tcp_ipv4.c 66.4 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>
L
Linus Torvalds 已提交
75 76 77 78 79 80 81

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

82 83 84
#include <linux/crypto.h>
#include <linux/scatterlist.h>

85 86
int sysctl_tcp_tw_reuse __read_mostly;
int sysctl_tcp_low_latency __read_mostly;
E
Eric Dumazet 已提交
87
EXPORT_SYMBOL(sysctl_tcp_low_latency);
L
Linus Torvalds 已提交
88 89


90
#ifdef CONFIG_TCP_MD5SIG
91 92
static struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk,
						   __be32 addr);
93 94
static int tcp_v4_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
			       __be32 daddr, __be32 saddr, struct tcphdr *th);
95 96 97 98 99 100
#else
static inline
struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk, __be32 addr)
{
	return NULL;
}
101 102
#endif

103
struct inet_hashinfo tcp_hashinfo;
E
Eric Dumazet 已提交
104
EXPORT_SYMBOL(tcp_hashinfo);
L
Linus Torvalds 已提交
105

106
static inline __u32 tcp_v4_init_sequence(struct sk_buff *skb)
L
Linus Torvalds 已提交
107
{
108 109
	return secure_tcp_sequence_number(ip_hdr(skb)->daddr,
					  ip_hdr(skb)->saddr,
110 111
					  tcp_hdr(skb)->dest,
					  tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
112 113
}

114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
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 &&
132
			     get_seconds() - tcptw->tw_ts_recent_stamp > 1))) {
133 134 135 136 137 138 139 140 141 142 143 144 145
		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
146 147 148
/* This will initiate an outgoing connection. */
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
{
149
	struct sockaddr_in *usin = (struct sockaddr_in *)uaddr;
L
Linus Torvalds 已提交
150 151
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
152
	__be16 orig_sport, orig_dport;
153
	__be32 daddr, nexthop;
154
	struct flowi4 *fl4;
155
	struct rtable *rt;
L
Linus Torvalds 已提交
156
	int err;
157
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
158 159 160 161 162 163 164 165

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

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

	nexthop = daddr = usin->sin_addr.s_addr;
166 167 168
	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
169 170
		if (!daddr)
			return -EINVAL;
171
		nexthop = inet_opt->opt.faddr;
L
Linus Torvalds 已提交
172 173
	}

174 175
	orig_sport = inet->inet_sport;
	orig_dport = usin->sin_port;
176 177
	fl4 = &inet->cork.fl.u.ip4;
	rt = ip_route_connect(fl4, nexthop, inet->inet_saddr,
178 179 180 181 182 183
			      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)
184
			IP_INC_STATS_BH(sock_net(sk), IPSTATS_MIB_OUTNOROUTES);
185
		return err;
186
	}
L
Linus Torvalds 已提交
187 188 189 190 191 192

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

193
	if (!inet_opt || !inet_opt->opt.srr)
194
		daddr = fl4->daddr;
L
Linus Torvalds 已提交
195

E
Eric Dumazet 已提交
196
	if (!inet->inet_saddr)
197
		inet->inet_saddr = fl4->saddr;
E
Eric Dumazet 已提交
198
	inet->inet_rcv_saddr = inet->inet_saddr;
L
Linus Torvalds 已提交
199

E
Eric Dumazet 已提交
200
	if (tp->rx_opt.ts_recent_stamp && inet->inet_daddr != daddr) {
L
Linus Torvalds 已提交
201 202 203 204 205 206
		/* Reset inherited state */
		tp->rx_opt.ts_recent	   = 0;
		tp->rx_opt.ts_recent_stamp = 0;
		tp->write_seq		   = 0;
	}

207
	if (tcp_death_row.sysctl_tw_recycle &&
208
	    !tp->rx_opt.ts_recent_stamp && fl4->daddr == daddr) {
209
		struct inet_peer *peer = rt_get_peer(rt, fl4->daddr);
210 211 212 213 214
		/*
		 * 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
215
		 */
216 217 218 219 220 221
		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
222 223 224
		}
	}

E
Eric Dumazet 已提交
225 226
	inet->inet_dport = usin->sin_port;
	inet->inet_daddr = daddr;
L
Linus Torvalds 已提交
227

228
	inet_csk(sk)->icsk_ext_hdr_len = 0;
229 230
	if (inet_opt)
		inet_csk(sk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
L
Linus Torvalds 已提交
231

232
	tp->rx_opt.mss_clamp = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
233 234 235 236 237 238 239

	/* 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);
240
	err = inet_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
241 242 243
	if (err)
		goto failure;

244
	rt = ip_route_newports(fl4, rt, orig_sport, orig_dport,
245 246 247 248
			       inet->inet_sport, inet->inet_dport, sk);
	if (IS_ERR(rt)) {
		err = PTR_ERR(rt);
		rt = NULL;
L
Linus Torvalds 已提交
249
		goto failure;
250
	}
L
Linus Torvalds 已提交
251
	/* OK, now commit destination to socket.  */
252
	sk->sk_gso_type = SKB_GSO_TCPV4;
253
	sk_setup_caps(sk, &rt->dst);
L
Linus Torvalds 已提交
254 255

	if (!tp->write_seq)
E
Eric Dumazet 已提交
256 257 258
		tp->write_seq = secure_tcp_sequence_number(inet->inet_saddr,
							   inet->inet_daddr,
							   inet->inet_sport,
L
Linus Torvalds 已提交
259 260
							   usin->sin_port);

E
Eric Dumazet 已提交
261
	inet->inet_id = tp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
262 263 264 265 266 267 268 269 270

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

	return 0;

failure:
271 272 273 274
	/*
	 * This unhashes the socket and releases the local port,
	 * if necessary.
	 */
L
Linus Torvalds 已提交
275 276 277
	tcp_set_state(sk, TCP_CLOSE);
	ip_rt_put(rt);
	sk->sk_route_caps = 0;
E
Eric Dumazet 已提交
278
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
279 280
	return err;
}
E
Eric Dumazet 已提交
281
EXPORT_SYMBOL(tcp_v4_connect);
L
Linus Torvalds 已提交
282 283 284 285

/*
 * This routine does path mtu discovery as defined in RFC1191.
 */
286
static void do_pmtu_discovery(struct sock *sk, const struct iphdr *iph, u32 mtu)
L
Linus Torvalds 已提交
287 288 289 290 291 292 293 294 295 296 297 298 299 300
{
	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.
301
	 * There is a small race when the user changes this flag in the
L
Linus Torvalds 已提交
302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317
	 * 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 &&
318
	    inet_csk(sk)->icsk_pmtu_cookie > mtu) {
L
Linus Torvalds 已提交
319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345
		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.
 *
 */

346
void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
L
Linus Torvalds 已提交
347
{
348
	const struct iphdr *iph = (const struct iphdr *)icmp_skb->data;
349
	struct tcphdr *th = (struct tcphdr *)(icmp_skb->data + (iph->ihl << 2));
350
	struct inet_connection_sock *icsk;
L
Linus Torvalds 已提交
351 352
	struct tcp_sock *tp;
	struct inet_sock *inet;
353 354
	const int type = icmp_hdr(icmp_skb)->type;
	const int code = icmp_hdr(icmp_skb)->code;
L
Linus Torvalds 已提交
355
	struct sock *sk;
356
	struct sk_buff *skb;
L
Linus Torvalds 已提交
357
	__u32 seq;
358
	__u32 remaining;
L
Linus Torvalds 已提交
359
	int err;
360
	struct net *net = dev_net(icmp_skb->dev);
L
Linus Torvalds 已提交
361

