tcp_ipv4.c 66.7 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;
594 595 596
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
597
	struct net *net;
L
Linus Torvalds 已提交
598 599 600 601 602

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

E
Eric Dumazet 已提交
603
	if (skb_rtable(skb)->rt_type != RTN_LOCAL)
L
Linus Torvalds 已提交
604 605 606
		return;

	/* Swap the send and the receive. */
607 608 609 610 611
	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
612 613

	if (th->ack) {
614
		rep.th.seq = th->ack_seq;
L
Linus Torvalds 已提交
615
	} else {
616 617 618
		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
619 620
	}

621
	memset(&arg, 0, sizeof(arg));
622 623 624 625
	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);

#ifdef CONFIG_TCP_MD5SIG
E
Eric Dumazet 已提交
626 627 628
	key = sk ? tcp_md5_do_lookup(sk,
				     (union tcp_md5_addr *)&ip_hdr(skb)->saddr,
				     AF_INET) : NULL;
629 630 631 632 633 634 635 636 637
	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;

638
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
639 640
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
641 642
	}
#endif
643 644
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
645
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
646
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
647
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
L
Linus Torvalds 已提交
648

E
Eric Dumazet 已提交
649
	net = dev_net(skb_dst(skb)->dev);
650
	arg.tos = ip_hdr(skb)->tos;
651
	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
652
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
653

654 655
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
L
Linus Torvalds 已提交
656 657 658 659 660 661
}

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

662 663
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
			    u32 win, u32 ts, int oif,
664
			    struct tcp_md5sig_key *key,
665
			    int reply_flags, u8 tos)
L
Linus Torvalds 已提交
666
{
667
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
668 669
	struct {
		struct tcphdr th;
670
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
671
#ifdef CONFIG_TCP_MD5SIG
672
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
673 674
#endif
			];
L
Linus Torvalds 已提交
675 676
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
677
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
678 679

	memset(&rep.th, 0, sizeof(struct tcphdr));
680
	memset(&arg, 0, sizeof(arg));
L
Linus Torvalds 已提交
681 682 683 684

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
	if (ts) {
685 686 687 688 689
		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);
690
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
691 692 693 694 695 696 697 698 699 700 701
	}

	/* 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);

702 703 704 705 706 707 708 709 710 711 712
#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;

713
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[offset],
714 715
				    key, ip_hdr(skb)->saddr,
				    ip_hdr(skb)->daddr, &rep.th);
716 717
	}
#endif
718
	arg.flags = reply_flags;
719 720
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
L
Linus Torvalds 已提交
721 722
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
723 724
	if (oif)
		arg.bound_dev_if = oif;
725
	arg.tos = tos;
726
	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
727
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
728

729
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
730 731 732 733
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
734
	struct inet_timewait_sock *tw = inet_twsk(sk);
735
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
736

737
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
738
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
739 740
			tcptw->tw_ts_recent,
			tw->tw_bound_dev_if,
741
			tcp_twsk_md5_key(tcptw),
742 743
			tw->tw_transparent ? IP_REPLY_ARG_NOSRCCHECK : 0,
			tw->tw_tos
744
			);
L
Linus Torvalds 已提交
745

746
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
747 748
}

749
static void tcp_v4_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
750
				  struct request_sock *req)
L
Linus Torvalds 已提交
751
{
752
	tcp_v4_send_ack(skb, tcp_rsk(req)->snt_isn + 1,
753
			tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd,
754 755
			req->ts_recent,
			0,
E
Eric Dumazet 已提交
756 757
			tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&ip_hdr(skb)->daddr,
					  AF_INET),
758 759
			inet_rsk(req)->no_srccheck ? IP_REPLY_ARG_NOSRCCHECK : 0,
			ip_hdr(skb)->tos);
L
Linus Torvalds 已提交
760 761 762
}

/*
763
 *	Send a SYN-ACK after having received a SYN.
764
 *	This still operates on a request_sock only, not on a big
L
Linus Torvalds 已提交
765 766
 *	socket.
 */
767 768 769
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
770
{
771
	const struct inet_request_sock *ireq = inet_rsk(req);
772
	struct flowi4 fl4;
L
Linus Torvalds 已提交
773 774 775 776
	int err = -1;
	struct sk_buff * skb;

	/* First, grab a route. */
777
	if (!dst && (dst = inet_csk_route_req(sk, &fl4, req)) == NULL)
778
		return -1;
L
Linus Torvalds 已提交
779

780
	skb = tcp_make_synack(sk, dst, req, rvp);
L
Linus Torvalds 已提交
781 782

	if (skb) {
783
		__tcp_v4_send_check(skb, ireq->loc_addr, ireq->rmt_addr);
L
Linus Torvalds 已提交
784

785 786 787
		err = ip_build_and_send_pkt(skb, sk, ireq->loc_addr,
					    ireq->rmt_addr,
					    ireq->opt);
788
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
789 790 791 792 793 794
	}

	dst_release(dst);
	return err;
}

795
static int tcp_v4_rtx_synack(struct sock *sk, struct request_sock *req,
796
			      struct request_values *rvp)
797
{
798 799
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
	return tcp_v4_send_synack(sk, NULL, req, rvp);
800 801
}

L
Linus Torvalds 已提交
802
/*
803
 *	IPv4 request_sock destructor.
L
Linus Torvalds 已提交
804
 */
805
static void tcp_v4_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
806
{
J
Jesper Juhl 已提交
807
	kfree(inet_rsk(req)->opt);
L
Linus Torvalds 已提交
808 809
}

810 811 812 813 814 815
/*
 * 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
816
{
817 818 819 820 821
	const char *msg = "Dropping request";
	int want_cookie = 0;
	struct listen_sock *lopt;


L
Linus Torvalds 已提交
822

823
#ifdef CONFIG_SYN_COOKIES
824
	if (sysctl_tcp_syncookies) {
825
		msg = "Sending cookies";
826 827 828
		want_cookie = 1;
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDOCOOKIES);
	} else
829
#endif
830 831 832 833 834 835 836 837 838 839
		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;
840
}
841
EXPORT_SYMBOL(tcp_syn_flood_action);
L
Linus Torvalds 已提交
842 843

