL;
L
Linus Torvalds 已提交
1447 1448 1449 1450
		if (newnp->pktoptions)
			skb_set_owner_r(newnp->pktoptions, newsk);
	}
	newnp->opt	  = NULL;
1451
	newnp->mcast_oif  = inet6_iif(skb);
1452
	newnp->mcast_hops = ipv6_hdr(skb)->hop_limit;
L
Linus Torvalds 已提交
1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465

	/* 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);
	}

1466
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
1467
	if (newnp->opt)
1468 1469
		inet_csk(newsk)->icsk_ext_hdr_len = (newnp->opt->opt_nflen +
						     newnp->opt->opt_flen);
L
Linus Torvalds 已提交
1470

J
John Heffner 已提交
1471
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1472 1473 1474 1475
	tcp_sync_mss(newsk, dst_mtu(dst));
	newtp->advmss = dst_metric(dst, RTAX_ADVMSS);
	tcp_initialize_rcv_mss(newsk);

E
Eric Dumazet 已提交
1476 1477
	newinet->inet_daddr = newinet->inet_saddr = LOOPBACK4_IPV6;
	newinet->inet_rcv_saddr = LOOPBACK4_IPV6;
L
Linus Torvalds 已提交
1478

1479 1480 1481 1482 1483 1484 1485 1486
#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.
		 */
A
Arnaldo Carvalho de Melo 已提交
1487 1488
		char *newkey = kmemdup(key->key, key->keylen, GFP_ATOMIC);
		if (newkey != NULL)
1489
			tcp_v6_md5_do_add(newsk, &newnp->daddr,
1490 1491 1492 1493
					  newkey, key->keylen);
	}
#endif

1494 1495 1496 1497
	if (__inet_inherit_port(sk, newsk) < 0) {
		sock_put(newsk);
		goto out;
	}
1498
	__inet6_hash(newsk, NULL);
L
Linus Torvalds 已提交
1499 1500 1501 1502

	return newsk;

out_overflow:
1503
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1504
out_nonewsk:
L
Linus Torvalds 已提交
1505 1506 1507
	if (opt && opt != np->opt)
		sock_kfree_s(sk, opt, opt->tot_len);
	dst_release(dst);
1508 1509
out:
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1510 1511 1512
	return NULL;
}

1513
static __sum16 tcp_v6_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1514
{
1515
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
H
Herbert Xu 已提交
1516
		if (!tcp_v6_check(skb->len, &ipv6_hdr(skb)->saddr,
1517
				  &ipv6_hdr(skb)->daddr, skb->csum)) {
1518
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1519
			return 0;
1520
		}
L
Linus Torvalds 已提交
1521
	}
1522

H
Herbert Xu 已提交
1523
	skb->csum = ~csum_unfold(tcp_v6_check(skb->len,
1524 1525
					      &ipv6_hdr(skb)->saddr,
					      &ipv6_hdr(skb)->daddr, 0));
1526

L
Linus Torvalds 已提交
1527
	if (skb->len <= 76) {
1528
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557
	}
	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);

1558 1559 1560 1561 1562
#ifdef CONFIG_TCP_MD5SIG
	if (tcp_v6_inbound_md5_hash (sk, skb))
		goto discard;
#endif

1563
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581
		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.
1582
					       --ANK (980728)
L
Linus Torvalds 已提交
1583 1584 1585 1586 1587 1588
	 */
	if (np->rxopt.all)
		opt_skb = skb_clone(skb, GFP_ATOMIC);

	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
		TCP_CHECK_TIMER(sk);
1589
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len))
L
Linus Torvalds 已提交
1590 1591 1592 1593 1594 1595 1596
			goto reset;
		TCP_CHECK_TIMER(sk);
		if (opt_skb)
			goto ipv6_pktoptions;
		return 0;
	}

1597
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1598 1599
		goto csum_err;

1600
	if (sk->sk_state == TCP_LISTEN) {
L
Linus Torvalds 已提交
1601 1602 1603 1604 1605 1606 1607 1608 1609
		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..
		 */
1610
		if(nsk != sk) {
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616 1617 1618 1619
			if (tcp_child_process(sk, nsk, skb))
				goto reset;
			if (opt_skb)
				__kfree_skb(opt_skb);
			return 0;
		}
	}

