tcp_ipv4.c 68.0 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 53
 *		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.
 */


H
Herbert Xu 已提交
54
#include <linux/bottom_half.h>
L
Linus Torvalds 已提交
55 56 57 58 59 60 61 62
#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>
63
#include <linux/slab.h>
L
Linus Torvalds 已提交
64

65
#include <net/net_namespace.h>
L
Linus Torvalds 已提交
66
#include <net/icmp.h>
67
#include <net/inet_hashtables.h>
L
Linus Torvalds 已提交
68
#include <net/tcp.h>
69
#include <net/transp_v6.h>
L
Linus Torvalds 已提交
70 71
#include <net/ipv6.h>
#include <net/inet_common.h>
72
#include <net/timewait_sock.h>
L
Linus Torvalds 已提交
73
#include <net/xfrm.h>
C
Chris Leech 已提交
74
#include <net/netdma.h>
75
#include <net/secure_seq.h>
G
Glauber Costa 已提交
76
#include <net/tcp_memcontrol.h>
L
Linus Torvalds 已提交
77 78 79 80 81 82 83

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

84 85 86
#include <linux/crypto.h>
#include <linux/scatterlist.h>

87 88
int sysctl_tcp_tw_reuse __read_mostly;
int sysctl_tcp_low_latency __read_mostly;
E
Eric Dumazet 已提交
89
EXPORT_SYMBOL(sysctl_tcp_low_latency);
L
Linus Torvalds 已提交
90 91


92
#ifdef CONFIG_TCP_MD5SIG
E
Eric Dumazet 已提交
93
static int tcp_v4_md5_hash_hdr(char *md5_hash, const struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
94
			       __be32 daddr, __be32 saddr, const struct tcphdr *th);
95 96
#endif

97
struct inet_hashinfo tcp_hashinfo;
E
Eric Dumazet 已提交
98
EXPORT_SYMBOL(tcp_hashinfo);
L
Linus Torvalds 已提交
99

100
static inline __u32 tcp_v4_init_sequence(const struct sk_buff *skb)
L
Linus Torvalds 已提交
101
{
102 103
	return secure_tcp_sequence_number(ip_hdr(skb)->daddr,
					  ip_hdr(skb)->saddr,
104 105
					  tcp_hdr(skb)->dest,
					  tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
106 107
}

108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125
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 &&
126
			     get_seconds() - tcptw->tw_ts_recent_stamp > 1))) {
127 128 129 130 131 132 133 134 135 136 137 138 139
		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
140 141 142
/* This will initiate an outgoing connection. */
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
{
143
	struct sockaddr_in *usin = (struct sockaddr_in *)uaddr;
L
Linus Torvalds 已提交
144 145
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
146
	__be16 orig_sport, orig_dport;
147
	__be32 daddr, nexthop;
148
	struct flowi4 *fl4;
149
	struct rtable *rt;
L
Linus Torvalds 已提交
150
	int err;
151
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
152 153 154 155 156 157 158 159

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

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

	nexthop = daddr = usin->sin_addr.s_addr;
160 161 162
	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
163 164
		if (!daddr)
			return -EINVAL;
165
		nexthop = inet_opt->opt.faddr;
L
Linus Torvalds 已提交
166 167
	}

168 169
	orig_sport = inet->inet_sport;
	orig_dport = usin->sin_port;
170 171
	fl4 = &inet->cork.fl.u.ip4;
	rt = ip_route_connect(fl4, nexthop, inet->inet_saddr,
172 173 174 175 176 177
			      RT_CONN_FLAGS(sk), sk->sk_bound_dev_if,
			      IPPROTO_TCP,
			      orig_sport, orig_dport, sk, true);
	if (IS_ERR(rt)) {
		err = PTR_ERR(rt);
		if (err == -ENETUNREACH)
178
			IP_INC_STATS_BH(sock_net(sk), IPSTATS_MIB_OUTNOROUTES);
179
		return err;
180
	}
L
Linus Torvalds 已提交
181 182 183 184 185 186

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

187
	if (!inet_opt || !inet_opt->opt.srr)
188
		daddr = fl4->daddr;
L
Linus Torvalds 已提交
189

E
Eric Dumazet 已提交
190
	if (!inet->inet_saddr)
191
		inet->inet_saddr = fl4->saddr;
E
Eric Dumazet 已提交
192
	inet->inet_rcv_saddr = inet->inet_saddr;
L
Linus Torvalds 已提交
193

E
Eric Dumazet 已提交
194
	if (tp->rx_opt.ts_recent_stamp && inet->inet_daddr != daddr) {
L
Linus Torvalds 已提交
195 196 197 198 199 200
		/* Reset inherited state */
		tp->rx_opt.ts_recent	   = 0;
		tp->rx_opt.ts_recent_stamp = 0;
		tp->write_seq		   = 0;
	}

201
	if (tcp_death_row.sysctl_tw_recycle &&
202
	    !tp->rx_opt.ts_recent_stamp && fl4->daddr == daddr) {
203
		struct inet_peer *peer = rt_get_peer(rt, fl4->daddr);
204 205 206 207 208
		/*
		 * VJ's idea. We save last timestamp seen from
		 * the destination in peer table, when entering state
		 * TIME-WAIT * and initialize rx_opt.ts_recent from it,
		 * when trying new connection.
L
Linus Torvalds 已提交
209
		 */
210 211 212 213 214 215
		if (peer) {
			inet_peer_refcheck(peer);
			if ((u32)get_seconds() - peer->tcp_ts_stamp <= TCP_PAWS_MSL) {
				tp->rx_opt.ts_recent_stamp = peer->tcp_ts_stamp;
				tp->rx_opt.ts_recent = peer->tcp_ts;
			}
L
Linus Torvalds 已提交
216 217 218
		}
	}

E
Eric Dumazet 已提交
219 220
	inet->inet_dport = usin->sin_port;
	inet->inet_daddr = daddr;
L
Linus Torvalds 已提交
221

222
	inet_csk(sk)->icsk_ext_hdr_len = 0;
223 224
	if (inet_opt)
		inet_csk(sk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
L
Linus Torvalds 已提交
225

226
	tp->rx_opt.mss_clamp = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
227 228 229 230 231 232 233

	/* 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);
234
	err = inet_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
235 236 237
	if (err)
		goto failure;

238
	rt = ip_route_newports(fl4, rt, orig_sport, orig_dport,
239 240 241 242
			       inet->inet_sport, inet->inet_dport, sk);
	if (IS_ERR(rt)) {
		err = PTR_ERR(rt);
		rt = NULL;
L
Linus Torvalds 已提交
243
		goto failure;
244
	}
L
Linus Torvalds 已提交
245
	/* OK, now commit destination to socket.  */
246
	sk->sk_gso_type = SKB_GSO_TCPV4;
247
	sk_setup_caps(sk, &rt->dst);
L
Linus Torvalds 已提交
248 249

	if (!tp->write_seq)
E
Eric Dumazet 已提交
250 251 252
		tp->write_seq = secure_tcp_sequence_number(inet->inet_saddr,
							   inet->inet_daddr,
							   inet->inet_sport,
L
Linus Torvalds 已提交
253 254
							   usin->sin_port);

E
Eric Dumazet 已提交
255
	inet->inet_id = tp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
256 257 258 259 260 261 262 263 264

	err = tcp_connect(sk);
	rt = NULL;
	if (err)
		goto failure;

	return 0;

failure:
265 266 267 268
	/*
	 * This unhashes the socket and releases the local port,
	 * if necessary.
	 */
L
Linus Torvalds 已提交
269 270 271
	tcp_set_state(sk, TCP_CLOSE);
	ip_rt_put(rt);
	sk->sk_route_caps = 0;
E
Eric Dumazet 已提交
272
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
273 274
	return err;
}
E
Eric Dumazet 已提交
275
EXPORT_SYMBOL(tcp_v4_connect);
L
Linus Torvalds 已提交
276 277 278 279

/*
 * This routine does path mtu discovery as defined in RFC1191.
 */
280
static void do_pmtu_discovery(struct sock *sk, const struct iphdr *iph, u32 mtu)
L
Linus Torvalds 已提交
281 282 283 284 285 286 287 288 289 290 291 292 293 294
{
	struct dst_entry *dst;
	struct inet_sock *inet = inet_sk(sk);

	/* 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)
		return;

	/* We don't check in the destentry if pmtu discovery is forbidden
	 * on this route. We just assume that no packet_to_big packets
	 * are send back when pmtu discovery is not active.
295
	 * There is a small race when the user changes this flag in the
L
Linus Torvalds 已提交
296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
	 * route, but I think that's acceptable.
	 */
	if ((dst = __sk_dst_check(sk, 0)) == NULL)
		return;

	dst->ops->update_pmtu(dst, mtu);

