tcp_ipv4.c 67.0 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>
L
Linus Torvalds 已提交
76 77 78 79 80 81 82

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

83 84 85
#include <linux/crypto.h>
#include <linux/scatterlist.h>

86 87
int sysctl_tcp_tw_reuse __read_mostly;
int sysctl_tcp_low_latency __read_mostly;
E
Eric Dumazet 已提交
88
EXPORT_SYMBOL(sysctl_tcp_low_latency);
L
Linus Torvalds 已提交
89 90


91
#ifdef CONFIG_TCP_MD5SIG
92 93
static struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk,
						   __be32 addr);
94
static int tcp_v4_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
95
			       __be32 daddr, __be32 saddr, const struct tcphdr *th);
96 97 98 99 100 101
#else
static inline
struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk, __be32 addr)
{
	return NULL;
}
102 103
#endif

104
struct inet_hashinfo tcp_hashinfo;
E
Eric Dumazet 已提交
105
EXPORT_SYMBOL(tcp_hashinfo);
L
Linus Torvalds 已提交
106

107
static inline __u32 tcp_v4_init_sequence(const struct sk_buff *skb)
L
Linus Torvalds 已提交
108
{
109 110
	return secure_tcp_sequence_number(ip_hdr(skb)->daddr,
					  ip_hdr(skb)->saddr,
111 112
					  tcp_hdr(skb)->dest,
					  tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
113 114
}

115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
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 &&
133
			     get_seconds() - tcptw->tw_ts_recent_stamp > 1))) {
134 135 136 137 138 139 140 141 142 143 144 145 146
		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
147 148 149
/* This will initiate an outgoing connection. */
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
{
150
	struct sockaddr_in *usin = (struct sockaddr_in *)uaddr;
L
Linus Torvalds 已提交
151 152
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
153
	__be16 orig_sport, orig_dport;
154
	__be32 daddr, nexthop;
155
	struct flowi4 *fl4;
156
	struct rtable *rt;
L
Linus Torvalds 已提交
157
	int err;
158
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
159 160 161 162 163 164 165 166

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

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

	nexthop = daddr = usin->sin_addr.s_addr;
167 168 169
	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
170 171
		if (!daddr)
			return -EINVAL;
172
		nexthop = inet_opt->opt.faddr;
L
Linus Torvalds 已提交
173 174
	}

175 176
	orig_sport = inet->inet_sport;
	orig_dport = usin->sin_port;
177 178
	fl4 = &inet->cork.fl.u.ip4;
	rt = ip_route_connect(fl4, nexthop, inet->inet_saddr,
179 180 181 182 183 184
			      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)
185
			IP_INC_STATS_BH(sock_net(sk), IPSTATS_MIB_OUTNOROUTES);
186
		return err;
187
	}
L
Linus Torvalds 已提交
188 189 190 191 192 193

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

194
	if (!inet_opt || !inet_opt->opt.srr)
195
		daddr = fl4->daddr;
L
Linus Torvalds 已提交
196

E
Eric Dumazet 已提交
197
	if (!inet->inet_saddr)
198
		inet->inet_saddr = fl4->saddr;
E
Eric Dumazet 已提交
199
	inet->inet_rcv_saddr = inet->inet_saddr;
L
Linus Torvalds 已提交
200

E
Eric Dumazet 已提交
201
	if (tp->rx_opt.ts_recent_stamp && inet->inet_daddr != daddr) {
L
Linus Torvalds 已提交
202 203 204 205 206 207
		/* Reset inherited state */
		tp->rx_opt.ts_recent	   = 0;
		tp->rx_opt.ts_recent_stamp = 0;
		tp->write_seq		   = 0;
	}

208
	if (tcp_death_row.sysctl_tw_recycle &&
209
	    !tp->rx_opt.ts_recent_stamp && fl4->daddr == daddr) {
210
		struct inet_peer *peer = rt_get_peer(rt, fl4->daddr);
211 212 213 214 215
		/*
		 * 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
216
		 */
217 218 219 220 221 222
		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
223 224 225
		}
	}

E
Eric Dumazet 已提交
226 227
	inet->inet_dport = usin->sin_port;
	inet->inet_daddr = daddr;
L
Linus Torvalds 已提交
228

229
	inet_csk(sk)->icsk_ext_hdr_len = 0;
230 231
	if (inet_opt)
		inet_csk(sk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
L
Linus Torvalds 已提交
232

233
	tp->rx_opt.mss_clamp = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
234 235 236 237 238 239 240

	/* 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);
241
	err = inet_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
242 243 244
	if (err)
		goto failure;

245
	rt = ip_route_newports(fl4, rt, orig_sport, orig_dport,
246 247 248 249
			       inet->inet_sport, inet->inet_dport, sk);
	if (IS_ERR(rt)) {
		err = PTR_ERR(rt);
		rt = NULL;
L
Linus Torvalds 已提交
250
		goto failure;
251
	}
L
Linus Torvalds 已提交
252
	/* OK, now commit destination to socket.  */
253
	sk->sk_gso_type = SKB_GSO_TCPV4;
254
	sk_setup_caps(sk, &rt->dst);
L
Linus Torvalds 已提交
255 256

	if (!tp->write_seq)
E
Eric Dumazet 已提交
257 258 259
		tp->write_seq = secure_tcp_sequence_number(inet->inet_saddr,
							   inet->inet_daddr,
							   inet->inet_sport,
L
Linus Torvalds 已提交
260 261
							   usin->sin_port);

E
Eric Dumazet 已提交
262
	inet->inet_id = tp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
263 264 265 266 267 268 269 270 271

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

	return 0;

failure:
272 273 274 275
	/*
	 * This unhashes the socket and releases the local port,
	 * if necessary.
	 */
L
Linus Torvalds 已提交
276 277 278
	tcp_set_state(sk, TCP_CLOSE);
	ip_rt_put(rt);
	sk->sk_route_caps = 0;
E
Eric Dumazet 已提交
279
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
280 281
	return err;
}
E
Eric Dumazet 已提交
282
EXPORT_SYMBOL(tcp_v4_connect);
L
Linus Torvalds 已提交
283 284 285 286

/*
 * This routine does path mtu discovery as defined in RFC1191.
 */
287
static void do_pmtu_discovery(struct sock *sk, const struct iphdr *iph, u32 mtu)
L
Linus Torvalds 已提交
288 289 290 291 292 293 294 295 296 297 298 299 300 301
{
	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.
302
	 * There is a small race when the user changes this flag in the
L
Linus Torvalds 已提交
303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318
	 * 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 &&
319
	    inet_csk(sk)->icsk_pmtu_cookie > mtu) {
L
Linus Torvalds 已提交
320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346
		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.
 *
 */

347
void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
L
Linus Torvalds 已提交
348
{
349
	const struct iphdr *iph = (const struct iphdr *)icmp_skb->data;
350
	struct tcphdr *th = (struct tcphdr *)(icmp_skb->data + (iph->ihl << 2));
351
	struct inet_connection_sock *icsk;
L
Linus Torvalds 已提交
352 353
	struct tcp_sock *tp;
	struct inet_sock *inet;
354 355
	const int type = icmp_hdr(icmp_skb)->type;
	const int code = icmp_hdr(icmp_skb)->code;
L
Linus Torvalds 已提交
356
	struct sock *sk;
357
	struct sk_buff *skb;
L
Linus Torvalds 已提交
358
	__u32 seq;
359
	__u32 remaining;
L
Linus Torvalds 已提交
360
	int err;
361
	struct net *net = dev_net(icmp_skb->dev);
L
Linus Torvalds 已提交
362

363
	if (icmp_skb->len < (iph->ihl << 2) + 8) {
364
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
365 366 367
		return;
	}