	TCP_CHECK_TIMER(sk);
1620
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len))
L
Linus Torvalds 已提交
1621 1622 1623 1624 1625 1626 1627
		goto reset;
	TCP_CHECK_TIMER(sk);
	if (opt_skb)
		goto ipv6_pktoptions;
	return 0;

reset:
1628
	tcp_v6_send_reset(sk, skb);
L
Linus Torvalds 已提交
1629 1630 1631 1632 1633 1634
discard:
	if (opt_skb)
		__kfree_skb(opt_skb);
	kfree_skb(skb);
	return 0;
csum_err:
1635
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649
	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))) {
1650
		if (np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo)
1651
			np->mcast_oif = inet6_iif(opt_skb);
1652
		if (np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim)
1653
			np->mcast_hops = ipv6_hdr(opt_skb)->hop_limit;
L
Linus Torvalds 已提交
1654 1655 1656 1657 1658 1659 1660 1661 1662
		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);
		}
	}

1663
	kfree_skb(opt_skb);
L
Linus Torvalds 已提交
1664 1665 1666
	return 0;
}

1667
static int tcp_v6_rcv(struct sk_buff *skb)
L
Linus Torvalds 已提交
1668
{
1669
	struct tcphdr *th;
1670
	struct ipv6hdr *hdr;
L
Linus Torvalds 已提交
1671 1672
	struct sock *sk;
	int ret;
1673
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1674 1675 1676 1677 1678 1679 1680

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

	/*
	 *	Count it even if it's bad.
	 */
1681
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1682 1683 1684 1685

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

1686
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1687 1688 1689 1690 1691 1692

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

1693
	if (!skb_csum_unnecessary(skb) && tcp_v6_checksum_init(skb))
L
Linus Torvalds 已提交
1694 1695
		goto bad_packet;

1696
	th = tcp_hdr(skb);
1697
	hdr = ipv6_hdr(skb);
L
Linus Torvalds 已提交
1698 1699 1700 1701 1702
	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;
1703
	TCP_SKB_CB(skb)->flags = ipv6_get_dsfield(hdr);
L
Linus Torvalds 已提交
1704 1705
	TCP_SKB_CB(skb)->sacked = 0;

1706
	sk = __inet6_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1707 1708 1709 1710 1711 1712 1713
	if (!sk)
		goto no_tcp_socket;

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

1714 1715 1716 1717 1718
	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
1719 1720 1721
	if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;

1722
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1723 1724 1725 1726
		goto discard_and_relse;

	skb->dev = NULL;

1727
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1728 1729
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1730
#ifdef CONFIG_NET_DMA
1731
		struct tcp_sock *tp = tcp_sk(sk);
1732
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1733
			tp->ucopy.dma_chan = dma_find_channel(DMA_MEMCPY);
1734 1735 1736
		if (tp->ucopy.dma_chan)
			ret = tcp_v6_do_rcv(sk, skb);
		else
C
Chris Leech 已提交
1737 1738 1739 1740 1741
#endif
		{
			if (!tcp_prequeue(sk, skb))
				ret = tcp_v6_do_rcv(sk, skb);
		}
1742
	} else if (unlikely(sk_add_backlog(sk, skb))) {
Z
Zhu Yi 已提交
1743
		bh_unlock_sock(sk);
1744
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1745 1746
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
	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:
1758
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1759
	} else {
1760
		tcp_v6_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
	}

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)) {
1778
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1779 1780 1781 1782
		goto discard_it;
	}

	if (skb->len < (th->doff<<2) || tcp_checksum_complete(skb)) {
1783
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1784
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1785 1786 1787
		goto discard_it;
	}

1788
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1789 1790 1791 1792
	case TCP_TW_SYN:
	{
		struct sock *sk2;

1793
		sk2 = inet6_lookup_listener(dev_net(skb->dev), &tcp_hashinfo,
1794
					    &ipv6_hdr(skb)->daddr,
1795
					    ntohs(th->dest), inet6_iif(skb));
L
Linus Torvalds 已提交
1796
		if (sk2 != NULL) {
1797 1798 1799
			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
1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814
			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;
}