	/* 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 &&
312
	    inet_csk(sk)->icsk_pmtu_cookie > mtu) {
L
Linus Torvalds 已提交
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339
		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 */
}

/*
 * 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.
 *
 */

340
void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
L
Linus Torvalds 已提交
341
{
342
	const struct iphdr *iph = (const struct iphdr *)icmp_skb->data;
343
	struct tcphdr *th = (struct tcphdr *)(icmp_skb->data + (iph->ihl << 2));
344
	struct inet_connection_sock *icsk;
L
Linus Torvalds 已提交
345 346
	struct tcp_sock *tp;
	struct inet_sock *inet;
347 348
	const int type = icmp_hdr(icmp_skb)->type;
	const int code = icmp_hdr(icmp_skb)->code;
L
Linus Torvalds 已提交
349
	struct sock *sk;
350
	struct sk_buff *skb;
L
Linus Torvalds 已提交
351
	__u32 seq;
352
	__u32 remaining;
L
Linus Torvalds 已提交
353
	int err;
354
	struct net *net = dev_net(icmp_skb->dev);
L
Linus Torvalds 已提交
355

356
	if (icmp_skb->len < (iph->ihl << 2) + 8) {
357
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
358 359 360
		return;
	}

361
	sk = inet_lookup(net, &tcp_hashinfo, iph->daddr, th->dest,
362
			iph->saddr, th->source, inet_iif(icmp_skb));
L
Linus Torvalds 已提交
363
	if (!sk) {
364
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
365 366 367
		return;
	}
	if (sk->sk_state == TCP_TIME_WAIT) {
368
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
369 370 371 372 373 374 375 376
		return;
	}

	bh_lock_sock(sk);
	/* If too many ICMPs get dropped on busy
	 * servers this needs to be solved differently.
	 */
	if (sock_owned_by_user(sk))
377
		NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
L
Linus Torvalds 已提交
378 379 380 381

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

382 383 384 385 386
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto out;
	}

387
	icsk = inet_csk(sk);
L
Linus Torvalds 已提交
388 389 390 391
	tp = tcp_sk(sk);
	seq = ntohl(th->seq);
	if (sk->sk_state != TCP_LISTEN &&
	    !between(seq, tp->snd_una, tp->snd_nxt)) {
392
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
		goto out;
	}

	switch (type) {
	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) */
			if (!sock_owned_by_user(sk))
				do_pmtu_discovery(sk, iph, info);
			goto out;
		}

		err = icmp_err_convert[code].errno;
414 415 416 417 418 419 420 421
		/* 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 ||
		    !icsk->icsk_backoff)
			break;

422 423 424
		if (sock_owned_by_user(sk))
			break;

425
		icsk->icsk_backoff--;
426 427
		inet_csk(sk)->icsk_rto = (tp->srtt ? __tcp_set_rto(tp) :
			TCP_TIMEOUT_INIT) << icsk->icsk_backoff;
428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
		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
445 446 447 448 449 450 451 452 453
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

	switch (sk->sk_state) {
454
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
455 456 457 458
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

459 460
		req = inet_csk_search_req(sk, &prev, th->dest,
					  iph->daddr, iph->saddr);
L
Linus Torvalds 已提交
461 462 463 464 465 466
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		   an established socket here.
		 */
467
		WARN_ON(req->sk);
L
Linus Torvalds 已提交
468

469
		if (seq != tcp_rsk(req)->snt_isn) {
470
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
471 472 473 474 475 476 477 478 479
			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().
		 */
480
		inet_csk_reqsk_queue_drop(sk, req, prev);
L
Linus Torvalds 已提交
481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
		goto out;

	case TCP_SYN_SENT:
	case TCP_SYN_RECV:  /* Cannot happen.
			       It can f.e. if SYNs crossed.
			     */
		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);
}

528 529
static void __tcp_v4_send_check(struct sk_buff *skb,
				__be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
530
{
531
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
532

533
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
534
		th->check = ~tcp_v4_check(skb->len, saddr, daddr, 0);
535
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
536
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
537
	} else {
538
		th->check = tcp_v4_check(skb->len, saddr, daddr,
539
					 csum_partial(th,
L
Linus Torvalds 已提交
540 541 542 543 544
						      th->doff << 2,
						      skb->csum));
	}
}

545
/* This routine computes an IPv4 TCP checksum. */
546
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb)
547
{
548
	const struct inet_sock *inet = inet_sk(sk);
549 550 551

	__tcp_v4_send_check(skb, inet->inet_saddr, inet->inet_daddr);
}
E
Eric Dumazet 已提交
552
EXPORT_SYMBOL(tcp_v4_send_check);
553

554 555
int tcp_v4_gso_send_check(struct sk_buff *skb)
{
556
	const struct iphdr *iph;
557 558 559 560 561
	struct tcphdr *th;

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

562
	iph = ip_hdr(skb);
563
	th = tcp_hdr(skb);
564 565

	th->check = 0;
566
	skb->ip_summed = CHECKSUM_PARTIAL;
567
	__tcp_v4_send_check(skb, iph->saddr, iph->daddr);
568 569 570
	return 0;
}

L
Linus Torvalds 已提交
571 572 573 574 575 576 577 578 579 580 581 582 583
/*
 *	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.
 */

584
static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
585
{
586
	const struct tcphdr *th = tcp_hdr(skb);
587 588 589
	struct {
		struct tcphdr th;
#ifdef CONFIG_TCP_MD5SIG
590
		__be32 opt[(TCPOLEN_MD5SIG_ALIGNED >> 2)];
591 592
#endif
	} rep;
L
Linus Torvalds 已提交
593
	struct ip_reply_arg arg;
594 595
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
596 597 598 599
	const __u8 *hash_location = NULL;
	unsigned char newhash[16];
	int genhash;
	struct sock *sk1 = NULL;
600
#endif
601
	struct net *net;
L
Linus Torvalds 已提交
602 603 604 605 606

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

E
Eric Dumazet 已提交
607
	if (skb_rtable(skb)->rt_type != RTN_LOCAL)
L
Linus Torvalds 已提交
608 609 610
		return;

	/* Swap the send and the receive. */
611 612 613 614 615
	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
616 617

	if (th->ack) {
618
		rep.th.seq = th->ack_seq;
L
Linus Torvalds 已提交
619
	} else {
620 621 622
		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
623 624
	}

625
	memset(&arg, 0, sizeof(arg));
626 627 628 629
	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);

#ifdef CONFIG_TCP_MD5SIG
630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659
	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),
					     &tcp_hashinfo, ip_hdr(skb)->daddr,
					     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;
	}

660 661 662 663 664 665 666 667 668
	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;

669
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
670 671
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
672 673
	}
#endif
674 675
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
676
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
677
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
678
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
679 680 681 682 683
	/* When socket is gone, all binding information is lost.
	 * routing might fail in this case. using iif for oif to
	 * make sure we can deliver it
	 */
	arg.bound_dev_if = sk ? sk->sk_bound_dev_if : inet_iif(skb);
L
Linus Torvalds 已提交
684

E
Eric Dumazet 已提交
685
	net = dev_net(skb_dst(skb)->dev);
686
	arg.tos = ip_hdr(skb)->tos;
687
	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
688
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
689

690 691
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
692 693 694 695 696 697 698 699

#ifdef CONFIG_TCP_MD5SIG
release_sk1:
	if (sk1) {
		rcu_read_unlock();
		sock_put(sk1);
	}
#endif
L
Linus Torvalds 已提交
700 701 702 703 704 705
}

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

706 707
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
			    u32 win, u32 ts, int oif,
708
			    struct tcp_md5sig_key *key,
709
			    int reply_flags, u8 tos)
L
Linus Torvalds 已提交
710
{
711
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
712 713
	struct {
		struct tcphdr th;
714
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
715
#ifdef CONFIG_TCP_MD5SIG
716
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
717 718
#endif
			];
L
Linus Torvalds 已提交
719 720
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
721
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
722 723

	memset(&rep.th, 0, sizeof(struct tcphdr));
724
	memset(&arg, 0, sizeof(arg));
L
Linus Torvalds 已提交
725 726 727 728

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
	if (ts) {
729 730 731 732 733
		rep.opt[0] = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				   (TCPOPT_TIMESTAMP << 8) |
				   TCPOLEN_TIMESTAMP);
		rep.opt[1] = htonl(tcp_time_stamp);
		rep.opt[2] = htonl(ts);
734
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
735 736 737 738 739 740 741 742 743 744 745
	}

	/* 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);

746 747 748 749 750 751 752 753 754 755 756
#ifdef CONFIG_TCP_MD5SIG
	if (key) {
		int offset = (ts) ? 3 : 0;

		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;