368
	sk = inet_lookup(net, &tcp_hashinfo, iph->daddr, th->dest,
369
			iph->saddr, th->source, inet_iif(icmp_skb));
L
Linus Torvalds 已提交
370
	if (!sk) {
371
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
372 373 374
		return;
	}
	if (sk->sk_state == TCP_TIME_WAIT) {
375
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
376 377 378 379 380 381 382 383
		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))
384
		NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
L
Linus Torvalds 已提交
385 386 387 388

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

389 390 391 392 393
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto out;
	}

394
	icsk = inet_csk(sk);
L
Linus Torvalds 已提交
395 396 397 398
	tp = tcp_sk(sk);
	seq = ntohl(th->seq);
	if (sk->sk_state != TCP_LISTEN &&
	    !between(seq, tp->snd_una, tp->snd_nxt)) {
399
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420
		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;
421 422 423 424 425 426 427 428
		/* check if icmp_skb allows revert of backoff
		 * (see draft-zimmermann-tcp-lcd) */
		if (code != ICMP_NET_UNREACH && code != ICMP_HOST_UNREACH)
			break;
		if (seq != tp->snd_una  || !icsk->icsk_retransmits ||
		    !icsk->icsk_backoff)
			break;

429 430 431
		if (sock_owned_by_user(sk))
			break;

432
		icsk->icsk_backoff--;
433 434
		inet_csk(sk)->icsk_rto = (tp->srtt ? __tcp_set_rto(tp) :
			TCP_TIMEOUT_INIT) << icsk->icsk_backoff;
435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
		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
452 453 454 455 456 457 458 459 460
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

	switch (sk->sk_state) {
461
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
462 463 464 465
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

466 467
		req = inet_csk_search_req(sk, &prev, th->dest,
					  iph->daddr, iph->saddr);
L
Linus Torvalds 已提交
468 469 470 471 472 473
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		   an established socket here.
		 */
474
		WARN_ON(req->sk);
L
Linus Torvalds 已提交
475

476
		if (seq != tcp_rsk(req)->snt_isn) {
477
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
478 479 480 481 482 483 484 485 486
			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().
		 */
487
		inet_csk_reqsk_queue_drop(sk, req, prev);
L
Linus Torvalds 已提交
488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534
		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);
}

535 536
static void __tcp_v4_send_check(struct sk_buff *skb,
				__be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
537
{
538
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
539

540
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
541
		th->check = ~tcp_v4_check(skb->len, saddr, daddr, 0);
542
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
543
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
544
	} else {
545
		th->check = tcp_v4_check(skb->len, saddr, daddr,
546
					 csum_partial(th,
L
Linus Torvalds 已提交
547 548 549 550 551
						      th->doff << 2,
						      skb->csum));
	}
}

552
/* This routine computes an IPv4 TCP checksum. */
553
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb)
554
{
555
	const struct inet_sock *inet = inet_sk(sk);
556 557 558

	__tcp_v4_send_check(skb, inet->inet_saddr, inet->inet_daddr);
}
E
Eric Dumazet 已提交
559
EXPORT_SYMBOL(tcp_v4_send_check);
560

561 562
int tcp_v4_gso_send_check(struct sk_buff *skb)
{
563
	const struct iphdr *iph;
564 565 566 567 568
	struct tcphdr *th;

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

569
	iph = ip_hdr(skb);
570
	th = tcp_hdr(skb);
571 572

	th->check = 0;
573
	skb->ip_summed = CHECKSUM_PARTIAL;
574
	__tcp_v4_send_check(skb, iph->saddr, iph->daddr);
575 576 577
	return 0;
}

L
Linus Torvalds 已提交
578 579 580 581 582 583 584 585 586 587 588 589 590
/*
 *	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.
 */

591
static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
592
{
593
	const struct tcphdr *th = tcp_hdr(skb);
594 595 596
	struct {
		struct tcphdr th;
#ifdef CONFIG_TCP_MD5SIG
597
		__be32 opt[(TCPOLEN_MD5SIG_ALIGNED >> 2)];
598 599
#endif
	} rep;
L
Linus Torvalds 已提交
600
	struct ip_reply_arg arg;
601 602 603
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
604
	struct net *net;
L
Linus Torvalds 已提交
605 606 607 608 609

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

E
Eric Dumazet 已提交
610
	if (skb_rtable(skb)->rt_type != RTN_LOCAL)
L
Linus Torvalds 已提交
611 612 613
		return;

	/* Swap the send and the receive. */
614 615 616 617 618
	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
619 620

	if (th->ack) {
621
		rep.th.seq = th->ack_seq;
L
Linus Torvalds 已提交
622
	} else {
623 624 625
		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
626 627
	}

628
	memset(&arg, 0, sizeof(arg));
629 630 631 632
	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);

#ifdef CONFIG_TCP_MD5SIG
633
	key = sk ? tcp_v4_md5_do_lookup(sk, ip_hdr(skb)->daddr) : NULL;
634 635 636 637 638 639 640 641 642
	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;

643
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
644 645
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
646 647
	}
#endif
648 649
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
650
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
651
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
652
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
L
Linus Torvalds 已提交
653

E
Eric Dumazet 已提交
654
	net = dev_net(skb_dst(skb)->dev);
655
	arg.tos = ip_hdr(skb)->tos;
656
	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
657
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
658

659 660
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
L
Linus Torvalds 已提交
661 662 663 664 665 666
}

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

667 668
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
			    u32 win, u32 ts, int oif,
669
			    struct tcp_md5sig_key *key,
670
			    int reply_flags, u8 tos)
L
Linus Torvalds 已提交
671
{
672
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
673 674
	struct {
		struct tcphdr th;
675
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
676
#ifdef CONFIG_TCP_MD5SIG
677
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
678 679
#endif
			];
L
Linus Torvalds 已提交
680 681
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
682
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
683 684

	memset(&rep.th, 0, sizeof(struct tcphdr));
685
	memset(&arg, 0, sizeof(arg));
L
Linus Torvalds 已提交
686 687 688 689

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
	if (ts) {
690 691 692 693 694
		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);
695
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
696 697 698 699 700 701 702 703 704 705 706
	}

	/* 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);

707 708 709 710 711 712 713 714 715 716 717
#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;

718
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[offset],
719 720
				    key, ip_hdr(skb)->saddr,
				    ip_hdr(skb)->daddr, &rep.th);
721 722
	}
#endif
723
	arg.flags = reply_flags;
724 725
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
L
Linus Torvalds 已提交
726 727
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
728 729
	if (oif)
		arg.bound_dev_if = oif;
730
	arg.tos = tos;
731
	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
732
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
733

734
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
735 736 737 738
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
739
	struct inet_timewait_sock *tw = inet_twsk(sk);
740
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
741

742
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
743
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
744 745
			tcptw->tw_ts_recent,
			tw->tw_bound_dev_if,
746
			tcp_twsk_md5_key(tcptw),
747 748
			tw->tw_transparent ? IP_REPLY_ARG_NOSRCCHECK : 0,
			tw->tw_tos
749
			);
L
Linus Torvalds 已提交
750

751
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
752 753
}

754
static void tcp_v4_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
755
				  struct request_sock *req)
L
Linus Torvalds 已提交
756
{
757
	tcp_v4_send_ack(skb, tcp_rsk(req)->snt_isn + 1,
758
			tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd,
759 760
			req->ts_recent,
			0,
761
			tcp_v4_md5_do_lookup(sk, ip_hdr(skb)->daddr),
762 763
			inet_rsk(req)->no_srccheck ? IP_REPLY_ARG_NOSRCCHECK : 0,
			ip_hdr(skb)->tos);
L
Linus Torvalds 已提交
764 765 766
}

/*
767
 *	Send a SYN-ACK after having received a SYN.
768
 *	This still operates on a request_sock only, not on a big
L
Linus Torvalds 已提交
769 770
 *	socket.
 */
771 772 773
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
774
{
775
	const struct inet_request_sock *ireq = inet_rsk(req);
776
	struct flowi4 fl4;
L
Linus Torvalds 已提交
777 778 779 780
	int err = -1;
	struct sk_buff * skb;