1815 1816 1817 1818 1819 1820 1821
static struct inet_peer *tcp_v6_get_peer(struct sock *sk, bool *release_it)
{
	/* Alas, not yet... */
	return NULL;
}

static void *tcp_v6_tw_get_peer(struct sock *sk)
L
Linus Torvalds 已提交
1822
{
1823 1824 1825 1826 1827
	struct inet_timewait_sock *tw = inet_twsk(sk);

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

L
Linus Torvalds 已提交
1828
	/* Alas, not yet... */
1829
	return NULL;
L
Linus Torvalds 已提交
1830 1831
}

1832 1833 1834 1835 1836 1837 1838
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,
};

1839
static const struct inet_connection_sock_af_ops ipv6_specific = {
1840 1841 1842 1843 1844
	.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,
1845
	.get_peer	   = tcp_v6_get_peer,
1846 1847 1848 1849 1850
	.net_header_len	   = sizeof(struct ipv6hdr),
	.setsockopt	   = ipv6_setsockopt,
	.getsockopt	   = ipv6_getsockopt,
	.addr2sockaddr	   = inet6_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in6),
1851
	.bind_conflict	   = inet6_csk_bind_conflict,
1852
#ifdef CONFIG_COMPAT
1853 1854
	.compat_setsockopt = compat_ipv6_setsockopt,
	.compat_getsockopt = compat_ipv6_getsockopt,
1855
#endif
L
Linus Torvalds 已提交
1856 1857
};

1858
#ifdef CONFIG_TCP_MD5SIG
1859
static const struct tcp_sock_af_ops tcp_sock_ipv6_specific = {
1860
	.md5_lookup	=	tcp_v6_md5_lookup,
1861
	.calc_md5_hash	=	tcp_v6_md5_hash_skb,
1862 1863 1864
	.md5_add	=	tcp_v6_md5_add_func,
	.md5_parse	=	tcp_v6_parse_md5_keys,
};
1865
#endif
1866

L
Linus Torvalds 已提交
1867 1868 1869 1870
/*
 *	TCP over IPv4 via INET6 API
 */

1871
static const struct inet_connection_sock_af_ops ipv6_mapped = {
1872 1873 1874 1875 1876
	.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,
1877
	.get_peer	   = tcp_v4_get_peer,
1878 1879 1880 1881 1882
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ipv6_setsockopt,
	.getsockopt	   = ipv6_getsockopt,
	.addr2sockaddr	   = inet6_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in6),
1883
	.bind_conflict	   = inet6_csk_bind_conflict,
1884
#ifdef CONFIG_COMPAT
1885 1886
	.compat_setsockopt = compat_ipv6_setsockopt,
	.compat_getsockopt = compat_ipv6_getsockopt,
1887
#endif
L
Linus Torvalds 已提交
1888 1889
};

1890
#ifdef CONFIG_TCP_MD5SIG
1891
static const struct tcp_sock_af_ops tcp_sock_ipv6_mapped_specific = {
1892
	.md5_lookup	=	tcp_v4_md5_lookup,
1893
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1894 1895 1896
	.md5_add	=	tcp_v6_md5_add_func,
	.md5_parse	=	tcp_v6_parse_md5_keys,
};
1897
#endif
1898

L
Linus Torvalds 已提交
1899 1900 1901 1902 1903
/* 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)
{
1904
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1905 1906 1907 1908 1909 1910
	struct tcp_sock *tp = tcp_sk(sk);

	skb_queue_head_init(&tp->out_of_order_queue);
	tcp_init_xmit_timers(sk);
	tcp_prequeue_init(tp);

1911
	icsk->icsk_rto = TCP_TIMEOUT_INIT;
L
Linus Torvalds 已提交
1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923
	tp->mdev = TCP_TIMEOUT_INIT;

	/* So many TCP implementations out there (incorrectly) count the
	 * initial SYN frame in their delayed-ACK and congestion control
	 * algorithms that we must have the following bandaid to talk
	 * efficiently to them.  -DaveM
	 */
	tp->snd_cwnd = 2;

