tcp_ipv4.c 66.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Implementation of the Transmission Control Protocol(TCP).
 *
 *		IPv4 specific functions
 *
 *
 *		code split from:
 *		linux/ipv4/tcp.c
 *		linux/ipv4/tcp_input.c
 *		linux/ipv4/tcp_output.c
 *
 *		See tcp.c for author information
 *
 *	This program is free software; you can redistribute it and/or
 *      modify it under the terms of the GNU General Public License
 *      as published by the Free Software Foundation; either version
 *      2 of the License, or (at your option) any later version.
 */

/*
 * Changes:
 *		David S. Miller	:	New socket lookup architecture.
 *					This code is dedicated to John Dyson.
 *		David S. Miller :	Change semantics of established hash,
 *					half is devoted to TIME_WAIT sockets
 *					and the rest go in the other half.
 *		Andi Kleen :		Add support for syncookies and fixed
 *					some bugs: ip options weren't passed to
 *					the TCP layer, missed a check for an
 *					ACK bit.
 *		Andi Kleen :		Implemented fast path mtu discovery.
 *	     				Fixed many serious bugs in the
37
 *					request_sock handling and moved
L
Linus Torvalds 已提交
38 39
 *					most of it into the af independent code.
 *					Added tail drop and some other bugfixes.
S
Stephen Hemminger 已提交
40
 *					Added new listen semantics.
L
Linus Torvalds 已提交
41 42 43 44 45 46 47 48 49 50 51 52
 *		Mike McLagan	:	Routing by source
 *	Juan Jose Ciarlante:		ip_dynaddr bits
 *		Andi Kleen:		various fixes.
 *	Vitaly E. Lavrov	:	Transparent proxy revived after year
 *					coma.
 *	Andi Kleen		:	Fix new listen.
 *	Andi Kleen		:	Fix accept error reporting.
 *	YOSHIFUJI Hideaki @USAGI and:	Support IPV6_V6ONLY socket option, which
 *	Alexey Kuznetsov		allow both IPv4 and IPv6 sockets to bind
 *					a single port at the same time.
 */

53
#define pr_fmt(fmt) "TCP: " fmt
L
Linus Torvalds 已提交
54

H
Herbert Xu 已提交
55
#include <linux/bottom_half.h>
L
Linus Torvalds 已提交
56 57 58 59 60 61 62 63
#include <linux/types.h>
#include <linux/fcntl.h>
#include <linux/module.h>
#include <linux/random.h>
#include <linux/cache.h>
#include <linux/jhash.h>
#include <linux/init.h>
#include <linux/times.h>
64
#include <linux/slab.h>
L
Linus Torvalds 已提交
65

66
#include <net/net_namespace.h>
L
Linus Torvalds 已提交
67
#include <net/icmp.h>
68
#include <net/inet_hashtables.h>
L
Linus Torvalds 已提交
69
#include <net/tcp.h>
70
#include <net/transp_v6.h>
L
Linus Torvalds 已提交
71 72
#include <net/ipv6.h>
#include <net/inet_common.h>
73
#include <net/timewait_sock.h>
L
Linus Torvalds 已提交
74
#include <net/xfrm.h>
C
Chris Leech 已提交
75
#include <net/netdma.h>
76
#include <net/secure_seq.h>
G
Glauber Costa 已提交
77
#include <net/tcp_memcontrol.h>
78
#include <net/busy_poll.h>
L
Linus Torvalds 已提交
79 80 81 82 83 84 85

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

86 87 88
#include <linux/crypto.h>
#include <linux/scatterlist.h>

89 90
int sysctl_tcp_tw_reuse __read_mostly;
int sysctl_tcp_low_latency __read_mostly;
E
Eric Dumazet 已提交
91
EXPORT_SYMBOL(sysctl_tcp_low_latency);
L
Linus Torvalds 已提交
92 93


94
#ifdef CONFIG_TCP_MD5SIG
E
Eric Dumazet 已提交
95
static int tcp_v4_md5_hash_hdr(char *md5_hash, const struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
96
			       __be32 daddr, __be32 saddr, const struct tcphdr *th);
97 98
#endif

99
struct inet_hashinfo tcp_hashinfo;
E
Eric Dumazet 已提交
100
EXPORT_SYMBOL(tcp_hashinfo);
L
Linus Torvalds 已提交
101

102
static inline __u32 tcp_v4_init_sequence(const struct sk_buff *skb)
L
Linus Torvalds 已提交
103
{
104 105
	return secure_tcp_sequence_number(ip_hdr(skb)->daddr,
					  ip_hdr(skb)->saddr,
106 107
					  tcp_hdr(skb)->dest,
					  tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
108 109
}

110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
int tcp_twsk_unique(struct sock *sk, struct sock *sktw, void *twp)
{
	const struct tcp_timewait_sock *tcptw = tcp_twsk(sktw);
	struct tcp_sock *tp = tcp_sk(sk);

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

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

	   If TW bucket has been already destroyed we fall back to VJ's scheme
	   and use initial timestamp retrieved from peer table.
	 */
	if (tcptw->tw_ts_recent_stamp &&
	    (twp == NULL || (sysctl_tcp_tw_reuse &&
128
			     get_seconds() - tcptw->tw_ts_recent_stamp > 1))) {
129 130 131 132 133 134 135 136 137 138 139 140 141
		tp->write_seq = tcptw->tw_snd_nxt + 65535 + 2;
		if (tp->write_seq == 0)
			tp->write_seq = 1;
		tp->rx_opt.ts_recent	   = tcptw->tw_ts_recent;
		tp->rx_opt.ts_recent_stamp = tcptw->tw_ts_recent_stamp;
		sock_hold(sktw);
		return 1;
	}

	return 0;
}
EXPORT_SYMBOL_GPL(tcp_twsk_unique);

L
Linus Torvalds 已提交
142 143 144
/* This will initiate an outgoing connection. */
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
{
145
	struct sockaddr_in *usin = (struct sockaddr_in *)uaddr;
L
Linus Torvalds 已提交
146 147
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
148
	__be16 orig_sport, orig_dport;
149
	__be32 daddr, nexthop;
150
	struct flowi4 *fl4;
151
	struct rtable *rt;
L
Linus Torvalds 已提交
152
	int err;
153
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
154 155 156 157 158 159 160 161

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

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

	nexthop = daddr = usin->sin_addr.s_addr;
162 163 164
	inet_opt = rcu_dereference_protected(inet->inet_opt,
					     sock_owned_by_user(sk));
	if (inet_opt && inet_opt->opt.srr) {
L
Linus Torvalds 已提交
165 166
		if (!daddr)
			return -EINVAL;
167
		nexthop = inet_opt->opt.faddr;
L
Linus Torvalds 已提交
168 169
	}

170 171
	orig_sport = inet->inet_sport;
	orig_dport = usin->sin_port;
172 173
	fl4 = &inet->cork.fl.u.ip4;
	rt = ip_route_connect(fl4, nexthop, inet->inet_saddr,
174 175
			      RT_CONN_FLAGS(sk), sk->sk_bound_dev_if,
			      IPPROTO_TCP,
176
			      orig_sport, orig_dport, sk);
177 178 179
	if (IS_ERR(rt)) {
		err = PTR_ERR(rt);
		if (err == -ENETUNREACH)
180
			IP_INC_STATS(sock_net(sk), IPSTATS_MIB_OUTNOROUTES);
181
		return err;
182
	}
L
Linus Torvalds 已提交
183 184 185 186 187 188

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

189
	if (!inet_opt || !inet_opt->opt.srr)
190
		daddr = fl4->daddr;
L
Linus Torvalds 已提交
191

E
Eric Dumazet 已提交
192
	if (!inet->inet_saddr)
193
		inet->inet_saddr = fl4->saddr;
E
Eric Dumazet 已提交
194
	inet->inet_rcv_saddr = inet->inet_saddr;
L
Linus Torvalds 已提交
195

E
Eric Dumazet 已提交
196
	if (tp->rx_opt.ts_recent_stamp && inet->inet_daddr != daddr) {
L
Linus Torvalds 已提交
197 198 199
		/* Reset inherited state */
		tp->rx_opt.ts_recent	   = 0;
		tp->rx_opt.ts_recent_stamp = 0;
P
Pavel Emelyanov 已提交
200 201
		if (likely(!tp->repair))
			tp->write_seq	   = 0;
L
Linus Torvalds 已提交
202 203
	}

204
	if (tcp_death_row.sysctl_tw_recycle &&
205 206
	    !tp->rx_opt.ts_recent_stamp && fl4->daddr == daddr)
		tcp_fetch_timewait_stamp(sk, &rt->dst);
L
Linus Torvalds 已提交
207

E
Eric Dumazet 已提交
208 209
	inet->inet_dport = usin->sin_port;
	inet->inet_daddr = daddr;
L
Linus Torvalds 已提交
210

211
	inet_csk(sk)->icsk_ext_hdr_len = 0;
212 213
	if (inet_opt)
		inet_csk(sk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
L
Linus Torvalds 已提交
214

215
	tp->rx_opt.mss_clamp = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
216 217 218 219 220 221 222

	/* Socket identity is still unknown (sport may be zero).
	 * However we set state to SYN-SENT and not releasing socket
	 * lock select source port, enter ourselves into the hash tables and
	 * complete initialization after this.
	 */
	tcp_set_state(sk, TCP_SYN_SENT);
223
	err = inet_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
224 225 226
	if (err)
		goto failure;

227
	rt = ip_route_newports(fl4, rt, orig_sport, orig_dport,
228 229 230 231
			       inet->inet_sport, inet->inet_dport, sk);
	if (IS_ERR(rt)) {
		err = PTR_ERR(rt);
		rt = NULL;
L
Linus Torvalds 已提交
232
		goto failure;
233
	}
L
Linus Torvalds 已提交
234
	/* OK, now commit destination to socket.  */
235
	sk->sk_gso_type = SKB_GSO_TCPV4;
236
	sk_setup_caps(sk, &rt->dst);
L
Linus Torvalds 已提交
237

P
Pavel Emelyanov 已提交
238
	if (!tp->write_seq && likely(!tp->repair))
E
Eric Dumazet 已提交
239 240 241
		tp->write_seq = secure_tcp_sequence_number(inet->inet_saddr,
							   inet->inet_daddr,
							   inet->inet_sport,
L
Linus Torvalds 已提交
242 243
							   usin->sin_port);

E
Eric Dumazet 已提交
244
	inet->inet_id = tp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
245

A
Andrey Vagin 已提交
246
	err = tcp_connect(sk);
P
Pavel Emelyanov 已提交
247

L
Linus Torvalds 已提交
248 249 250 251 252 253 254
	rt = NULL;
	if (err)
		goto failure;

	return 0;

failure:
255 256 257 258
	/*
	 * This unhashes the socket and releases the local port,
	 * if necessary.
	 */
L
Linus Torvalds 已提交
259 260 261
	tcp_set_state(sk, TCP_CLOSE);
	ip_rt_put(rt);
	sk->sk_route_caps = 0;
E
Eric Dumazet 已提交
262
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
263 264
	return err;
}
E
Eric Dumazet 已提交
265
EXPORT_SYMBOL(tcp_v4_connect);
L
Linus Torvalds 已提交
266 267