757
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[offset],
758 759
				    key, ip_hdr(skb)->saddr,
				    ip_hdr(skb)->daddr, &rep.th);
760 761
	}
#endif
762
	arg.flags = reply_flags;
763 764
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
L
Linus Torvalds 已提交
765 766
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
767 768
	if (oif)
		arg.bound_dev_if = oif;
769
	arg.tos = tos;
770
	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
771
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
772

773
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
774 775 776 777
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
778
	struct inet_timewait_sock *tw = inet_twsk(sk);
779
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
780

781
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
782
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
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
	tcp_v4_send_ack(skb, tcp_rsk(req)->snt_isn + 1,
797
			tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd,
798 799
			req->ts_recent,
			0,
E
Eric Dumazet 已提交
800 801
			tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&ip_hdr(skb)->daddr,
					  AF_INET),
802 803
			inet_rsk(req)->no_srccheck ? IP_REPLY_ARG_NOSRCCHECK : 0,
			ip_hdr(skb)->tos);
L
Linus Torvalds 已提交
804 805 806
}

/*
807
 *	Send a SYN-ACK after having received a SYN.
808
 *	This still operates on a request_sock only, not on a big
L
Linus Torvalds 已提交
809 810
 *	socket.
 */
811 812 813
static int tcp_v4_send_synack(struct sock *sk, struct dst_entry *dst,
			      struct request_sock *req,
			      struct request_values *rvp)
L
Linus Torvalds 已提交
814
{
815
	const struct inet_request_sock *ireq = inet_rsk(req);
816
	struct flowi4 fl4;
L
Linus Torvalds 已提交
817 818 819 820
	int err = -1;
	struct sk_buff * skb;

	/* First, grab a route. */
821
	if (!dst && (dst = inet_csk_route_req(sk, &fl4, req)) == NULL)
822
		return -1;
L
Linus Torvalds 已提交
823

824
	skb = tcp_make_synack(sk, dst, req, rvp);
L
Linus Torvalds 已提交
825 826

	if (skb) {
827
		__tcp_v4_send_check(skb, ireq->loc_addr, ireq->rmt_addr);
L
Linus Torvalds 已提交
828

829 830 831
		err = ip_build_and_send_pkt(skb, sk, ireq->loc_addr,
					    ireq->rmt_addr,
					    ireq->opt);
832
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
833 834 835 836 837 838
	}

	dst_release(dst);
	return err;
}

839
static int tcp_v4_rtx_synack(struct sock *sk, struct request_sock *req,
840
			      struct request_values *rvp)
841
{
842 843
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
	return tcp_v4_send_synack(sk, NULL, req, rvp);
844 845
}

L
Linus Torvalds 已提交
846
/*
847
 *	IPv4 request_sock destructor.
L
Linus Torvalds 已提交
848
 */
849
static void tcp_v4_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
850
{
J
Jesper Juhl 已提交
851
	kfree(inet_rsk(req)->opt);
L
Linus Torvalds 已提交
852 853
}

854 855 856 857 858 859
/*
 * Return 1 if a syncookie should be sent
 */
int tcp_syn_flood_action(struct sock *sk,
			 const struct sk_buff *skb,
			 const char *proto)
L
Linus Torvalds 已提交
860
{
861 862 863 864 865
	const char *msg = "Dropping request";
	int want_cookie = 0;
	struct listen_sock *lopt;


L
Linus Torvalds 已提交
866

867
#ifdef CONFIG_SYN_COOKIES
868
	if (sysctl_tcp_syncookies) {
869
		msg = "Sending cookies";
870 871 872
		want_cookie = 1;
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDOCOOKIES);
	} else
873
#endif
874 875 876 877 878 879 880 881 882 883
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDROP);

	lopt = inet_csk(sk)->icsk_accept_queue.listen_opt;
	if (!lopt->synflood_warned) {
		lopt->synflood_warned = 1;
		pr_info("%s: Possible SYN flooding on port %d. %s. "
			" Check SNMP counters.\n",
			proto, ntohs(tcp_hdr(skb)->dest), msg);
	}
	return want_cookie;
884
}
885
EXPORT_SYMBOL(tcp_syn_flood_action);
L
Linus Torvalds 已提交
886 887

/*
888
 * Save and compile IPv4 options into the request_sock if needed.
L
Linus Torvalds 已提交
889
 */
890 891
static struct ip_options_rcu *tcp_v4_save_options(struct sock *sk,
						  struct sk_buff *skb)
L
Linus Torvalds 已提交
892
{
893 894
	const struct ip_options *opt = &(IPCB(skb)->opt);
	struct ip_options_rcu *dopt = NULL;
L
Linus Torvalds 已提交
895 896

	if (opt && opt->optlen) {
897 898
		int opt_size = sizeof(*dopt) + opt->optlen;

L
Linus Torvalds 已提交
899 900
		dopt = kmalloc(opt_size, GFP_ATOMIC);
		if (dopt) {
901
			if (ip_options_echo(&dopt->opt, skb)) {
L
Linus Torvalds 已提交
902 903 904 905 906 907 908 909
				kfree(dopt);
				dopt = NULL;
			}
		}
	}
	return dopt;
}

910 911 912 913 914 915 916 917
#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
918 919 920
struct tcp_md5sig_key *tcp_md5_do_lookup(struct sock *sk,
					 const union tcp_md5_addr *addr,
					 int family)
921 922
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
923 924 925
	struct tcp_md5sig_key *key;
	struct hlist_node *pos;
	unsigned int size = sizeof(struct in_addr);
926
	struct tcp_md5sig_info *md5sig;
927

928 929
	/* caller either holds rcu_read_lock() or socket lock */
	md5sig = rcu_dereference_check(tp->md5sig_info,
930 931
				       sock_owned_by_user(sk) ||
				       lockdep_is_held(&sk->sk_lock.slock));
932
	if (!md5sig)
933
		return NULL;
E
Eric Dumazet 已提交
934 935 936 937
#if IS_ENABLED(CONFIG_IPV6)
	if (family == AF_INET6)
		size = sizeof(struct in6_addr);
#endif
938
	hlist_for_each_entry_rcu(key, pos, &md5sig->head, node) {
E
Eric Dumazet 已提交
939 940 941 942
		if (key->family != family)
			continue;
		if (!memcmp(&key->addr, addr, size))
			return key;
943 944 945
	}
	return NULL;
}
E
Eric Dumazet 已提交
946
EXPORT_SYMBOL(tcp_md5_do_lookup);
947 948 949 950

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
951 952 953 954
	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);
955 956 957
}
EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
958 959
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
960
{
E
Eric Dumazet 已提交
961 962 963 964
	union tcp_md5_addr *addr;

	addr = (union tcp_md5_addr *)&inet_rsk(req)->rmt_addr;
	return tcp_md5_do_lookup(sk, addr, AF_INET);
965 966 967
}

/* This can be called on a newly created socket, from other files */
E
Eric Dumazet 已提交
968 969
int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
		   int family, const u8 *newkey, u8 newkeylen, gfp_t gfp)
970 971
{
	/* Add Key to the list */
972
	struct tcp_md5sig_key *key;
973
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
974
	struct tcp_md5sig_info *md5sig;
975

E
Eric Dumazet 已提交
976
	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&addr, AF_INET);
977 978
	if (key) {
		/* Pre-existing entry - just update that one. */
E
Eric Dumazet 已提交
979
		memcpy(key->key, newkey, newkeylen);
980
		key->keylen = newkeylen;
E
Eric Dumazet 已提交
981 982
		return 0;
	}
983

984 985
	md5sig = rcu_dereference_protected(tp->md5sig_info,
					   sock_owned_by_user(sk));
E
Eric Dumazet 已提交
986 987 988
	if (!md5sig) {
		md5sig = kmalloc(sizeof(*md5sig), gfp);
		if (!md5sig)
989 990
			return -ENOMEM;

E
Eric Dumazet 已提交
991 992
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
		INIT_HLIST_HEAD(&md5sig->head);
993
		rcu_assign_pointer(tp->md5sig_info, md5sig);
E
Eric Dumazet 已提交
994
	}
995

996
	key = sock_kmalloc(sk, sizeof(*key), gfp);
E
Eric Dumazet 已提交
997 998 999
	if (!key)
		return -ENOMEM;
	if (hlist_empty(&md5sig->head) && !tcp_alloc_md5sig_pool(sk)) {
1000
		sock_kfree_s(sk, key, sizeof(*key));
E
Eric Dumazet 已提交
1001
		return -ENOMEM;
1002
	}
E
Eric Dumazet 已提交
1003 1004 1005 1006 1007 1008 1009 1010

	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);
1011 1012
	return 0;
}
E
Eric Dumazet 已提交
1013
EXPORT_SYMBOL(tcp_md5_do_add);
1014