362
	if (icmp_skb->len < (iph->ihl << 2) + 8) {
363
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
364 365 366
		return;
	}

367
	sk = inet_lookup(net, &tcp_hashinfo, iph->daddr, th->dest,
368
			iph->saddr, th->source, inet_iif(icmp_skb));
L
Linus Torvalds 已提交
369
	if (!sk) {
370
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
371 372 373
		return;
	}
	if (sk->sk_state == TCP_TIME_WAIT) {
374
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
375 376 377 378 379 380 381 382
		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))
383
		NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
L
Linus Torvalds 已提交
384 385 386 387

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

388 389 390 391 392
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto out;
	}

393
	icsk = inet_csk(sk);
L
Linus Torvalds 已提交
394 395 396 397
	tp = tcp_sk(sk);
	seq = ntohl(th->seq);
	if (sk->sk_state != TCP_LISTEN &&
	    !between(seq, tp->snd_una, tp->snd_nxt)) {
398
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419
		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;
420 421 422 423 424 425 426 427
		/* 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;

428 429 430
		if (sock_owned_by_user(sk))
			break;

431
		icsk->icsk_backoff--;
432 433
		inet_csk(sk)->icsk_rto = (tp->srtt ? __tcp_set_rto(tp) :
			TCP_TIMEOUT_INIT) << icsk->icsk_backoff;
434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450
		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
451 452 453 454 455 456 457 458 459
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

	switch (sk->sk_state) {
460
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
461 462 463 464
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

465 466
		req = inet_csk_search_req(sk, &prev, th->dest,
					  iph->daddr, iph->saddr);
L
Linus Torvalds 已提交
467 468 469 470 471 472
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		   an established socket here.
		 */
473
		WARN_ON(req->sk);
L
Linus Torvalds 已提交
474

475
		if (seq != tcp_rsk(req)->snt_isn) {
476
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
477 478 479 480 481 482 483 484 485
			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().
		 */
486
		inet_csk_reqsk_queue_drop(sk, req, prev);
L
Linus Torvalds 已提交
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533
		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);
}

534 535
static void __tcp_v4_send_check(struct sk_buff *skb,
				__be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
536
{
537
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
538

539
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
540
		th->check = ~tcp_v4_check(skb->len, saddr, daddr, 0);
541
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
542
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
543
	} else {
544
		th->check = tcp_v4_check(skb->len, saddr, daddr,
545
					 csum_partial(th,
L
Linus Torvalds 已提交
546 547 548 549 550
						      th->doff << 2,
						      skb->csum));
	}
}

551
/* This routine computes an IPv4 TCP checksum. */
552
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb)
553 554 555 556 557
{
	struct inet_sock *inet = inet_sk(sk);

	__tcp_v4_send_check(skb, inet->inet_saddr, inet->inet_daddr);
}
E
Eric Dumazet 已提交
558
EXPORT_SYMBOL(tcp_v4_send_check);
559

560 561
int tcp_v4_gso_send_check(struct sk_buff *skb)
{
562
	const struct iphdr *iph;
563 564 565 566 567
	struct tcphdr *th;

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

568
	iph = ip_hdr(skb);
569
	th = tcp_hdr(skb);
570 571

	th->check = 0;
572
	skb->ip_summed = CHECKSUM_PARTIAL;
573
	__tcp_v4_send_check(skb, iph->saddr, iph->daddr);
574 575 576
	return 0;
}

L
Linus Torvalds 已提交
577 578 579 580 581 582 583 584 585 586 587 588 589
/*
 *	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.
 */

590
static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
591
{
592
	struct tcphdr *th = tcp_hdr(skb);
593 594 595
	struct {
		struct tcphdr th;
#ifdef CONFIG_TCP_MD5SIG
596
		__be32 opt[(TCPOLEN_MD5SIG_ALIGNED >> 2)];
597 598
#endif
	} rep;
L
Linus Torvalds 已提交
599
	struct ip_reply_arg arg;
600 601 602
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
603
	struct net *net;
L
Linus Torvalds 已提交
604 605 606 607 608

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

E
Eric Dumazet 已提交
609
	if (skb_rtable(skb)->rt_type != RTN_LOCAL)
L
Linus Torvalds 已提交
610 611 612
		return;

	/* Swap the send and the receive. */
613 614 615 616 617
	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
618 619

	if (th->ack) {
620
		rep.th.seq = th->ack_seq;
L
Linus Torvalds 已提交
621
	} else {
622 623 624
		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
625 626
	}

627
	memset(&arg, 0, sizeof(arg));
628 629 630 631
	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);

#ifdef CONFIG_TCP_MD5SIG
632
	key = sk ? tcp_v4_md5_do_lookup(sk, ip_hdr(skb)->daddr) : NULL;
633 634 635 636 637 638 639 640 641
	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;

642
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
643 644
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
645 646
	}
#endif
647 648
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
649
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
650
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
651
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
L
Linus Torvalds 已提交
652

E
Eric Dumazet 已提交
653
	net = dev_net(skb_dst(skb)->dev);
654
	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
655
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
656

657 658
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
L
Linus Torvalds 已提交
659 660 661 662 663 664
}

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

665 666
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
			    u32 win, u32 ts, int oif,
667 668
			    struct tcp_md5sig_key *key,
			    int reply_flags)
L
Linus Torvalds 已提交
669
{
670
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
671 672
	struct {
		struct tcphdr th;
673
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
674
#ifdef CONFIG_TCP_MD5SIG
675
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
676 677
#endif
			];
L
Linus Torvalds 已提交
678 679
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
680
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
681 682

	memset(&rep.th, 0, sizeof(struct tcphdr));
683
	memset(&arg, 0, sizeof(arg));
L
Linus Torvalds 已提交
684 685 686 687

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
	if (ts) {
688 689 690 691 692
		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);
693
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
694 695 696 697 698 699 700 701 702 703 704
	}

	/* 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);

705 706 707 708 709 710 711 712 713 714 715
#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;

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

729
	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
730
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
731

732
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
733 734 735 736
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
737
	struct inet_timewait_sock *tw = inet_twsk(sk);
738
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
739

740
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
741
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
742 743
			tcptw->tw_ts_recent,
			tw->tw_bound_dev_if,
744 745
			tcp_twsk_md5_key(tcptw),
			tw->tw_transparent ? IP_REPLY_ARG_NOSRCCHECK : 0
746
			);
L
Linus Torvalds 已提交
747

748
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
749 750
}

751
static void tcp_v4_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
752
				  struct request_sock *req)
L
Linus Torvalds 已提交
753
{
754
	tcp_v4_send_ack(skb, tcp_rsk(req)->snt_isn + 1,
755
			tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd,
756 757
			req->ts_recent,
			0,
758 759
			tcp_v4_md5_do_lookup(sk, ip_hdr(skb)->daddr),
			inet_rsk(req)->no_srccheck ? IP_REPLY_ARG_NOSRCCHECK : 0);
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
static void syn_flood_warning(const struct sk_buff *skb)
L
Linus Torvalds 已提交
811
{
812
	const char *msg;
L
Linus Torvalds 已提交
813