/*
268 269 270
 * This routine reacts to ICMP_FRAG_NEEDED mtu indications as defined in RFC1191.
 * It can be called through tcp_release_cb() if socket was owned by user
 * at the time tcp_v4_err() was called to handle ICMP message.
L
Linus Torvalds 已提交
271
 */
272
static void tcp_v4_mtu_reduced(struct sock *sk)
L
Linus Torvalds 已提交
273 274 275
{
	struct dst_entry *dst;
	struct inet_sock *inet = inet_sk(sk);
276
	u32 mtu = tcp_sk(sk)->mtu_info;
L
Linus Torvalds 已提交
277

278 279
	dst = inet_csk_update_pmtu(sk, mtu);
	if (!dst)
L
Linus Torvalds 已提交
280 281 282 283 284 285 286 287 288 289 290
		return;

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

	mtu = dst_mtu(dst);

	if (inet->pmtudisc != IP_PMTUDISC_DONT &&
291
	    ip_sk_accept_pmtu(sk) &&
292
	    inet_csk(sk)->icsk_pmtu_cookie > mtu) {
L
Linus Torvalds 已提交
293 294 295 296 297 298 299 300 301 302 303
		tcp_sync_mss(sk, mtu);

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

304 305 306 307
static void do_redirect(struct sk_buff *skb, struct sock *sk)
{
	struct dst_entry *dst = __sk_dst_check(sk, 0);

308
	if (dst)
309
		dst->ops->redirect(dst, sk, skb);
310 311
}

L
Linus Torvalds 已提交
312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327
/*
 * This routine is called by the ICMP module when it gets some
 * sort of error condition.  If err < 0 then the socket should
 * be closed and the error returned to the user.  If err > 0
 * it's just the icmp type << 8 | icmp code.  After adjustment
 * header points to the first 8 bytes of the tcp header.  We need
 * to find the appropriate port.
 *
 * The locking strategy used here is very "optimistic". When
 * someone else accesses the socket the ICMP is just dropped
 * and for some paths there is no check at all.
 * A more general error queue to queue errors for later handling
 * is probably better.
 *
 */

328
void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
L
Linus Torvalds 已提交
329
{
330
	const struct iphdr *iph = (const struct iphdr *)icmp_skb->data;
331
	struct tcphdr *th = (struct tcphdr *)(icmp_skb->data + (iph->ihl << 2));
332
	struct inet_connection_sock *icsk;
L
Linus Torvalds 已提交
333 334
	struct tcp_sock *tp;
	struct inet_sock *inet;
335 336
	const int type = icmp_hdr(icmp_skb)->type;
	const int code = icmp_hdr(icmp_skb)->code;
L
Linus Torvalds 已提交
337
	struct sock *sk;
338
	struct sk_buff *skb;
339 340
	struct request_sock *fastopen;
	__u32 seq, snd_una;
341
	__u32 remaining;
L
Linus Torvalds 已提交
342
	int err;
343
	struct net *net = dev_net(icmp_skb->dev);
L
Linus Torvalds 已提交
344

345
	if (icmp_skb->len < (iph->ihl << 2) + 8) {
346
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
347 348 349
		return;
	}

350
	sk = inet_lookup(net, &tcp_hashinfo, iph->daddr, th->dest,
351
			iph->saddr, th->source, inet_iif(icmp_skb));
L
Linus Torvalds 已提交
352
	if (!sk) {
353
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
354 355 356
		return;
	}
	if (sk->sk_state == TCP_TIME_WAIT) {
357
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
358 359 360 361 362 363
		return;
	}

	bh_lock_sock(sk);
	/* If too many ICMPs get dropped on busy
	 * servers this needs to be solved differently.
364 365
	 * We do take care of PMTU discovery (RFC1191) special case :
	 * we can receive locally generated ICMP messages while socket is held.
L
Linus Torvalds 已提交
366
	 */
367 368 369 370
	if (sock_owned_by_user(sk)) {
		if (!(type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED))
			NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
	}
L
Linus Torvalds 已提交
371 372 373
	if (sk->sk_state == TCP_CLOSE)
		goto out;

374 375 376 377 378
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto out;
	}

379
	icsk = inet_csk(sk);
L
Linus Torvalds 已提交
380 381
	tp = tcp_sk(sk);
	seq = ntohl(th->seq);
382 383 384
	/* XXX (TFO) - tp->snd_una should be ISN (tcp_create_openreq_child() */
	fastopen = tp->fastopen_rsk;
	snd_una = fastopen ? tcp_rsk(fastopen)->snt_isn : tp->snd_una;
L
Linus Torvalds 已提交
385
	if (sk->sk_state != TCP_LISTEN &&
386
	    !between(seq, snd_una, tp->snd_nxt)) {
387
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
388 389 390 391
		goto out;
	}

	switch (type) {
392 393 394
	case ICMP_REDIRECT:
		do_redirect(icmp_skb, sk);
		goto out;
L
Linus Torvalds 已提交
395 396 397 398 399 400 401 402 403 404 405
	case ICMP_SOURCE_QUENCH:
		/* Just silently ignore these. */
		goto out;
	case ICMP_PARAMETERPROB:
		err = EPROTO;
		break;
	case ICMP_DEST_UNREACH:
		if (code > NR_ICMP_UNREACH)
			goto out;

		if (code == ICMP_FRAG_NEEDED) { /* PMTU discovery (RFC1191) */
406 407 408 409 410 411 412
			/* We are not interested in TCP_LISTEN and open_requests
			 * (SYN-ACKs send out by Linux are always <576bytes so
			 * they should go through unfragmented).
			 */
			if (sk->sk_state == TCP_LISTEN)
				goto out;

413
			tp->mtu_info = info;
414
			if (!sock_owned_by_user(sk)) {
415
				tcp_v4_mtu_reduced(sk);
416 417 418 419
			} else {
				if (!test_and_set_bit(TCP_MTU_REDUCED_DEFERRED, &tp->tsq_flags))
					sock_hold(sk);
			}
L
Linus Torvalds 已提交
420 421 422 423
			goto out;
		}

		err = icmp_err_convert[code].errno;
424 425 426 427 428
		/* check if icmp_skb allows revert of backoff
		 * (see draft-zimmermann-tcp-lcd) */
		if (code != ICMP_NET_UNREACH && code != ICMP_HOST_UNREACH)
			break;
		if (seq != tp->snd_una  || !icsk->icsk_retransmits ||
429
		    !icsk->icsk_backoff || fastopen)
430 431
			break;

432 433 434
		if (sock_owned_by_user(sk))
			break;

435
		icsk->icsk_backoff--;
436
		inet_csk(sk)->icsk_rto = (tp->srtt_us ? __tcp_set_rto(tp) :
437
			TCP_TIMEOUT_INIT) << icsk->icsk_backoff;
438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454
		tcp_bound_rto(sk);

		skb = tcp_write_queue_head(sk);
		BUG_ON(!skb);

		remaining = icsk->icsk_rto - min(icsk->icsk_rto,
				tcp_time_stamp - TCP_SKB_CB(skb)->when);

		if (remaining) {
			inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
						  remaining, TCP_RTO_MAX);
		} else {
			/* RTO revert clocked out retransmission.
			 * Will retransmit now */
			tcp_retransmit_timer(sk);
		}

L
Linus Torvalds 已提交
455 456 457 458 459 460 461 462 463
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

	switch (sk->sk_state) {
464
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
465 466 467 468
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

469 470
		req = inet_csk_search_req(sk, &prev, th->dest,
					  iph->daddr, iph->saddr);
L
Linus Torvalds 已提交
471 472 473 474 475 476
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		   an established socket here.
		 */
477
		WARN_ON(req->sk);
L
Linus Torvalds 已提交
478

479
		if (seq != tcp_rsk(req)->snt_isn) {
480
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
481 482 483 484 485 486 487 488 489
			goto out;
		}

		/*
		 * Still in SYN_RECV, just remove it silently.
		 * There is no good way to pass the error to the newly
		 * created socket, and POSIX does not want network
		 * errors returned from accept().
		 */
490
		inet_csk_reqsk_queue_drop(sk, req, prev);
491
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
492 493 494
		goto out;

	case TCP_SYN_SENT:
495 496 497 498 499 500 501
	case TCP_SYN_RECV:
		/* Only in fast or simultaneous open. If a fast open socket is
		 * is already accepted it is treated as a connected one below.
		 */
		if (fastopen && fastopen->sk == NULL)
			break;

L
Linus Torvalds 已提交
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542
		if (!sock_owned_by_user(sk)) {
			sk->sk_err = err;

			sk->sk_error_report(sk);

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

	/* If we've already connected we will keep trying
	 * until we time out, or the user gives up.
	 *
	 * rfc1122 4.2.3.9 allows to consider as hard errors
	 * only PROTO_UNREACH and PORT_UNREACH (well, FRAG_FAILED too,
	 * but it is obsoleted by pmtu discovery).
	 *
	 * Note, that in modern internet, where routing is unreliable
	 * and in each dark corner broken firewalls sit, sending random
	 * errors ordered by their masters even this two messages finally lose
	 * their original sense (even Linux sends invalid PORT_UNREACHs)
	 *
	 * Now we are in compliance with RFCs.
	 *							--ANK (980905)
	 */

	inet = inet_sk(sk);
	if (!sock_owned_by_user(sk) && inet->recverr) {
		sk->sk_err = err;
		sk->sk_error_report(sk);
	} else	{ /* Only an error on timeout */
		sk->sk_err_soft = err;
	}

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

543
void __tcp_v4_send_check(struct sk_buff *skb, __be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
544
{
545
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
546

547
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
548
		th->check = ~tcp_v4_check(skb->len, saddr, daddr, 0);
549
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
550
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
551
	} else {
552
		th->check = tcp_v4_check(skb->len, saddr, daddr,
553
					 csum_partial(th,
L
Linus Torvalds 已提交
554 555 556 557 558
						      th->doff << 2,
						      skb->csum));
	}
}

559
/* This routine computes an IPv4 TCP checksum. */
560
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb)
561
{
562
	const struct inet_sock *inet = inet_sk(sk);
563 564 565

	__tcp_v4_send_check(skb, inet->inet_saddr, inet->inet_daddr);
}
E
Eric Dumazet 已提交
566
EXPORT_SYMBOL(tcp_v4_send_check);
567

L
Linus Torvalds 已提交
568 569 570 571 572 573 574 575 576 577 578 579 580
/*
 *	This routine will send an RST to the other tcp.
 *
 *	Someone asks: why I NEVER use socket parameters (TOS, TTL etc.)
 *		      for reset.
 *	Answer: if a packet caused RST, it is not for a socket
 *		existing in our system, if it is matched to a socket,
 *		it is just duplicate segment or bug in other side's TCP.
 *		So that we build reply only basing on parameters
 *		arrived with segment.
 *	Exception: precedence violation. We do not implement it in any case.
 */

581
static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
582
{
583
	const struct tcphdr *th = tcp_hdr(skb);
584 585 586
	struct {
		struct tcphdr th;
#ifdef CONFIG_TCP_MD5SIG
587
		__be32 opt[(TCPOLEN_MD5SIG_ALIGNED >> 2)];
588 589
#endif
	} rep;
L
Linus Torvalds 已提交
590
	struct ip_reply_arg arg;
591 592
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
593 594 595 596
	const __u8 *hash_location = NULL;
	unsigned char newhash[16];
	int genhash;
	struct sock *sk1 = NULL;
597
#endif
598
	struct net *net;
L
Linus Torvalds 已提交
599 600 601 602 603