/*
844
 * Save and compile IPv4 options into the request_sock if needed.
L
Linus Torvalds 已提交
845
 */
846 847
static struct ip_options_rcu *tcp_v4_save_options(struct sock *sk,
						  struct sk_buff *skb)
L
Linus Torvalds 已提交
848
{
849 850
	const struct ip_options *opt = &(IPCB(skb)->opt);
	struct ip_options_rcu *dopt = NULL;
L
Linus Torvalds 已提交
851 852

	if (opt && opt->optlen) {
853 854
		int opt_size = sizeof(*dopt) + opt->optlen;

L
Linus Torvalds 已提交
855 856
		dopt = kmalloc(opt_size, GFP_ATOMIC);
		if (dopt) {
857
			if (ip_options_echo(&dopt->opt, skb)) {
L
Linus Torvalds 已提交
858 859 860 861 862 863 864 865
				kfree(dopt);
				dopt = NULL;
			}
		}
	}
	return dopt;
}

866 867 868 869 870 871 872 873
#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
874 875 876
struct tcp_md5sig_key *tcp_md5_do_lookup(struct sock *sk,
					 const union tcp_md5_addr *addr,
					 int family)
877 878
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
879 880 881
	struct tcp_md5sig_key *key;
	struct hlist_node *pos;
	unsigned int size = sizeof(struct in_addr);
882
	struct tcp_md5sig_info *md5sig;
883

884 885 886 887
	/* caller either holds rcu_read_lock() or socket lock */
	md5sig = rcu_dereference_check(tp->md5sig_info,
				       sock_owned_by_user(sk));
	if (!md5sig)
888
		return NULL;
E
Eric Dumazet 已提交
889 890 891 892
#if IS_ENABLED(CONFIG_IPV6)
	if (family == AF_INET6)
		size = sizeof(struct in6_addr);
#endif
893
	hlist_for_each_entry_rcu(key, pos, &md5sig->head, node) {
E
Eric Dumazet 已提交
894 895 896 897
		if (key->family != family)
			continue;
		if (!memcmp(&key->addr, addr, size))
			return key;
898 899 900
	}
	return NULL;
}
E
Eric Dumazet 已提交
901
EXPORT_SYMBOL(tcp_md5_do_lookup);
902 903 904 905

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
906 907 908 909
	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);
910 911 912
}
EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
913 914
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
915
{
E
Eric Dumazet 已提交
916 917 918 919
	union tcp_md5_addr *addr;

	addr = (union tcp_md5_addr *)&inet_rsk(req)->rmt_addr;
	return tcp_md5_do_lookup(sk, addr, AF_INET);
920 921 922
}

/* This can be called on a newly created socket, from other files */
E
Eric Dumazet 已提交
923 924
int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
		   int family, const u8 *newkey, u8 newkeylen, gfp_t gfp)
925 926
{
	/* Add Key to the list */
927
	struct tcp_md5sig_key *key;
928
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
929
	struct tcp_md5sig_info *md5sig;
930

E
Eric Dumazet 已提交
931
	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&addr, AF_INET);
932 933
	if (key) {
		/* Pre-existing entry - just update that one. */
E
Eric Dumazet 已提交
934
		memcpy(key->key, newkey, newkeylen);
935
		key->keylen = newkeylen;
E
Eric Dumazet 已提交
936 937
		return 0;
	}
938

939 940
	md5sig = rcu_dereference_protected(tp->md5sig_info,
					   sock_owned_by_user(sk));
E
Eric Dumazet 已提交
941 942 943
	if (!md5sig) {
		md5sig = kmalloc(sizeof(*md5sig), gfp);
		if (!md5sig)
944 945
			return -ENOMEM;

E
Eric Dumazet 已提交
946 947
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
		INIT_HLIST_HEAD(&md5sig->head);
948
		rcu_assign_pointer(tp->md5sig_info, md5sig);
E
Eric Dumazet 已提交
949
	}
950

951
	key = sock_kmalloc(sk, sizeof(*key), gfp);
E
Eric Dumazet 已提交
952 953 954
	if (!key)
		return -ENOMEM;
	if (hlist_empty(&md5sig->head) && !tcp_alloc_md5sig_pool(sk)) {
955
		sock_kfree_s(sk, key, sizeof(*key));
E
Eric Dumazet 已提交
956
		return -ENOMEM;
957
	}
E
Eric Dumazet 已提交
958 959 960 961 962 963 964 965

	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);
966 967
	return 0;
}
E
Eric Dumazet 已提交
968
EXPORT_SYMBOL(tcp_md5_do_add);
969

E
Eric Dumazet 已提交
970
int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr, int family)
971 972
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
973
	struct tcp_md5sig_key *key;
974
	struct tcp_md5sig_info *md5sig;
E
Eric Dumazet 已提交
975 976 977 978 979

	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&addr, AF_INET);
	if (!key)
		return -ENOENT;
	hlist_del_rcu(&key->node);
980
	atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
981
	kfree_rcu(key, rcu);
982 983 984
	md5sig = rcu_dereference_protected(tp->md5sig_info,
					   sock_owned_by_user(sk));
	if (hlist_empty(&md5sig->head))
E
Eric Dumazet 已提交
985 986
		tcp_free_md5sig_pool();
	return 0;
987
}
E
Eric Dumazet 已提交
988
EXPORT_SYMBOL(tcp_md5_do_del);
989

E
Eric Dumazet 已提交
990
void tcp_clear_md5_list(struct sock *sk)
991 992
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
993 994
	struct tcp_md5sig_key *key;
	struct hlist_node *pos, *n;
995
	struct tcp_md5sig_info *md5sig;
996

997 998 999
	md5sig = rcu_dereference_protected(tp->md5sig_info, 1);

	if (!hlist_empty(&md5sig->head))
1000
		tcp_free_md5sig_pool();
1001
	hlist_for_each_entry_safe(key, pos, n, &md5sig->head, node) {
E
Eric Dumazet 已提交
1002
		hlist_del_rcu(&key->node);
1003
		atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1004
		kfree_rcu(key, rcu);
1005 1006 1007
	}
}

1008 1009
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
1010 1011 1012 1013 1014 1015 1016
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;

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

1017
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
1018 1019 1020 1021 1022
		return -EFAULT;

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

1023
	if (!cmd.tcpm_key || !cmd.tcpm_keylen)
E
Eric Dumazet 已提交
1024 1025
		return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin->sin_addr.s_addr,
				      AF_INET);
1026 1027 1028 1029