814 815 816 817
#ifdef CONFIG_SYN_COOKIES
	if (sysctl_tcp_syncookies)
		msg = "Sending cookies";
	else
818
#endif
819 820 821 822 823
		msg = "Dropping request";

	pr_info("TCP: Possible SYN flooding on port %d. %s.\n",
				ntohs(tcp_hdr(skb)->dest), msg);
}
L
Linus Torvalds 已提交
824 825

/*
826
 * Save and compile IPv4 options into the request_sock if needed.
L
Linus Torvalds 已提交
827
 */
828 829
static struct ip_options_rcu *tcp_v4_save_options(struct sock *sk,
						  struct sk_buff *skb)
L
Linus Torvalds 已提交
830
{
831 832
	const struct ip_options *opt = &(IPCB(skb)->opt);
	struct ip_options_rcu *dopt = NULL;
L
Linus Torvalds 已提交
833 834

	if (opt && opt->optlen) {
835 836
		int opt_size = sizeof(*dopt) + opt->optlen;

L
Linus Torvalds 已提交
837 838
		dopt = kmalloc(opt_size, GFP_ATOMIC);
		if (dopt) {
839
			if (ip_options_echo(&dopt->opt, skb)) {
L
Linus Torvalds 已提交
840 841 842 843 844 845 846 847
				kfree(dopt);
				dopt = NULL;
			}
		}
	}
	return dopt;
}

848 849 850 851 852 853 854 855
#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.  */
856 857
static struct tcp_md5sig_key *
			tcp_v4_md5_do_lookup(struct sock *sk, __be32 addr)
858 859 860 861 862 863 864 865
{
	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)
866
			return &tp->md5sig_info->keys4[i].base;
867 868 869 870 871 872 873
	}
	return NULL;
}

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
874
	return tcp_v4_md5_do_lookup(sk, inet_sk(addr_sk)->inet_daddr);
875 876 877
}
EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
878 879
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
880 881 882 883 884 885 886 887 888
{
	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 */
889
	struct tcp_md5sig_key *key;
890 891 892
	struct tcp_sock *tp = tcp_sk(sk);
	struct tcp4_md5sig_key *keys;

893
	key = tcp_v4_md5_do_lookup(sk, addr);
894 895
	if (key) {
		/* Pre-existing entry - just update that one. */
896 897 898
		kfree(key->key);
		key->key = newkey;
		key->keylen = newkeylen;
899
	} else {
900 901
		struct tcp_md5sig_info *md5sig;

902
		if (!tp->md5sig_info) {
903 904
			tp->md5sig_info = kzalloc(sizeof(*tp->md5sig_info),
						  GFP_ATOMIC);
905 906 907 908
			if (!tp->md5sig_info) {
				kfree(newkey);
				return -ENOMEM;
			}
E
Eric Dumazet 已提交
909
			sk_nocaps_add(sk, NETIF_F_GSO_MASK);
910
		}
911
		if (tcp_alloc_md5sig_pool(sk) == NULL) {
912 913 914
			kfree(newkey);
			return -ENOMEM;
		}
915 916 917 918
		md5sig = tp->md5sig_info;

		if (md5sig->alloced4 == md5sig->entries4) {
			keys = kmalloc((sizeof(*keys) *
919
					(md5sig->entries4 + 1)), GFP_ATOMIC);
920 921 922 923 924 925
			if (!keys) {
				kfree(newkey);
				tcp_free_md5sig_pool();
				return -ENOMEM;
			}

926 927 928
			if (md5sig->entries4)
				memcpy(keys, md5sig->keys4,
				       sizeof(*keys) * md5sig->entries4);
929 930

			/* Free old key list, and reference new one */
931
			kfree(md5sig->keys4);
932 933
			md5sig->keys4 = keys;
			md5sig->alloced4++;
934
		}
935
		md5sig->entries4++;
936 937 938
		md5sig->keys4[md5sig->entries4 - 1].addr        = addr;
		md5sig->keys4[md5sig->entries4 - 1].base.key    = newkey;
		md5sig->keys4[md5sig->entries4 - 1].base.keylen = newkeylen;
939 940 941 942 943 944 945 946
	}
	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
947
	return tcp_v4_md5_do_add(sk, inet_sk(addr_sk)->inet_daddr,
948 949 950 951 952 953 954 955 956 957 958
				 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 */
959
			kfree(tp->md5sig_info->keys4[i].base.key);
960 961 962 963 964
			tp->md5sig_info->entries4--;

			if (tp->md5sig_info->entries4 == 0) {
				kfree(tp->md5sig_info->keys4);
				tp->md5sig_info->keys4 = NULL;
965
				tp->md5sig_info->alloced4 = 0;
966
			} else if (tp->md5sig_info->entries4 != i) {
967
				/* Need to do some manipulation */
968 969 970 971
				memmove(&tp->md5sig_info->keys4[i],
					&tp->md5sig_info->keys4[i+1],
					(tp->md5sig_info->entries4 - i) *
					 sizeof(struct tcp4_md5sig_key));
972 973 974 975 976 977 978 979 980
			}
			tcp_free_md5sig_pool();
			return 0;
		}
	}
	return -ENOENT;
}
EXPORT_SYMBOL(tcp_v4_md5_do_del);

981
static void tcp_v4_clear_md5_list(struct sock *sk)
982 983 984 985 986 987 988 989 990 991
{
	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++)
992
			kfree(tp->md5sig_info->keys4[i].base.key);
993 994 995 996 997 998 999 1000 1001 1002
		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;
	}
}

1003 1004
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
1005 1006 1007 1008 1009 1010 1011 1012
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;
	u8 *newkey;

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

1013
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029
		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);
1030
		struct tcp_md5sig_info *p;
1031

1032
		p = kzalloc(sizeof(*p), sk->sk_allocation);
1033 1034 1035 1036
		if (!p)
			return -EINVAL;

		tp->md5sig_info = p;
E
Eric Dumazet 已提交
1037
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
1038 1039
	}

1040
	newkey = kmemdup(cmd.tcpm_key, cmd.tcpm_keylen, sk->sk_allocation);
1041 1042 1043 1044 1045 1046
	if (!newkey)
		return -ENOMEM;
	return tcp_v4_md5_do_add(sk, sin->sin_addr.s_addr,
				 newkey, cmd.tcpm_keylen);
}

1047 1048
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1049 1050
{
	struct tcp4_pseudohdr *bp;
1051
	struct scatterlist sg;
1052 1053 1054 1055

	bp = &hp->md5_blk.ip4;

	/*
1056
	 * 1. the TCP pseudo-header (in the order: source IP address,
1057 1058 1059 1060 1061 1062
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1063
	bp->protocol = IPPROTO_TCP;
1064
	bp->len = cpu_to_be16(nbytes);
1065

1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
	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,
			       __be32 daddr, __be32 saddr, struct tcphdr *th)
{
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;

	hp = tcp_get_md5sig_pool();
	if (!hp)
		goto clear_hash_noput;
	desc = &hp->md5_desc;

	if (crypto_hash_init(desc))
		goto clear_hash;
	if (tcp_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))
1090 1091 1092 1093
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1094

1095 1096 1097 1098
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1099
	return 1;
1100 1101
}