	/* Never send a reset in response to a reset. */
	if (th->rst)
		return;

E
Eric Dumazet 已提交
604
	if (skb_rtable(skb)->rt_type != RTN_LOCAL)
L
Linus Torvalds 已提交
605 606 607
		return;

	/* Swap the send and the receive. */
608 609 610 611 612
	memset(&rep, 0, sizeof(rep));
	rep.th.dest   = th->source;
	rep.th.source = th->dest;
	rep.th.doff   = sizeof(struct tcphdr) / 4;
	rep.th.rst    = 1;
L
Linus Torvalds 已提交
613 614

	if (th->ack) {
615
		rep.th.seq = th->ack_seq;
L
Linus Torvalds 已提交
616
	} else {
617 618 619
		rep.th.ack = 1;
		rep.th.ack_seq = htonl(ntohl(th->seq) + th->syn + th->fin +
				       skb->len - (th->doff << 2));
L
Linus Torvalds 已提交
620 621
	}

622
	memset(&arg, 0, sizeof(arg));
623 624 625 626
	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);

#ifdef CONFIG_TCP_MD5SIG
627 628 629 630 631 632 633 634 635 636
	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 = __inet_lookup_listener(dev_net(skb_dst(skb)->dev),
637 638
					     &tcp_hashinfo, ip_hdr(skb)->saddr,
					     th->source, ip_hdr(skb)->daddr,
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657
					     ntohs(th->source), inet_iif(skb));
		/* don't send rst if it can't find key */
		if (!sk1)
			return;
		rcu_read_lock();
		key = tcp_md5_do_lookup(sk1, (union tcp_md5_addr *)
					&ip_hdr(skb)->saddr, AF_INET);
		if (!key)
			goto release_sk1;

		genhash = tcp_v4_md5_hash_skb(newhash, key, NULL, NULL, skb);
		if (genhash || memcmp(hash_location, newhash, 16) != 0)
			goto release_sk1;
	} else {
		key = sk ? tcp_md5_do_lookup(sk, (union tcp_md5_addr *)
					     &ip_hdr(skb)->saddr,
					     AF_INET) : NULL;
	}

658 659 660 661 662 663 664 665 666
	if (key) {
		rep.opt[0] = htonl((TCPOPT_NOP << 24) |
				   (TCPOPT_NOP << 16) |
				   (TCPOPT_MD5SIG << 8) |
				   TCPOLEN_MD5SIG);
		/* Update length and the length the header thinks exists */
		arg.iov[0].iov_len += TCPOLEN_MD5SIG_ALIGNED;
		rep.th.doff = arg.iov[0].iov_len / 4;

667
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
668 669
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
670 671
	}
#endif
672 673
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
674
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
675
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
676
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
677
	/* When socket is gone, all binding information is lost.
A
Alexey Kuznetsov 已提交
678 679
	 * routing might fail in this case. No choice here, if we choose to force
	 * input interface, we will misroute in case of asymmetric route.
680
	 */
A
Alexey Kuznetsov 已提交
681 682
	if (sk)
		arg.bound_dev_if = sk->sk_bound_dev_if;
L
Linus Torvalds 已提交
683

E
Eric Dumazet 已提交
684
	net = dev_net(skb_dst(skb)->dev);
685
	arg.tos = ip_hdr(skb)->tos;
686
	ip_send_unicast_reply(net, skb, ip_hdr(skb)->saddr,
687
			      ip_hdr(skb)->daddr, &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
688

689 690
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
691 692 693 694 695 696 697 698

#ifdef CONFIG_TCP_MD5SIG
release_sk1:
	if (sk1) {
		rcu_read_unlock();
		sock_put(sk1);
	}
#endif
L
Linus Torvalds 已提交
699 700 701 702 703 704
}

/* The code following below sending ACKs in SYN-RECV and TIME-WAIT states
   outside socket context is ugly, certainly. What can I do?
 */

705
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
706
			    u32 win, u32 tsval, u32 tsecr, int oif,
707
			    struct tcp_md5sig_key *key,
708
			    int reply_flags, u8 tos)
L
Linus Torvalds 已提交
709
{
710
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
711 712
	struct {
		struct tcphdr th;
713
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
714
#ifdef CONFIG_TCP_MD5SIG
715
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
716 717
#endif
			];
L
Linus Torvalds 已提交
718 719
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
720
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
721 722

	memset(&rep.th, 0, sizeof(struct tcphdr));
723
	memset(&arg, 0, sizeof(arg));
L
Linus Torvalds 已提交
724 725 726

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
727
	if (tsecr) {
728 729 730
		rep.opt[0] = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				   (TCPOPT_TIMESTAMP << 8) |
				   TCPOLEN_TIMESTAMP);
731 732
		rep.opt[1] = htonl(tsval);
		rep.opt[2] = htonl(tsecr);
733
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
734 735 736 737 738 739 740 741 742 743 744
	}

	/* Swap the send and the receive. */
	rep.th.dest    = th->source;
	rep.th.source  = th->dest;
	rep.th.doff    = arg.iov[0].iov_len / 4;
	rep.th.seq     = htonl(seq);
	rep.th.ack_seq = htonl(ack);
	rep.th.ack     = 1;
	rep.th.window  = htons(win);

745 746
#ifdef CONFIG_TCP_MD5SIG
	if (key) {
747
		int offset = (tsecr) ? 3 : 0;
748 749 750 751 752 753 754 755

		rep.opt[offset++] = htonl((TCPOPT_NOP << 24) |
					  (TCPOPT_NOP << 16) |
					  (TCPOPT_MD5SIG << 8) |
					  TCPOLEN_MD5SIG);
		arg.iov[0].iov_len += TCPOLEN_MD5SIG_ALIGNED;
		rep.th.doff = arg.iov[0].iov_len/4;

756
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[offset],
757 758
				    key, ip_hdr(skb)->saddr,
				    ip_hdr(skb)->daddr, &rep.th);
759 760
	}
#endif
761
	arg.flags = reply_flags;
762 763
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
L
Linus Torvalds 已提交
764 765
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
766 767
	if (oif)
		arg.bound_dev_if = oif;
768
	arg.tos = tos;
769
	ip_send_unicast_reply(net, skb, ip_hdr(skb)->saddr,
770
			      ip_hdr(skb)->daddr, &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
771

772
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
773 774 775 776
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
777
	struct inet_timewait_sock *tw = inet_twsk(sk);
778
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
779

780
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
781
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
782
			tcp_time_stamp + tcptw->tw_ts_offset,
783 784
			tcptw->tw_ts_recent,
			tw->tw_bound_dev_if,
785
			tcp_twsk_md5_key(tcptw),
786 787
			tw->tw_transparent ? IP_REPLY_ARG_NOSRCCHECK : 0,
			tw->tw_tos
788
			);
L
Linus Torvalds 已提交
789

790
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
791 792
}

793
static void tcp_v4_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
794
				  struct request_sock *req)
L
Linus Torvalds 已提交
795
{
796 797 798 799 800 801
	/* sk->sk_state == TCP_LISTEN -> for regular TCP_SYN_RECV
	 * sk->sk_state == TCP_SYN_RECV -> for Fast Open.
	 */
	tcp_v4_send_ack(skb, (sk->sk_state == TCP_LISTEN) ?
			tcp_rsk(req)->snt_isn + 1 : tcp_sk(sk)->snd_nxt,
			tcp_rsk(req)->rcv_nxt, req->rcv_wnd,
802
			tcp_time_stamp,
803 804
			req->ts_recent,
			0,
E
Eric Dumazet 已提交
805 806
			tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&ip_hdr(skb)->daddr,
					  AF_INET),
807 808
			inet_rsk(req)->no_srccheck ? IP_REPLY_ARG_NOSRCCHECK : 0,
			ip_hdr(skb)->tos);
L
Linus Torvalds 已提交
809 810 811
}

/*
812
 *	Send a SYN-ACK after having received a SYN.
813
 *	This still operates on a request_sock only, not on a big
L
Linus Torvalds 已提交
814 815
 *	socket.
 */
816 817
static int tcp_v4_send_synack(struct sock *sk, struct dst_entry *dst,
			      struct request_sock *req,
818 819
			      u16 queue_mapping,
			      struct tcp_fastopen_cookie *foc)
L
Linus Torvalds 已提交
820
{
821
	const struct inet_request_sock *ireq = inet_rsk(req);
822
	struct flowi4 fl4;
L
Linus Torvalds 已提交
823
	int err = -1;
824
	struct sk_buff *skb;
L
Linus Torvalds 已提交
825 826

	/* First, grab a route. */
827
	if (!dst && (dst = inet_csk_route_req(sk, &fl4, req)) == NULL)
828
		return -1;
L
Linus Torvalds 已提交
829

830
	skb = tcp_make_synack(sk, dst, req, foc);
L
Linus Torvalds 已提交
831 832

	if (skb) {
833
		__tcp_v4_send_check(skb, ireq->ir_loc_addr, ireq->ir_rmt_addr);
L
Linus Torvalds 已提交
834

835
		skb_set_queue_mapping(skb, queue_mapping);
836 837
		err = ip_build_and_send_pkt(skb, sk, ireq->ir_loc_addr,
					    ireq->ir_rmt_addr,
838
					    ireq->opt);
839
		err = net_xmit_eval(err);
840 841
		if (!tcp_rsk(req)->snt_synack && !err)
			tcp_rsk(req)->snt_synack = tcp_time_stamp;
L
Linus Torvalds 已提交
842 843 844 845 846
	}

	return err;
}

C
Christoph Paasch 已提交
847
static int tcp_v4_rtx_synack(struct sock *sk, struct request_sock *req)
848
{
849
	int res = tcp_v4_send_synack(sk, NULL, req, 0, NULL);
850

851
	if (!res) {
852
		TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
853 854
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPSYNRETRANS);
	}
855
	return res;
856 857
}

L
Linus Torvalds 已提交
858
/*
859
 *	IPv4 request_sock destructor.
L
Linus Torvalds 已提交
860
 */
861
static void tcp_v4_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
862
{
J
Jesper Juhl 已提交
863
	kfree(inet_rsk(req)->opt);
L
Linus Torvalds 已提交
864 865
}

866
/*
E
Eric Dumazet 已提交
867
 * Return true if a syncookie should be sent
868
 */
E
Eric Dumazet 已提交
869
bool tcp_syn_flood_action(struct sock *sk,
870 871
			 const struct sk_buff *skb,
			 const char *proto)
L
Linus Torvalds 已提交
872
{
873
	const char *msg = "Dropping request";
E
Eric Dumazet 已提交
874
	bool want_cookie = false;
875 876
	struct listen_sock *lopt;