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

E
Eric Dumazet 已提交
1030 1031 1032
	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);
1033 1034
}

1035 1036
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1037 1038
{
	struct tcp4_pseudohdr *bp;
1039
	struct scatterlist sg;
1040 1041 1042 1043

	bp = &hp->md5_blk.ip4;

	/*
1044
	 * 1. the TCP pseudo-header (in the order: source IP address,
1045 1046 1047 1048 1049 1050
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1051
	bp->protocol = IPPROTO_TCP;
1052
	bp->len = cpu_to_be16(nbytes);
1053

1054 1055 1056 1057
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}

E
Eric Dumazet 已提交
1058
static int tcp_v4_md5_hash_hdr(char *md5_hash, const struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1059
			       __be32 daddr, __be32 saddr, const struct tcphdr *th)
1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
{
	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))
1078 1079 1080 1081
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1082

1083 1084 1085 1086
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1087
	return 1;
1088 1089
}

1090
int tcp_v4_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1091 1092
			const struct sock *sk, const struct request_sock *req,
			const struct sk_buff *skb)
1093
{
1094 1095
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
E
Eric Dumazet 已提交
1096
	const struct tcphdr *th = tcp_hdr(skb);
1097 1098 1099
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1100 1101
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1102 1103 1104
	} else if (req) {
		saddr = inet_rsk(req)->loc_addr;
		daddr = inet_rsk(req)->rmt_addr;
1105
	} else {
1106 1107 1108
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
1109
	}
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137

	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;
1138
}
1139
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1140

E
Eric Dumazet 已提交
1141
static int tcp_v4_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
1142 1143 1144 1145 1146 1147 1148 1149 1150
{
	/*
	 * 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.
	 */
1151
	const __u8 *hash_location = NULL;
1152
	struct tcp_md5sig_key *hash_expected;
1153
	const struct iphdr *iph = ip_hdr(skb);
1154
	const struct tcphdr *th = tcp_hdr(skb);
1155 1156 1157
	int genhash;
	unsigned char newhash[16];

E
Eric Dumazet 已提交
1158 1159
	hash_expected = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&iph->saddr,
					  AF_INET);
1160
	hash_location = tcp_parse_md5sig_option(th);
1161 1162 1163 1164 1165 1166

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

	if (hash_expected && !hash_location) {
1167
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
1168 1169 1170 1171
		return 1;
	}

	if (!hash_expected && hash_location) {
1172
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
1173 1174 1175 1176 1177 1178
		return 1;
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1179 1180 1181
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1182 1183 1184

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
		if (net_ratelimit()) {
1185 1186 1187
			printk(KERN_INFO "MD5 Hash failed for (%pI4, %d)->(%pI4, %d)%s\n",
			       &iph->saddr, ntohs(th->source),
			       &iph->daddr, ntohs(th->dest),
1188 1189 1190 1191 1192 1193 1194 1195 1196
			       genhash ? " tcp_v4_calc_md5_hash failed" : "");
		}
		return 1;
	}
	return 0;
}

#endif

1197
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1198
	.family		=	PF_INET,
1199
	.obj_size	=	sizeof(struct tcp_request_sock),
1200
	.rtx_syn_ack	=	tcp_v4_rtx_synack,
1201 1202
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1203
	.send_reset	=	tcp_v4_send_reset,
1204
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
1205 1206
};

1207
#ifdef CONFIG_TCP_MD5SIG
1208
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1209
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1210
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1211
};
1212
#endif
1213

L
Linus Torvalds 已提交
1214 1215
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
1216
	struct tcp_extend_values tmp_ext;
L
Linus Torvalds 已提交
1217
	struct tcp_options_received tmp_opt;
1218
	const u8 *hash_location;
1219
	struct request_sock *req;
1220
	struct inet_request_sock *ireq;
1221
	struct tcp_sock *tp = tcp_sk(sk);
1222
	struct dst_entry *dst = NULL;
1223 1224
	__be32 saddr = ip_hdr(skb)->saddr;
	__be32 daddr = ip_hdr(skb)->daddr;
L
Linus Torvalds 已提交
1225 1226 1227 1228
	__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
1229
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1230 1231 1232 1233 1234 1235
		goto drop;

	/* TW buckets are converted to open requests without
	 * limitations, they conserve resources and peer is
	 * evidently real one.
	 */
1236
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
1237 1238 1239
		want_cookie = tcp_syn_flood_action(sk, skb, "TCP");
		if (!want_cookie)
			goto drop;
L
Linus Torvalds 已提交
1240 1241 1242 1243 1244 1245 1246
	}

	/* 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.
	 */
1247
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1248 1249
		goto drop;

1250
	req = inet_reqsk_alloc(&tcp_request_sock_ops);
L
Linus Torvalds 已提交
1251 1252 1253
	if (!req)
		goto drop;

1254 1255 1256 1257
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv4_ops;
#endif

L
Linus Torvalds 已提交
1258
	tcp_clear_options(&tmp_opt);
1259
	tmp_opt.mss_clamp = TCP_MSS_DEFAULT;
1260
	tmp_opt.user_mss  = tp->rx_opt.user_mss;
1261
	tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276

	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 */
1277 1278
		*mess++ ^= (__force u32)daddr;
		*mess++ ^= (__force u32)saddr;
L
Linus Torvalds 已提交
1279

1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
		/* 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
1296

1297
	if (want_cookie && !tmp_opt.saw_tstamp)
L
Linus Torvalds 已提交
1298 1299 1300 1301 1302
		tcp_clear_options(&tmp_opt);

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

1303 1304 1305 1306 1307 1308
	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);

1309
	if (security_inet_conn_request(sk, skb, req))
1310
		goto drop_and_free;
1311

1312
	if (!want_cookie || tmp_opt.tstamp_ok)
1313
		TCP_ECN_create_request(req, tcp_hdr(skb));
L
Linus Torvalds 已提交
1314 1315 1316

	if (want_cookie) {
		isn = cookie_v4_init_sequence(sk, skb, &req->mss);
1317
		req->cookie_ts = tmp_opt.tstamp_ok;
L
Linus Torvalds 已提交
1318 1319
	} else if (!isn) {
		struct inet_peer *peer = NULL;
1320
		struct flowi4 fl4;
L
Linus Torvalds 已提交
1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331