1102 1103 1104
int tcp_v4_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
			struct sock *sk, struct request_sock *req,
			struct sk_buff *skb)
1105
{
1106 1107 1108
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
	struct tcphdr *th = tcp_hdr(skb);
1109 1110 1111
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1112 1113
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1114 1115 1116
	} else if (req) {
		saddr = inet_rsk(req)->loc_addr;
		daddr = inet_rsk(req)->rmt_addr;
1117
	} else {
1118 1119 1120
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
1121
	}
1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149

	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;
1150
}
1151
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1152

1153
static int tcp_v4_inbound_md5_hash(struct sock *sk, struct sk_buff *skb)
1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
{
	/*
	 * 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.
	 */
	__u8 *hash_location = NULL;
	struct tcp_md5sig_key *hash_expected;
1165
	const struct iphdr *iph = ip_hdr(skb);
1166
	struct tcphdr *th = tcp_hdr(skb);
1167 1168 1169 1170
	int genhash;
	unsigned char newhash[16];

	hash_expected = tcp_v4_md5_do_lookup(sk, iph->saddr);
1171
	hash_location = tcp_parse_md5sig_option(th);
1172 1173 1174 1175 1176 1177

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

	if (hash_expected && !hash_location) {
1178
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
1179 1180 1181 1182
		return 1;
	}

	if (!hash_expected && hash_location) {
1183
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
1184 1185 1186 1187 1188 1189
		return 1;
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1190 1191 1192
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1193 1194 1195

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
		if (net_ratelimit()) {
1196 1197 1198
			printk(KERN_INFO "MD5 Hash failed for (%pI4, %d)->(%pI4, %d)%s\n",
			       &iph->saddr, ntohs(th->source),
			       &iph->daddr, ntohs(th->dest),
1199 1200 1201 1202 1203 1204 1205 1206 1207
			       genhash ? " tcp_v4_calc_md5_hash failed" : "");
		}
		return 1;
	}
	return 0;
}

#endif

1208
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1209
	.family		=	PF_INET,
1210
	.obj_size	=	sizeof(struct tcp_request_sock),
1211
	.rtx_syn_ack	=	tcp_v4_rtx_synack,
1212 1213
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1214
	.send_reset	=	tcp_v4_send_reset,
1215
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
1216 1217
};

1218
#ifdef CONFIG_TCP_MD5SIG
1219
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1220
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1221
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1222
};
1223
#endif
1224

L
Linus Torvalds 已提交
1225 1226
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
1227
	struct tcp_extend_values tmp_ext;
L
Linus Torvalds 已提交
1228
	struct tcp_options_received tmp_opt;
1229
	u8 *hash_location;
1230
	struct request_sock *req;
1231
	struct inet_request_sock *ireq;
1232
	struct tcp_sock *tp = tcp_sk(sk);
1233
	struct dst_entry *dst = NULL;
1234 1235
	__be32 saddr = ip_hdr(skb)->saddr;
	__be32 daddr = ip_hdr(skb)->daddr;
L
Linus Torvalds 已提交
1236 1237 1238 1239 1240 1241 1242 1243
	__u32 isn = TCP_SKB_CB(skb)->when;
#ifdef CONFIG_SYN_COOKIES
	int want_cookie = 0;
#else
#define want_cookie 0 /* Argh, why doesn't gcc optimize this :( */
#endif

	/* Never answer to SYNs send to broadcast or multicast */
E
Eric Dumazet 已提交
1244
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1245 1246 1247 1248 1249 1250
		goto drop;

	/* TW buckets are converted to open requests without
	 * limitations, they conserve resources and peer is
	 * evidently real one.
	 */
1251
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
1252 1253
		if (net_ratelimit())
			syn_flood_warning(skb);
L
Linus Torvalds 已提交
1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
#ifdef CONFIG_SYN_COOKIES
		if (sysctl_tcp_syncookies) {
			want_cookie = 1;
		} else
#endif
		goto drop;
	}

	/* 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.
	 */
1267
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1268 1269
		goto drop;

1270
	req = inet_reqsk_alloc(&tcp_request_sock_ops);
L
Linus Torvalds 已提交
1271 1272 1273
	if (!req)
		goto drop;

1274 1275 1276 1277
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv4_ops;
#endif

L
Linus Torvalds 已提交
1278
	tcp_clear_options(&tmp_opt);
1279
	tmp_opt.mss_clamp = TCP_MSS_DEFAULT;
1280
	tmp_opt.user_mss  = tp->rx_opt.user_mss;
1281
	tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296

	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 */
1297 1298
		*mess++ ^= (__force u32)daddr;
		*mess++ ^= (__force u32)saddr;
L
Linus Torvalds 已提交
1299

1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317
		/* plus variable length Initiator Cookie */
		c = (u8 *)mess;
		while (l-- > 0)
			*c++ ^= *hash_location++;

#ifdef CONFIG_SYN_COOKIES
		want_cookie = 0;	/* not our kind of cookie */
#endif
		tmp_ext.cookie_out_never = 0; /* false */
		tmp_ext.cookie_plus = tmp_opt.cookie_plus;
	} else if (!tp->rx_opt.cookie_in_always) {
		/* redundant indications, but ensure initialization. */
		tmp_ext.cookie_out_never = 1; /* true */
		tmp_ext.cookie_plus = 0;
	} else {
		goto drop_and_release;
	}
	tmp_ext.cookie_in_always = tp->rx_opt.cookie_in_always;
L
Linus Torvalds 已提交
1318

1319
	if (want_cookie && !tmp_opt.saw_tstamp)
L
Linus Torvalds 已提交
1320 1321 1322 1323 1324
		tcp_clear_options(&tmp_opt);

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

1325 1326 1327 1328 1329 1330
	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);

1331
	if (security_inet_conn_request(sk, skb, req))
1332
		goto drop_and_free;
1333

1334
	if (!want_cookie || tmp_opt.tstamp_ok)
1335
		TCP_ECN_create_request(req, tcp_hdr(skb));
L
Linus Torvalds 已提交
1336 1337 1338

	if (want_cookie) {
		isn = cookie_v4_init_sequence(sk, skb, &req->mss);
1339
		req->cookie_ts = tmp_opt.tstamp_ok;
L
Linus Torvalds 已提交
1340 1341
	} else if (!isn) {
		struct inet_peer *peer = NULL;
1342
		struct flowi4 fl4;
L
Linus Torvalds 已提交
1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353

		/* 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 &&
1354
		    tcp_death_row.sysctl_tw_recycle &&
1355
		    (dst = inet_csk_route_req(sk, &fl4, req)) != NULL &&
1356 1357
		    fl4.daddr == saddr &&
		    (peer = rt_get_peer((struct rtable *)dst, fl4.daddr)) != NULL) {
1358
			inet_peer_refcheck(peer);
E
Eric Dumazet 已提交
1359
			if ((u32)get_seconds() - peer->tcp_ts_stamp < TCP_PAWS_MSL &&
L
Linus Torvalds 已提交
1360 1361
			    (s32)(peer->tcp_ts - req->ts_recent) >
							TCP_PAWS_WINDOW) {
1362
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1363
				goto drop_and_release;
L
Linus Torvalds 已提交
1364 1365 1366 1367
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
1368
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
L
Linus Torvalds 已提交
1369 1370 1371 1372 1373 1374 1375 1376 1377 1378
			  (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.
			 */
1379 1380
			LIMIT_NETDEBUG(KERN_DEBUG "TCP: drop open request from %pI4/%u\n",
				       &saddr, ntohs(tcp_hdr(skb)->source));
1381
			goto drop_and_release;
L
Linus Torvalds 已提交
1382 1383
		}