	/* First, grab a route. */
781
	if (!dst && (dst = inet_csk_route_req(sk, &fl4, req)) == NULL)
782
		return -1;
L
Linus Torvalds 已提交
783

784
	skb = tcp_make_synack(sk, dst, req, rvp);
L
Linus Torvalds 已提交
785 786

	if (skb) {
787
		__tcp_v4_send_check(skb, ireq->loc_addr, ireq->rmt_addr);
L
Linus Torvalds 已提交
788

789 790 791
		err = ip_build_and_send_pkt(skb, sk, ireq->loc_addr,
					    ireq->rmt_addr,
					    ireq->opt);
792
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
793 794 795 796 797 798
	}

	dst_release(dst);
	return err;
}

799
static int tcp_v4_rtx_synack(struct sock *sk, struct request_sock *req,
800
			      struct request_values *rvp)
801
{
802 803
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
	return tcp_v4_send_synack(sk, NULL, req, rvp);
804 805
}

L
Linus Torvalds 已提交
806
/*
807
 *	IPv4 request_sock destructor.
L
Linus Torvalds 已提交
808
 */
809
static void tcp_v4_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
810
{
J
Jesper Juhl 已提交
811
	kfree(inet_rsk(req)->opt);
L
Linus Torvalds 已提交
812 813
}

814 815 816 817 818 819
/*
 * 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
820
{
821 822 823 824 825
	const char *msg = "Dropping request";
	int want_cookie = 0;
	struct listen_sock *lopt;


L
Linus Torvalds 已提交
826

827
#ifdef CONFIG_SYN_COOKIES
828
	if (sysctl_tcp_syncookies) {
829
		msg = "Sending cookies";
830 831 832
		want_cookie = 1;
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDOCOOKIES);
	} else
833
#endif
834 835 836 837 838 839 840 841 842 843
		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;
844
}
845
EXPORT_SYMBOL(tcp_syn_flood_action);
L
Linus Torvalds 已提交
846 847

/*
848
 * Save and compile IPv4 options into the request_sock if needed.
L
Linus Torvalds 已提交
849
 */
850 851
static struct ip_options_rcu *tcp_v4_save_options(struct sock *sk,
						  struct sk_buff *skb)
L
Linus Torvalds 已提交
852
{
853 854
	const struct ip_options *opt = &(IPCB(skb)->opt);
	struct ip_options_rcu *dopt = NULL;
L
Linus Torvalds 已提交
855 856

	if (opt && opt->optlen) {
857 858
		int opt_size = sizeof(*dopt) + opt->optlen;

L
Linus Torvalds 已提交
859 860
		dopt = kmalloc(opt_size, GFP_ATOMIC);
		if (dopt) {
861
			if (ip_options_echo(&dopt->opt, skb)) {
L
Linus Torvalds 已提交
862 863 864 865 866 867 868 869
				kfree(dopt);
				dopt = NULL;
			}
		}
	}
	return dopt;
}

870 871 872 873 874 875 876 877
#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.  */
878 879
static struct tcp_md5sig_key *
			tcp_v4_md5_do_lookup(struct sock *sk, __be32 addr)
880 881 882 883 884 885 886 887
{
	struct tcp_sock *tp = tcp_sk(sk);
	int i;

	if (!tp->md5sig_info || !tp->md5sig_info->entries4)
		return NULL;
	for (i = 0; i < tp->md5sig_info->entries4; i++) {
		if (tp->md5sig_info->keys4[i].addr == addr)
888
			return &tp->md5sig_info->keys4[i].base;
889 890 891 892 893 894 895
	}
	return NULL;
}

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
896
	return tcp_v4_md5_do_lookup(sk, inet_sk(addr_sk)->inet_daddr);
897 898 899
}
EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
900 901
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
902 903 904 905 906 907 908 909 910
{
	return tcp_v4_md5_do_lookup(sk, inet_rsk(req)->rmt_addr);
}

/* This can be called on a newly created socket, from other files */
int tcp_v4_md5_do_add(struct sock *sk, __be32 addr,
		      u8 *newkey, u8 newkeylen)
{
	/* Add Key to the list */
911
	struct tcp_md5sig_key *key;
912 913 914
	struct tcp_sock *tp = tcp_sk(sk);
	struct tcp4_md5sig_key *keys;

915
	key = tcp_v4_md5_do_lookup(sk, addr);
916 917
	if (key) {
		/* Pre-existing entry - just update that one. */
918 919 920
		kfree(key->key);
		key->key = newkey;
		key->keylen = newkeylen;
921
	} else {
922 923
		struct tcp_md5sig_info *md5sig;

924
		if (!tp->md5sig_info) {
925 926
			tp->md5sig_info = kzalloc(sizeof(*tp->md5sig_info),
						  GFP_ATOMIC);
927 928 929 930
			if (!tp->md5sig_info) {
				kfree(newkey);
				return -ENOMEM;
			}
E
Eric Dumazet 已提交
931
			sk_nocaps_add(sk, NETIF_F_GSO_MASK);
932
		}
933 934 935 936

		md5sig = tp->md5sig_info;
		if (md5sig->entries4 == 0 &&
		    tcp_alloc_md5sig_pool(sk) == NULL) {
937 938 939
			kfree(newkey);
			return -ENOMEM;
		}
940 941 942

		if (md5sig->alloced4 == md5sig->entries4) {
			keys = kmalloc((sizeof(*keys) *
943
					(md5sig->entries4 + 1)), GFP_ATOMIC);
944 945
			if (!keys) {
				kfree(newkey);
946 947
				if (md5sig->entries4 == 0)
					tcp_free_md5sig_pool();
948 949 950
				return -ENOMEM;
			}

951 952 953
			if (md5sig->entries4)
				memcpy(keys, md5sig->keys4,
				       sizeof(*keys) * md5sig->entries4);
954 955

			/* Free old key list, and reference new one */
956
			kfree(md5sig->keys4);
957 958
			md5sig->keys4 = keys;
			md5sig->alloced4++;
959
		}
960
		md5sig->entries4++;
961 962 963
		md5sig->keys4[md5sig->entries4 - 1].addr        = addr;
		md5sig->keys4[md5sig->entries4 - 1].base.key    = newkey;
		md5sig->keys4[md5sig->entries4 - 1].base.keylen = newkeylen;
964 965 966 967 968 969 970 971
	}
	return 0;
}
EXPORT_SYMBOL(tcp_v4_md5_do_add);

static int tcp_v4_md5_add_func(struct sock *sk, struct sock *addr_sk,
			       u8 *newkey, u8 newkeylen)
{
E
Eric Dumazet 已提交
972
	return tcp_v4_md5_do_add(sk, inet_sk(addr_sk)->inet_daddr,
973 974 975 976 977 978 979 980 981 982 983
				 newkey, newkeylen);
}

int tcp_v4_md5_do_del(struct sock *sk, __be32 addr)
{
	struct tcp_sock *tp = tcp_sk(sk);
	int i;

	for (i = 0; i < tp->md5sig_info->entries4; i++) {
		if (tp->md5sig_info->keys4[i].addr == addr) {
			/* Free the key */
984
			kfree(tp->md5sig_info->keys4[i].base.key);
985 986 987 988 989
			tp->md5sig_info->entries4--;

			if (tp->md5sig_info->entries4 == 0) {
				kfree(tp->md5sig_info->keys4);
				tp->md5sig_info->keys4 = NULL;
990
				tp->md5sig_info->alloced4 = 0;
991
				tcp_free_md5sig_pool();
992
			} else if (tp->md5sig_info->entries4 != i) {
993
				/* Need to do some manipulation */
994 995 996 997
				memmove(&tp->md5sig_info->keys4[i],
					&tp->md5sig_info->keys4[i+1],
					(tp->md5sig_info->entries4 - i) *
					 sizeof(struct tcp4_md5sig_key));
998 999 1000 1001 1002 1003 1004 1005
			}
			return 0;
		}
	}
	return -ENOENT;
}
EXPORT_SYMBOL(tcp_v4_md5_do_del);