877
#ifdef CONFIG_SYN_COOKIES
878
	if (sysctl_tcp_syncookies) {
879
		msg = "Sending cookies";
E
Eric Dumazet 已提交
880
		want_cookie = true;
881 882
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDOCOOKIES);
	} else
883
#endif
884 885 886
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDROP);

	lopt = inet_csk(sk)->icsk_accept_queue.listen_opt;
887
	if (!lopt->synflood_warned && sysctl_tcp_syncookies != 2) {
888
		lopt->synflood_warned = 1;
889
		pr_info("%s: Possible SYN flooding on port %d. %s.  Check SNMP counters.\n",
890 891 892
			proto, ntohs(tcp_hdr(skb)->dest), msg);
	}
	return want_cookie;
893
}
894
EXPORT_SYMBOL(tcp_syn_flood_action);
L
Linus Torvalds 已提交
895 896

/*
897
 * Save and compile IPv4 options into the request_sock if needed.
L
Linus Torvalds 已提交
898
 */
899
static struct ip_options_rcu *tcp_v4_save_options(struct sk_buff *skb)
L
Linus Torvalds 已提交
900
{
901 902
	const struct ip_options *opt = &(IPCB(skb)->opt);
	struct ip_options_rcu *dopt = NULL;
L
Linus Torvalds 已提交
903 904

	if (opt && opt->optlen) {
905 906
		int opt_size = sizeof(*dopt) + opt->optlen;

L
Linus Torvalds 已提交
907 908
		dopt = kmalloc(opt_size, GFP_ATOMIC);
		if (dopt) {
909
			if (ip_options_echo(&dopt->opt, skb)) {
L
Linus Torvalds 已提交
910 911 912 913 914 915 916 917
				kfree(dopt);
				dopt = NULL;
			}
		}
	}
	return dopt;
}

918 919 920 921 922 923 924 925
#ifdef CONFIG_TCP_MD5SIG
/*
 * RFC2385 MD5 checksumming requires a mapping of
 * IP address->MD5 Key.
 * We need to maintain these in the sk structure.
 */

/* Find the Key structure for an address.  */
E
Eric Dumazet 已提交
926 927 928
struct tcp_md5sig_key *tcp_md5_do_lookup(struct sock *sk,
					 const union tcp_md5_addr *addr,
					 int family)
929 930
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
931 932
	struct tcp_md5sig_key *key;
	unsigned int size = sizeof(struct in_addr);
933
	struct tcp_md5sig_info *md5sig;
934

935 936
	/* caller either holds rcu_read_lock() or socket lock */
	md5sig = rcu_dereference_check(tp->md5sig_info,
937 938
				       sock_owned_by_user(sk) ||
				       lockdep_is_held(&sk->sk_lock.slock));
939
	if (!md5sig)
940
		return NULL;
E
Eric Dumazet 已提交
941 942 943 944
#if IS_ENABLED(CONFIG_IPV6)
	if (family == AF_INET6)
		size = sizeof(struct in6_addr);
#endif
945
	hlist_for_each_entry_rcu(key, &md5sig->head, node) {
E
Eric Dumazet 已提交
946 947 948 949
		if (key->family != family)
			continue;
		if (!memcmp(&key->addr, addr, size))
			return key;
950 951 952
	}
	return NULL;
}
E
Eric Dumazet 已提交
953
EXPORT_SYMBOL(tcp_md5_do_lookup);
954 955 956 957

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
958 959 960 961
	union tcp_md5_addr *addr;

	addr = (union tcp_md5_addr *)&inet_sk(addr_sk)->inet_daddr;
	return tcp_md5_do_lookup(sk, addr, AF_INET);
962 963 964
}
EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
965 966
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
967
{
E
Eric Dumazet 已提交
968 969
	union tcp_md5_addr *addr;

970
	addr = (union tcp_md5_addr *)&inet_rsk(req)->ir_rmt_addr;
E
Eric Dumazet 已提交
971
	return tcp_md5_do_lookup(sk, addr, AF_INET);
972 973 974
}

/* This can be called on a newly created socket, from other files */
E
Eric Dumazet 已提交
975 976
int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
		   int family, const u8 *newkey, u8 newkeylen, gfp_t gfp)
977 978
{
	/* Add Key to the list */
979
	struct tcp_md5sig_key *key;
980
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
981
	struct tcp_md5sig_info *md5sig;
982

983
	key = tcp_md5_do_lookup(sk, addr, family);
984 985
	if (key) {
		/* Pre-existing entry - just update that one. */
E
Eric Dumazet 已提交
986
		memcpy(key->key, newkey, newkeylen);
987
		key->keylen = newkeylen;
E
Eric Dumazet 已提交
988 989
		return 0;
	}
990

991 992
	md5sig = rcu_dereference_protected(tp->md5sig_info,
					   sock_owned_by_user(sk));
E
Eric Dumazet 已提交
993 994 995
	if (!md5sig) {
		md5sig = kmalloc(sizeof(*md5sig), gfp);
		if (!md5sig)
996 997
			return -ENOMEM;

E
Eric Dumazet 已提交
998 999
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
		INIT_HLIST_HEAD(&md5sig->head);
1000
		rcu_assign_pointer(tp->md5sig_info, md5sig);
E
Eric Dumazet 已提交
1001
	}
1002

1003
	key = sock_kmalloc(sk, sizeof(*key), gfp);
E
Eric Dumazet 已提交
1004 1005
	if (!key)
		return -ENOMEM;
1006
	if (!tcp_alloc_md5sig_pool()) {
1007
		sock_kfree_s(sk, key, sizeof(*key));
E
Eric Dumazet 已提交
1008
		return -ENOMEM;
1009
	}
E
Eric Dumazet 已提交
1010 1011 1012 1013 1014 1015 1016 1017

	memcpy(key->key, newkey, newkeylen);
	key->keylen = newkeylen;
	key->family = family;
	memcpy(&key->addr, addr,
	       (family == AF_INET6) ? sizeof(struct in6_addr) :
				      sizeof(struct in_addr));
	hlist_add_head_rcu(&key->node, &md5sig->head);
1018 1019
	return 0;
}
E
Eric Dumazet 已提交
1020
EXPORT_SYMBOL(tcp_md5_do_add);
1021

E
Eric Dumazet 已提交
1022
int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr, int family)
1023
{
E
Eric Dumazet 已提交
1024 1025
	struct tcp_md5sig_key *key;

1026
	key = tcp_md5_do_lookup(sk, addr, family);
E
Eric Dumazet 已提交
1027 1028 1029
	if (!key)
		return -ENOENT;
	hlist_del_rcu(&key->node);
1030
	atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1031 1032
	kfree_rcu(key, rcu);
	return 0;
1033
}
E
Eric Dumazet 已提交
1034
EXPORT_SYMBOL(tcp_md5_do_del);
1035

1036
static void tcp_clear_md5_list(struct sock *sk)
1037 1038
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
1039
	struct tcp_md5sig_key *key;
1040
	struct hlist_node *n;
1041
	struct tcp_md5sig_info *md5sig;
1042

1043 1044
	md5sig = rcu_dereference_protected(tp->md5sig_info, 1);

1045
	hlist_for_each_entry_safe(key, n, &md5sig->head, node) {
E
Eric Dumazet 已提交
1046
		hlist_del_rcu(&key->node);
1047
		atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1048
		kfree_rcu(key, rcu);
1049 1050 1051
	}
}

1052 1053
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
1054 1055 1056 1057 1058 1059 1060
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;

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

1061
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
1062 1063 1064 1065 1066
		return -EFAULT;

	if (sin->sin_family != AF_INET)
		return -EINVAL;

1067
	if (!cmd.tcpm_key || !cmd.tcpm_keylen)
E
Eric Dumazet 已提交
1068 1069
		return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin->sin_addr.s_addr,
				      AF_INET);
1070 1071 1072 1073

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

E
Eric Dumazet 已提交
1074 1075 1076
	return tcp_md5_do_add(sk, (union tcp_md5_addr *)&sin->sin_addr.s_addr,
			      AF_INET, cmd.tcpm_key, cmd.tcpm_keylen,
			      GFP_KERNEL);
1077 1078
}

1079 1080
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1081 1082
{
	struct tcp4_pseudohdr *bp;
1083
	struct scatterlist sg;
1084 1085 1086 1087

	bp = &hp->md5_blk.ip4;

	/*
1088
	 * 1. the TCP pseudo-header (in the order: source IP address,
1089 1090 1091 1092 1093 1094
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1095
	bp->protocol = IPPROTO_TCP;
1096
	bp->len = cpu_to_be16(nbytes);
1097

1098 1099 1100 1101
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}

E
Eric Dumazet 已提交
1102
static int tcp_v4_md5_hash_hdr(char *md5_hash, const struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1103
			       __be32 daddr, __be32 saddr, const struct tcphdr *th)
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121
{
	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_v4_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))
1122 1123 1124 1125
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1126

1127 1128 1129 1130
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1131
	return 1;
1132 1133
}

1134
int tcp_v4_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1135 1136
			const struct sock *sk, const struct request_sock *req,
			const struct sk_buff *skb)
1137
{
1138 1139
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
E
Eric Dumazet 已提交
1140
	const struct tcphdr *th = tcp_hdr(skb);
1141 1142 1143
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1144 1145
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1146
	} else if (req) {
1147 1148
		saddr = inet_rsk(req)->ir_loc_addr;
		daddr = inet_rsk(req)->ir_rmt_addr;
1149
	} else {
1150 1151 1152
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
1153
	}
1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181

	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_v4_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;
1182
}
1183
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1184

E
Eric Dumazet 已提交
1185
static bool tcp_v4_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
1186 1187 1188 1189 1190 1191 1192 1193 1194
{
	/*
	 * This gets called for each TCP segment that arrives
	 * so we want to be efficient.
	 * We have 3 drop cases:
	 * o No MD5 hash and one expected.
	 * o MD5 hash and we're not expecting one.
	 * o MD5 hash and its wrong.
	 */
1195
	const __u8 *hash_location = NULL;
1196
	struct tcp_md5sig_key *hash_expected;
1197
	const struct iphdr *iph = ip_hdr(skb);
1198
	const struct tcphdr *th = tcp_hdr(skb);
1199 1200 1201
	int genhash;
	unsigned char newhash[16];

E
Eric Dumazet 已提交
1202 1203
	hash_expected = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&iph->saddr,
					  AF_INET);
1204
	hash_location = tcp_parse_md5sig_option(th);
1205 1206 1207

	/* We've parsed the options - do we have a hash? */
	if (!hash_expected && !hash_location)
E
Eric Dumazet 已提交
1208
		return false;
1209 1210

	if (hash_expected && !hash_location) {
1211
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
E
Eric Dumazet 已提交
1212
		return true;
1213 1214 1215
	}

	if (!hash_expected && hash_location) {
1216
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
E
Eric Dumazet 已提交
1217
		return true;
1218 1219 1220 1221 1222
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1223 1224 1225
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1226 1227

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
1228 1229 1230 1231 1232
		net_info_ratelimited("MD5 Hash failed for (%pI4, %d)->(%pI4, %d)%s\n",
				     &iph->saddr, ntohs(th->source),
				     &iph->daddr, ntohs(th->dest),
				     genhash ? " tcp_v4_calc_md5_hash failed"
				     : "");
E
Eric Dumazet 已提交
1233
		return true;
1234
	}
E
Eric Dumazet 已提交
1235
	return false;
1236 1237 1238 1239
}