E
Eric Dumazet 已提交
1015
int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr, int family)
1016 1017
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
1018
	struct tcp_md5sig_key *key;
1019
	struct tcp_md5sig_info *md5sig;
E
Eric Dumazet 已提交
1020 1021 1022 1023 1024

	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&addr, AF_INET);
	if (!key)
		return -ENOENT;
	hlist_del_rcu(&key->node);
1025
	atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1026
	kfree_rcu(key, rcu);
1027 1028 1029
	md5sig = rcu_dereference_protected(tp->md5sig_info,
					   sock_owned_by_user(sk));
	if (hlist_empty(&md5sig->head))
E
Eric Dumazet 已提交
1030 1031
		tcp_free_md5sig_pool();
	return 0;
1032
}
E
Eric Dumazet 已提交
1033
EXPORT_SYMBOL(tcp_md5_do_del);
1034

E
Eric Dumazet 已提交
1035
void tcp_clear_md5_list(struct sock *sk)
1036 1037
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
1038 1039
	struct tcp_md5sig_key *key;
	struct hlist_node *pos, *n;
1040
	struct tcp_md5sig_info *md5sig;
1041

1042 1043 1044
	md5sig = rcu_dereference_protected(tp->md5sig_info, 1);

	if (!hlist_empty(&md5sig->head))
1045
		tcp_free_md5sig_pool();
1046
	hlist_for_each_entry_safe(key, pos, n, &md5sig->head, node) {
E
Eric Dumazet 已提交
1047
		hlist_del_rcu(&key->node);
1048
		atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1049
		kfree_rcu(key, rcu);
1050 1051 1052
	}
}

1053 1054
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
1055 1056 1057 1058 1059 1060 1061
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;

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

1062
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
1063 1064 1065 1066 1067
		return -EFAULT;

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

1068
	if (!cmd.tcpm_key || !cmd.tcpm_keylen)
E
Eric Dumazet 已提交
1069 1070
		return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin->sin_addr.s_addr,
				      AF_INET);
1071 1072 1073 1074

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

E
Eric Dumazet 已提交
1075 1076 1077
	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);
1078 1079
}

1080 1081
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1082 1083
{
	struct tcp4_pseudohdr *bp;
1084
	struct scatterlist sg;
1085 1086 1087 1088

	bp = &hp->md5_blk.ip4;

	/*
1089
	 * 1. the TCP pseudo-header (in the order: source IP address,
1090 1091 1092 1093 1094 1095
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1096
	bp->protocol = IPPROTO_TCP;
1097
	bp->len = cpu_to_be16(nbytes);
1098

1099 1100 1101 1102
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}

E
Eric Dumazet 已提交
1103
static int tcp_v4_md5_hash_hdr(char *md5_hash, const struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1104
			       __be32 daddr, __be32 saddr, const struct tcphdr *th)
1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122
{
	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))
1123 1124 1125 1126
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1127

1128 1129 1130 1131
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1132
	return 1;
1133 1134
}

1135
int tcp_v4_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1136 1137
			const struct sock *sk, const struct request_sock *req,
			const struct sk_buff *skb)
1138
{
1139 1140
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
E
Eric Dumazet 已提交
1141
	const struct tcphdr *th = tcp_hdr(skb);
1142 1143 1144
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1145 1146
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1147 1148 1149
	} else if (req) {
		saddr = inet_rsk(req)->loc_addr;
		daddr = inet_rsk(req)->rmt_addr;
1150
	} else {
1151 1152 1153
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
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 1182

	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;
1183
}
1184
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1185

E
Eric Dumazet 已提交
1186
static int tcp_v4_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
1187 1188 1189 1190 1191 1192 1193 1194 1195
{
	/*
	 * 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.
	 */
1196
	const __u8 *hash_location = NULL;
1197
	struct tcp_md5sig_key *hash_expected;
1198
	const struct iphdr *iph = ip_hdr(skb);
1199
	const struct tcphdr *th = tcp_hdr(skb);
1200 1201 1202
	int genhash;
	unsigned char newhash[16];

E
Eric Dumazet 已提交
1203 1204
	hash_expected = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&iph->saddr,
					  AF_INET);
1205
	hash_location = tcp_parse_md5sig_option(th);
1206 1207 1208 1209 1210 1211

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

	if (hash_expected && !hash_location) {
1212
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
1213 1214 1215 1216
		return 1;
	}

	if (!hash_expected && hash_location) {
1217
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
1218 1219 1220 1221 1222 1223
		return 1;
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1224 1225 1226
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1227 1228 1229

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
		if (net_ratelimit()) {
J
Joe Perches 已提交
1230 1231 1232 1233
			pr_info("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" : "");
1234 1235 1236 1237 1238 1239 1240 1241
		}
		return 1;
	}
	return 0;
}

#endif

1242
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1243
	.family		=	PF_INET,
1244
	.obj_size	=	sizeof(struct tcp_request_sock),
1245
	.rtx_syn_ack	=	tcp_v4_rtx_synack,
1246 1247
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1248
	.send_reset	=	tcp_v4_send_reset,
1249
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
1250 1251
};

1252
#ifdef CONFIG_TCP_MD5SIG
1253
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1254
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1255
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1256
};
1257
#endif
1258

L
Linus Torvalds 已提交
1259 1260
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
1261
	struct tcp_extend_values tmp_ext;
L
Linus Torvalds 已提交
1262
	struct tcp_options_received tmp_opt;
1263
	const u8 *hash_location;
1264
	struct request_sock *req;
1265
	struct inet_request_sock *ireq;
1266
	struct tcp_sock *tp = tcp_sk(sk);
1267
	struct dst_entry *dst = NULL;
1268 1269
	__be32 saddr = ip_hdr(skb)->saddr;
	__be32 daddr = ip_hdr(skb)->daddr;
L
Linus Torvalds 已提交
1270 1271 1272 1273
	__u32 isn = TCP_SKB_CB(skb)->when;
	int want_cookie = 0;

	/* Never answer to SYNs send to broadcast or multicast */
E
Eric Dumazet 已提交
1274
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1275 1276 1277 1278 1279 1280
		goto drop;

	/* TW buckets are converted to open requests without
	 * limitations, they conserve resources and peer is
	 * evidently real one.
	 */
1281
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
1282 1283 1284
		want_cookie = tcp_syn_flood_action(sk, skb, "TCP");
		if (!want_cookie)
			goto drop;
L
Linus Torvalds 已提交
1285 1286 1287 1288 1289 1290 1291
	}

	/* 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.
	 */
1292
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1293 1294
		goto drop;

1295
	req = inet_reqsk_alloc(&tcp_request_sock_ops);
L
Linus Torvalds 已提交
1296 1297 1298
	if (!req)
		goto drop;

1299 1300 1301 1302
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv4_ops;
#endif

L
Linus Torvalds 已提交
1303
	tcp_clear_options(&tmp_opt);
1304
	tmp_opt.mss_clamp = TCP_MSS_DEFAULT;
1305
	tmp_opt.user_mss  = tp->rx_opt.user_mss;
1306
	tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321

	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 *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_release;

		/* Secret recipe starts with IP addresses */
1322 1323
		*mess++ ^= (__force u32)daddr;
		*mess++ ^= (__force u32)saddr;
L
Linus Torvalds 已提交
1324

1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340
		/* plus variable length Initiator Cookie */
		c = (u8 *)mess;
		while (l-- > 0)
			*c++ ^= *hash_location++;

		want_cookie = 0;	/* not our kind of cookie */
		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_release;
	}
	tmp_ext.cookie_in_always = tp->rx_opt.cookie_in_always;
L
Linus Torvalds 已提交
1341

1342
	if (want_cookie && !tmp_opt.saw_tstamp)
L
Linus Torvalds 已提交
1343 1344 1345 1346 1347
		tcp_clear_options(&tmp_opt);

	tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;
	tcp_openreq_init(req, &tmp_opt, skb);

1348 1349 1350 1351 1352 1353
	ireq = inet_rsk(req);
	ireq->loc_addr = daddr;
	ireq->rmt_addr = saddr;
	ireq->no_srccheck = inet_sk(sk)->transparent;
	ireq->opt = tcp_v4_save_options(sk, skb);

1354
	if (security_inet_conn_request(sk, skb, req))
1355
		goto drop_and_free;
1356

1357
	if (!want_cookie || tmp_opt.tstamp_ok)
1358
		TCP_ECN_create_request(req, tcp_hdr(skb));
L
Linus Torvalds 已提交
1359 1360 1361

	if (want_cookie) {
		isn = cookie_v4_init_sequence(sk, skb, &req->mss);
1362
		req->cookie_ts = tmp_opt.tstamp_ok;
L
Linus Torvalds 已提交
1363 1364
	} else if (!isn) {
		struct inet_peer *peer = NULL;
1365
		struct flowi4 fl4;
L
Linus Torvalds 已提交
1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376