1006
static void tcp_v4_clear_md5_list(struct sock *sk)
1007 1008 1009 1010 1011 1012 1013 1014 1015 1016
{
	struct tcp_sock *tp = tcp_sk(sk);

	/* Free each key, then the set of key keys,
	 * the crypto element, and then decrement our
	 * hold on the last resort crypto.
	 */
	if (tp->md5sig_info->entries4) {
		int i;
		for (i = 0; i < tp->md5sig_info->entries4; i++)
1017
			kfree(tp->md5sig_info->keys4[i].base.key);
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027
		tp->md5sig_info->entries4 = 0;
		tcp_free_md5sig_pool();
	}
	if (tp->md5sig_info->keys4) {
		kfree(tp->md5sig_info->keys4);
		tp->md5sig_info->keys4 = NULL;
		tp->md5sig_info->alloced4  = 0;
	}
}

1028 1029
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
1030 1031 1032 1033 1034 1035 1036 1037
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;
	u8 *newkey;

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

1038
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
		return -EFAULT;

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

	if (!cmd.tcpm_key || !cmd.tcpm_keylen) {
		if (!tcp_sk(sk)->md5sig_info)
			return -ENOENT;
		return tcp_v4_md5_do_del(sk, sin->sin_addr.s_addr);
	}

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

	if (!tcp_sk(sk)->md5sig_info) {
		struct tcp_sock *tp = tcp_sk(sk);
1055
		struct tcp_md5sig_info *p;
1056

1057
		p = kzalloc(sizeof(*p), sk->sk_allocation);
1058 1059 1060 1061
		if (!p)
			return -EINVAL;

		tp->md5sig_info = p;
E
Eric Dumazet 已提交
1062
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
1063 1064
	}

1065
	newkey = kmemdup(cmd.tcpm_key, cmd.tcpm_keylen, sk->sk_allocation);
1066 1067 1068 1069 1070 1071
	if (!newkey)
		return -ENOMEM;
	return tcp_v4_md5_do_add(sk, sin->sin_addr.s_addr,
				 newkey, cmd.tcpm_keylen);
}

1072 1073
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1074 1075
{
	struct tcp4_pseudohdr *bp;
1076
	struct scatterlist sg;
1077 1078 1079 1080

	bp = &hp->md5_blk.ip4;

	/*
1081
	 * 1. the TCP pseudo-header (in the order: source IP address,
1082 1083 1084 1085 1086 1087
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1088
	bp->protocol = IPPROTO_TCP;
1089
	bp->len = cpu_to_be16(nbytes);
1090

1091 1092 1093 1094 1095
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}

static int tcp_v4_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1096
			       __be32 daddr, __be32 saddr, const struct tcphdr *th)
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
{
	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))
1115 1116 1117 1118
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1119

1120 1121 1122 1123
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1124
	return 1;
1125 1126
}

1127
int tcp_v4_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1128 1129
			const struct sock *sk, const struct request_sock *req,
			const struct sk_buff *skb)
1130
{
1131 1132
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
E
Eric Dumazet 已提交
1133
	const struct tcphdr *th = tcp_hdr(skb);
1134 1135 1136
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1137 1138
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1139 1140 1141
	} else if (req) {
		saddr = inet_rsk(req)->loc_addr;
		daddr = inet_rsk(req)->rmt_addr;
1142
	} else {
1143 1144 1145
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
1146
	}
1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174

	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;
1175
}
1176
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1177

E
Eric Dumazet 已提交
1178
static int tcp_v4_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
1179 1180 1181 1182 1183 1184 1185 1186 1187
{
	/*
	 * 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.
	 */
1188
	const __u8 *hash_location = NULL;
1189
	struct tcp_md5sig_key *hash_expected;
1190
	const struct iphdr *iph = ip_hdr(skb);
1191
	const struct tcphdr *th = tcp_hdr(skb);
1192 1193 1194 1195
	int genhash;
	unsigned char newhash[16];

	hash_expected = tcp_v4_md5_do_lookup(sk, iph->saddr);
1196
	hash_location = tcp_parse_md5sig_option(th);
1197 1198 1199 1200 1201 1202

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

	if (hash_expected && !hash_location) {
1203
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
1204 1205 1206 1207
		return 1;
	}

	if (!hash_expected && hash_location) {
1208
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
1209 1210 1211 1212 1213 1214
		return 1;
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1215 1216 1217
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1218 1219 1220

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
		if (net_ratelimit()) {
1221 1222 1223
			printk(KERN_INFO "MD5 Hash failed for (%pI4, %d)->(%pI4, %d)%s\n",
			       &iph->saddr, ntohs(th->source),
			       &iph->daddr, ntohs(th->dest),
1224 1225 1226 1227 1228 1229 1230 1231 1232
			       genhash ? " tcp_v4_calc_md5_hash failed" : "");
		}
		return 1;
	}
	return 0;
}

#endif

1233
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1234
	.family		=	PF_INET,
1235
	.obj_size	=	sizeof(struct tcp_request_sock),
1236
	.rtx_syn_ack	=	tcp_v4_rtx_synack,
1237 1238
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1239
	.send_reset	=	tcp_v4_send_reset,
1240
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
1241 1242
};

1243
#ifdef CONFIG_TCP_MD5SIG
1244
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1245
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1246
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1247
};
1248
#endif
1249

L
Linus Torvalds 已提交
1250 1251
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
1252
	struct tcp_extend_values tmp_ext;
L
Linus Torvalds 已提交
1253
	struct tcp_options_received tmp_opt;
1254
	const u8 *hash_location;
1255
	struct request_sock *req;
1256
	struct inet_request_sock *ireq;
1257
	struct tcp_sock *tp = tcp_sk(sk);
1258
	struct dst_entry *dst = NULL;
1259 1260
	__be32 saddr = ip_hdr(skb)->saddr;
	__be32 daddr = ip_hdr(skb)->daddr;
L
Linus Torvalds 已提交
1261 1262 1263 1264
	__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
1265
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1266 1267 1268 1269 1270 1271
		goto drop;

	/* TW buckets are converted to open requests without
	 * limitations, they conserve resources and peer is
	 * evidently real one.
	 */
1272
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
1273 1274 1275
		want_cookie = tcp_syn_flood_action(sk, skb, "TCP");
		if (!want_cookie)
			goto drop;
L
Linus Torvalds 已提交
1276 1277 1278 1279 1280 1281 1282
	}

	/* 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.
	 */
1283
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1284 1285
		goto drop;

1286
	req = inet_reqsk_alloc(&tcp_request_sock_ops);
L
Linus Torvalds 已提交
1287 1288 1289
	if (!req)
		goto drop;

1290 1291 1292 1293
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv4_ops;
#endif

L
Linus Torvalds 已提交
1294
	tcp_clear_options(&tmp_opt);
1295
	tmp_opt.mss_clamp = TCP_MSS_DEFAULT;
1296
	tmp_opt.user_mss  = tp->rx_opt.user_mss;
1297
	tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312

	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 */
1313 1314
		*mess++ ^= (__force u32)daddr;
		*mess++ ^= (__force u32)saddr;
L
Linus Torvalds 已提交
1315

1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331
		/* 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
1332

1333
	if (want_cookie && !tmp_opt.saw_tstamp)
L
Linus Torvalds 已提交
1334 1335 1336 1337 1338
		tcp_clear_options(&tmp_opt);

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

1339 1340 1341 1342 1343 1344
	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);

1345
	if (security_inet_conn_request(sk, skb, req))
1346
		goto drop_and_free;
1347