	/* See draft-stevens-tcpca-spec-01 for discussion of the
	 * initialization of these values.
	 */
I
Ilpo Järvinen 已提交
1924
	tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
L
Linus Torvalds 已提交
1925
	tp->snd_cwnd_clamp = ~0;
1926
	tp->mss_cache = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
1927 1928 1929 1930 1931

	tp->reordering = sysctl_tcp_reordering;

	sk->sk_state = TCP_CLOSE;

1932
	icsk->icsk_af_ops = &ipv6_specific;
1933
	icsk->icsk_ca_ops = &tcp_init_congestion_ops;
1934
	icsk->icsk_sync_mss = tcp_sync_mss;
L
Linus Torvalds 已提交
1935 1936 1937
	sk->sk_write_space = sk_stream_write_space;
	sock_set_flag(sk, SOCK_USE_WRITE_QUEUE);

1938 1939 1940 1941
#ifdef CONFIG_TCP_MD5SIG
	tp->af_specific = &tcp_sock_ipv6_specific;
#endif

1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954
	/* TCP Cookie Transactions */
	if (sysctl_tcp_cookie_size > 0) {
		/* Default, cookies without s_data_payload. */
		tp->cookie_values =
			kzalloc(sizeof(*tp->cookie_values),
				sk->sk_allocation);
		if (tp->cookie_values != NULL)
			kref_init(&tp->cookie_values->kref);
	}
	/* Presumed zeroed, in order of appearance:
	 *	cookie_in_always, cookie_out_never,
	 *	s_data_constant, s_data_in, s_data_out
	 */
L
Linus Torvalds 已提交
1955 1956 1957
	sk->sk_sndbuf = sysctl_tcp_wmem[1];
	sk->sk_rcvbuf = sysctl_tcp_rmem[1];

H
Herbert Xu 已提交
1958
	local_bh_disable();
1959
	percpu_counter_inc(&tcp_sockets_allocated);
H
Herbert Xu 已提交
1960
	local_bh_enable();
L
Linus Torvalds 已提交
1961 1962 1963 1964

	return 0;
}

1965
static void tcp_v6_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1966
{
1967 1968 1969 1970 1971
#ifdef CONFIG_TCP_MD5SIG
	/* Clean up the MD5 key list */
	if (tcp_sk(sk)->md5sig_info)
		tcp_v6_clear_md5_list(sk);
#endif
L
Linus Torvalds 已提交
1972
	tcp_v4_destroy_sock(sk);
1973
	inet6_destroy_sock(sk);
L
Linus Torvalds 已提交
1974 1975
}

1976
#ifdef CONFIG_PROC_FS
L
Linus Torvalds 已提交
1977
/* Proc filesystem TCPv6 sock list dumping. */
1978
static void get_openreq6(struct seq_file *seq,
1979
			 struct sock *sk, struct request_sock *req, int i, int uid)
L
Linus Torvalds 已提交
1980 1981
{
	int ttd = req->expires - jiffies;
1982 1983
	struct in6_addr *src = &inet6_rsk(req)->loc_addr;
	struct in6_addr *dest = &inet6_rsk(req)->rmt_addr;
L
Linus Torvalds 已提交
1984 1985 1986 1987 1988 1989 1990 1991 1992 1993

	if (ttd < 0)
		ttd = 0;

	seq_printf(seq,
		   "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X "
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %p\n",
		   i,
		   src->s6_addr32[0], src->s6_addr32[1],
		   src->s6_addr32[2], src->s6_addr32[3],
1994
		   ntohs(inet_rsk(req)->loc_port),
L
Linus Torvalds 已提交
1995 1996
		   dest->s6_addr32[0], dest->s6_addr32[1],
		   dest->s6_addr32[2], dest->s6_addr32[3],
1997
		   ntohs(inet_rsk(req)->rmt_port),
L
Linus Torvalds 已提交
1998 1999
		   TCP_SYN_RECV,
		   0,0, /* could print option size, but that is af dependent. */
2000 2001
		   1,   /* timers active (only the expire timer) */
		   jiffies_to_clock_t(ttd),
L
Linus Torvalds 已提交
2002 2003
		   req->retrans,
		   uid,
2004
		   0,  /* non standard timer */
L
Linus Torvalds 已提交
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
		   0, /* open_requests have no inode */
		   0, req);
}