		/* 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 &&
1332
		    tcp_death_row.sysctl_tw_recycle &&
1333
		    (dst = inet_csk_route_req(sk, &fl4, req)) != NULL &&
1334 1335
		    fl4.daddr == saddr &&
		    (peer = rt_get_peer((struct rtable *)dst, fl4.daddr)) != NULL) {
1336
			inet_peer_refcheck(peer);
E
Eric Dumazet 已提交
1337
			if ((u32)get_seconds() - peer->tcp_ts_stamp < TCP_PAWS_MSL &&
L
Linus Torvalds 已提交
1338 1339
			    (s32)(peer->tcp_ts - req->ts_recent) >
							TCP_PAWS_WINDOW) {
1340
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1341
				goto drop_and_release;
L
Linus Torvalds 已提交
1342 1343 1344 1345
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
1346
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
L
Linus Torvalds 已提交
1347 1348 1349 1350 1351 1352 1353 1354 1355 1356
			  (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.
			 */
1357 1358
			LIMIT_NETDEBUG(KERN_DEBUG "TCP: drop open request from %pI4/%u\n",
				       &saddr, ntohs(tcp_hdr(skb)->source));
1359
			goto drop_and_release;
L
Linus Torvalds 已提交
1360 1361
		}

1362
		isn = tcp_v4_init_sequence(skb);
L
Linus Torvalds 已提交
1363
	}
1364
	tcp_rsk(req)->snt_isn = isn;
1365
	tcp_rsk(req)->snt_synack = tcp_time_stamp;
L
Linus Torvalds 已提交
1366

1367 1368
	if (tcp_v4_send_synack(sk, dst, req,
			       (struct request_values *)&tmp_ext) ||
1369
	    want_cookie)
L
Linus Torvalds 已提交
1370 1371
		goto drop_and_free;

1372
	inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
L
Linus Torvalds 已提交
1373 1374
	return 0;

1375 1376
drop_and_release:
	dst_release(dst);
L
Linus Torvalds 已提交
1377
drop_and_free:
1378
	reqsk_free(req);
L
Linus Torvalds 已提交
1379 1380 1381
drop:
	return 0;
}
E
Eric Dumazet 已提交
1382
EXPORT_SYMBOL(tcp_v4_conn_request);
L
Linus Torvalds 已提交
1383 1384 1385 1386 1387 1388 1389


/*
 * 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,
1390
				  struct request_sock *req,
L
Linus Torvalds 已提交
1391 1392
				  struct dst_entry *dst)
{
1393
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1394 1395 1396
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1397 1398 1399
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
1400
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
1401 1402 1403 1404 1405 1406

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (!newsk)
1407
		goto exit_nonewsk;
L
Linus Torvalds 已提交
1408

1409
	newsk->sk_gso_type = SKB_GSO_TCPV4;
L
Linus Torvalds 已提交
1410 1411 1412

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1413
	ireq		      = inet_rsk(req);
E
Eric Dumazet 已提交
1414 1415 1416
	newinet->inet_daddr   = ireq->rmt_addr;
	newinet->inet_rcv_saddr = ireq->loc_addr;
	newinet->inet_saddr	      = ireq->loc_addr;
1417 1418
	inet_opt	      = ireq->opt;
	rcu_assign_pointer(newinet->inet_opt, inet_opt);
1419
	ireq->opt	      = NULL;
1420
	newinet->mc_index     = inet_iif(skb);
1421
	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
1422
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
1423 1424
	if (inet_opt)
		inet_csk(newsk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
E
Eric Dumazet 已提交
1425
	newinet->inet_id = newtp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
1426

1427 1428 1429 1430 1431
	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
1432
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1433
	tcp_sync_mss(newsk, dst_mtu(dst));
1434
	newtp->advmss = dst_metric_advmss(dst);
1435 1436 1437 1438
	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
1439
	tcp_initialize_rcv_mss(newsk);
1440 1441 1442 1443
	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
1444

1445 1446
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1447 1448
	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&newinet->inet_daddr,
				AF_INET);
E
Eric Dumazet 已提交
1449
	if (key != NULL) {
1450 1451 1452 1453 1454 1455
		/*
		 * 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
1456 1457
		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
1458
		sk_nocaps_add(newsk, NETIF_F_GSO_MASK);
1459 1460 1461
	}
#endif

1462 1463
	if (__inet_inherit_port(sk, newsk) < 0)
		goto put_and_exit;
1464
	__inet_hash_nolisten(newsk, NULL);
L
Linus Torvalds 已提交
1465 1466 1467 1468

	return newsk;

exit_overflow:
1469
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1470 1471
exit_nonewsk:
	dst_release(dst);
L
Linus Torvalds 已提交
1472
exit:
1473
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1474
	return NULL;
1475
put_and_exit:
1476
	tcp_clear_xmit_timers(newsk);
1477
	tcp_cleanup_congestion_control(newsk);
1478
	bh_unlock_sock(newsk);
1479 1480
	sock_put(newsk);
	goto exit;
L
Linus Torvalds 已提交
1481
}
E
Eric Dumazet 已提交
1482
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
L
Linus Torvalds 已提交
1483 1484 1485

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1486
	struct tcphdr *th = tcp_hdr(skb);
1487
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1488
	struct sock *nsk;
1489
	struct request_sock **prev;
L
Linus Torvalds 已提交
1490
	/* Find possible connection requests. */
1491 1492
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1493 1494 1495
	if (req)
		return tcp_check_req(sk, skb, req, prev);

1496
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1497
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1498 1499 1500 1501 1502 1503

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1504
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1505 1506 1507 1508
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
1509
	if (!th->syn)
L
Linus Torvalds 已提交
1510 1511 1512 1513 1514
		sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
#endif
	return sk;
}

1515
static __sum16 tcp_v4_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1516
{
1517 1518
	const struct iphdr *iph = ip_hdr(skb);

1519
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
1520 1521
		if (!tcp_v4_check(skb->len, iph->saddr,
				  iph->daddr, skb->csum)) {
1522
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1523
			return 0;
1524
		}
L
Linus Torvalds 已提交
1525
	}
1526

1527
	skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
1528 1529
				       skb->len, IPPROTO_TCP, 0);