1348
	if (!want_cookie || tmp_opt.tstamp_ok)
1349
		TCP_ECN_create_request(req, tcp_hdr(skb));
L
Linus Torvalds 已提交
1350 1351 1352

	if (want_cookie) {
		isn = cookie_v4_init_sequence(sk, skb, &req->mss);
1353
		req->cookie_ts = tmp_opt.tstamp_ok;
L
Linus Torvalds 已提交
1354 1355
	} else if (!isn) {
		struct inet_peer *peer = NULL;
1356
		struct flowi4 fl4;
L
Linus Torvalds 已提交
1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367

		/* 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 &&
1368
		    tcp_death_row.sysctl_tw_recycle &&
1369
		    (dst = inet_csk_route_req(sk, &fl4, req)) != NULL &&
1370 1371
		    fl4.daddr == saddr &&
		    (peer = rt_get_peer((struct rtable *)dst, fl4.daddr)) != NULL) {
1372
			inet_peer_refcheck(peer);
E
Eric Dumazet 已提交
1373
			if ((u32)get_seconds() - peer->tcp_ts_stamp < TCP_PAWS_MSL &&
L
Linus Torvalds 已提交
1374 1375
			    (s32)(peer->tcp_ts - req->ts_recent) >
							TCP_PAWS_WINDOW) {
1376
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1377
				goto drop_and_release;
L
Linus Torvalds 已提交
1378 1379 1380 1381
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
1382
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
L
Linus Torvalds 已提交
1383 1384 1385 1386 1387 1388 1389 1390 1391 1392
			  (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.
			 */
1393 1394
			LIMIT_NETDEBUG(KERN_DEBUG "TCP: drop open request from %pI4/%u\n",
				       &saddr, ntohs(tcp_hdr(skb)->source));
1395
			goto drop_and_release;
L
Linus Torvalds 已提交
1396 1397
		}

1398
		isn = tcp_v4_init_sequence(skb);
L
Linus Torvalds 已提交
1399
	}
1400
	tcp_rsk(req)->snt_isn = isn;
1401
	tcp_rsk(req)->snt_synack = tcp_time_stamp;
L
Linus Torvalds 已提交
1402

1403 1404
	if (tcp_v4_send_synack(sk, dst, req,
			       (struct request_values *)&tmp_ext) ||
1405
	    want_cookie)
L
Linus Torvalds 已提交
1406 1407
		goto drop_and_free;

1408
	inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
L
Linus Torvalds 已提交
1409 1410
	return 0;

1411 1412
drop_and_release:
	dst_release(dst);
L
Linus Torvalds 已提交
1413
drop_and_free:
1414
	reqsk_free(req);
L
Linus Torvalds 已提交
1415 1416 1417
drop:
	return 0;
}
E
Eric Dumazet 已提交
1418
EXPORT_SYMBOL(tcp_v4_conn_request);
L
Linus Torvalds 已提交
1419 1420 1421 1422 1423 1424 1425


/*
 * 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,
1426
				  struct request_sock *req,
L
Linus Torvalds 已提交
1427 1428
				  struct dst_entry *dst)
{
1429
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1430 1431 1432
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1433 1434 1435
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
1436
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
1437 1438 1439 1440 1441 1442

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (!newsk)
1443
		goto exit_nonewsk;
L
Linus Torvalds 已提交
1444

1445
	newsk->sk_gso_type = SKB_GSO_TCPV4;
L
Linus Torvalds 已提交
1446 1447 1448

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1449
	ireq		      = inet_rsk(req);
E
Eric Dumazet 已提交
1450 1451 1452
	newinet->inet_daddr   = ireq->rmt_addr;
	newinet->inet_rcv_saddr = ireq->loc_addr;
	newinet->inet_saddr	      = ireq->loc_addr;
1453 1454
	inet_opt	      = ireq->opt;
	rcu_assign_pointer(newinet->inet_opt, inet_opt);
1455
	ireq->opt	      = NULL;
1456
	newinet->mc_index     = inet_iif(skb);
1457
	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
1458
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
1459 1460
	if (inet_opt)
		inet_csk(newsk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
E
Eric Dumazet 已提交
1461
	newinet->inet_id = newtp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
1462

1463 1464 1465 1466 1467
	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
1468
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1469
	tcp_sync_mss(newsk, dst_mtu(dst));
1470
	newtp->advmss = dst_metric_advmss(dst);
1471 1472 1473 1474
	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
1475
	tcp_initialize_rcv_mss(newsk);
1476 1477 1478 1479
	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
1480

1481 1482
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1483 1484
	key = tcp_v4_md5_do_lookup(sk, newinet->inet_daddr);
	if (key != NULL) {
1485 1486 1487 1488 1489 1490
		/*
		 * 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.
		 */
1491 1492
		char *newkey = kmemdup(key->key, key->keylen, GFP_ATOMIC);
		if (newkey != NULL)
E
Eric Dumazet 已提交
1493
			tcp_v4_md5_do_add(newsk, newinet->inet_daddr,
1494
					  newkey, key->keylen);
E
Eric Dumazet 已提交
1495
		sk_nocaps_add(newsk, NETIF_F_GSO_MASK);
1496 1497 1498
	}
#endif

1499 1500
	if (__inet_inherit_port(sk, newsk) < 0)
		goto put_and_exit;
1501
	__inet_hash_nolisten(newsk, NULL);
L
Linus Torvalds 已提交
1502 1503 1504 1505

	return newsk;

exit_overflow:
1506
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1507 1508
exit_nonewsk:
	dst_release(dst);
L
Linus Torvalds 已提交
1509
exit:
1510
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1511
	return NULL;
1512 1513 1514
put_and_exit:
	sock_put(newsk);
	goto exit;
L
Linus Torvalds 已提交
1515
}
E
Eric Dumazet 已提交
1516
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
L
Linus Torvalds 已提交
1517 1518 1519

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1520
	struct tcphdr *th = tcp_hdr(skb);
1521
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1522
	struct sock *nsk;
1523
	struct request_sock **prev;
L
Linus Torvalds 已提交
1524
	/* Find possible connection requests. */
1525 1526
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1527 1528 1529
	if (req)
		return tcp_check_req(sk, skb, req, prev);

1530
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1531
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1532 1533 1534 1535 1536 1537

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1538
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1539 1540 1541 1542
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
1543
	if (!th->syn)
L
Linus Torvalds 已提交
1544 1545 1546 1547 1548
		sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
#endif
	return sk;
}

1549
static __sum16 tcp_v4_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1550
{
1551 1552
	const struct iphdr *iph = ip_hdr(skb);

1553
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
1554 1555
		if (!tcp_v4_check(skb->len, iph->saddr,
				  iph->daddr, skb->csum)) {
1556
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1557
			return 0;
1558
		}
L
Linus Torvalds 已提交
1559
	}
1560

1561
	skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
1562 1563
				       skb->len, IPPROTO_TCP, 0);

L
Linus Torvalds 已提交
1564
	if (skb->len <= 76) {
1565
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580
	}
	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)
{
1581 1582 1583 1584 1585 1586 1587 1588
	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
	 */
1589
	if (tcp_v4_inbound_md5_hash(sk, skb))
1590 1591 1592
		goto discard;
#endif

L
Linus Torvalds 已提交
1593
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1594
		sock_rps_save_rxhash(sk, skb);
1595
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
1596
			rsk = sk;
L
Linus Torvalds 已提交
1597
			goto reset;
1598
		}
L
Linus Torvalds 已提交
1599 1600 1601
		return 0;
	}

1602
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1603 1604 1605 1606 1607 1608 1609 1610
		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) {
1611
			sock_rps_save_rxhash(nsk, skb);
1612 1613
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1614
				goto reset;
1615
			}
L
Linus Torvalds 已提交
1616 1617
			return 0;
		}
1618
	} else
1619
		sock_rps_save_rxhash(sk, skb);
1620

1621
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1622
		rsk = sk;
L
Linus Torvalds 已提交
1623
		goto reset;
1624
	}
L
Linus Torvalds 已提交
1625 1626 1627
	return 0;

reset:
1628
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1629 1630 1631 1632 1633 1634 1635 1636 1637 1638
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:
1639
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1640 1641
	goto discard;
}
E
Eric Dumazet 已提交
1642
EXPORT_SYMBOL(tcp_v4_do_rcv);
L
Linus Torvalds 已提交
1643 1644 1645 1646 1647 1648 1649