		/* VJ's idea. We save last timestamp seen
		 * from the destination in peer table, when entering
		 * state TIME-WAIT, and check against it before
		 * accepting new connection request.
		 *
		 * If "isn" is not zero, this request hit alive
		 * timewait bucket, so that all the necessary checks
		 * are made in the function processing timewait state.
		 */
		if (tmp_opt.saw_tstamp &&
1377
		    tcp_death_row.sysctl_tw_recycle &&
1378
		    (dst = inet_csk_route_req(sk, &fl4, req)) != NULL &&
1379 1380
		    fl4.daddr == saddr &&
		    (peer = rt_get_peer((struct rtable *)dst, fl4.daddr)) != NULL) {
1381
			inet_peer_refcheck(peer);
E
Eric Dumazet 已提交
1382
			if ((u32)get_seconds() - peer->tcp_ts_stamp < TCP_PAWS_MSL &&
L
Linus Torvalds 已提交
1383 1384
			    (s32)(peer->tcp_ts - req->ts_recent) >
							TCP_PAWS_WINDOW) {
1385
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1386
				goto drop_and_release;
L
Linus Torvalds 已提交
1387 1388 1389 1390
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
1391
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
L
Linus Torvalds 已提交
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401
			  (sysctl_max_syn_backlog >> 2)) &&
			 (!peer || !peer->tcp_ts_stamp) &&
			 (!dst || !dst_metric(dst, RTAX_RTT))) {
			/* Without syncookies last quarter of
			 * backlog is filled with destinations,
			 * proven to be alive.
			 * It means that we continue to communicate
			 * to destinations, already remembered
			 * to the moment of synflood.
			 */
1402 1403
			LIMIT_NETDEBUG(KERN_DEBUG "TCP: drop open request from %pI4/%u\n",
				       &saddr, ntohs(tcp_hdr(skb)->source));
1404
			goto drop_and_release;
L
Linus Torvalds 已提交
1405 1406
		}

1407
		isn = tcp_v4_init_sequence(skb);
L
Linus Torvalds 已提交
1408
	}
1409
	tcp_rsk(req)->snt_isn = isn;
1410
	tcp_rsk(req)->snt_synack = tcp_time_stamp;
L
Linus Torvalds 已提交
1411

1412 1413
	if (tcp_v4_send_synack(sk, dst, req,
			       (struct request_values *)&tmp_ext) ||
1414
	    want_cookie)
L
Linus Torvalds 已提交
1415 1416
		goto drop_and_free;

1417
	inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
L
Linus Torvalds 已提交
1418 1419
	return 0;

1420 1421
drop_and_release:
	dst_release(dst);
L
Linus Torvalds 已提交
1422
drop_and_free:
1423
	reqsk_free(req);
L
Linus Torvalds 已提交
1424 1425 1426
drop:
	return 0;
}
E
Eric Dumazet 已提交
1427
EXPORT_SYMBOL(tcp_v4_conn_request);
L
Linus Torvalds 已提交
1428 1429 1430 1431 1432 1433 1434


/*
 * 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,
1435
				  struct request_sock *req,
L
Linus Torvalds 已提交
1436 1437
				  struct dst_entry *dst)
{
1438
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1439 1440 1441
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1442 1443 1444
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
1445
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
1446 1447 1448 1449 1450 1451

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (!newsk)
1452
		goto exit_nonewsk;
L
Linus Torvalds 已提交
1453

1454
	newsk->sk_gso_type = SKB_GSO_TCPV4;
L
Linus Torvalds 已提交
1455 1456 1457

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1458
	ireq		      = inet_rsk(req);
E
Eric Dumazet 已提交
1459 1460 1461
	newinet->inet_daddr   = ireq->rmt_addr;
	newinet->inet_rcv_saddr = ireq->loc_addr;
	newinet->inet_saddr	      = ireq->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

1473 1474 1475 1476 1477
	if (!dst && (dst = inet_csk_route_child_sock(sk, newsk, req)) == NULL)
		goto put_and_exit;

	sk_setup_caps(newsk, dst);

J
John Heffner 已提交
1478
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1479
	tcp_sync_mss(newsk, dst_mtu(dst));
1480
	newtp->advmss = dst_metric_advmss(dst);
1481 1482 1483 1484
	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
1485
	tcp_initialize_rcv_mss(newsk);
1486 1487 1488 1489
	if (tcp_rsk(req)->snt_synack)
		tcp_valid_rtt_meas(newsk,
		    tcp_time_stamp - tcp_rsk(req)->snt_synack);
	newtp->total_retrans = req->retrans;
L
Linus Torvalds 已提交
1490

1491 1492
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1493 1494
	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&newinet->inet_daddr,
				AF_INET);
E
Eric Dumazet 已提交
1495
	if (key != NULL) {
1496 1497 1498 1499 1500 1501
		/*
		 * 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
1502 1503
		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
1504
		sk_nocaps_add(newsk, NETIF_F_GSO_MASK);
1505 1506 1507
	}
#endif

1508 1509
	if (__inet_inherit_port(sk, newsk) < 0)
		goto put_and_exit;
1510
	__inet_hash_nolisten(newsk, NULL);
L
Linus Torvalds 已提交
1511 1512 1513 1514

	return newsk;

exit_overflow:
1515
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1516 1517
exit_nonewsk:
	dst_release(dst);
L
Linus Torvalds 已提交
1518
exit:
1519
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1520
	return NULL;
1521
put_and_exit:
1522
	tcp_clear_xmit_timers(newsk);
1523
	tcp_cleanup_congestion_control(newsk);
1524
	bh_unlock_sock(newsk);
1525 1526
	sock_put(newsk);
	goto exit;
L
Linus Torvalds 已提交
1527
}
E
Eric Dumazet 已提交
1528
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
L
Linus Torvalds 已提交
1529 1530 1531

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1532
	struct tcphdr *th = tcp_hdr(skb);
1533
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1534
	struct sock *nsk;
1535
	struct request_sock **prev;
L
Linus Torvalds 已提交
1536
	/* Find possible connection requests. */
1537 1538
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1539 1540 1541
	if (req)
		return tcp_check_req(sk, skb, req, prev);

1542
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1543
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1544 1545 1546 1547 1548 1549

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1550
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1551 1552 1553 1554
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
1555
	if (!th->syn)
L
Linus Torvalds 已提交
1556 1557 1558 1559 1560
		sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
#endif
	return sk;
}

1561
static __sum16 tcp_v4_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1562
{
1563 1564
	const struct iphdr *iph = ip_hdr(skb);

1565
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
1566 1567
		if (!tcp_v4_check(skb->len, iph->saddr,
				  iph->daddr, skb->csum)) {
1568
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1569
			return 0;
1570
		}
L
Linus Torvalds 已提交
1571
	}
1572

1573
	skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
1574 1575
				       skb->len, IPPROTO_TCP, 0);

L
Linus Torvalds 已提交
1576
	if (skb->len <= 76) {
1577
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592
	}
	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.
 */
int tcp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
{
1593 1594 1595 1596 1597 1598 1599 1600
	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
	 */
1601
	if (tcp_v4_inbound_md5_hash(sk, skb))
1602 1603 1604
		goto discard;
#endif

L
Linus Torvalds 已提交
1605
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1606
		sock_rps_save_rxhash(sk, skb);
1607
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
1608
			rsk = sk;
L
Linus Torvalds 已提交
1609
			goto reset;
1610
		}
L
Linus Torvalds 已提交
1611 1612 1613
		return 0;
	}

1614
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1615 1616 1617 1618 1619 1620 1621 1622
		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) {
1623
			sock_rps_save_rxhash(nsk, skb);
1624 1625
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1626
				goto reset;
1627
			}
L
Linus Torvalds 已提交
1628 1629
			return 0;
		}
1630
	} else
1631
		sock_rps_save_rxhash(sk, skb);
1632

1633
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1634
		rsk = sk;
L
Linus Torvalds 已提交
1635
		goto reset;
1636
	}
L
Linus Torvalds 已提交
1637 1638 1639
	return 0;

reset:
1640
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1641 1642 1643 1644 1645 1646 1647 1648 1649 1650
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:
1651
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1652 1653
	goto discard;
}
E
Eric Dumazet 已提交
1654
EXPORT_SYMBOL(tcp_v4_do_rcv);
L
Linus Torvalds 已提交
1655 1656 1657 1658 1659 1660 1661

/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1662
	const struct iphdr *iph;
1663
	const struct tcphdr *th;
L
Linus Torvalds 已提交
1664 1665
	struct sock *sk;
	int ret;
1666
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1667 1668 1669 1670 1671