1384
		isn = tcp_v4_init_sequence(skb);
L
Linus Torvalds 已提交
1385
	}
1386
	tcp_rsk(req)->snt_isn = isn;
1387
	tcp_rsk(req)->snt_synack = tcp_time_stamp;
L
Linus Torvalds 已提交
1388

1389 1390
	if (tcp_v4_send_synack(sk, dst, req,
			       (struct request_values *)&tmp_ext) ||
1391
	    want_cookie)
L
Linus Torvalds 已提交
1392 1393
		goto drop_and_free;

1394
	inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
L
Linus Torvalds 已提交
1395 1396
	return 0;

1397 1398
drop_and_release:
	dst_release(dst);
L
Linus Torvalds 已提交
1399
drop_and_free:
1400
	reqsk_free(req);
L
Linus Torvalds 已提交
1401 1402 1403
drop:
	return 0;
}
E
Eric Dumazet 已提交
1404
EXPORT_SYMBOL(tcp_v4_conn_request);
L
Linus Torvalds 已提交
1405 1406 1407 1408 1409 1410 1411


/*
 * 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,
1412
				  struct request_sock *req,
L
Linus Torvalds 已提交
1413 1414
				  struct dst_entry *dst)
{
1415
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1416 1417 1418
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1419 1420 1421
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
1422
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
1423 1424 1425 1426 1427 1428

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (!newsk)
1429
		goto exit_nonewsk;
L
Linus Torvalds 已提交
1430

1431
	newsk->sk_gso_type = SKB_GSO_TCPV4;
L
Linus Torvalds 已提交
1432 1433 1434

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1435
	ireq		      = inet_rsk(req);
E
Eric Dumazet 已提交
1436 1437 1438
	newinet->inet_daddr   = ireq->rmt_addr;
	newinet->inet_rcv_saddr = ireq->loc_addr;
	newinet->inet_saddr	      = ireq->loc_addr;
1439 1440
	inet_opt	      = ireq->opt;
	rcu_assign_pointer(newinet->inet_opt, inet_opt);
1441
	ireq->opt	      = NULL;
1442
	newinet->mc_index     = inet_iif(skb);
1443
	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
1444
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
1445 1446
	if (inet_opt)
		inet_csk(newsk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
E
Eric Dumazet 已提交
1447
	newinet->inet_id = newtp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
1448

1449 1450 1451 1452 1453
	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
1454
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1455
	tcp_sync_mss(newsk, dst_mtu(dst));
1456
	newtp->advmss = dst_metric_advmss(dst);
1457 1458 1459 1460
	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
1461
	tcp_initialize_rcv_mss(newsk);
1462 1463 1464 1465
	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
1466

1467 1468
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1469 1470
	key = tcp_v4_md5_do_lookup(sk, newinet->inet_daddr);
	if (key != NULL) {
1471 1472 1473 1474 1475 1476
		/*
		 * 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.
		 */
1477 1478
		char *newkey = kmemdup(key->key, key->keylen, GFP_ATOMIC);
		if (newkey != NULL)
E
Eric Dumazet 已提交
1479
			tcp_v4_md5_do_add(newsk, newinet->inet_daddr,
1480
					  newkey, key->keylen);
E
Eric Dumazet 已提交
1481
		sk_nocaps_add(newsk, NETIF_F_GSO_MASK);
1482 1483 1484
	}
#endif

1485 1486
	if (__inet_inherit_port(sk, newsk) < 0)
		goto put_and_exit;
1487
	__inet_hash_nolisten(newsk, NULL);
L
Linus Torvalds 已提交
1488 1489 1490 1491

	return newsk;

exit_overflow:
1492
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1493 1494
exit_nonewsk:
	dst_release(dst);
L
Linus Torvalds 已提交
1495
exit:
1496
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1497
	return NULL;
1498 1499 1500
put_and_exit:
	sock_put(newsk);
	goto exit;
L
Linus Torvalds 已提交
1501
}
E
Eric Dumazet 已提交
1502
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
L
Linus Torvalds 已提交
1503 1504 1505

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1506
	struct tcphdr *th = tcp_hdr(skb);
1507
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1508
	struct sock *nsk;
1509
	struct request_sock **prev;
L
Linus Torvalds 已提交
1510
	/* Find possible connection requests. */
1511 1512
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1513 1514 1515
	if (req)
		return tcp_check_req(sk, skb, req, prev);

1516
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1517
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1518 1519 1520 1521 1522 1523

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1524
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1525 1526 1527 1528
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
1529
	if (!th->syn)
L
Linus Torvalds 已提交
1530 1531 1532 1533 1534
		sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
#endif
	return sk;
}

1535
static __sum16 tcp_v4_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1536
{
1537 1538
	const struct iphdr *iph = ip_hdr(skb);

1539
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
1540 1541
		if (!tcp_v4_check(skb->len, iph->saddr,
				  iph->daddr, skb->csum)) {
1542
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1543
			return 0;
1544
		}
L
Linus Torvalds 已提交
1545
	}
1546

1547
	skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
1548 1549
				       skb->len, IPPROTO_TCP, 0);

L
Linus Torvalds 已提交
1550
	if (skb->len <= 76) {
1551
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566
	}
	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)
{
1567 1568 1569 1570 1571 1572 1573 1574
	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
	 */
1575
	if (tcp_v4_inbound_md5_hash(sk, skb))
1576 1577 1578
		goto discard;
#endif

L
Linus Torvalds 已提交
1579
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1580
		sock_rps_save_rxhash(sk, skb->rxhash);
1581
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
1582
			rsk = sk;
L
Linus Torvalds 已提交
1583
			goto reset;
1584
		}
L
Linus Torvalds 已提交
1585 1586 1587
		return 0;
	}

1588
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1589 1590 1591 1592 1593 1594 1595 1596
		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) {
1597
			sock_rps_save_rxhash(nsk, skb->rxhash);
1598 1599
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1600
				goto reset;
1601
			}
L
Linus Torvalds 已提交
1602 1603
			return 0;
		}
1604 1605 1606
	} else
		sock_rps_save_rxhash(sk, skb->rxhash);

1607
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1608
		rsk = sk;
L
Linus Torvalds 已提交
1609
		goto reset;
1610
	}
L
Linus Torvalds 已提交
1611 1612 1613
	return 0;

reset:
1614
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1615 1616 1617 1618 1619 1620 1621 1622 1623 1624
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:
1625
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1626 1627
	goto discard;
}
E
Eric Dumazet 已提交
1628
EXPORT_SYMBOL(tcp_v4_do_rcv);
L
Linus Torvalds 已提交
1629 1630 1631 1632 1633 1634 1635

/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1636
	const struct iphdr *iph;
L
Linus Torvalds 已提交
1637 1638 1639
	struct tcphdr *th;
	struct sock *sk;
	int ret;
1640
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1641 1642 1643 1644 1645

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

	/* Count it even if it's bad */
1646
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1647 1648 1649 1650

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

1651
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1652 1653 1654 1655 1656 1657 1658 1659

	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
1660
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1661
	 * So, we defer the checks. */
1662
	if (!skb_csum_unnecessary(skb) && tcp_v4_checksum_init(skb))
L
Linus Torvalds 已提交
1663 1664
		goto bad_packet;