/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1650
	const struct iphdr *iph;
1651
	const struct tcphdr *th;
L
Linus Torvalds 已提交
1652 1653
	struct sock *sk;
	int ret;
1654
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1655 1656 1657 1658 1659

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

	/* Count it even if it's bad */
1660
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1661 1662 1663 1664

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

1665
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1666 1667 1668 1669 1670 1671 1672 1673

	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
1674
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1675
	 * So, we defer the checks. */
1676
	if (!skb_csum_unnecessary(skb) && tcp_v4_checksum_init(skb))
L
Linus Torvalds 已提交
1677 1678
		goto bad_packet;

1679
	th = tcp_hdr(skb);
1680
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1681 1682 1683 1684 1685
	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;
1686
	TCP_SKB_CB(skb)->ip_dsfield = ipv4_get_dsfield(iph);
L
Linus Torvalds 已提交
1687 1688
	TCP_SKB_CB(skb)->sacked	 = 0;

1689
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1690 1691 1692
	if (!sk)
		goto no_tcp_socket;

E
Eric Dumazet 已提交
1693 1694 1695 1696
process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1697 1698
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1699
		goto discard_and_relse;
1700
	}
1701

L
Linus Torvalds 已提交
1702 1703
	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1704
	nf_reset(skb);
L
Linus Torvalds 已提交
1705

1706
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1707 1708 1709 1710
		goto discard_and_relse;

	skb->dev = NULL;

1711
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1712 1713
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1714 1715 1716
#ifdef CONFIG_NET_DMA
		struct tcp_sock *tp = tcp_sk(sk);
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1717
			tp->ucopy.dma_chan = dma_find_channel(DMA_MEMCPY);
C
Chris Leech 已提交
1718
		if (tp->ucopy.dma_chan)
L
Linus Torvalds 已提交
1719
			ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1720 1721 1722 1723
		else
#endif
		{
			if (!tcp_prequeue(sk, skb))
S
Shan Wei 已提交
1724
				ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1725
		}
1726
	} else if (unlikely(sk_add_backlog(sk, skb))) {
Z
Zhu Yi 已提交
1727
		bh_unlock_sock(sk);
1728
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1729 1730
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
	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:
1743
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1744
	} else {
1745
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1746 1747 1748 1749 1750
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
1751
	return 0;
L
Linus Torvalds 已提交
1752 1753 1754 1755 1756 1757 1758

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1759
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1760 1761 1762 1763
		goto discard_it;
	}

	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
1764
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1765
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1766 1767
		goto discard_it;
	}
1768
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1769
	case TCP_TW_SYN: {
1770
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
1771
							&tcp_hashinfo,
1772
							iph->daddr, th->dest,
1773
							inet_iif(skb));
L
Linus Torvalds 已提交
1774
		if (sk2) {
1775 1776
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791
			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;
}

1792
struct inet_peer *tcp_v4_get_peer(struct sock *sk, bool *release_it)
L
Linus Torvalds 已提交
1793
{
1794
	struct rtable *rt = (struct rtable *) __sk_dst_get(sk);
L
Linus Torvalds 已提交
1795
	struct inet_sock *inet = inet_sk(sk);
1796
	struct inet_peer *peer;
L
Linus Torvalds 已提交
1797

1798 1799
	if (!rt ||
	    inet->cork.fl.u.ip4.daddr != inet->inet_daddr) {
1800
		peer = inet_getpeer_v4(inet->inet_daddr, 1);
1801
		*release_it = true;
L
Linus Torvalds 已提交
1802 1803
	} else {
		if (!rt->peer)
1804
			rt_bind_peer(rt, inet->inet_daddr, 1);
L
Linus Torvalds 已提交
1805
		peer = rt->peer;
1806
		*release_it = false;
L
Linus Torvalds 已提交
1807 1808
	}

1809
	return peer;
L
Linus Torvalds 已提交
1810
}
1811
EXPORT_SYMBOL(tcp_v4_get_peer);
L
Linus Torvalds 已提交
1812

1813
void *tcp_v4_tw_get_peer(struct sock *sk)
L
Linus Torvalds 已提交
1814
{
1815
	const struct inet_timewait_sock *tw = inet_twsk(sk);
L
Linus Torvalds 已提交
1816

1817
	return inet_getpeer_v4(tw->tw_daddr, 1);
L
Linus Torvalds 已提交
1818
}
1819 1820 1821 1822 1823 1824 1825 1826
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
1827

1828
const struct inet_connection_sock_af_ops ipv4_specific = {
1829 1830 1831 1832 1833
	.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,
1834
	.get_peer	   = tcp_v4_get_peer,
1835 1836 1837 1838 1839
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ip_setsockopt,
	.getsockopt	   = ip_getsockopt,
	.addr2sockaddr	   = inet_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in),
1840
	.bind_conflict	   = inet_csk_bind_conflict,
1841
#ifdef CONFIG_COMPAT
1842 1843
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
1844
#endif
L
Linus Torvalds 已提交
1845
};
E
Eric Dumazet 已提交
1846
EXPORT_SYMBOL(ipv4_specific);
L
Linus Torvalds 已提交
1847

1848
#ifdef CONFIG_TCP_MD5SIG
1849
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
1850
	.md5_lookup		= tcp_v4_md5_lookup,
1851
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
1852 1853 1854
	.md5_add		= tcp_v4_md5_add_func,
	.md5_parse		= tcp_v4_parse_md5_keys,
};
1855
#endif
1856

L
Linus Torvalds 已提交
1857 1858 1859 1860 1861
/* 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)
{
1862
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1863 1864 1865 1866 1867 1868
	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);

1869
	icsk->icsk_rto = TCP_TIMEOUT_INIT;
L
Linus Torvalds 已提交
1870 1871 1872 1873 1874 1875 1876
	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
	 */
1877
	tp->snd_cwnd = TCP_INIT_CWND;
L
Linus Torvalds 已提交
1878 1879 1880 1881

	/* See draft-stevens-tcpca-spec-01 for discussion of the
	 * initialization of these values.
	 */
I
Ilpo Järvinen 已提交
1882
	tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
L
Linus Torvalds 已提交
1883
	tp->snd_cwnd_clamp = ~0;
1884
	tp->mss_cache = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
1885 1886

	tp->reordering = sysctl_tcp_reordering;
1887
	icsk->icsk_ca_ops = &tcp_init_congestion_ops;
L
Linus Torvalds 已提交
1888 1889 1890 1891 1892 1893

	sk->sk_state = TCP_CLOSE;

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

1894
	icsk->icsk_af_ops = &ipv4_specific;
1895
	icsk->icsk_sync_mss = tcp_sync_mss;
1896 1897 1898
#ifdef CONFIG_TCP_MD5SIG
	tp->af_specific = &tcp_sock_ipv4_specific;
#endif
L
Linus Torvalds 已提交
1899

1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912
	/* 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
1913 1914 1915
	sk->sk_sndbuf = sysctl_tcp_wmem[1];
	sk->sk_rcvbuf = sysctl_tcp_rmem[1];

H
Herbert Xu 已提交
1916
	local_bh_disable();
1917
	percpu_counter_inc(&tcp_sockets_allocated);
H
Herbert Xu 已提交
1918
	local_bh_enable();
L
Linus Torvalds 已提交
1919 1920 1921 1922

	return 0;
}

1923
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1924 1925 1926 1927 1928
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

1929
	tcp_cleanup_congestion_control(sk);
1930

L
Linus Torvalds 已提交
1931
	/* Cleanup up the write buffer. */
1932
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
1933 1934

	/* Cleans up our, hopefully empty, out_of_order_queue. */
1935
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
1936