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

	/* Count it even if it's bad */
1672
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1673 1674 1675 1676

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

1677
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1678 1679 1680 1681 1682 1683 1684 1685

	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
1686
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1687
	 * So, we defer the checks. */
1688
	if (!skb_csum_unnecessary(skb) && tcp_v4_checksum_init(skb))
L
Linus Torvalds 已提交
1689 1690
		goto bad_packet;

1691
	th = tcp_hdr(skb);
1692
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1693 1694 1695 1696 1697
	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;
1698
	TCP_SKB_CB(skb)->ip_dsfield = ipv4_get_dsfield(iph);
L
Linus Torvalds 已提交
1699 1700
	TCP_SKB_CB(skb)->sacked	 = 0;

1701
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1702 1703 1704
	if (!sk)
		goto no_tcp_socket;

E
Eric Dumazet 已提交
1705 1706 1707 1708
process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1709 1710
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1711
		goto discard_and_relse;
1712
	}
1713

L
Linus Torvalds 已提交
1714 1715
	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1716
	nf_reset(skb);
L
Linus Torvalds 已提交
1717

1718
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1719 1720 1721 1722
		goto discard_and_relse;

	skb->dev = NULL;

1723
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1724 1725
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1726 1727 1728
#ifdef CONFIG_NET_DMA
		struct tcp_sock *tp = tcp_sk(sk);
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1729
			tp->ucopy.dma_chan = dma_find_channel(DMA_MEMCPY);
C
Chris Leech 已提交
1730
		if (tp->ucopy.dma_chan)
L
Linus Torvalds 已提交
1731
			ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1732 1733 1734 1735
		else
#endif
		{
			if (!tcp_prequeue(sk, skb))
S
Shan Wei 已提交
1736
				ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1737
		}
1738
	} else if (unlikely(sk_add_backlog(sk, skb))) {
Z
Zhu Yi 已提交
1739
		bh_unlock_sock(sk);
1740
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1741 1742
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754
	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)) {
bad_packet:
1755
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1756
	} else {
1757
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1758 1759 1760 1761 1762
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
1763
	return 0;
L
Linus Torvalds 已提交
1764 1765 1766 1767 1768 1769 1770

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1771
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1772 1773 1774 1775
		goto discard_it;
	}

	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
1776
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1777
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1778 1779
		goto discard_it;
	}
1780
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1781
	case TCP_TW_SYN: {
1782
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
1783
							&tcp_hashinfo,
1784
							iph->daddr, th->dest,
1785
							inet_iif(skb));
L
Linus Torvalds 已提交
1786
		if (sk2) {
1787 1788
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803
			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;
}

1804
struct inet_peer *tcp_v4_get_peer(struct sock *sk, bool *release_it)
L
Linus Torvalds 已提交
1805
{
1806
	struct rtable *rt = (struct rtable *) __sk_dst_get(sk);
L
Linus Torvalds 已提交
1807
	struct inet_sock *inet = inet_sk(sk);
1808
	struct inet_peer *peer;
L
Linus Torvalds 已提交
1809

1810 1811
	if (!rt ||
	    inet->cork.fl.u.ip4.daddr != inet->inet_daddr) {
1812
		peer = inet_getpeer_v4(inet->inet_daddr, 1);
1813
		*release_it = true;
L
Linus Torvalds 已提交
1814 1815
	} else {
		if (!rt->peer)
1816
			rt_bind_peer(rt, inet->inet_daddr, 1);
L
Linus Torvalds 已提交
1817
		peer = rt->peer;
1818
		*release_it = false;
L
Linus Torvalds 已提交
1819 1820
	}

1821
	return peer;
L
Linus Torvalds 已提交
1822
}
1823
EXPORT_SYMBOL(tcp_v4_get_peer);
L
Linus Torvalds 已提交
1824

1825
void *tcp_v4_tw_get_peer(struct sock *sk)
L
Linus Torvalds 已提交
1826
{
1827
	const struct inet_timewait_sock *tw = inet_twsk(sk);
L
Linus Torvalds 已提交
1828

1829
	return inet_getpeer_v4(tw->tw_daddr, 1);
L
Linus Torvalds 已提交
1830
}
1831 1832 1833 1834 1835 1836 1837 1838
EXPORT_SYMBOL(tcp_v4_tw_get_peer);

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,
	.twsk_getpeer	= tcp_v4_tw_get_peer,
};
L
Linus Torvalds 已提交
1839

1840
const struct inet_connection_sock_af_ops ipv4_specific = {
1841 1842 1843 1844 1845
	.queue_xmit	   = ip_queue_xmit,
	.send_check	   = tcp_v4_send_check,
	.rebuild_header	   = inet_sk_rebuild_header,
	.conn_request	   = tcp_v4_conn_request,
	.syn_recv_sock	   = tcp_v4_syn_recv_sock,
1846
	.get_peer	   = tcp_v4_get_peer,
1847 1848 1849 1850 1851
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ip_setsockopt,
	.getsockopt	   = ip_getsockopt,
	.addr2sockaddr	   = inet_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in),
1852
	.bind_conflict	   = inet_csk_bind_conflict,
1853
#ifdef CONFIG_COMPAT
1854 1855
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
1856
#endif
L
Linus Torvalds 已提交
1857
};
E
Eric Dumazet 已提交
1858
EXPORT_SYMBOL(ipv4_specific);
L
Linus Torvalds 已提交
1859

1860
#ifdef CONFIG_TCP_MD5SIG
1861
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
1862
	.md5_lookup		= tcp_v4_md5_lookup,
1863
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
1864 1865
	.md5_parse		= tcp_v4_parse_md5_keys,
};
1866
#endif
1867

L
Linus Torvalds 已提交
1868 1869 1870 1871 1872
/* 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)
{
1873
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1874 1875 1876 1877 1878 1879
	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);

1880
	icsk->icsk_rto = TCP_TIMEOUT_INIT;
L
Linus Torvalds 已提交
1881 1882 1883 1884 1885 1886 1887
	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
	 */
1888
	tp->snd_cwnd = TCP_INIT_CWND;
L
Linus Torvalds 已提交
1889 1890 1891 1892

	/* See draft-stevens-tcpca-spec-01 for discussion of the
	 * initialization of these values.
	 */
I
Ilpo Järvinen 已提交
1893
	tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
L
Linus Torvalds 已提交
1894
	tp->snd_cwnd_clamp = ~0;
1895
	tp->mss_cache = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
1896 1897

	tp->reordering = sysctl_tcp_reordering;
1898
	icsk->icsk_ca_ops = &tcp_init_congestion_ops;
L
Linus Torvalds 已提交
1899 1900 1901 1902 1903 1904

	sk->sk_state = TCP_CLOSE;

	sk->sk_write_space = sk_stream_write_space;
	sock_set_flag(sk, SOCK_USE_WRITE_QUEUE);

1905
	icsk->icsk_af_ops = &ipv4_specific;
1906
	icsk->icsk_sync_mss = tcp_sync_mss;
1907 1908 1909
#ifdef CONFIG_TCP_MD5SIG
	tp->af_specific = &tcp_sock_ipv4_specific;
#endif
L
Linus Torvalds 已提交
1910

1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923
	/* 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
1924 1925 1926
	sk->sk_sndbuf = sysctl_tcp_wmem[1];
	sk->sk_rcvbuf = sysctl_tcp_rmem[1];

H
Herbert Xu 已提交
1927
	local_bh_disable();
G
Glauber Costa 已提交
1928
	sock_update_memcg(sk);
1929
	sk_sockets_allocated_inc(sk);
H
Herbert Xu 已提交
1930
	local_bh_enable();
L
Linus Torvalds 已提交
1931 1932 1933 1934

	return 0;
}

1935
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1936 1937 1938 1939 1940
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

1941
	tcp_cleanup_congestion_control(sk);
1942

L
Linus Torvalds 已提交
1943
	/* Cleanup up the write buffer. */
1944
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
1945 1946

	/* Cleans up our, hopefully empty, out_of_order_queue. */
1947
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
1948

1949 1950 1951
#ifdef CONFIG_TCP_MD5SIG
	/* Clean up the MD5 key list, if any */
	if (tp->md5sig_info) {
E
Eric Dumazet 已提交
1952
		tcp_clear_md5_list(sk);
1953
		kfree_rcu(tp->md5sig_info, rcu);
1954 1955 1956 1957
		tp->md5sig_info = NULL;
	}
#endif

C
Chris Leech 已提交
1958 1959
#ifdef CONFIG_NET_DMA
	/* Cleans up our sk_async_wait_queue */
1960
	__skb_queue_purge(&sk->sk_async_wait_queue);
C
Chris Leech 已提交
1961 1962
#endif