1665
	th = tcp_hdr(skb);
1666
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1667 1668 1669 1670 1671
	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;
1672
	TCP_SKB_CB(skb)->flags	 = iph->tos;
L
Linus Torvalds 已提交
1673 1674
	TCP_SKB_CB(skb)->sacked	 = 0;

1675
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1676 1677 1678
	if (!sk)
		goto no_tcp_socket;

E
Eric Dumazet 已提交
1679 1680 1681 1682
process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1683 1684
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1685
		goto discard_and_relse;
1686
	}
1687

L
Linus Torvalds 已提交
1688 1689
	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1690
	nf_reset(skb);
L
Linus Torvalds 已提交
1691

1692
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1693 1694 1695 1696
		goto discard_and_relse;

	skb->dev = NULL;

1697
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1698 1699
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1700 1701 1702
#ifdef CONFIG_NET_DMA
		struct tcp_sock *tp = tcp_sk(sk);
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1703
			tp->ucopy.dma_chan = dma_find_channel(DMA_MEMCPY);
C
Chris Leech 已提交
1704
		if (tp->ucopy.dma_chan)
L
Linus Torvalds 已提交
1705
			ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1706 1707 1708 1709
		else
#endif
		{
			if (!tcp_prequeue(sk, skb))
S
Shan Wei 已提交
1710
				ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1711
		}
1712
	} else if (unlikely(sk_add_backlog(sk, skb))) {
Z
Zhu Yi 已提交
1713
		bh_unlock_sock(sk);
1714
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1715 1716
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728
	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:
1729
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1730
	} else {
1731
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1732 1733 1734 1735 1736
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
1737
	return 0;
L
Linus Torvalds 已提交
1738 1739 1740 1741 1742 1743 1744

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1745
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1746 1747 1748 1749
		goto discard_it;
	}

	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
1750
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1751
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1752 1753
		goto discard_it;
	}
1754
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1755
	case TCP_TW_SYN: {
1756
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
1757
							&tcp_hashinfo,
1758
							iph->daddr, th->dest,
1759
							inet_iif(skb));
L
Linus Torvalds 已提交
1760
		if (sk2) {
1761 1762
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
			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;
}

1778
struct inet_peer *tcp_v4_get_peer(struct sock *sk, bool *release_it)
L
Linus Torvalds 已提交
1779
{
1780
	struct rtable *rt = (struct rtable *) __sk_dst_get(sk);
L
Linus Torvalds 已提交
1781
	struct inet_sock *inet = inet_sk(sk);
1782
	struct inet_peer *peer;
L
Linus Torvalds 已提交
1783

1784 1785
	if (!rt ||
	    inet->cork.fl.u.ip4.daddr != inet->inet_daddr) {
1786
		peer = inet_getpeer_v4(inet->inet_daddr, 1);
1787
		*release_it = true;
L
Linus Torvalds 已提交
1788 1789
	} else {
		if (!rt->peer)
1790
			rt_bind_peer(rt, inet->inet_daddr, 1);
L
Linus Torvalds 已提交
1791
		peer = rt->peer;
1792
		*release_it = false;
L
Linus Torvalds 已提交
1793 1794
	}

1795
	return peer;
L
Linus Torvalds 已提交
1796
}
1797
EXPORT_SYMBOL(tcp_v4_get_peer);
L
Linus Torvalds 已提交
1798

1799
void *tcp_v4_tw_get_peer(struct sock *sk)
L
Linus Torvalds 已提交
1800
{
1801
	struct inet_timewait_sock *tw = inet_twsk(sk);
L
Linus Torvalds 已提交
1802

1803
	return inet_getpeer_v4(tw->tw_daddr, 1);
L
Linus Torvalds 已提交
1804
}
1805 1806 1807 1808 1809 1810 1811 1812
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
1813

1814
const struct inet_connection_sock_af_ops ipv4_specific = {
1815 1816 1817 1818 1819
	.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,
1820
	.get_peer	   = tcp_v4_get_peer,
1821 1822 1823 1824 1825
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ip_setsockopt,
	.getsockopt	   = ip_getsockopt,
	.addr2sockaddr	   = inet_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in),
1826
	.bind_conflict	   = inet_csk_bind_conflict,
1827
#ifdef CONFIG_COMPAT
1828 1829
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
1830
#endif
L
Linus Torvalds 已提交
1831
};
E
Eric Dumazet 已提交
1832
EXPORT_SYMBOL(ipv4_specific);
L
Linus Torvalds 已提交
1833

1834
#ifdef CONFIG_TCP_MD5SIG
1835
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
1836
	.md5_lookup		= tcp_v4_md5_lookup,
1837
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
1838 1839 1840
	.md5_add		= tcp_v4_md5_add_func,
	.md5_parse		= tcp_v4_parse_md5_keys,
};
1841
#endif
1842

L
Linus Torvalds 已提交
1843 1844 1845 1846 1847
/* 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)
{
1848
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1849 1850 1851 1852 1853 1854
	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);

1855
	icsk->icsk_rto = TCP_TIMEOUT_INIT;
L
Linus Torvalds 已提交
1856 1857 1858 1859 1860 1861 1862
	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
	 */
1863
	tp->snd_cwnd = TCP_INIT_CWND;
L
Linus Torvalds 已提交
1864 1865 1866 1867

	/* See draft-stevens-tcpca-spec-01 for discussion of the
	 * initialization of these values.
	 */
I
Ilpo Järvinen 已提交
1868
	tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
L
Linus Torvalds 已提交
1869
	tp->snd_cwnd_clamp = ~0;
1870
	tp->mss_cache = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
1871 1872

	tp->reordering = sysctl_tcp_reordering;
1873
	icsk->icsk_ca_ops = &tcp_init_congestion_ops;
L
Linus Torvalds 已提交
1874 1875 1876 1877 1878 1879

	sk->sk_state = TCP_CLOSE;

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

1880
	icsk->icsk_af_ops = &ipv4_specific;
1881
	icsk->icsk_sync_mss = tcp_sync_mss;
1882 1883 1884
#ifdef CONFIG_TCP_MD5SIG
	tp->af_specific = &tcp_sock_ipv4_specific;
#endif
L
Linus Torvalds 已提交
1885

1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898
	/* 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
1899 1900 1901
	sk->sk_sndbuf = sysctl_tcp_wmem[1];
	sk->sk_rcvbuf = sysctl_tcp_rmem[1];

H
Herbert Xu 已提交
1902
	local_bh_disable();
1903
	percpu_counter_inc(&tcp_sockets_allocated);
H
Herbert Xu 已提交
1904
	local_bh_enable();
L
Linus Torvalds 已提交
1905 1906 1907 1908

	return 0;
}

1909
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1910 1911 1912 1913 1914
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

1915
	tcp_cleanup_congestion_control(sk);
1916

L
Linus Torvalds 已提交
1917
	/* Cleanup up the write buffer. */
1918
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
1919 1920

	/* Cleans up our, hopefully empty, out_of_order_queue. */
1921
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
1922

1923 1924 1925 1926 1927 1928 1929 1930 1931
#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
1932 1933
#ifdef CONFIG_NET_DMA
	/* Cleans up our sk_async_wait_queue */
1934
	__skb_queue_purge(&sk->sk_async_wait_queue);
C
Chris Leech 已提交
1935 1936
#endif