L
Linus Torvalds 已提交
1530
	if (skb->len <= 76) {
1531
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546
	}
	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)
{
1547 1548 1549 1550 1551 1552 1553 1554
	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
	 */
1555
	if (tcp_v4_inbound_md5_hash(sk, skb))
1556 1557 1558
		goto discard;
#endif

L
Linus Torvalds 已提交
1559
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1560
		sock_rps_save_rxhash(sk, skb);
1561
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
1562
			rsk = sk;
L
Linus Torvalds 已提交
1563
			goto reset;
1564
		}
L
Linus Torvalds 已提交
1565 1566 1567
		return 0;
	}

1568
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1569 1570 1571 1572 1573 1574 1575 1576
		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) {
1577
			sock_rps_save_rxhash(nsk, skb);
1578 1579
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1580
				goto reset;
1581
			}
L
Linus Torvalds 已提交
1582 1583
			return 0;
		}
1584
	} else
1585
		sock_rps_save_rxhash(sk, skb);
1586

1587
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1588
		rsk = sk;
L
Linus Torvalds 已提交
1589
		goto reset;
1590
	}
L
Linus Torvalds 已提交
1591 1592 1593
	return 0;

reset:
1594
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1595 1596 1597 1598 1599 1600 1601 1602 1603 1604
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:
1605
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1606 1607
	goto discard;
}
E
Eric Dumazet 已提交
1608
EXPORT_SYMBOL(tcp_v4_do_rcv);
L
Linus Torvalds 已提交
1609 1610 1611 1612 1613 1614 1615

/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1616
	const struct iphdr *iph;
1617
	const struct tcphdr *th;
L
Linus Torvalds 已提交
1618 1619
	struct sock *sk;
	int ret;
1620
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1621 1622 1623 1624 1625

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

	/* Count it even if it's bad */
1626
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1627 1628 1629 1630

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

1631
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1632 1633 1634 1635 1636 1637 1638 1639

	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
1640
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1641
	 * So, we defer the checks. */
1642
	if (!skb_csum_unnecessary(skb) && tcp_v4_checksum_init(skb))
L
Linus Torvalds 已提交
1643 1644
		goto bad_packet;

1645
	th = tcp_hdr(skb);
1646
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1647 1648 1649 1650 1651
	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;
1652
	TCP_SKB_CB(skb)->ip_dsfield = ipv4_get_dsfield(iph);
L
Linus Torvalds 已提交
1653 1654
	TCP_SKB_CB(skb)->sacked	 = 0;

1655
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1656 1657 1658
	if (!sk)
		goto no_tcp_socket;

E
Eric Dumazet 已提交
1659 1660 1661 1662
process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1663 1664
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1665
		goto discard_and_relse;
1666
	}
1667

L
Linus Torvalds 已提交
1668 1669
	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1670
	nf_reset(skb);
L
Linus Torvalds 已提交
1671

1672
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1673 1674 1675 1676
		goto discard_and_relse;

	skb->dev = NULL;

1677
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1678 1679
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1680 1681 1682
#ifdef CONFIG_NET_DMA
		struct tcp_sock *tp = tcp_sk(sk);
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1683
			tp->ucopy.dma_chan = dma_find_channel(DMA_MEMCPY);
C
Chris Leech 已提交
1684
		if (tp->ucopy.dma_chan)
L
Linus Torvalds 已提交
1685
			ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1686 1687 1688 1689
		else
#endif
		{
			if (!tcp_prequeue(sk, skb))
S
Shan Wei 已提交
1690
				ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1691
		}
1692
	} else if (unlikely(sk_add_backlog(sk, skb))) {
Z
Zhu Yi 已提交
1693
		bh_unlock_sock(sk);
1694
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1695 1696
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708
	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:
1709
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1710
	} else {
1711
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1712 1713 1714 1715 1716
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
1717
	return 0;
L
Linus Torvalds 已提交
1718 1719 1720 1721 1722 1723 1724

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1725
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1726 1727 1728 1729
		goto discard_it;
	}

	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
1730
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1731
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1732 1733
		goto discard_it;
	}
1734
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1735
	case TCP_TW_SYN: {
1736
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
1737
							&tcp_hashinfo,
1738
							iph->daddr, th->dest,
1739
							inet_iif(skb));
L
Linus Torvalds 已提交
1740
		if (sk2) {
1741 1742
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
			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;
}

1758
struct inet_peer *tcp_v4_get_peer(struct sock *sk, bool *release_it)
L
Linus Torvalds 已提交
1759
{
1760
	struct rtable *rt = (struct rtable *) __sk_dst_get(sk);
L
Linus Torvalds 已提交
1761
	struct inet_sock *inet = inet_sk(sk);
1762
	struct inet_peer *peer;
L
Linus Torvalds 已提交
1763

1764 1765
	if (!rt ||
	    inet->cork.fl.u.ip4.daddr != inet->inet_daddr) {
1766
		peer = inet_getpeer_v4(inet->inet_daddr, 1);
1767
		*release_it = true;
L
Linus Torvalds 已提交
1768 1769
	} else {
		if (!rt->peer)
1770
			rt_bind_peer(rt, inet->inet_daddr, 1);
L
Linus Torvalds 已提交
1771
		peer = rt->peer;
1772
		*release_it = false;
L
Linus Torvalds 已提交
1773 1774
	}

1775
	return peer;
L
Linus Torvalds 已提交
1776
}
1777
EXPORT_SYMBOL(tcp_v4_get_peer);
L
Linus Torvalds 已提交
1778

1779
void *tcp_v4_tw_get_peer(struct sock *sk)
L
Linus Torvalds 已提交
1780
{
1781
	const struct inet_timewait_sock *tw = inet_twsk(sk);
L
Linus Torvalds 已提交
1782

1783
	return inet_getpeer_v4(tw->tw_daddr, 1);
L
Linus Torvalds 已提交
1784
}
1785 1786 1787 1788 1789 1790 1791 1792
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
1793