L
Linus Torvalds 已提交
1963 1964 1965 1966
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
1967
	if (inet_csk(sk)->icsk_bind_hash)
1968
		inet_put_port(sk);
L
Linus Torvalds 已提交
1969 1970 1971 1972 1973 1974 1975 1976 1977

	/*
	 * If sendmsg cached page exists, toss it.
	 */
	if (sk->sk_sndmsg_page) {
		__free_page(sk->sk_sndmsg_page);
		sk->sk_sndmsg_page = NULL;
	}

1978 1979 1980 1981 1982 1983 1984
	/* TCP Cookie Transactions */
	if (tp->cookie_values != NULL) {
		kref_put(&tp->cookie_values->kref,
			 tcp_cookie_values_release);
		tp->cookie_values = NULL;
	}

1985
	sk_sockets_allocated_dec(sk);
G
Glauber Costa 已提交
1986
	sock_release_memcg(sk);
L
Linus Torvalds 已提交
1987 1988 1989 1990 1991 1992
}
EXPORT_SYMBOL(tcp_v4_destroy_sock);

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

1993
static inline struct inet_timewait_sock *tw_head(struct hlist_nulls_head *head)
L
Linus Torvalds 已提交
1994
{
1995
	return hlist_nulls_empty(head) ? NULL :
1996
		list_entry(head->first, struct inet_timewait_sock, tw_node);
L
Linus Torvalds 已提交
1997 1998
}

1999
static inline struct inet_timewait_sock *tw_next(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
2000
{
2001 2002
	return !is_a_nulls(tw->tw_node.next) ?
		hlist_nulls_entry(tw->tw_node.next, typeof(*tw), tw_node) : NULL;
L
Linus Torvalds 已提交
2003 2004
}

2005 2006 2007 2008 2009
/*
 * 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
2010 2011
static void *listening_get_next(struct seq_file *seq, void *cur)
{
2012
	struct inet_connection_sock *icsk;
2013
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
2014
	struct sock *sk = cur;
2015
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
2016
	struct tcp_iter_state *st = seq->private;
2017
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2018 2019

	if (!sk) {
2020
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
2021
		spin_lock_bh(&ilb->lock);
2022
		sk = sk_nulls_head(&ilb->head);
2023
		st->offset = 0;
L
Linus Torvalds 已提交
2024 2025
		goto get_sk;
	}
2026
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
2027
	++st->num;
2028
	++st->offset;
L
Linus Torvalds 已提交
2029 2030

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
2031
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
2032

2033
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
2034 2035 2036
		req = req->dl_next;
		while (1) {
			while (req) {
2037
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
2038 2039 2040 2041 2042
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
2043
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
2044 2045
				break;
get_req:
2046
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
2047
		}
E
Eric Dumazet 已提交
2048
		sk	  = sk_nulls_next(st->syn_wait_sk);
L
Linus Torvalds 已提交
2049
		st->state = TCP_SEQ_STATE_LISTENING;
2050
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2051
	} else {
2052
		icsk = inet_csk(sk);
2053 2054
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
2055
			goto start_req;
2056
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
E
Eric Dumazet 已提交
2057
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2058 2059
	}
get_sk:
2060
	sk_nulls_for_each_from(sk, node) {
2061 2062 2063
		if (!net_eq(sock_net(sk), net))
			continue;
		if (sk->sk_family == st->family) {
L
Linus Torvalds 已提交
2064 2065 2066
			cur = sk;
			goto out;
		}
2067
		icsk = inet_csk(sk);
2068 2069
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
2070 2071 2072 2073 2074 2075 2076
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;
		}
2077
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2078
	}
2079
	spin_unlock_bh(&ilb->lock);
2080
	st->offset = 0;
2081
	if (++st->bucket < INET_LHTABLE_SIZE) {
2082 2083
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
2084
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
2085 2086 2087 2088 2089 2090 2091 2092 2093
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
2094 2095 2096 2097 2098 2099
	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
2100 2101 2102 2103 2104 2105 2106 2107

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

2108 2109
static inline int empty_bucket(struct tcp_iter_state *st)
{
2110 2111
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain) &&
		hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].twchain);
2112 2113
}

2114 2115 2116 2117
/*
 * 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
2118 2119
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
2120
	struct tcp_iter_state *st = seq->private;
2121
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2122 2123
	void *rc = NULL;

2124 2125
	st->offset = 0;
	for (; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
2126
		struct sock *sk;
2127
		struct hlist_nulls_node *node;
2128
		struct inet_timewait_sock *tw;
2129
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
2130

2131 2132 2133 2134
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

2135
		spin_lock_bh(lock);
2136
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
2137
			if (sk->sk_family != st->family ||
2138
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2139 2140 2141 2142 2143 2144
				continue;
			}
			rc = sk;
			goto out;
		}
		st->state = TCP_SEQ_STATE_TIME_WAIT;
2145
		inet_twsk_for_each(tw, node,
2146
				   &tcp_hashinfo.ehash[st->bucket].twchain) {
2147
			if (tw->tw_family != st->family ||
2148
			    !net_eq(twsk_net(tw), net)) {
L
Linus Torvalds 已提交
2149 2150 2151 2152 2153
				continue;
			}
			rc = tw;
			goto out;
		}
2154
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2155 2156 2157 2158 2159 2160 2161 2162 2163
		st->state = TCP_SEQ_STATE_ESTABLISHED;
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2164
	struct inet_timewait_sock *tw;
2165
	struct hlist_nulls_node *node;
J
Jianjun Kong 已提交
2166
	struct tcp_iter_state *st = seq->private;
2167
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2168 2169

	++st->num;
2170
	++st->offset;
L
Linus Torvalds 已提交
2171 2172 2173 2174 2175

	if (st->state == TCP_SEQ_STATE_TIME_WAIT) {
		tw = cur;
		tw = tw_next(tw);
get_tw:
2176
		while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
L
Linus Torvalds 已提交
2177 2178 2179 2180 2181 2182
			tw = tw_next(tw);
		}
		if (tw) {
			cur = tw;
			goto out;
		}
2183
		spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2184 2185
		st->state = TCP_SEQ_STATE_ESTABLISHED;

2186
		/* Look for next non empty bucket */
2187
		st->offset = 0;
2188
		while (++st->bucket <= tcp_hashinfo.ehash_mask &&
2189 2190
				empty_bucket(st))
			;
2191
		if (st->bucket > tcp_hashinfo.ehash_mask)
2192 2193
			return NULL;

2194
		spin_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
2195
		sk = sk_nulls_head(&tcp_hashinfo.ehash[st->bucket].chain);
L
Linus Torvalds 已提交
2196
	} else
2197
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2198

2199
	sk_nulls_for_each_from(sk, node) {
2200
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
L
Linus Torvalds 已提交
2201 2202 2203 2204
			goto found;
	}

	st->state = TCP_SEQ_STATE_TIME_WAIT;
2205
	tw = tw_head(&tcp_hashinfo.ehash[st->bucket].twchain);
L
Linus Torvalds 已提交
2206 2207 2208 2209 2210 2211 2212 2213 2214
	goto get_tw;
found:
	cur = sk;
out:
	return cur;
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
2215 2216 2217 2218 2219
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	rc = established_get_first(seq);
L
Linus Torvalds 已提交
2220 2221 2222 2223

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2224
	}
L
Linus Torvalds 已提交
2225 2226 2227 2228 2229 2230
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2231
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243

	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;
}

2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278
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;
		/* Fallthrough */
	case TCP_SEQ_STATE_ESTABLISHED:
	case TCP_SEQ_STATE_TIME_WAIT:
		st->state = 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
2279 2280
static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2281
	struct tcp_iter_state *st = seq->private;
2282 2283 2284 2285 2286 2287 2288 2289
	void *rc;

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

L
Linus Torvalds 已提交
2290 2291
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
2292 2293 2294 2295 2296 2297 2298
	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
2299 2300 2301 2302
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2303
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316
	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;
2317 2318
			st->bucket = 0;
			st->offset = 0;
L
Linus Torvalds 已提交
2319 2320 2321 2322 2323 2324 2325 2326 2327 2328
			rc	  = established_get_first(seq);
		}
		break;
	case TCP_SEQ_STATE_ESTABLISHED:
	case TCP_SEQ_STATE_TIME_WAIT:
		rc = established_get_next(seq, v);
		break;
	}
out:
	++*pos;
2329
	st->last_pos = *pos;
L
Linus Torvalds 已提交
2330 2331 2332 2333 2334
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2335
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2336 2337 2338 2339

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2340 2341
			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
2342 2343 2344
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2345
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2346 2347 2348 2349
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2350
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2351 2352 2353 2354
		break;
	}
}