L
Linus Torvalds 已提交
1937 1938 1939 1940
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
1941
	if (inet_csk(sk)->icsk_bind_hash)
1942
		inet_put_port(sk);
L
Linus Torvalds 已提交
1943 1944 1945 1946 1947 1948 1949 1950 1951

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

1952 1953 1954 1955 1956 1957 1958
	/* TCP Cookie Transactions */
	if (tp->cookie_values != NULL) {
		kref_put(&tp->cookie_values->kref,
			 tcp_cookie_values_release);
		tp->cookie_values = NULL;
	}

1959
	percpu_counter_dec(&tcp_sockets_allocated);
L
Linus Torvalds 已提交
1960 1961 1962 1963 1964 1965
}
EXPORT_SYMBOL(tcp_v4_destroy_sock);

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

1966
static inline struct inet_timewait_sock *tw_head(struct hlist_nulls_head *head)
L
Linus Torvalds 已提交
1967
{
1968
	return hlist_nulls_empty(head) ? NULL :
1969
		list_entry(head->first, struct inet_timewait_sock, tw_node);
L
Linus Torvalds 已提交
1970 1971
}

1972
static inline struct inet_timewait_sock *tw_next(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
1973
{
1974 1975
	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
1976 1977
}

1978 1979 1980 1981 1982
/*
 * 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
1983 1984
static void *listening_get_next(struct seq_file *seq, void *cur)
{
1985
	struct inet_connection_sock *icsk;
1986
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
1987
	struct sock *sk = cur;
1988
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
1989
	struct tcp_iter_state *st = seq->private;
1990
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
1991 1992

	if (!sk) {
1993
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
1994
		spin_lock_bh(&ilb->lock);
1995
		sk = sk_nulls_head(&ilb->head);
1996
		st->offset = 0;
L
Linus Torvalds 已提交
1997 1998
		goto get_sk;
	}
1999
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
2000
	++st->num;
2001
	++st->offset;
L
Linus Torvalds 已提交
2002 2003

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
2004
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
2005

2006
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
2007 2008 2009
		req = req->dl_next;
		while (1) {
			while (req) {
2010
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
2011 2012 2013 2014 2015
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
2016
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
2017 2018
				break;
get_req:
2019
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
2020
		}
E
Eric Dumazet 已提交
2021
		sk	  = sk_nulls_next(st->syn_wait_sk);
L
Linus Torvalds 已提交
2022
		st->state = TCP_SEQ_STATE_LISTENING;
2023
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2024
	} else {
2025
		icsk = inet_csk(sk);
2026 2027
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
2028
			goto start_req;
2029
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
E
Eric Dumazet 已提交
2030
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2031 2032
	}
get_sk:
2033
	sk_nulls_for_each_from(sk, node) {
2034 2035 2036
		if (!net_eq(sock_net(sk), net))
			continue;
		if (sk->sk_family == st->family) {
L
Linus Torvalds 已提交
2037 2038 2039
			cur = sk;
			goto out;
		}
2040
		icsk = inet_csk(sk);
2041 2042
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
2043 2044 2045 2046 2047 2048 2049
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;
		}
2050
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2051
	}
2052
	spin_unlock_bh(&ilb->lock);
2053
	st->offset = 0;
2054
	if (++st->bucket < INET_LHTABLE_SIZE) {
2055 2056
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
2057
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
2058 2059 2060 2061 2062 2063 2064 2065 2066
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
2067 2068 2069 2070 2071 2072
	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
2073 2074 2075 2076 2077 2078 2079 2080

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

2081 2082
static inline int empty_bucket(struct tcp_iter_state *st)
{
2083 2084
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain) &&
		hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].twchain);
2085 2086
}

2087 2088 2089 2090
/*
 * 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
2091 2092
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
2093
	struct tcp_iter_state *st = seq->private;
2094
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2095 2096
	void *rc = NULL;

2097 2098
	st->offset = 0;
	for (; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
2099
		struct sock *sk;
2100
		struct hlist_nulls_node *node;
2101
		struct inet_timewait_sock *tw;
2102
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
2103

2104 2105 2106 2107
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

2108
		spin_lock_bh(lock);
2109
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
2110
			if (sk->sk_family != st->family ||
2111
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2112 2113 2114 2115 2116 2117
				continue;
			}
			rc = sk;
			goto out;
		}
		st->state = TCP_SEQ_STATE_TIME_WAIT;
2118
		inet_twsk_for_each(tw, node,
2119
				   &tcp_hashinfo.ehash[st->bucket].twchain) {
2120
			if (tw->tw_family != st->family ||
2121
			    !net_eq(twsk_net(tw), net)) {
L
Linus Torvalds 已提交
2122 2123 2124 2125 2126
				continue;
			}
			rc = tw;
			goto out;
		}
2127
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2128 2129 2130 2131 2132 2133 2134 2135 2136
		st->state = TCP_SEQ_STATE_ESTABLISHED;
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2137
	struct inet_timewait_sock *tw;
2138
	struct hlist_nulls_node *node;
J
Jianjun Kong 已提交
2139
	struct tcp_iter_state *st = seq->private;
2140
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2141 2142

	++st->num;
2143
	++st->offset;
L
Linus Torvalds 已提交
2144 2145 2146 2147 2148

	if (st->state == TCP_SEQ_STATE_TIME_WAIT) {
		tw = cur;
		tw = tw_next(tw);
get_tw:
2149
		while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
L
Linus Torvalds 已提交
2150 2151 2152 2153 2154 2155
			tw = tw_next(tw);
		}
		if (tw) {
			cur = tw;
			goto out;
		}
2156
		spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2157 2158
		st->state = TCP_SEQ_STATE_ESTABLISHED;

2159
		/* Look for next non empty bucket */
2160
		st->offset = 0;
2161
		while (++st->bucket <= tcp_hashinfo.ehash_mask &&
2162 2163
				empty_bucket(st))
			;
2164
		if (st->bucket > tcp_hashinfo.ehash_mask)
2165 2166
			return NULL;

2167
		spin_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
2168
		sk = sk_nulls_head(&tcp_hashinfo.ehash[st->bucket].chain);
L
Linus Torvalds 已提交
2169
	} else
2170
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2171

2172
	sk_nulls_for_each_from(sk, node) {
2173
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
L
Linus Torvalds 已提交
2174 2175 2176 2177
			goto found;
	}

	st->state = TCP_SEQ_STATE_TIME_WAIT;
2178
	tw = tw_head(&tcp_hashinfo.ehash[st->bucket].twchain);
L
Linus Torvalds 已提交
2179 2180 2181 2182 2183 2184 2185 2186 2187
	goto get_tw;
found:
	cur = sk;
out:
	return cur;
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
2188 2189 2190 2191 2192
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	rc = established_get_first(seq);
L
Linus Torvalds 已提交
2193 2194 2195 2196

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2197
	}
L
Linus Torvalds 已提交
2198 2199 2200 2201 2202 2203
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2204
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216

	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;
}

2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251
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
2252 2253
static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2254
	struct tcp_iter_state *st = seq->private;
2255 2256 2257 2258 2259 2260 2261 2262
	void *rc;