1937 1938 1939 1940 1941 1942 1943 1944 1945
#ifdef CONFIG_TCP_MD5SIG
	/* Clean up the MD5 key list, if any */
	if (tp->md5sig_info) {
		tcp_v4_clear_md5_list(sk);
		kfree(tp->md5sig_info);
		tp->md5sig_info = NULL;
	}
#endif

C
Chris Leech 已提交
1946 1947
#ifdef CONFIG_NET_DMA
	/* Cleans up our sk_async_wait_queue */
1948
	__skb_queue_purge(&sk->sk_async_wait_queue);
C
Chris Leech 已提交
1949 1950
#endif

L
Linus Torvalds 已提交
1951 1952 1953 1954
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
1955
	if (inet_csk(sk)->icsk_bind_hash)
1956
		inet_put_port(sk);
L
Linus Torvalds 已提交
1957 1958 1959 1960 1961 1962 1963 1964 1965

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

1966 1967 1968 1969 1970 1971 1972
	/* TCP Cookie Transactions */
	if (tp->cookie_values != NULL) {
		kref_put(&tp->cookie_values->kref,
			 tcp_cookie_values_release);
		tp->cookie_values = NULL;
	}

1973
	percpu_counter_dec(&tcp_sockets_allocated);
L
Linus Torvalds 已提交
1974 1975 1976 1977 1978 1979
}
EXPORT_SYMBOL(tcp_v4_destroy_sock);

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

1980
static inline struct inet_timewait_sock *tw_head(struct hlist_nulls_head *head)
L
Linus Torvalds 已提交
1981
{
1982
	return hlist_nulls_empty(head) ? NULL :
1983
		list_entry(head->first, struct inet_timewait_sock, tw_node);
L
Linus Torvalds 已提交
1984 1985
}

1986
static inline struct inet_timewait_sock *tw_next(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
1987
{
1988 1989
	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
1990 1991
}

1992 1993 1994 1995 1996
/*
 * 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
1997 1998
static void *listening_get_next(struct seq_file *seq, void *cur)
{
1999
	struct inet_connection_sock *icsk;
2000
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
2001
	struct sock *sk = cur;
2002
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
2003
	struct tcp_iter_state *st = seq->private;
2004
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2005 2006

	if (!sk) {
2007
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
2008
		spin_lock_bh(&ilb->lock);
2009
		sk = sk_nulls_head(&ilb->head);
2010
		st->offset = 0;
L
Linus Torvalds 已提交
2011 2012
		goto get_sk;
	}
2013
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
2014
	++st->num;
2015
	++st->offset;
L
Linus Torvalds 已提交
2016 2017

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
2018
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
2019

2020
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
2021 2022 2023
		req = req->dl_next;
		while (1) {
			while (req) {
2024
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
2025 2026 2027 2028 2029
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
2030
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
2031 2032
				break;
get_req:
2033
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
2034
		}
E
Eric Dumazet 已提交
2035
		sk	  = sk_nulls_next(st->syn_wait_sk);
L
Linus Torvalds 已提交
2036
		st->state = TCP_SEQ_STATE_LISTENING;
2037
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2038
	} else {
2039
		icsk = inet_csk(sk);
2040 2041
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
2042
			goto start_req;
2043
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
E
Eric Dumazet 已提交
2044
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2045 2046
	}
get_sk:
2047
	sk_nulls_for_each_from(sk, node) {
2048 2049 2050
		if (!net_eq(sock_net(sk), net))
			continue;
		if (sk->sk_family == st->family) {
L
Linus Torvalds 已提交
2051 2052 2053
			cur = sk;
			goto out;
		}
2054
		icsk = inet_csk(sk);
2055 2056
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
2057 2058 2059 2060 2061 2062 2063
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;
		}
2064
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2065
	}
2066
	spin_unlock_bh(&ilb->lock);
2067
	st->offset = 0;
2068
	if (++st->bucket < INET_LHTABLE_SIZE) {
2069 2070
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
2071
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
2072 2073 2074 2075 2076 2077 2078 2079 2080
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
2081 2082 2083 2084 2085 2086
	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
2087 2088 2089 2090 2091 2092 2093 2094

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

2095 2096
static inline int empty_bucket(struct tcp_iter_state *st)
{
2097 2098
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain) &&
		hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].twchain);
2099 2100
}

2101 2102 2103 2104
/*
 * 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
2105 2106
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
2107
	struct tcp_iter_state *st = seq->private;
2108
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2109 2110
	void *rc = NULL;

2111 2112
	st->offset = 0;
	for (; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
2113
		struct sock *sk;
2114
		struct hlist_nulls_node *node;
2115
		struct inet_timewait_sock *tw;
2116
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
2117

2118 2119 2120 2121
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

2122
		spin_lock_bh(lock);
2123
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
2124
			if (sk->sk_family != st->family ||
2125
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2126 2127 2128 2129 2130 2131
				continue;
			}
			rc = sk;
			goto out;
		}
		st->state = TCP_SEQ_STATE_TIME_WAIT;
2132
		inet_twsk_for_each(tw, node,
2133
				   &tcp_hashinfo.ehash[st->bucket].twchain) {
2134
			if (tw->tw_family != st->family ||
2135
			    !net_eq(twsk_net(tw), net)) {
L
Linus Torvalds 已提交
2136 2137 2138 2139 2140
				continue;
			}
			rc = tw;
			goto out;
		}
2141
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2142 2143 2144 2145 2146 2147 2148 2149 2150
		st->state = TCP_SEQ_STATE_ESTABLISHED;
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2151
	struct inet_timewait_sock *tw;
2152
	struct hlist_nulls_node *node;
J
Jianjun Kong 已提交
2153
	struct tcp_iter_state *st = seq->private;
2154
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2155 2156

	++st->num;
2157
	++st->offset;
L
Linus Torvalds 已提交
2158 2159 2160 2161 2162

	if (st->state == TCP_SEQ_STATE_TIME_WAIT) {
		tw = cur;
		tw = tw_next(tw);
get_tw:
2163
		while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
L
Linus Torvalds 已提交
2164 2165 2166 2167 2168 2169
			tw = tw_next(tw);
		}
		if (tw) {
			cur = tw;
			goto out;
		}
2170
		spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2171 2172
		st->state = TCP_SEQ_STATE_ESTABLISHED;

2173
		/* Look for next non empty bucket */
2174
		st->offset = 0;
2175
		while (++st->bucket <= tcp_hashinfo.ehash_mask &&
2176 2177
				empty_bucket(st))
			;
2178
		if (st->bucket > tcp_hashinfo.ehash_mask)
2179 2180
			return NULL;

2181
		spin_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
2182
		sk = sk_nulls_head(&tcp_hashinfo.ehash[st->bucket].chain);
L
Linus Torvalds 已提交
2183
	} else
2184
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2185

2186
	sk_nulls_for_each_from(sk, node) {
2187
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
L
Linus Torvalds 已提交
2188 2189 2190 2191
			goto found;
	}

	st->state = TCP_SEQ_STATE_TIME_WAIT;
2192
	tw = tw_head(&tcp_hashinfo.ehash[st->bucket].twchain);
L
Linus Torvalds 已提交
2193 2194 2195 2196 2197 2198 2199 2200 2201
	goto get_tw;
found:
	cur = sk;
out:
	return cur;
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
2202 2203 2204 2205 2206
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	rc = established_get_first(seq);
L
Linus Torvalds 已提交
2207 2208 2209 2210

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2211
	}
L
Linus Torvalds 已提交
2212 2213 2214 2215 2216 2217
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2218
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230

	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;
}

2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265
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
2266 2267
static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2268
	struct tcp_iter_state *st = seq->private;
2269 2270 2271 2272 2273 2274 2275 2276
	void *rc;