2355
int tcp_seq_open(struct inode *inode, struct file *file)
L
Linus Torvalds 已提交
2356 2357 2358
{
	struct tcp_seq_afinfo *afinfo = PDE(inode)->data;
	struct tcp_iter_state *s;
2359
	int err;
L
Linus Torvalds 已提交
2360

2361 2362 2363 2364
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2365

2366
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2367
	s->family		= afinfo->family;
2368
	s->last_pos 		= 0;
2369 2370
	return 0;
}
2371
EXPORT_SYMBOL(tcp_seq_open);
2372

2373
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2374 2375 2376 2377
{
	int rc = 0;
	struct proc_dir_entry *p;

2378 2379 2380 2381
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2382
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
2383
			     afinfo->seq_fops, afinfo);
2384
	if (!p)
L
Linus Torvalds 已提交
2385 2386 2387
		rc = -ENOMEM;
	return rc;
}
E
Eric Dumazet 已提交
2388
EXPORT_SYMBOL(tcp_proc_register);
L
Linus Torvalds 已提交
2389

2390
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2391
{
2392
	proc_net_remove(net, afinfo->name);
L
Linus Torvalds 已提交
2393
}
E
Eric Dumazet 已提交
2394
EXPORT_SYMBOL(tcp_proc_unregister);
L
Linus Torvalds 已提交
2395

2396
static void get_openreq4(const struct sock *sk, const struct request_sock *req,
2397
			 struct seq_file *f, int i, int uid, int *len)
L
Linus Torvalds 已提交
2398
{
2399
	const struct inet_request_sock *ireq = inet_rsk(req);
L
Linus Torvalds 已提交
2400 2401
	int ttd = req->expires - jiffies;

2402
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2403
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %pK%n",
L
Linus Torvalds 已提交
2404
		i,
2405
		ireq->loc_addr,
E
Eric Dumazet 已提交
2406
		ntohs(inet_sk(sk)->inet_sport),
2407 2408
		ireq->rmt_addr,
		ntohs(ireq->rmt_port),
L
Linus Torvalds 已提交
2409 2410 2411 2412 2413 2414 2415 2416 2417
		TCP_SYN_RECV,
		0, 0, /* could print option size, but that is af dependent. */
		1,    /* timers active (only the expire timer) */
		jiffies_to_clock_t(ttd),
		req->retrans,
		uid,
		0,  /* non standard timer */
		0, /* open_requests have no inode */
		atomic_read(&sk->sk_refcnt),
2418 2419
		req,
		len);
L
Linus Torvalds 已提交
2420 2421
}

2422
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2423 2424 2425
{
	int timer_active;
	unsigned long timer_expires;
2426
	const struct tcp_sock *tp = tcp_sk(sk);
2427
	const struct inet_connection_sock *icsk = inet_csk(sk);
2428
	const struct inet_sock *inet = inet_sk(sk);
E
Eric Dumazet 已提交
2429 2430 2431 2432
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
2433
	int rx_queue;
L
Linus Torvalds 已提交
2434

2435
	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
2436
		timer_active	= 1;
2437 2438
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2439
		timer_active	= 4;
2440
		timer_expires	= icsk->icsk_timeout;
2441
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2442
		timer_active	= 2;
2443
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2444 2445 2446 2447 2448
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2449 2450 2451 2452 2453 2454 2455 2456
	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);

2457
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
D
Dan Rosenberg 已提交
2458
			"%08X %5d %8d %lu %d %pK %lu %lu %u %u %d%n",
2459
		i, src, srcp, dest, destp, sk->sk_state,
2460
		tp->write_seq - tp->snd_una,
2461
		rx_queue,
L
Linus Torvalds 已提交
2462 2463
		timer_active,
		jiffies_to_clock_t(timer_expires - jiffies),
2464
		icsk->icsk_retransmits,
2465
		sock_i_uid(sk),
2466
		icsk->icsk_probes_out,
2467 2468
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2469 2470
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2471
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2472
		tp->snd_cwnd,
I
Ilpo Järvinen 已提交
2473
		tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh,
2474
		len);
L
Linus Torvalds 已提交
2475 2476
}

2477
static void get_timewait4_sock(const struct inet_timewait_sock *tw,
2478
			       struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2479
{
2480
	__be32 dest, src;
L
Linus Torvalds 已提交
2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491
	__u16 destp, srcp;
	int ttd = tw->tw_ttd - jiffies;

	if (ttd < 0)
		ttd = 0;

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

2492
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2493
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK%n",
L
Linus Torvalds 已提交
2494 2495
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
		3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
2496
		atomic_read(&tw->tw_refcnt), tw, len);
L
Linus Torvalds 已提交
2497 2498 2499 2500 2501 2502
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2503
	struct tcp_iter_state *st;
2504
	int len;
L
Linus Torvalds 已提交
2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517

	if (v == SEQ_START_TOKEN) {
		seq_printf(seq, "%-*s\n", TMPSZ - 1,
			   "  sl  local_address rem_address   st tx_queue "
			   "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:
2518
		get_tcp4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2519 2520
		break;
	case TCP_SEQ_STATE_OPENREQ:
2521
		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid, &len);
L
Linus Torvalds 已提交
2522 2523
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
2524
		get_timewait4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2525 2526
		break;
	}
2527
	seq_printf(seq, "%*s\n", TMPSZ - 1 - len, "");
L
Linus Torvalds 已提交
2528 2529 2530 2531
out:
	return 0;
}

2532 2533 2534 2535 2536 2537 2538 2539
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
2540 2541 2542
static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2543
	.seq_fops	= &tcp_afinfo_seq_fops,
2544 2545 2546
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2547 2548
};

2549
static int __net_init tcp4_proc_init_net(struct net *net)
2550 2551 2552 2553
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

2554
static void __net_exit tcp4_proc_exit_net(struct net *net)
2555 2556 2557 2558 2559 2560 2561 2562 2563
{
	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
2564 2565
int __init tcp4_proc_init(void)
{
2566
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2567 2568 2569 2570
}

void tcp4_proc_exit(void)
{
2571
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2572 2573 2574
}
#endif /* CONFIG_PROC_FS */

H
Herbert Xu 已提交
2575 2576
struct sk_buff **tcp4_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
2577
	const struct iphdr *iph = skb_gro_network_header(skb);
H
Herbert Xu 已提交
2578 2579 2580

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
2581
		if (!tcp_v4_check(skb_gro_len(skb), iph->saddr, iph->daddr,
H
Herbert Xu 已提交
2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597
				  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);
}

int tcp4_gro_complete(struct sk_buff *skb)
{
2598
	const struct iphdr *iph = ip_hdr(skb);
H
Herbert Xu 已提交
2599 2600 2601 2602 2603 2604 2605 2606 2607
	struct tcphdr *th = tcp_hdr(skb);

	th->check = ~tcp_v4_check(skb->len - skb_transport_offset(skb),
				  iph->saddr, iph->daddr, 0);
	skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;

	return tcp_gro_complete(skb);
}

L
Linus Torvalds 已提交
2608 2609 2610 2611 2612 2613
struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2614
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2615 2616 2617 2618 2619 2620 2621
	.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,
2622 2623
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2624
	.backlog_rcv		= tcp_v4_do_rcv,
2625 2626 2627
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2628 2629
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
2630
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2631 2632 2633 2634 2635 2636
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp_sock),
2637
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2638
	.twsk_prot		= &tcp_timewait_sock_ops,
2639
	.rsk_prot		= &tcp_request_sock_ops,
2640
	.h.hashinfo		= &tcp_hashinfo,
2641
	.no_autobind		= true,
2642 2643 2644 2645
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
G
Glauber Costa 已提交
2646 2647 2648 2649 2650
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
	.init_cgroup		= tcp_init_cgroup,
	.destroy_cgroup		= tcp_destroy_cgroup,
	.proto_cgroup		= tcp_proto_cgroup,
#endif
L
Linus Torvalds 已提交
2651
};
E
Eric Dumazet 已提交
2652
EXPORT_SYMBOL(tcp_prot);
L
Linus Torvalds 已提交
2653

2654 2655 2656 2657 2658 2659 2660 2661 2662
static int __net_init tcp_sk_init(struct net *net)
{
	return inet_ctl_sock_create(&net->ipv4.tcp_sock,
				    PF_INET, SOCK_RAW, IPPROTO_TCP, net);
}

static void __net_exit tcp_sk_exit(struct net *net)
{
	inet_ctl_sock_destroy(net->ipv4.tcp_sock);
E
Eric W. Biederman 已提交
2663 2664 2665 2666 2667
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2668 2669 2670
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2671 2672 2673
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2674 2675
};

2676
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2677
{
2678
	inet_hashinfo_init(&tcp_hashinfo);
2679
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2680 2681
		panic("Failed to create the TCP control socket.\n");
}