#endif

1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
static void tcp_v4_init_req(struct request_sock *req, struct sock *sk,
			    struct sk_buff *skb)
{
	struct inet_request_sock *ireq = inet_rsk(req);

	ireq->ir_loc_addr = ip_hdr(skb)->daddr;
	ireq->ir_rmt_addr = ip_hdr(skb)->saddr;
	ireq->no_srccheck = inet_sk(sk)->transparent;
	ireq->opt = tcp_v4_save_options(skb);
}

1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
static struct dst_entry *tcp_v4_route_req(struct sock *sk, struct flowi *fl,
					  const struct request_sock *req,
					  bool *strict)
{
	struct dst_entry *dst = inet_csk_route_req(sk, &fl->u.ip4, req);

	if (strict) {
		if (fl->u.ip4.daddr == inet_rsk(req)->ir_rmt_addr)
			*strict = true;
		else
			*strict = false;
	}

	return dst;
}

1267
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1268
	.family		=	PF_INET,
1269
	.obj_size	=	sizeof(struct tcp_request_sock),
1270
	.rtx_syn_ack	=	tcp_v4_rtx_synack,
1271 1272
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1273
	.send_reset	=	tcp_v4_send_reset,
1274
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
1275 1276
};

1277
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1278
#ifdef CONFIG_TCP_MD5SIG
1279
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1280
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1281
#endif
1282
	.init_req	=	tcp_v4_init_req,
1283 1284 1285
#ifdef CONFIG_SYN_COOKIES
	.cookie_init_seq =	cookie_v4_init_sequence,
#endif
1286
	.route_req	=	tcp_v4_route_req,
1287
};
1288

L
Linus Torvalds 已提交
1289 1290 1291
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
	struct tcp_options_received tmp_opt;
1292
	struct request_sock *req;
1293
	struct tcp_sock *tp = tcp_sk(sk);
1294
	struct dst_entry *dst = NULL;
1295
	__be32 saddr = ip_hdr(skb)->saddr;
L
Linus Torvalds 已提交
1296
	__u32 isn = TCP_SKB_CB(skb)->when;
1297
	bool want_cookie = false, fastopen;
1298 1299
	struct flowi4 fl4;
	struct tcp_fastopen_cookie foc = { .len = -1 };
1300
	const struct tcp_request_sock_ops *af_ops;
1301
	int err;
L
Linus Torvalds 已提交
1302 1303

	/* Never answer to SYNs send to broadcast or multicast */
E
Eric Dumazet 已提交
1304
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1305 1306 1307 1308 1309 1310
		goto drop;

	/* TW buckets are converted to open requests without
	 * limitations, they conserve resources and peer is
	 * evidently real one.
	 */
1311 1312
	if ((sysctl_tcp_syncookies == 2 ||
	     inet_csk_reqsk_queue_is_full(sk)) && !isn) {
1313 1314 1315
		want_cookie = tcp_syn_flood_action(sk, skb, "TCP");
		if (!want_cookie)
			goto drop;
L
Linus Torvalds 已提交
1316 1317 1318 1319 1320 1321 1322
	}

	/* Accept backlog is full. If we have already queued enough
	 * of warm entries in syn queue, drop request. It is better than
	 * clogging syn queue with openreqs with exponentially increasing
	 * timeout.
	 */
1323 1324
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1) {
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
L
Linus Torvalds 已提交
1325
		goto drop;
1326
	}
L
Linus Torvalds 已提交
1327

1328
	req = inet_reqsk_alloc(&tcp_request_sock_ops);
L
Linus Torvalds 已提交
1329 1330 1331
	if (!req)
		goto drop;

1332
	af_ops = tcp_rsk(req)->af_specific = &tcp_request_sock_ipv4_ops;
1333

L
Linus Torvalds 已提交
1334
	tcp_clear_options(&tmp_opt);
1335
	tmp_opt.mss_clamp = TCP_MSS_DEFAULT;
1336
	tmp_opt.user_mss  = tp->rx_opt.user_mss;
C
Christoph Paasch 已提交
1337
	tcp_parse_options(skb, &tmp_opt, 0, want_cookie ? NULL : &foc);
L
Linus Torvalds 已提交
1338

1339
	if (want_cookie && !tmp_opt.saw_tstamp)
L
Linus Torvalds 已提交
1340 1341 1342
		tcp_clear_options(&tmp_opt);

	tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;
1343
	tcp_openreq_init(req, &tmp_opt, skb, sk);
L
Linus Torvalds 已提交
1344

1345
	af_ops->init_req(req, sk, skb);
1346

1347
	if (security_inet_conn_request(sk, skb, req))
1348
		goto drop_and_free;
1349

1350
	if (!want_cookie || tmp_opt.tstamp_ok)
1351
		TCP_ECN_create_request(req, skb, sock_net(sk));
L
Linus Torvalds 已提交
1352 1353

	if (want_cookie) {
1354
		isn = cookie_init_sequence(af_ops, sk, skb, &req->mss);
1355
		req->cookie_ts = tmp_opt.tstamp_ok;
L
Linus Torvalds 已提交
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365
	} else if (!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.
		 */
1366 1367 1368 1369 1370 1371 1372
		if (tmp_opt.saw_tstamp && tcp_death_row.sysctl_tw_recycle) {
			bool strict;

			dst = af_ops->route_req(sk, (struct flowi *)&fl4, req,
						&strict);
			if (dst && strict &&
			    !tcp_peer_is_proven(req, dst, true)) {
1373
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1374
				goto drop_and_release;
L
Linus Torvalds 已提交
1375 1376 1377 1378
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
1379
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
L
Linus Torvalds 已提交
1380
			  (sysctl_max_syn_backlog >> 2)) &&
1381
			 !tcp_peer_is_proven(req, dst, false)) {
L
Linus Torvalds 已提交
1382 1383 1384 1385 1386 1387 1388
			/* 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.
			 */
1389
			LIMIT_NETDEBUG(KERN_DEBUG pr_fmt("drop open request from %pI4/%u\n"),
1390
				       &saddr, ntohs(tcp_hdr(skb)->source));
1391
			goto drop_and_release;
L
Linus Torvalds 已提交
1392 1393
		}

1394
		isn = tcp_v4_init_sequence(skb);
L
Linus Torvalds 已提交
1395
	}
1396 1397 1398 1399 1400
	if (!dst) {
		dst = af_ops->route_req(sk, (struct flowi *)&fl4, req, NULL);
		if (!dst)
			goto drop_and_free;
	}
1401

1402 1403 1404 1405 1406 1407 1408
	tcp_rsk(req)->snt_isn = isn;
	tcp_openreq_init_rwin(req, sk, dst);
	fastopen = !want_cookie &&
		   tcp_try_fastopen(sk, skb, req, &foc, dst);
	err = tcp_v4_send_synack(sk, dst, req,
				 skb_get_queue_mapping(skb), &foc);
	if (!fastopen) {
1409 1410 1411 1412 1413
		if (err || want_cookie)
			goto drop_and_free;

		tcp_rsk(req)->listener = NULL;
		inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
1414
	}
L
Linus Torvalds 已提交
1415 1416 1417

	return 0;

1418 1419
drop_and_release:
	dst_release(dst);
L
Linus Torvalds 已提交
1420
drop_and_free:
1421
	reqsk_free(req);
L
Linus Torvalds 已提交
1422
drop:
1423
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1424 1425
	return 0;
}
E
Eric Dumazet 已提交
1426
EXPORT_SYMBOL(tcp_v4_conn_request);
L
Linus Torvalds 已提交
1427 1428 1429 1430 1431 1432 1433


/*
 * The three way handshake has completed - we got a valid synack -
 * now create the new socket.
 */
struct sock *tcp_v4_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
1434
				  struct request_sock *req,
L
Linus Torvalds 已提交
1435 1436
				  struct dst_entry *dst)
{
1437
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1438 1439 1440
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1441 1442 1443
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
1444
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
1445 1446 1447 1448 1449 1450

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (!newsk)
1451
		goto exit_nonewsk;
L
Linus Torvalds 已提交
1452

1453
	newsk->sk_gso_type = SKB_GSO_TCPV4;
1454
	inet_sk_rx_dst_set(newsk, skb);
L
Linus Torvalds 已提交
1455 1456 1457

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1458
	ireq		      = inet_rsk(req);
1459 1460 1461
	newinet->inet_daddr   = ireq->ir_rmt_addr;
	newinet->inet_rcv_saddr = ireq->ir_loc_addr;
	newinet->inet_saddr	      = ireq->ir_loc_addr;
1462 1463
	inet_opt	      = ireq->opt;
	rcu_assign_pointer(newinet->inet_opt, inet_opt);
1464
	ireq->opt	      = NULL;
1465
	newinet->mc_index     = inet_iif(skb);
1466
	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
1467
	newinet->rcv_tos      = ip_hdr(skb)->tos;
1468
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
1469 1470
	if (inet_opt)
		inet_csk(newsk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
E
Eric Dumazet 已提交
1471
	newinet->inet_id = newtp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
1472

E
Eric Dumazet 已提交
1473 1474 1475 1476 1477 1478 1479
	if (!dst) {
		dst = inet_csk_route_child_sock(sk, newsk, req);
		if (!dst)
			goto put_and_exit;
	} else {
		/* syncookie case : see end of cookie_v4_check() */
	}
1480 1481
	sk_setup_caps(newsk, dst);

L
Linus Torvalds 已提交
1482
	tcp_sync_mss(newsk, dst_mtu(dst));
1483
	newtp->advmss = dst_metric_advmss(dst);
1484 1485 1486 1487
	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
1488 1489
	tcp_initialize_rcv_mss(newsk);

1490 1491
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1492 1493
	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&newinet->inet_daddr,
				AF_INET);
E
Eric Dumazet 已提交
1494
	if (key != NULL) {
1495 1496 1497 1498 1499 1500
		/*
		 * 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
1501 1502
		tcp_md5_do_add(newsk, (union tcp_md5_addr *)&newinet->inet_daddr,
			       AF_INET, key->key, key->keylen, GFP_ATOMIC);
E
Eric Dumazet 已提交
1503
		sk_nocaps_add(newsk, NETIF_F_GSO_MASK);
1504 1505 1506
	}
#endif

1507 1508
	if (__inet_inherit_port(sk, newsk) < 0)
		goto put_and_exit;
1509
	__inet_hash_nolisten(newsk, NULL);
L
Linus Torvalds 已提交
1510 1511 1512 1513

	return newsk;

exit_overflow:
1514
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1515 1516
exit_nonewsk:
	dst_release(dst);
L
Linus Torvalds 已提交
1517
exit:
1518
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1519
	return NULL;
1520
put_and_exit:
1521 1522
	inet_csk_prepare_forced_close(newsk);
	tcp_done(newsk);
1523
	goto exit;
L
Linus Torvalds 已提交
1524
}
E
Eric Dumazet 已提交
1525
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
L
Linus Torvalds 已提交
1526 1527 1528

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1529
	struct tcphdr *th = tcp_hdr(skb);
1530
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1531
	struct sock *nsk;
1532
	struct request_sock **prev;
L
Linus Torvalds 已提交
1533
	/* Find possible connection requests. */
1534 1535
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1536
	if (req)
1537
		return tcp_check_req(sk, skb, req, prev, false);
L
Linus Torvalds 已提交
1538

1539
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1540
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1541 1542 1543 1544 1545 1546