1794
const struct inet_connection_sock_af_ops ipv4_specific = {
1795 1796 1797 1798 1799
	.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,
1800
	.get_peer	   = tcp_v4_get_peer,
1801 1802 1803 1804 1805
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ip_setsockopt,
	.getsockopt	   = ip_getsockopt,
	.addr2sockaddr	   = inet_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in),
1806
	.bind_conflict	   = inet_csk_bind_conflict,
1807
#ifdef CONFIG_COMPAT
1808 1809
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
1810
#endif
L
Linus Torvalds 已提交
1811
};
E
Eric Dumazet 已提交
1812
EXPORT_SYMBOL(ipv4_specific);
L
Linus Torvalds 已提交
1813

1814
#ifdef CONFIG_TCP_MD5SIG
1815
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
1816
	.md5_lookup		= tcp_v4_md5_lookup,
1817
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
1818 1819
	.md5_parse		= tcp_v4_parse_md5_keys,
};
1820
#endif
1821

L
Linus Torvalds 已提交
1822 1823 1824 1825 1826
/* NOTE: A lot of things set to zero explicitly by call to
 *       sk_alloc() so need not be done here.
 */
static int tcp_v4_init_sock(struct sock *sk)
{
1827
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1828 1829 1830 1831 1832 1833
	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);

1834
	icsk->icsk_rto = TCP_TIMEOUT_INIT;
L
Linus Torvalds 已提交
1835 1836 1837 1838 1839 1840 1841
	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
	 */
1842
	tp->snd_cwnd = TCP_INIT_CWND;
L
Linus Torvalds 已提交
1843 1844 1845 1846

	/* See draft-stevens-tcpca-spec-01 for discussion of the
	 * initialization of these values.
	 */
I
Ilpo Järvinen 已提交
1847
	tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
L
Linus Torvalds 已提交
1848
	tp->snd_cwnd_clamp = ~0;
1849
	tp->mss_cache = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
1850 1851

	tp->reordering = sysctl_tcp_reordering;
1852
	icsk->icsk_ca_ops = &tcp_init_congestion_ops;
L
Linus Torvalds 已提交
1853 1854 1855 1856 1857 1858

	sk->sk_state = TCP_CLOSE;

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

1859
	icsk->icsk_af_ops = &ipv4_specific;
1860
	icsk->icsk_sync_mss = tcp_sync_mss;
1861 1862 1863
#ifdef CONFIG_TCP_MD5SIG
	tp->af_specific = &tcp_sock_ipv4_specific;
#endif
L
Linus Torvalds 已提交
1864

1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
	/* 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
1878 1879 1880
	sk->sk_sndbuf = sysctl_tcp_wmem[1];
	sk->sk_rcvbuf = sysctl_tcp_rmem[1];

H
Herbert Xu 已提交
1881
	local_bh_disable();
G
Glauber Costa 已提交
1882
	sock_update_memcg(sk);
1883
	sk_sockets_allocated_inc(sk);
H
Herbert Xu 已提交
1884
	local_bh_enable();
L
Linus Torvalds 已提交
1885 1886 1887 1888

	return 0;
}

1889
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1890 1891 1892 1893 1894
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

1895
	tcp_cleanup_congestion_control(sk);
1896

L
Linus Torvalds 已提交
1897
	/* Cleanup up the write buffer. */
1898
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
1899 1900

	/* Cleans up our, hopefully empty, out_of_order_queue. */
1901
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
1902

1903 1904 1905
#ifdef CONFIG_TCP_MD5SIG
	/* Clean up the MD5 key list, if any */
	if (tp->md5sig_info) {
E
Eric Dumazet 已提交
1906
		tcp_clear_md5_list(sk);
1907
		kfree_rcu(tp->md5sig_info, rcu);
1908 1909 1910 1911
		tp->md5sig_info = NULL;
	}
#endif

C
Chris Leech 已提交
1912 1913
#ifdef CONFIG_NET_DMA
	/* Cleans up our sk_async_wait_queue */
1914
	__skb_queue_purge(&sk->sk_async_wait_queue);
C
Chris Leech 已提交
1915 1916
#endif

L
Linus Torvalds 已提交
1917 1918 1919 1920
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
1921
	if (inet_csk(sk)->icsk_bind_hash)
1922
		inet_put_port(sk);
L
Linus Torvalds 已提交
1923 1924 1925 1926 1927 1928 1929 1930 1931

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

1932 1933 1934 1935 1936 1937 1938
	/* TCP Cookie Transactions */
	if (tp->cookie_values != NULL) {
		kref_put(&tp->cookie_values->kref,
			 tcp_cookie_values_release);
		tp->cookie_values = NULL;
	}

1939
	sk_sockets_allocated_dec(sk);
G
Glauber Costa 已提交
1940
	sock_release_memcg(sk);
L
Linus Torvalds 已提交
1941 1942 1943 1944 1945 1946
}
EXPORT_SYMBOL(tcp_v4_destroy_sock);

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

1947
static inline struct inet_timewait_sock *tw_head(struct hlist_nulls_head *head)
L
Linus Torvalds 已提交
1948
{
1949
	return hlist_nulls_empty(head) ? NULL :
1950
		list_entry(head->first, struct inet_timewait_sock, tw_node);
L
Linus Torvalds 已提交
1951 1952
}

1953
static inline struct inet_timewait_sock *tw_next(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
1954
{
1955 1956
	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
1957 1958
}

1959 1960 1961 1962 1963
/*
 * 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
1964 1965
static void *listening_get_next(struct seq_file *seq, void *cur)
{
1966
	struct inet_connection_sock *icsk;
1967
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
1968
	struct sock *sk = cur;
1969
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
1970
	struct tcp_iter_state *st = seq->private;
1971
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
1972 1973

	if (!sk) {
1974
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
1975
		spin_lock_bh(&ilb->lock);
1976
		sk = sk_nulls_head(&ilb->head);
1977
		st->offset = 0;
L
Linus Torvalds 已提交
1978 1979
		goto get_sk;
	}
1980
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
1981
	++st->num;
1982
	++st->offset;
L
Linus Torvalds 已提交
1983 1984

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
1985
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
1986