static void get_tcp6_sock(struct seq_file *seq, struct sock *sp, int i)
{
	struct in6_addr *dest, *src;
	__u16 destp, srcp;
	int timer_active;
	unsigned long timer_expires;
	struct inet_sock *inet = inet_sk(sp);
	struct tcp_sock *tp = tcp_sk(sp);
2017
	const struct inet_connection_sock *icsk = inet_csk(sp);
L
Linus Torvalds 已提交
2018 2019 2020 2021
	struct ipv6_pinfo *np = inet6_sk(sp);

	dest  = &np->daddr;
	src   = &np->rcv_saddr;
E
Eric Dumazet 已提交
2022 2023
	destp = ntohs(inet->inet_dport);
	srcp  = ntohs(inet->inet_sport);
2024 2025

	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
2026
		timer_active	= 1;
2027 2028
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2029
		timer_active	= 4;
2030
		timer_expires	= icsk->icsk_timeout;
L
Linus Torvalds 已提交
2031 2032 2033 2034 2035 2036 2037 2038 2039 2040
	} 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 "
2041
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %lu %lu %u %u %d\n",
L
Linus Torvalds 已提交
2042 2043 2044 2045 2046
		   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,
2047
		   sp->sk_state,
2048 2049
		   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
2050 2051
		   timer_active,
		   jiffies_to_clock_t(timer_expires - jiffies),
2052
		   icsk->icsk_retransmits,
L
Linus Torvalds 已提交
2053
		   sock_i_uid(sp),
2054
		   icsk->icsk_probes_out,
L
Linus Torvalds 已提交
2055 2056
		   sock_i_ino(sp),
		   atomic_read(&sp->sk_refcnt), sp,
2057 2058
		   jiffies_to_clock_t(icsk->icsk_rto),
		   jiffies_to_clock_t(icsk->icsk_ack.ato),
2059
		   (icsk->icsk_ack.quick << 1 ) | icsk->icsk_ack.pingpong,
I
Ilpo Järvinen 已提交
2060 2061
		   tp->snd_cwnd,
		   tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh
L
Linus Torvalds 已提交
2062 2063 2064
		   );
}

2065
static void get_timewait6_sock(struct seq_file *seq,
2066
			       struct inet_timewait_sock *tw, int i)
L
Linus Torvalds 已提交
2067 2068 2069
{
	struct in6_addr *dest, *src;
	__u16 destp, srcp;
2070
	struct inet6_timewait_sock *tw6 = inet6_twsk((struct sock *)tw);
L
Linus Torvalds 已提交
2071 2072 2073 2074 2075
	int ttd = tw->tw_ttd - jiffies;

	if (ttd < 0)
		ttd = 0;

2076 2077
	dest = &tw6->tw_v6_daddr;
	src  = &tw6->tw_v6_rcv_saddr;
L
Linus Torvalds 已提交
2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127
	destp = ntohs(tw->tw_dport);
	srcp  = ntohs(tw->tw_sport);

	seq_printf(seq,
		   "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X "
		   "%02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %p\n",
		   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;
}

static struct tcp_seq_afinfo tcp6_seq_afinfo = {
	.name		= "tcp6",
	.family		= AF_INET6,
2128 2129 2130
	.seq_fops	= {
		.owner		= THIS_MODULE,
	},
2131 2132 2133
	.seq_ops	= {
		.show		= tcp6_seq_show,
	},
L
Linus Torvalds 已提交
2134 2135
};

2136
int __net_init tcp6_proc_init(struct net *net)
L
Linus Torvalds 已提交
2137
{
2138
	return tcp_proc_register(net, &tcp6_seq_afinfo);
L
Linus Torvalds 已提交
2139 2140
}