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

L
Linus Torvalds 已提交
2277 2278
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
2279 2280 2281 2282 2283 2284 2285
	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
2286 2287 2288 2289
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2290
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303
	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;
2304 2305
			st->bucket = 0;
			st->offset = 0;
L
Linus Torvalds 已提交
2306 2307 2308 2309 2310 2311 2312 2313 2314 2315
			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;
2316
	st->last_pos = *pos;
L
Linus Torvalds 已提交
2317 2318 2319 2320 2321
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2322
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2323 2324 2325 2326

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2327 2328
			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
2329 2330 2331
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2332
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2333 2334 2335 2336
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2337
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2338 2339 2340 2341 2342 2343 2344 2345
		break;
	}
}

static int tcp_seq_open(struct inode *inode, struct file *file)
{
	struct tcp_seq_afinfo *afinfo = PDE(inode)->data;
	struct tcp_iter_state *s;
2346
	int err;
L
Linus Torvalds 已提交
2347

2348 2349 2350 2351
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2352

2353
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2354
	s->family		= afinfo->family;
2355
	s->last_pos 		= 0;
2356 2357 2358
	return 0;
}

2359
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2360 2361 2362 2363
{
	int rc = 0;
	struct proc_dir_entry *p;

2364 2365 2366 2367
	afinfo->seq_fops.open		= tcp_seq_open;
	afinfo->seq_fops.read		= seq_read;
	afinfo->seq_fops.llseek		= seq_lseek;
	afinfo->seq_fops.release	= seq_release_net;
2368

2369 2370 2371 2372
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2373 2374 2375
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
			     &afinfo->seq_fops, afinfo);
	if (!p)
L
Linus Torvalds 已提交
2376 2377 2378
		rc = -ENOMEM;
	return rc;
}
E
Eric Dumazet 已提交
2379
EXPORT_SYMBOL(tcp_proc_register);
L
Linus Torvalds 已提交
2380

2381
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2382
{
2383
	proc_net_remove(net, afinfo->name);
L
Linus Torvalds 已提交
2384
}
E
Eric Dumazet 已提交
2385
EXPORT_SYMBOL(tcp_proc_unregister);
L
Linus Torvalds 已提交
2386

2387
static void get_openreq4(const struct sock *sk, const struct request_sock *req,
2388
			 struct seq_file *f, int i, int uid, int *len)
L
Linus Torvalds 已提交
2389
{
2390
	const struct inet_request_sock *ireq = inet_rsk(req);
L
Linus Torvalds 已提交
2391 2392
	int ttd = req->expires - jiffies;

2393
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2394
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %pK%n",
L
Linus Torvalds 已提交
2395
		i,
2396
		ireq->loc_addr,
E
Eric Dumazet 已提交
2397
		ntohs(inet_sk(sk)->inet_sport),
2398 2399
		ireq->rmt_addr,
		ntohs(ireq->rmt_port),
L
Linus Torvalds 已提交
2400 2401 2402 2403 2404 2405 2406 2407 2408
		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),
2409 2410
		req,
		len);
L
Linus Torvalds 已提交
2411 2412
}

2413
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2414 2415 2416
{
	int timer_active;
	unsigned long timer_expires;
2417
	const struct tcp_sock *tp = tcp_sk(sk);
2418
	const struct inet_connection_sock *icsk = inet_csk(sk);
2419
	const struct inet_sock *inet = inet_sk(sk);
E
Eric Dumazet 已提交
2420 2421 2422 2423
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
2424
	int rx_queue;
L
Linus Torvalds 已提交
2425

2426
	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
2427
		timer_active	= 1;
2428 2429
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2430
		timer_active	= 4;
2431
		timer_expires	= icsk->icsk_timeout;
2432
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2433
		timer_active	= 2;
2434
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2435 2436 2437 2438 2439
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2440 2441 2442 2443 2444 2445 2446 2447
	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);

2448
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
D
Dan Rosenberg 已提交
2449
			"%08X %5d %8d %lu %d %pK %lu %lu %u %u %d%n",
2450
		i, src, srcp, dest, destp, sk->sk_state,
2451
		tp->write_seq - tp->snd_una,
2452
		rx_queue,
L
Linus Torvalds 已提交
2453 2454
		timer_active,
		jiffies_to_clock_t(timer_expires - jiffies),
2455
		icsk->icsk_retransmits,
2456
		sock_i_uid(sk),
2457
		icsk->icsk_probes_out,
2458 2459
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2460 2461
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2462
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2463
		tp->snd_cwnd,
I
Ilpo Järvinen 已提交
2464
		tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh,
2465
		len);
L
Linus Torvalds 已提交
2466 2467
}

2468
static void get_timewait4_sock(const struct inet_timewait_sock *tw,
2469
			       struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2470
{
2471
	__be32 dest, src;
L
Linus Torvalds 已提交
2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482
	__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);

2483
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2484
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK%n",
L
Linus Torvalds 已提交
2485 2486
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
		3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
2487
		atomic_read(&tw->tw_refcnt), tw, len);
L
Linus Torvalds 已提交
2488 2489 2490 2491 2492 2493
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2494
	struct tcp_iter_state *st;
2495
	int len;
L
Linus Torvalds 已提交
2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508

	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:
2509
		get_tcp4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2510 2511
		break;
	case TCP_SEQ_STATE_OPENREQ:
2512
		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid, &len);
L
Linus Torvalds 已提交
2513 2514
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
2515
		get_timewait4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2516 2517
		break;
	}
2518
	seq_printf(seq, "%*s\n", TMPSZ - 1 - len, "");
L
Linus Torvalds 已提交
2519 2520 2521 2522 2523 2524 2525
out:
	return 0;
}

static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2526 2527 2528
	.seq_fops	= {
		.owner		= THIS_MODULE,
	},
2529 2530 2531
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2532 2533
};

2534
static int __net_init tcp4_proc_init_net(struct net *net)
2535 2536 2537 2538
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

2539
static void __net_exit tcp4_proc_exit_net(struct net *net)
2540 2541 2542 2543 2544 2545 2546 2547 2548
{
	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
2549 2550
int __init tcp4_proc_init(void)
{
2551
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2552 2553 2554 2555
}

void tcp4_proc_exit(void)
{
2556
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2557 2558 2559
}
#endif /* CONFIG_PROC_FS */

H
Herbert Xu 已提交
2560 2561
struct sk_buff **tcp4_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
2562
	const struct iphdr *iph = skb_gro_network_header(skb);
H
Herbert Xu 已提交
2563 2564 2565

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
2566
		if (!tcp_v4_check(skb_gro_len(skb), iph->saddr, iph->daddr,
H
Herbert Xu 已提交
2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582
				  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)
{
2583
	const struct iphdr *iph = ip_hdr(skb);
H
Herbert Xu 已提交
2584 2585 2586 2587 2588 2589 2590 2591 2592
	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
2593 2594 2595 2596 2597 2598
struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2599
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2600 2601 2602 2603 2604 2605 2606
	.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,
2607 2608
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2609
	.backlog_rcv		= tcp_v4_do_rcv,
2610 2611 2612
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2613 2614
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
2615
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2616 2617 2618 2619 2620 2621 2622
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
	.sysctl_mem		= sysctl_tcp_mem,
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp_sock),
2623
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2624
	.twsk_prot		= &tcp_timewait_sock_ops,
2625
	.rsk_prot		= &tcp_request_sock_ops,
2626
	.h.hashinfo		= &tcp_hashinfo,
2627
	.no_autobind		= true,
2628 2629 2630 2631
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
L
Linus Torvalds 已提交
2632
};
E
Eric Dumazet 已提交
2633
EXPORT_SYMBOL(tcp_prot);
L
Linus Torvalds 已提交
2634

2635 2636 2637 2638 2639 2640 2641 2642 2643 2644

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
2645 2646 2647 2648 2649
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2650 2651 2652
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2653 2654 2655
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2656 2657
};

2658
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2659
{
2660
	inet_hashinfo_init(&tcp_hashinfo);
2661
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2662 2663
		panic("Failed to create the TCP control socket.\n");
}