1987
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
1988 1989 1990
		req = req->dl_next;
		while (1) {
			while (req) {
1991
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
1992 1993 1994 1995 1996
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
1997
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
1998 1999
				break;
get_req:
2000
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
2001
		}
E
Eric Dumazet 已提交
2002
		sk	  = sk_nulls_next(st->syn_wait_sk);
L
Linus Torvalds 已提交
2003
		st->state = TCP_SEQ_STATE_LISTENING;
2004
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2005
	} else {
2006
		icsk = inet_csk(sk);
2007 2008
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
2009
			goto start_req;
2010
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
E
Eric Dumazet 已提交
2011
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2012 2013
	}
get_sk:
2014
	sk_nulls_for_each_from(sk, node) {
2015 2016 2017
		if (!net_eq(sock_net(sk), net))
			continue;
		if (sk->sk_family == st->family) {
L
Linus Torvalds 已提交
2018 2019 2020
			cur = sk;
			goto out;
		}
2021
		icsk = inet_csk(sk);
2022 2023
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
2024 2025 2026 2027 2028 2029 2030
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;
		}
2031
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2032
	}
2033
	spin_unlock_bh(&ilb->lock);
2034
	st->offset = 0;
2035
	if (++st->bucket < INET_LHTABLE_SIZE) {
2036 2037
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
2038
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
2039 2040 2041 2042 2043 2044 2045 2046 2047
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
2048 2049 2050 2051 2052 2053
	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
2054 2055 2056 2057 2058 2059 2060 2061

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

2062 2063
static inline int empty_bucket(struct tcp_iter_state *st)
{
2064 2065
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain) &&
		hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].twchain);
2066 2067
}

2068 2069 2070 2071
/*
 * 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
2072 2073
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
2074
	struct tcp_iter_state *st = seq->private;
2075
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2076 2077
	void *rc = NULL;

2078 2079
	st->offset = 0;
	for (; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
2080
		struct sock *sk;
2081
		struct hlist_nulls_node *node;
2082
		struct inet_timewait_sock *tw;
2083
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
2084

2085 2086 2087 2088
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

2089
		spin_lock_bh(lock);
2090
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
2091
			if (sk->sk_family != st->family ||
2092
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2093 2094 2095 2096 2097 2098
				continue;
			}
			rc = sk;
			goto out;
		}
		st->state = TCP_SEQ_STATE_TIME_WAIT;
2099
		inet_twsk_for_each(tw, node,
2100
				   &tcp_hashinfo.ehash[st->bucket].twchain) {
2101
			if (tw->tw_family != st->family ||
2102
			    !net_eq(twsk_net(tw), net)) {
L
Linus Torvalds 已提交
2103 2104 2105 2106 2107
				continue;
			}
			rc = tw;
			goto out;
		}
2108
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2109 2110 2111 2112 2113 2114 2115 2116 2117
		st->state = TCP_SEQ_STATE_ESTABLISHED;
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2118
	struct inet_timewait_sock *tw;
2119
	struct hlist_nulls_node *node;
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

	++st->num;
2124
	++st->offset;
L
Linus Torvalds 已提交
2125 2126 2127 2128 2129

	if (st->state == TCP_SEQ_STATE_TIME_WAIT) {
		tw = cur;
		tw = tw_next(tw);
get_tw:
2130
		while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
L
Linus Torvalds 已提交
2131 2132 2133 2134 2135 2136
			tw = tw_next(tw);
		}
		if (tw) {
			cur = tw;
			goto out;
		}
2137
		spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2138 2139
		st->state = TCP_SEQ_STATE_ESTABLISHED;

2140
		/* Look for next non empty bucket */
2141
		st->offset = 0;
2142
		while (++st->bucket <= tcp_hashinfo.ehash_mask &&
2143 2144
				empty_bucket(st))
			;
2145
		if (st->bucket > tcp_hashinfo.ehash_mask)
2146 2147
			return NULL;

2148
		spin_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
2149
		sk = sk_nulls_head(&tcp_hashinfo.ehash[st->bucket].chain);
L
Linus Torvalds 已提交
2150
	} else
2151
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2152

2153
	sk_nulls_for_each_from(sk, node) {
2154
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
L
Linus Torvalds 已提交
2155 2156 2157 2158
			goto found;
	}

	st->state = TCP_SEQ_STATE_TIME_WAIT;
2159
	tw = tw_head(&tcp_hashinfo.ehash[st->bucket].twchain);
L
Linus Torvalds 已提交
2160 2161 2162 2163 2164 2165 2166 2167 2168
	goto get_tw;
found:
	cur = sk;
out:
	return cur;
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
2169 2170 2171 2172 2173
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	rc = established_get_first(seq);
L
Linus Torvalds 已提交
2174 2175 2176 2177

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2178
	}
L
Linus Torvalds 已提交
2179 2180 2181 2182 2183 2184
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2185
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197

	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;
}

2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232
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
2233 2234
static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2235
	struct tcp_iter_state *st = seq->private;
2236 2237 2238 2239 2240 2241 2242 2243
	void *rc;

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

L
Linus Torvalds 已提交
2244 2245
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
2246 2247 2248 2249 2250 2251 2252
	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
2253 2254 2255 2256
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2257
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270
	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;
2271 2272
			st->bucket = 0;
			st->offset = 0;
L
Linus Torvalds 已提交
2273 2274 2275 2276 2277 2278 2279 2280 2281 2282
			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;
2283
	st->last_pos = *pos;
L
Linus Torvalds 已提交
2284 2285 2286 2287 2288
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2289
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2290 2291 2292 2293

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2294 2295
			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
2296 2297 2298
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2299
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2300 2301 2302 2303
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2304
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2305 2306 2307 2308
		break;
	}
}

2309
int tcp_seq_open(struct inode *inode, struct file *file)
L
Linus Torvalds 已提交
2310 2311 2312
{
	struct tcp_seq_afinfo *afinfo = PDE(inode)->data;
	struct tcp_iter_state *s;
2313
	int err;
L
Linus Torvalds 已提交
2314

2315 2316 2317 2318
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2319

2320
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2321
	s->family		= afinfo->family;
2322
	s->last_pos 		= 0;
2323 2324
	return 0;
}
2325
EXPORT_SYMBOL(tcp_seq_open);
2326

2327
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2328 2329 2330 2331
{
	int rc = 0;
	struct proc_dir_entry *p;

2332 2333 2334 2335
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2336
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
2337
			     afinfo->seq_fops, afinfo);
2338
	if (!p)
L
Linus Torvalds 已提交
2339 2340 2341
		rc = -ENOMEM;
	return rc;
}
E
Eric Dumazet 已提交
2342
EXPORT_SYMBOL(tcp_proc_register);
L
Linus Torvalds 已提交
2343