2141
void tcp6_proc_exit(struct net *net)
L
Linus Torvalds 已提交
2142
{
2143
	tcp_proc_unregister(net, &tcp6_seq_afinfo);
L
Linus Torvalds 已提交
2144 2145 2146 2147 2148 2149 2150 2151 2152
}
#endif

struct proto tcpv6_prot = {
	.name			= "TCPv6",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v6_connect,
	.disconnect		= tcp_disconnect,
2153
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2154 2155 2156 2157 2158 2159 2160
	.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,
2161 2162
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2163 2164
	.backlog_rcv		= tcp_v6_do_rcv,
	.hash			= tcp_v6_hash,
2165 2166
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2167 2168 2169 2170
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
2171
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2172 2173 2174 2175 2176
	.sysctl_mem		= sysctl_tcp_mem,
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp6_sock),
2177
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2178
	.twsk_prot		= &tcp6_timewait_sock_ops,
2179
	.rsk_prot		= &tcp6_request_sock_ops,
2180
	.h.hashinfo		= &tcp_hashinfo,
2181
	.no_autobind		= true,
2182 2183 2184 2185
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
L
Linus Torvalds 已提交
2186 2187
};

2188
static const struct inet6_protocol tcpv6_protocol = {
L
Linus Torvalds 已提交
2189 2190
	.handler	=	tcp_v6_rcv,
	.err_handler	=	tcp_v6_err,
2191
	.gso_send_check	=	tcp_v6_gso_send_check,
H
Herbert Xu 已提交
2192
	.gso_segment	=	tcp_tso_segment,
H
Herbert Xu 已提交
2193 2194
	.gro_receive	=	tcp6_gro_receive,
	.gro_complete	=	tcp6_gro_complete,
L
Linus Torvalds 已提交
2195 2196 2197 2198 2199 2200 2201 2202 2203
	.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,
2204 2205
	.flags		=	INET_PROTOSW_PERMANENT |
				INET_PROTOSW_ICSK,
L
Linus Torvalds 已提交
2206 2207
};

2208
static int __net_init tcpv6_net_init(struct net *net)
2209
{
2210 2211
	return inet_ctl_sock_create(&net->ipv6.tcp_sk, PF_INET6,
				    SOCK_RAW, IPPROTO_TCP, net);
2212 2213
}

2214
static void __net_exit tcpv6_net_exit(struct net *net)
2215
{
2216
	inet_ctl_sock_destroy(net->ipv6.tcp_sk);
E
Eric W. Biederman 已提交
2217 2218
}

2219
static void __net_exit tcpv6_net_exit_batch(struct list_head *net_exit_list)
E
Eric W. Biederman 已提交
2220 2221
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET6);
2222 2223 2224
}

static struct pernet_operations tcpv6_net_ops = {
E
Eric W. Biederman 已提交
2225 2226 2227
	.init	    = tcpv6_net_init,
	.exit	    = tcpv6_net_exit,
	.exit_batch = tcpv6_net_exit_batch,
2228 2229
};

2230
int __init tcpv6_init(void)
L
Linus Torvalds 已提交
2231
{
2232 2233 2234 2235 2236 2237
	int ret;

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

L
Linus Torvalds 已提交
2238
	/* register inet6 protocol */
2239 2240 2241 2242
	ret = inet6_register_protosw(&tcpv6_protosw);
	if (ret)
		goto out_tcpv6_protocol;

2243
	ret = register_pernet_subsys(&tcpv6_net_ops);
2244 2245 2246 2247
	if (ret)
		goto out_tcpv6_protosw;
out:
	return ret;
2248

2249 2250 2251 2252 2253 2254 2255
out_tcpv6_protocol:
	inet6_del_protocol(&tcpv6_protocol, IPPROTO_TCP);
out_tcpv6_protosw:
	inet6_unregister_protosw(&tcpv6_protosw);
	goto out;
}

2256
void tcpv6_exit(void)
2257
{
2258
	unregister_pernet_subsys(&tcpv6_net_ops);
2259 2260
	inet6_unregister_protosw(&tcpv6_protosw);
	inet6_del_protocol(&tcpv6_protocol, IPPROTO_TCP);
L
Linus Torvalds 已提交
2261
}