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1547
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1548 1549 1550 1551
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
1552
	if (!th->syn)
L
Linus Torvalds 已提交
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
		sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
#endif
	return sk;
}

/* 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.
 */
int tcp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
{
1568 1569 1570 1571 1572 1573 1574 1575
	struct sock *rsk;
#ifdef CONFIG_TCP_MD5SIG
	/*
	 * We really want to reject the packet as early as possible
	 * if:
	 *  o We're expecting an MD5'd packet and this is no MD5 tcp option
	 *  o There is an MD5 option and we're not expecting one
	 */
1576
	if (tcp_v4_inbound_md5_hash(sk, skb))
1577 1578 1579
		goto discard;
#endif

L
Linus Torvalds 已提交
1580
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1581 1582
		struct dst_entry *dst = sk->sk_rx_dst;

1583
		sock_rps_save_rxhash(sk, skb);
1584
		if (dst) {
E
Eric Dumazet 已提交
1585 1586
			if (inet_sk(sk)->rx_dst_ifindex != skb->skb_iif ||
			    dst->ops->check(dst, 0) == NULL) {
1587 1588 1589 1590
				dst_release(dst);
				sk->sk_rx_dst = NULL;
			}
		}
1591
		tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len);
L
Linus Torvalds 已提交
1592 1593 1594
		return 0;
	}

1595
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1596 1597 1598 1599 1600 1601 1602 1603
		goto csum_err;

	if (sk->sk_state == TCP_LISTEN) {
		struct sock *nsk = tcp_v4_hnd_req(sk, skb);
		if (!nsk)
			goto discard;

		if (nsk != sk) {
1604
			sock_rps_save_rxhash(nsk, skb);
1605 1606
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1607
				goto reset;
1608
			}
L
Linus Torvalds 已提交
1609 1610
			return 0;
		}
1611
	} else
1612
		sock_rps_save_rxhash(sk, skb);
1613

1614
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1615
		rsk = sk;
L
Linus Torvalds 已提交
1616
		goto reset;
1617
	}
L
Linus Torvalds 已提交
1618 1619 1620
	return 0;

reset:
1621
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1622 1623 1624 1625 1626 1627 1628 1629 1630 1631
discard:
	kfree_skb(skb);
	/* Be careful here. If this function gets more complicated and
	 * gcc suffers from register pressure on the x86, sk (in %ebx)
	 * might be destroyed here. This current version compiles correctly,
	 * but you have been warned.
	 */
	return 0;

csum_err:
1632
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_CSUMERRORS);
1633
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1634 1635
	goto discard;
}
E
Eric Dumazet 已提交
1636
EXPORT_SYMBOL(tcp_v4_do_rcv);
L
Linus Torvalds 已提交
1637

1638
void tcp_v4_early_demux(struct sk_buff *skb)
D
David S. Miller 已提交
1639 1640 1641 1642 1643 1644
{
	const struct iphdr *iph;
	const struct tcphdr *th;
	struct sock *sk;

	if (skb->pkt_type != PACKET_HOST)
1645
		return;
D
David S. Miller 已提交
1646

1647
	if (!pskb_may_pull(skb, skb_transport_offset(skb) + sizeof(struct tcphdr)))
1648
		return;
D
David S. Miller 已提交
1649 1650

	iph = ip_hdr(skb);
1651
	th = tcp_hdr(skb);
D
David S. Miller 已提交
1652 1653

	if (th->doff < sizeof(struct tcphdr) / 4)
1654
		return;
D
David S. Miller 已提交
1655

1656
	sk = __inet_lookup_established(dev_net(skb->dev), &tcp_hashinfo,
D
David S. Miller 已提交
1657
				       iph->saddr, th->source,
1658
				       iph->daddr, ntohs(th->dest),
E
Eric Dumazet 已提交
1659
				       skb->skb_iif);
D
David S. Miller 已提交
1660 1661 1662 1663 1664
	if (sk) {
		skb->sk = sk;
		skb->destructor = sock_edemux;
		if (sk->sk_state != TCP_TIME_WAIT) {
			struct dst_entry *dst = sk->sk_rx_dst;
E
Eric Dumazet 已提交
1665

D
David S. Miller 已提交
1666 1667
			if (dst)
				dst = dst_check(dst, 0);
1668
			if (dst &&
E
Eric Dumazet 已提交
1669
			    inet_sk(sk)->rx_dst_ifindex == skb->skb_iif)
1670
				skb_dst_set_noref(skb, dst);
D
David S. Miller 已提交
1671 1672 1673 1674
		}
	}
}

E
Eric Dumazet 已提交
1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692
/* Packet is added to VJ-style prequeue for processing in process
 * context, if a reader task is waiting. Apparently, this exciting
 * idea (VJ's mail "Re: query about TCP header on tcp-ip" of 07 Sep 93)
 * failed somewhere. Latency? Burstiness? Well, at least now we will
 * see, why it failed. 8)8)				  --ANK
 *
 */
bool tcp_prequeue(struct sock *sk, struct sk_buff *skb)
{
	struct tcp_sock *tp = tcp_sk(sk);

	if (sysctl_tcp_low_latency || !tp->ucopy.task)
		return false;

	if (skb->len <= tcp_hdrlen(skb) &&
	    skb_queue_len(&tp->ucopy.prequeue) == 0)
		return false;

1693
	skb_dst_force(skb);
E
Eric Dumazet 已提交
1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719
	__skb_queue_tail(&tp->ucopy.prequeue, skb);
	tp->ucopy.memory += skb->truesize;
	if (tp->ucopy.memory > sk->sk_rcvbuf) {
		struct sk_buff *skb1;

		BUG_ON(sock_owned_by_user(sk));

		while ((skb1 = __skb_dequeue(&tp->ucopy.prequeue)) != NULL) {
			sk_backlog_rcv(sk, skb1);
			NET_INC_STATS_BH(sock_net(sk),
					 LINUX_MIB_TCPPREQUEUEDROPPED);
		}

		tp->ucopy.memory = 0;
	} else if (skb_queue_len(&tp->ucopy.prequeue) == 1) {
		wake_up_interruptible_sync_poll(sk_sleep(sk),
					   POLLIN | POLLRDNORM | POLLRDBAND);
		if (!inet_csk_ack_scheduled(sk))
			inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
						  (3 * tcp_rto_min(sk)) / 4,
						  TCP_RTO_MAX);
	}
	return true;
}
EXPORT_SYMBOL(tcp_prequeue);

L
Linus Torvalds 已提交
1720 1721 1722 1723 1724 1725
/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1726
	const struct iphdr *iph;
1727
	const struct tcphdr *th;
L
Linus Torvalds 已提交
1728 1729
	struct sock *sk;
	int ret;
1730
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1731 1732 1733 1734 1735

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

	/* Count it even if it's bad */
1736
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1737 1738 1739 1740

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

1741
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1742 1743 1744 1745 1746 1747 1748 1749

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

	/* An explanation is required here, I think.
	 * Packet length and doff are validated by header prediction,
S
Stephen Hemminger 已提交
1750
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1751
	 * So, we defer the checks. */
1752 1753

	if (skb_checksum_init(skb, IPPROTO_TCP, inet_compute_pseudo))
1754
		goto csum_error;
L
Linus Torvalds 已提交
1755

1756
	th = tcp_hdr(skb);
1757
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1758 1759 1760 1761 1762
	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;
1763
	TCP_SKB_CB(skb)->ip_dsfield = ipv4_get_dsfield(iph);
L
Linus Torvalds 已提交
1764 1765
	TCP_SKB_CB(skb)->sacked	 = 0;

1766
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1767 1768 1769
	if (!sk)
		goto no_tcp_socket;

E
Eric Dumazet 已提交
1770 1771 1772 1773
process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1774 1775
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1776
		goto discard_and_relse;
1777
	}
1778

L
Linus Torvalds 已提交
1779 1780
	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1781
	nf_reset(skb);
L
Linus Torvalds 已提交
1782

1783
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1784 1785
		goto discard_and_relse;

1786
	sk_mark_napi_id(sk, skb);
L
Linus Torvalds 已提交
1787 1788
	skb->dev = NULL;

1789
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1790 1791
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1792 1793 1794
#ifdef CONFIG_NET_DMA
		struct tcp_sock *tp = tcp_sk(sk);
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1795
			tp->ucopy.dma_chan = net_dma_find_channel();
C
Chris Leech 已提交
1796
		if (tp->ucopy.dma_chan)
L
Linus Torvalds 已提交
1797
			ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1798 1799 1800 1801
		else
#endif
		{
			if (!tcp_prequeue(sk, skb))
S
Shan Wei 已提交
1802
				ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1803
		}
1804 1805
	} else if (unlikely(sk_add_backlog(sk, skb,
					   sk->sk_rcvbuf + sk->sk_sndbuf))) {
Z
Zhu Yi 已提交
1806
		bh_unlock_sock(sk);
1807
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1808 1809
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820
	bh_unlock_sock(sk);

	sock_put(sk);

	return ret;

no_tcp_socket:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
		goto discard_it;

	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
1821 1822
csum_error:
		TCP_INC_STATS_BH(net, TCP_MIB_CSUMERRORS);
L
Linus Torvalds 已提交
1823
bad_packet:
1824
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1825
	} else {
1826
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1827 1828 1829 1830 1831
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
1832
	return 0;
L
Linus Torvalds 已提交
1833 1834 1835 1836 1837 1838 1839

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1840
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1841 1842 1843
		goto discard_it;
	}

1844
	if (skb->len < (th->doff << 2)) {
1845
		inet_twsk_put(inet_twsk(sk));
1846 1847 1848 1849 1850
		goto bad_packet;
	}
	if (tcp_checksum_complete(skb)) {
		inet_twsk_put(inet_twsk(sk));
		goto csum_error;
L
Linus Torvalds 已提交
1851
	}
1852
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1853
	case TCP_TW_SYN: {
1854
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
1855
							&tcp_hashinfo,
1856
							iph->saddr, th->source,
1857
							iph->daddr, th->dest,
1858
							inet_iif(skb));
L
Linus Torvalds 已提交
1859
		if (sk2) {
1860 1861
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876
			sk = sk2;
			goto process;
		}
		/* Fall through to ACK */
	}
	case TCP_TW_ACK:
		tcp_v4_timewait_ack(sk, skb);
		break;
	case TCP_TW_RST:
		goto no_tcp_socket;
	case TCP_TW_SUCCESS:;
	}
	goto discard_it;
}

1877 1878 1879 1880 1881
static struct timewait_sock_ops tcp_timewait_sock_ops = {
	.twsk_obj_size	= sizeof(struct tcp_timewait_sock),
	.twsk_unique	= tcp_twsk_unique,
	.twsk_destructor= tcp_twsk_destructor,
};
L
Linus Torvalds 已提交
1882

1883
void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
E
Eric Dumazet 已提交
1884 1885 1886 1887 1888 1889 1890
{
	struct dst_entry *dst = skb_dst(skb);

	dst_hold(dst);
	sk->sk_rx_dst = dst;
	inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
}
1891
EXPORT_SYMBOL(inet_sk_rx_dst_set);
E
Eric Dumazet 已提交
1892