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

L
Linus Torvalds 已提交
2263 2264
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
2265 2266 2267 2268 2269 2270 2271
	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
2272 2273 2274 2275
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2276
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289
	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;
2290 2291
			st->bucket = 0;
			st->offset = 0;
L
Linus Torvalds 已提交
2292 2293 2294 2295 2296 2297 2298 2299 2300 2301
			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;
2302
	st->last_pos = *pos;
L
Linus Torvalds 已提交
2303 2304 2305 2306 2307
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2308
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2309 2310 2311 2312

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2313 2314
			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
2315 2316 2317
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2318
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2319 2320 2321 2322
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2323
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2324 2325 2326 2327 2328 2329 2330 2331
		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;
2332
	int err;
L
Linus Torvalds 已提交
2333

2334 2335 2336 2337
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2338

2339
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2340
	s->family		= afinfo->family;
2341
	s->last_pos 		= 0;
2342 2343 2344
	return 0;
}

2345
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2346 2347 2348 2349
{
	int rc = 0;
	struct proc_dir_entry *p;

2350 2351 2352 2353
	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;
2354

2355 2356 2357 2358
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2359 2360 2361
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
			     &afinfo->seq_fops, afinfo);
	if (!p)
L
Linus Torvalds 已提交
2362 2363 2364
		rc = -ENOMEM;
	return rc;
}
E
Eric Dumazet 已提交
2365
EXPORT_SYMBOL(tcp_proc_register);
L
Linus Torvalds 已提交
2366

2367
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2368
{
2369
	proc_net_remove(net, afinfo->name);
L
Linus Torvalds 已提交
2370
}
E
Eric Dumazet 已提交
2371
EXPORT_SYMBOL(tcp_proc_unregister);
L
Linus Torvalds 已提交
2372

2373
static void get_openreq4(struct sock *sk, struct request_sock *req,
2374
			 struct seq_file *f, int i, int uid, int *len)
L
Linus Torvalds 已提交
2375
{
2376
	const struct inet_request_sock *ireq = inet_rsk(req);
L
Linus Torvalds 已提交
2377 2378
	int ttd = req->expires - jiffies;

2379
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2380
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %pK%n",
L
Linus Torvalds 已提交
2381
		i,
2382
		ireq->loc_addr,
E
Eric Dumazet 已提交
2383
		ntohs(inet_sk(sk)->inet_sport),
2384 2385
		ireq->rmt_addr,
		ntohs(ireq->rmt_port),
L
Linus Torvalds 已提交
2386 2387 2388 2389 2390 2391 2392 2393 2394
		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),
2395 2396
		req,
		len);
L
Linus Torvalds 已提交
2397 2398
}

2399
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2400 2401 2402
{
	int timer_active;
	unsigned long timer_expires;
2403 2404 2405
	struct tcp_sock *tp = tcp_sk(sk);
	const struct inet_connection_sock *icsk = inet_csk(sk);
	struct inet_sock *inet = inet_sk(sk);
E
Eric Dumazet 已提交
2406 2407 2408 2409
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
2410
	int rx_queue;
L
Linus Torvalds 已提交
2411

2412
	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
2413
		timer_active	= 1;
2414 2415
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2416
		timer_active	= 4;
2417
		timer_expires	= icsk->icsk_timeout;
2418
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2419
		timer_active	= 2;
2420
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2421 2422 2423 2424 2425
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2426 2427 2428 2429 2430 2431 2432 2433
	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);

2434
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
D
Dan Rosenberg 已提交
2435
			"%08X %5d %8d %lu %d %pK %lu %lu %u %u %d%n",
2436
		i, src, srcp, dest, destp, sk->sk_state,
2437
		tp->write_seq - tp->snd_una,
2438
		rx_queue,
L
Linus Torvalds 已提交
2439 2440
		timer_active,
		jiffies_to_clock_t(timer_expires - jiffies),
2441
		icsk->icsk_retransmits,
2442
		sock_i_uid(sk),
2443
		icsk->icsk_probes_out,
2444 2445
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2446 2447
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2448
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2449
		tp->snd_cwnd,
I
Ilpo Järvinen 已提交
2450
		tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh,
2451
		len);
L
Linus Torvalds 已提交
2452 2453
}

2454
static void get_timewait4_sock(struct inet_timewait_sock *tw,
2455
			       struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2456
{
2457
	__be32 dest, src;
L
Linus Torvalds 已提交
2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468
	__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);

2469
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2470
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK%n",
L
Linus Torvalds 已提交
2471 2472
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
		3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
2473
		atomic_read(&tw->tw_refcnt), tw, len);
L
Linus Torvalds 已提交
2474 2475 2476 2477 2478 2479
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2480
	struct tcp_iter_state *st;
2481
	int len;
L
Linus Torvalds 已提交
2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494

	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:
2495
		get_tcp4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2496 2497
		break;
	case TCP_SEQ_STATE_OPENREQ:
2498
		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid, &len);
L
Linus Torvalds 已提交
2499 2500
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
2501
		get_timewait4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2502 2503
		break;
	}
2504
	seq_printf(seq, "%*s\n", TMPSZ - 1 - len, "");
L
Linus Torvalds 已提交
2505 2506 2507 2508 2509 2510 2511
out:
	return 0;
}

static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2512 2513 2514
	.seq_fops	= {
		.owner		= THIS_MODULE,
	},
2515 2516 2517
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2518 2519
};

2520
static int __net_init tcp4_proc_init_net(struct net *net)
2521 2522 2523 2524
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

2525
static void __net_exit tcp4_proc_exit_net(struct net *net)
2526 2527 2528 2529 2530 2531 2532 2533 2534
{
	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
2535 2536
int __init tcp4_proc_init(void)
{
2537
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2538 2539 2540 2541
}

void tcp4_proc_exit(void)
{
2542
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2543 2544 2545
}
#endif /* CONFIG_PROC_FS */

H
Herbert Xu 已提交
2546 2547
struct sk_buff **tcp4_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
2548
	const struct iphdr *iph = skb_gro_network_header(skb);
H
Herbert Xu 已提交
2549 2550 2551

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
2552
		if (!tcp_v4_check(skb_gro_len(skb), iph->saddr, iph->daddr,
H
Herbert Xu 已提交
2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568
				  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)
{
2569
	const struct iphdr *iph = ip_hdr(skb);
H
Herbert Xu 已提交
2570 2571 2572 2573 2574 2575 2576 2577 2578
	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
2579 2580 2581 2582 2583 2584
struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2585
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2586 2587 2588 2589 2590 2591 2592
	.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,
2593 2594
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2595
	.backlog_rcv		= tcp_v4_do_rcv,
2596 2597 2598
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2599 2600
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
2601
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2602 2603 2604 2605 2606 2607 2608
	.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),
2609
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2610
	.twsk_prot		= &tcp_timewait_sock_ops,
2611
	.rsk_prot		= &tcp_request_sock_ops,
2612
	.h.hashinfo		= &tcp_hashinfo,
2613
	.no_autobind		= true,
2614 2615 2616 2617
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
L
Linus Torvalds 已提交
2618
};
E
Eric Dumazet 已提交
2619
EXPORT_SYMBOL(tcp_prot);
L
Linus Torvalds 已提交
2620

2621 2622 2623 2624 2625 2626 2627 2628 2629 2630

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
2631 2632 2633 2634 2635
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2636 2637 2638
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2639 2640 2641
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2642 2643
};

2644
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2645
{
2646
	inet_hashinfo_init(&tcp_hashinfo);
2647
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2648 2649
		panic("Failed to create the TCP control socket.\n");
}