2344
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2345
{
2346
	proc_net_remove(net, afinfo->name);
L
Linus Torvalds 已提交
2347
}
E
Eric Dumazet 已提交
2348
EXPORT_SYMBOL(tcp_proc_unregister);
L
Linus Torvalds 已提交
2349

2350
static void get_openreq4(const struct sock *sk, const struct request_sock *req,
2351
			 struct seq_file *f, int i, int uid, int *len)
L
Linus Torvalds 已提交
2352
{
2353
	const struct inet_request_sock *ireq = inet_rsk(req);
L
Linus Torvalds 已提交
2354 2355
	int ttd = req->expires - jiffies;

2356
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2357
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %pK%n",
L
Linus Torvalds 已提交
2358
		i,
2359
		ireq->loc_addr,
E
Eric Dumazet 已提交
2360
		ntohs(inet_sk(sk)->inet_sport),
2361 2362
		ireq->rmt_addr,
		ntohs(ireq->rmt_port),
L
Linus Torvalds 已提交
2363 2364 2365 2366 2367 2368 2369 2370 2371
		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),
2372 2373
		req,
		len);
L
Linus Torvalds 已提交
2374 2375
}

2376
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2377 2378 2379
{
	int timer_active;
	unsigned long timer_expires;
2380
	const struct tcp_sock *tp = tcp_sk(sk);
2381
	const struct inet_connection_sock *icsk = inet_csk(sk);
2382
	const struct inet_sock *inet = inet_sk(sk);
E
Eric Dumazet 已提交
2383 2384 2385 2386
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
2387
	int rx_queue;
L
Linus Torvalds 已提交
2388

2389
	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
2390
		timer_active	= 1;
2391 2392
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2393
		timer_active	= 4;
2394
		timer_expires	= icsk->icsk_timeout;
2395
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2396
		timer_active	= 2;
2397
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2398 2399 2400 2401 2402
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2403 2404 2405 2406 2407 2408 2409 2410
	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);

2411
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
D
Dan Rosenberg 已提交
2412
			"%08X %5d %8d %lu %d %pK %lu %lu %u %u %d%n",
2413
		i, src, srcp, dest, destp, sk->sk_state,
2414
		tp->write_seq - tp->snd_una,
2415
		rx_queue,
L
Linus Torvalds 已提交
2416 2417
		timer_active,
		jiffies_to_clock_t(timer_expires - jiffies),
2418
		icsk->icsk_retransmits,
2419
		sock_i_uid(sk),
2420
		icsk->icsk_probes_out,
2421 2422
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2423 2424
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2425
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2426
		tp->snd_cwnd,
I
Ilpo Järvinen 已提交
2427
		tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh,
2428
		len);
L
Linus Torvalds 已提交
2429 2430
}

2431
static void get_timewait4_sock(const struct inet_timewait_sock *tw,
2432
			       struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2433
{
2434
	__be32 dest, src;
L
Linus Torvalds 已提交
2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445
	__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);

2446
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2447
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK%n",
L
Linus Torvalds 已提交
2448 2449
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
		3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
2450
		atomic_read(&tw->tw_refcnt), tw, len);
L
Linus Torvalds 已提交
2451 2452 2453 2454 2455 2456
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2457
	struct tcp_iter_state *st;
2458
	int len;
L
Linus Torvalds 已提交
2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471

	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:
2472
		get_tcp4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2473 2474
		break;
	case TCP_SEQ_STATE_OPENREQ:
2475
		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid, &len);
L
Linus Torvalds 已提交
2476 2477
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
2478
		get_timewait4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2479 2480
		break;
	}
2481
	seq_printf(seq, "%*s\n", TMPSZ - 1 - len, "");
L
Linus Torvalds 已提交
2482 2483 2484 2485
out:
	return 0;
}

2486 2487 2488 2489 2490 2491 2492 2493
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
2494 2495 2496
static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2497
	.seq_fops	= &tcp_afinfo_seq_fops,
2498 2499 2500
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2501 2502
};

2503
static int __net_init tcp4_proc_init_net(struct net *net)
2504 2505 2506 2507
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

2508
static void __net_exit tcp4_proc_exit_net(struct net *net)
2509 2510 2511 2512 2513 2514 2515 2516 2517
{
	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
2518 2519
int __init tcp4_proc_init(void)
{
2520
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2521 2522 2523 2524
}

void tcp4_proc_exit(void)
{
2525
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2526 2527 2528
}
#endif /* CONFIG_PROC_FS */

H
Herbert Xu 已提交
2529 2530
struct sk_buff **tcp4_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
2531
	const struct iphdr *iph = skb_gro_network_header(skb);
H
Herbert Xu 已提交
2532 2533 2534

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
2535
		if (!tcp_v4_check(skb_gro_len(skb), iph->saddr, iph->daddr,
H
Herbert Xu 已提交
2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551
				  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)
{
2552
	const struct iphdr *iph = ip_hdr(skb);
H
Herbert Xu 已提交
2553 2554 2555 2556 2557 2558 2559 2560 2561
	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
2562 2563 2564 2565 2566 2567
struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2568
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2569 2570 2571 2572 2573 2574 2575
	.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,
2576 2577
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2578
	.backlog_rcv		= tcp_v4_do_rcv,
2579 2580 2581
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2582 2583
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
2584
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2585 2586 2587 2588 2589 2590
	.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),
2591
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2592
	.twsk_prot		= &tcp_timewait_sock_ops,
2593
	.rsk_prot		= &tcp_request_sock_ops,
2594
	.h.hashinfo		= &tcp_hashinfo,
2595
	.no_autobind		= true,
2596 2597 2598 2599
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
G
Glauber Costa 已提交
2600 2601 2602 2603 2604
#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
2605
};
E
Eric Dumazet 已提交
2606
EXPORT_SYMBOL(tcp_prot);
L
Linus Torvalds 已提交
2607

2608 2609 2610 2611 2612 2613 2614 2615 2616
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
2617 2618 2619 2620 2621
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2622 2623 2624
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2625 2626 2627
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2628 2629
};

2630
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2631
{
2632
	inet_hashinfo_init(&tcp_hashinfo);
2633
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2634 2635
		panic("Failed to create the TCP control socket.\n");
}