1893
const struct inet_connection_sock_af_ops ipv4_specific = {
1894 1895 1896
	.queue_xmit	   = ip_queue_xmit,
	.send_check	   = tcp_v4_send_check,
	.rebuild_header	   = inet_sk_rebuild_header,
E
Eric Dumazet 已提交
1897
	.sk_rx_dst_set	   = inet_sk_rx_dst_set,
1898 1899 1900 1901 1902 1903 1904
	.conn_request	   = tcp_v4_conn_request,
	.syn_recv_sock	   = tcp_v4_syn_recv_sock,
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ip_setsockopt,
	.getsockopt	   = ip_getsockopt,
	.addr2sockaddr	   = inet_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in),
1905
	.bind_conflict	   = inet_csk_bind_conflict,
1906
#ifdef CONFIG_COMPAT
1907 1908
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
1909
#endif
L
Linus Torvalds 已提交
1910
};
E
Eric Dumazet 已提交
1911
EXPORT_SYMBOL(ipv4_specific);
L
Linus Torvalds 已提交
1912

1913
#ifdef CONFIG_TCP_MD5SIG
1914
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
1915
	.md5_lookup		= tcp_v4_md5_lookup,
1916
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
1917 1918
	.md5_parse		= tcp_v4_parse_md5_keys,
};
1919
#endif
1920

L
Linus Torvalds 已提交
1921 1922 1923 1924 1925
/* NOTE: A lot of things set to zero explicitly by call to
 *       sk_alloc() so need not be done here.
 */
static int tcp_v4_init_sock(struct sock *sk)
{
1926
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1927

1928
	tcp_init_sock(sk);
L
Linus Torvalds 已提交
1929

1930
	icsk->icsk_af_ops = &ipv4_specific;
1931

1932
#ifdef CONFIG_TCP_MD5SIG
1933
	tcp_sk(sk)->af_specific = &tcp_sock_ipv4_specific;
1934
#endif
L
Linus Torvalds 已提交
1935 1936 1937 1938

	return 0;
}

1939
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1940 1941 1942 1943 1944
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

1945
	tcp_cleanup_congestion_control(sk);
1946

L
Linus Torvalds 已提交
1947
	/* Cleanup up the write buffer. */
1948
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
1949 1950

	/* Cleans up our, hopefully empty, out_of_order_queue. */
1951
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
1952

1953 1954 1955
#ifdef CONFIG_TCP_MD5SIG
	/* Clean up the MD5 key list, if any */
	if (tp->md5sig_info) {
E
Eric Dumazet 已提交
1956
		tcp_clear_md5_list(sk);
1957
		kfree_rcu(tp->md5sig_info, rcu);
1958 1959 1960 1961
		tp->md5sig_info = NULL;
	}
#endif

C
Chris Leech 已提交
1962 1963
#ifdef CONFIG_NET_DMA
	/* Cleans up our sk_async_wait_queue */
1964
	__skb_queue_purge(&sk->sk_async_wait_queue);
C
Chris Leech 已提交
1965 1966
#endif

L
Linus Torvalds 已提交
1967 1968 1969 1970
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
1971
	if (inet_csk(sk)->icsk_bind_hash)
1972
		inet_put_port(sk);
L
Linus Torvalds 已提交
1973

1974
	BUG_ON(tp->fastopen_rsk != NULL);
1975

1976 1977 1978
	/* If socket is aborted during connect operation */
	tcp_free_fastopen_req(tp);

1979
	sk_sockets_allocated_dec(sk);
G
Glauber Costa 已提交
1980
	sock_release_memcg(sk);
L
Linus Torvalds 已提交
1981 1982 1983 1984 1985 1986
}
EXPORT_SYMBOL(tcp_v4_destroy_sock);

#ifdef CONFIG_PROC_FS
/* Proc filesystem TCP sock list dumping. */

1987 1988 1989 1990 1991
/*
 * Get next listener socket follow cur.  If cur is NULL, get first socket
 * starting from bucket given in st->bucket; when st->bucket is zero the
 * very first socket in the hash table is returned.
 */
L
Linus Torvalds 已提交
1992 1993
static void *listening_get_next(struct seq_file *seq, void *cur)
{
1994
	struct inet_connection_sock *icsk;
1995
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
1996
	struct sock *sk = cur;
1997
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
1998
	struct tcp_iter_state *st = seq->private;
1999
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2000 2001

	if (!sk) {
2002
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
2003
		spin_lock_bh(&ilb->lock);
2004
		sk = sk_nulls_head(&ilb->head);
2005
		st->offset = 0;
L
Linus Torvalds 已提交
2006 2007
		goto get_sk;
	}
2008
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
2009
	++st->num;
2010
	++st->offset;
L
Linus Torvalds 已提交
2011 2012

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
2013
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
2014

2015
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
2016 2017 2018
		req = req->dl_next;
		while (1) {
			while (req) {
2019
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
2020 2021 2022 2023 2024
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
2025
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
2026 2027
				break;
get_req:
2028
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
2029
		}
E
Eric Dumazet 已提交
2030
		sk	  = sk_nulls_next(st->syn_wait_sk);
L
Linus Torvalds 已提交
2031
		st->state = TCP_SEQ_STATE_LISTENING;
2032
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2033
	} else {
2034
		icsk = inet_csk(sk);
2035 2036
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
2037
			goto start_req;
2038
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
E
Eric Dumazet 已提交
2039
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2040 2041
	}
get_sk:
2042
	sk_nulls_for_each_from(sk, node) {
2043 2044 2045
		if (!net_eq(sock_net(sk), net))
			continue;
		if (sk->sk_family == st->family) {
L
Linus Torvalds 已提交
2046 2047 2048
			cur = sk;
			goto out;
		}
2049
		icsk = inet_csk(sk);
2050 2051
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
2052 2053 2054 2055 2056 2057 2058
start_req:
			st->uid		= sock_i_uid(sk);
			st->syn_wait_sk = sk;
			st->state	= TCP_SEQ_STATE_OPENREQ;
			st->sbucket	= 0;
			goto get_req;
		}
2059
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2060
	}
2061
	spin_unlock_bh(&ilb->lock);
2062
	st->offset = 0;
2063
	if (++st->bucket < INET_LHTABLE_SIZE) {
2064 2065
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
2066
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
2067 2068 2069 2070 2071 2072 2073 2074 2075
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
2076 2077 2078 2079 2080 2081
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	st->offset = 0;
	rc = listening_get_next(seq, NULL);
L
Linus Torvalds 已提交
2082 2083 2084 2085 2086 2087 2088 2089

	while (rc && *pos) {
		rc = listening_get_next(seq, rc);
		--*pos;
	}
	return rc;
}

E
Eric Dumazet 已提交
2090
static inline bool empty_bucket(const struct tcp_iter_state *st)
2091
{
E
Eric Dumazet 已提交
2092
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain);
2093 2094
}

2095 2096 2097 2098
/*
 * Get first established socket starting from bucket given in st->bucket.
 * If st->bucket is zero, the very first socket in the hash is returned.
 */
L
Linus Torvalds 已提交
2099 2100
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
2101
	struct tcp_iter_state *st = seq->private;
2102
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2103 2104
	void *rc = NULL;

2105 2106
	st->offset = 0;
	for (; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
2107
		struct sock *sk;
2108
		struct hlist_nulls_node *node;
2109
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
2110

2111 2112 2113 2114
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

2115
		spin_lock_bh(lock);
2116
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
2117
			if (sk->sk_family != st->family ||
2118
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2119 2120 2121 2122 2123
				continue;
			}
			rc = sk;
			goto out;
		}
2124
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2125 2126 2127 2128 2129 2130 2131 2132
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2133
	struct hlist_nulls_node *node;
J
Jianjun Kong 已提交
2134
	struct tcp_iter_state *st = seq->private;
2135
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2136 2137

	++st->num;
2138
	++st->offset;
L
Linus Torvalds 已提交
2139

E
Eric Dumazet 已提交
2140
	sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2141

2142
	sk_nulls_for_each_from(sk, node) {
2143
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
E
Eric Dumazet 已提交
2144
			return sk;
L
Linus Torvalds 已提交
2145 2146
	}

E
Eric Dumazet 已提交
2147 2148 2149
	spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
	++st->bucket;
	return established_get_first(seq);
L
Linus Torvalds 已提交
2150 2151 2152 2153
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
2154 2155 2156 2157 2158
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	rc = established_get_first(seq);
L
Linus Torvalds 已提交
2159 2160 2161 2162

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2163
	}
L
Linus Torvalds 已提交
2164 2165 2166 2167 2168 2169
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2170
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182

	st->state = TCP_SEQ_STATE_LISTENING;
	rc	  = listening_get_idx(seq, &pos);

	if (!rc) {
		st->state = TCP_SEQ_STATE_ESTABLISHED;
		rc	  = established_get_idx(seq, pos);
	}

	return rc;
}

2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201
static void *tcp_seek_last_pos(struct seq_file *seq)
{
	struct tcp_iter_state *st = seq->private;
	int offset = st->offset;
	int orig_num = st->num;
	void *rc = NULL;

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
	case TCP_SEQ_STATE_LISTENING:
		if (st->bucket >= INET_LHTABLE_SIZE)
			break;
		st->state = TCP_SEQ_STATE_LISTENING;
		rc = listening_get_next(seq, NULL);
		while (offset-- && rc)
			rc = listening_get_next(seq, rc);
		if (rc)
			break;
		st->bucket = 0;
E
Eric Dumazet 已提交
2202
		st->state = TCP_SEQ_STATE_ESTABLISHED;
2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216
		/* Fallthrough */
	case TCP_SEQ_STATE_ESTABLISHED:
		if (st->bucket > tcp_hashinfo.ehash_mask)
			break;
		rc = established_get_first(seq);
		while (offset-- && rc)
			rc = established_get_next(seq, rc);
	}

	st->num = orig_num;

	return rc;
}

L
Linus Torvalds 已提交
2217 2218
static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2219
	struct tcp_iter_state *st = seq->private;
2220 2221 2222 2223 2224 2225 2226 2227
	void *rc;

	if (*pos && *pos == st->last_pos) {
		rc = tcp_seek_last_pos(seq);
		if (rc)
			goto out;
	}

L
Linus Torvalds 已提交
2228 2229
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
2230 2231 2232 2233 2234 2235 2236
	st->bucket = 0;
	st->offset = 0;
	rc = *pos ? tcp_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;

out:
	st->last_pos = *pos;
	return rc;
L
Linus Torvalds 已提交
2237 2238 2239 2240
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2241
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254
	void *rc = NULL;

	if (v == SEQ_START_TOKEN) {
		rc = tcp_get_idx(seq, 0);
		goto out;
	}

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
	case TCP_SEQ_STATE_LISTENING:
		rc = listening_get_next(seq, v);
		if (!rc) {
			st->state = TCP_SEQ_STATE_ESTABLISHED;
2255 2256
			st->bucket = 0;
			st->offset = 0;
L
Linus Torvalds 已提交
2257 2258 2259 2260 2261 2262 2263 2264 2265
			rc	  = established_get_first(seq);
		}
		break;
	case TCP_SEQ_STATE_ESTABLISHED:
		rc = established_get_next(seq, v);
		break;
	}
out:
	++*pos;
2266
	st->last_pos = *pos;
L
Linus Torvalds 已提交
2267 2268 2269 2270 2271
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2272
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2273 2274 2275 2276

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2277 2278
			struct inet_connection_sock *icsk = inet_csk(st->syn_wait_sk);
			read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2279 2280 2281
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2282
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2283 2284 2285
		break;
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2286
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2287 2288 2289 2290
		break;
	}
}

2291
int tcp_seq_open(struct inode *inode, struct file *file)
L
Linus Torvalds 已提交
2292
{
A
Al Viro 已提交
2293
	struct tcp_seq_afinfo *afinfo = PDE_DATA(inode);
L
Linus Torvalds 已提交
2294
	struct tcp_iter_state *s;
2295
	int err;
L
Linus Torvalds 已提交
2296

2297 2298 2299 2300
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2301

2302
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2303
	s->family		= afinfo->family;
2304
	s->last_pos 		= 0;
2305 2306
	return 0;
}
2307
EXPORT_SYMBOL(tcp_seq_open);
2308

2309
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2310 2311 2312 2313
{
	int rc = 0;
	struct proc_dir_entry *p;

2314 2315 2316 2317
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2318
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
2319
			     afinfo->seq_fops, afinfo);
2320
	if (!p)
L
Linus Torvalds 已提交
2321 2322 2323
		rc = -ENOMEM;
	return rc;
}
E
Eric Dumazet 已提交
2324
EXPORT_SYMBOL(tcp_proc_register);
L
Linus Torvalds 已提交
2325

2326
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2327
{
2328
	remove_proc_entry(afinfo->name, net->proc_net);
L
Linus Torvalds 已提交
2329
}
E
Eric Dumazet 已提交
2330
EXPORT_SYMBOL(tcp_proc_unregister);
L
Linus Torvalds 已提交
2331

2332
static void get_openreq4(const struct sock *sk, const struct request_sock *req,
2333
			 struct seq_file *f, int i, kuid_t uid)
L
Linus Torvalds 已提交
2334
{
2335
	const struct inet_request_sock *ireq = inet_rsk(req);
2336
	long delta = req->expires - jiffies;
L
Linus Torvalds 已提交
2337

2338
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
2339
		" %02X %08X:%08X %02X:%08lX %08X %5u %8d %u %d %pK",
L
Linus Torvalds 已提交
2340
		i,
2341
		ireq->ir_loc_addr,
E
Eric Dumazet 已提交
2342
		ntohs(inet_sk(sk)->inet_sport),
2343 2344
		ireq->ir_rmt_addr,
		ntohs(ireq->ir_rmt_port),
L
Linus Torvalds 已提交
2345 2346 2347
		TCP_SYN_RECV,
		0, 0, /* could print option size, but that is af dependent. */
		1,    /* timers active (only the expire timer) */
2348
		jiffies_delta_to_clock_t(delta),
2349
		req->num_timeout,
2350
		from_kuid_munged(seq_user_ns(f), uid),
L
Linus Torvalds 已提交
2351 2352 2353
		0,  /* non standard timer */
		0, /* open_requests have no inode */
		atomic_read(&sk->sk_refcnt),
2354
		req);
L
Linus Torvalds 已提交
2355 2356
}

2357
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i)
L
Linus Torvalds 已提交
2358 2359 2360
{
	int timer_active;
	unsigned long timer_expires;
2361
	const struct tcp_sock *tp = tcp_sk(sk);
2362
	const struct inet_connection_sock *icsk = inet_csk(sk);
2363
	const struct inet_sock *inet = inet_sk(sk);
2364
	struct fastopen_queue *fastopenq = icsk->icsk_accept_queue.fastopenq;
E
Eric Dumazet 已提交
2365 2366 2367 2368
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
2369
	int rx_queue;
L
Linus Torvalds 已提交
2370

N
Nandita Dukkipati 已提交
2371 2372 2373
	if (icsk->icsk_pending == ICSK_TIME_RETRANS ||
	    icsk->icsk_pending == ICSK_TIME_EARLY_RETRANS ||
	    icsk->icsk_pending == ICSK_TIME_LOSS_PROBE) {
L
Linus Torvalds 已提交
2374
		timer_active	= 1;
2375 2376
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2377
		timer_active	= 4;
2378
		timer_expires	= icsk->icsk_timeout;
2379
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2380
		timer_active	= 2;
2381
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2382 2383 2384 2385 2386
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2387 2388 2389 2390 2391 2392 2393 2394
	if (sk->sk_state == TCP_LISTEN)
		rx_queue = sk->sk_ack_backlog;
	else
		/*
		 * because we dont lock socket, we might find a transient negative value
		 */
		rx_queue = max_t(int, tp->rcv_nxt - tp->copied_seq, 0);

2395
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
2396
			"%08X %5u %8d %lu %d %pK %lu %lu %u %u %d",
2397
		i, src, srcp, dest, destp, sk->sk_state,
2398
		tp->write_seq - tp->snd_una,
2399
		rx_queue,
L
Linus Torvalds 已提交
2400
		timer_active,
2401
		jiffies_delta_to_clock_t(timer_expires - jiffies),
2402
		icsk->icsk_retransmits,
2403
		from_kuid_munged(seq_user_ns(f), sock_i_uid(sk)),
2404
		icsk->icsk_probes_out,
2405 2406
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2407 2408
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2409
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2410
		tp->snd_cwnd,
2411 2412
		sk->sk_state == TCP_LISTEN ?
		    (fastopenq ? fastopenq->max_qlen : 0) :
2413
		    (tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh));
L
Linus Torvalds 已提交
2414 2415
}

2416
static void get_timewait4_sock(const struct inet_timewait_sock *tw,
2417
			       struct seq_file *f, int i)
L
Linus Torvalds 已提交
2418
{
2419
	__be32 dest, src;
L
Linus Torvalds 已提交
2420
	__u16 destp, srcp;
2421
	s32 delta = tw->tw_ttd - inet_tw_time_stamp();
L
Linus Torvalds 已提交
2422 2423 2424 2425 2426 2427

	dest  = tw->tw_daddr;
	src   = tw->tw_rcv_saddr;
	destp = ntohs(tw->tw_dport);
	srcp  = ntohs(tw->tw_sport);

2428
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
2429
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK",
L
Linus Torvalds 已提交
2430
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
2431
		3, jiffies_delta_to_clock_t(delta), 0, 0, 0, 0,
2432
		atomic_read(&tw->tw_refcnt), tw);
L
Linus Torvalds 已提交
2433 2434 2435 2436 2437 2438
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2439
	struct tcp_iter_state *st;
E
Eric Dumazet 已提交
2440
	struct sock *sk = v;
L
Linus Torvalds 已提交
2441

2442
	seq_setwidth(seq, TMPSZ - 1);
L
Linus Torvalds 已提交
2443
	if (v == SEQ_START_TOKEN) {
2444
		seq_puts(seq, "  sl  local_address rem_address   st tx_queue "
L
Linus Torvalds 已提交
2445 2446 2447 2448 2449 2450 2451 2452 2453
			   "rx_queue tr tm->when retrnsmt   uid  timeout "
			   "inode");
		goto out;
	}
	st = seq->private;

	switch (st->state) {
	case TCP_SEQ_STATE_LISTENING:
	case TCP_SEQ_STATE_ESTABLISHED:
E
Eric Dumazet 已提交
2454
		if (sk->sk_state == TCP_TIME_WAIT)
2455
			get_timewait4_sock(v, seq, st->num);
E
Eric Dumazet 已提交
2456
		else
2457
			get_tcp4_sock(v, seq, st->num);
L
Linus Torvalds 已提交
2458 2459
		break;
	case TCP_SEQ_STATE_OPENREQ:
2460
		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid);
L
Linus Torvalds 已提交
2461 2462 2463
		break;
	}
out:
2464
	seq_pad(seq, '\n');
L
Linus Torvalds 已提交
2465 2466 2467
	return 0;
}

2468 2469 2470 2471 2472 2473 2474 2475
static const struct file_operations tcp_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
2476 2477 2478
static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2479
	.seq_fops	= &tcp_afinfo_seq_fops,
2480 2481 2482
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2483 2484
};

2485
static int __net_init tcp4_proc_init_net(struct net *net)
2486 2487 2488 2489
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

2490
static void __net_exit tcp4_proc_exit_net(struct net *net)
2491 2492 2493 2494 2495 2496 2497 2498 2499
{
	tcp_proc_unregister(net, &tcp4_seq_afinfo);
}

static struct pernet_operations tcp4_net_ops = {
	.init = tcp4_proc_init_net,
	.exit = tcp4_proc_exit_net,
};

L
Linus Torvalds 已提交
2500 2501
int __init tcp4_proc_init(void)
{
2502
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2503 2504 2505 2506
}

void tcp4_proc_exit(void)
{
2507
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2508 2509 2510 2511 2512 2513 2514 2515 2516
}
#endif /* CONFIG_PROC_FS */

struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2517
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2518 2519 2520 2521 2522 2523 2524
	.ioctl			= tcp_ioctl,
	.init			= tcp_v4_init_sock,
	.destroy		= tcp_v4_destroy_sock,
	.shutdown		= tcp_shutdown,
	.setsockopt		= tcp_setsockopt,
	.getsockopt		= tcp_getsockopt,
	.recvmsg		= tcp_recvmsg,
2525 2526
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2527
	.backlog_rcv		= tcp_v4_do_rcv,
E
Eric Dumazet 已提交
2528
	.release_cb		= tcp_release_cb,
2529
	.mtu_reduced		= tcp_v4_mtu_reduced,
2530 2531 2532
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2533
	.enter_memory_pressure	= tcp_enter_memory_pressure,
2534
	.stream_memory_free	= tcp_stream_memory_free,
L
Linus Torvalds 已提交
2535
	.sockets_allocated	= &tcp_sockets_allocated,
2536
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2537 2538
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
2539
	.sysctl_mem		= sysctl_tcp_mem,
L
Linus Torvalds 已提交
2540 2541 2542 2543
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp_sock),
2544
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2545
	.twsk_prot		= &tcp_timewait_sock_ops,
2546
	.rsk_prot		= &tcp_request_sock_ops,
2547
	.h.hashinfo		= &tcp_hashinfo,
2548
	.no_autobind		= true,
2549 2550 2551 2552
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
A
Andrew Morton 已提交
2553
#ifdef CONFIG_MEMCG_KMEM
G
Glauber Costa 已提交
2554 2555 2556 2557
	.init_cgroup		= tcp_init_cgroup,
	.destroy_cgroup		= tcp_destroy_cgroup,
	.proto_cgroup		= tcp_proto_cgroup,
#endif
L
Linus Torvalds 已提交
2558
};
E
Eric Dumazet 已提交
2559
EXPORT_SYMBOL(tcp_prot);
L
Linus Torvalds 已提交
2560

2561 2562
static int __net_init tcp_sk_init(struct net *net)
{
2563
	net->ipv4.sysctl_tcp_ecn = 2;
2564
	return 0;
2565 2566 2567 2568
}

static void __net_exit tcp_sk_exit(struct net *net)
{
E
Eric W. Biederman 已提交
2569 2570 2571 2572 2573
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2574 2575 2576
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2577 2578 2579
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2580 2581
};

2582
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2583
{
2584
	inet_hashinfo_init(&tcp_hashinfo);
2585
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2586 2587
		panic("Failed to create the TCP control socket.\n");
}