tcp_ipv4.c 74.4 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
 *		Mike McLagan	:	Routing by source
 *	Juan Jose Ciarlante:		ip_dynaddr bits
 *		Andi Kleen:		various fixes.
 *	Vitaly E. Lavrov	:	Transparent proxy revived after year
 *					coma.
 *	Andi Kleen		:	Fix new listen.
 *	Andi Kleen		:	Fix accept error reporting.
 *	YOSHIFUJI Hideaki @USAGI and:	Support IPV6_V6ONLY socket option, which
 *	Alexey Kuznetsov		allow both IPv4 and IPv6 sockets to bind
 *					a single port at the same time.
 */

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

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

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

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

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

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


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

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

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

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

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

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

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

	return 0;
}
EXPORT_SYMBOL_GPL(tcp_twsk_unique);

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

	return 0;

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

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

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

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

	mtu = dst_mtu(dst);

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

303 304 305 306
static void do_redirect(struct sk_buff *skb, struct sock *sk)
{
	struct dst_entry *dst = __sk_dst_check(sk, 0);

307
	if (dst)
308
		dst->ops->redirect(dst, sk, skb);
309 310
}

L
Linus Torvalds 已提交
311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326
/*
 * 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.
 *
 */

327
void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
L
Linus Torvalds 已提交
328
{
329
	const struct iphdr *iph = (const struct iphdr *)icmp_skb->data;
330
	struct tcphdr *th = (struct tcphdr *)(icmp_skb->data + (iph->ihl << 2));
331
	struct inet_connection_sock *icsk;
L
Linus Torvalds 已提交
332 333
	struct tcp_sock *tp;
	struct inet_sock *inet;
334 335
	const int type = icmp_hdr(icmp_skb)->type;
	const int code = icmp_hdr(icmp_skb)->code;
L
Linus Torvalds 已提交
336
	struct sock *sk;
337
	struct sk_buff *skb;
338
	struct request_sock *req;
L
Linus Torvalds 已提交
339
	__u32 seq;
340
	__u32 remaining;
L
Linus Torvalds 已提交
341
	int err;
342
	struct net *net = dev_net(icmp_skb->dev);
L
Linus Torvalds 已提交
343

344
	if (icmp_skb->len < (iph->ihl << 2) + 8) {
345
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
346 347 348
		return;
	}

349
	sk = inet_lookup(net, &tcp_hashinfo, iph->daddr, th->dest,
350
			iph->saddr, th->source, inet_iif(icmp_skb));
L
Linus Torvalds 已提交
351
	if (!sk) {
352
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
353 354 355
		return;
	}
	if (sk->sk_state == TCP_TIME_WAIT) {
356
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
357 358 359 360 361 362
		return;
	}

	bh_lock_sock(sk);
	/* If too many ICMPs get dropped on busy
	 * servers this needs to be solved differently.
363 364
	 * We do take care of PMTU discovery (RFC1191) special case :
	 * we can receive locally generated ICMP messages while socket is held.
L
Linus Torvalds 已提交
365
	 */
366 367 368 369
	if (sock_owned_by_user(sk)) {
		if (!(type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED))
			NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
	}
L
Linus Torvalds 已提交
370 371 372
	if (sk->sk_state == TCP_CLOSE)
		goto out;

373 374 375 376 377
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto out;
	}

378
	icsk = inet_csk(sk);
L
Linus Torvalds 已提交
379
	tp = tcp_sk(sk);
380
	req = tp->fastopen_rsk;
L
Linus Torvalds 已提交
381 382
	seq = ntohl(th->seq);
	if (sk->sk_state != TCP_LISTEN &&
383 384 385
	    !between(seq, tp->snd_una, tp->snd_nxt) &&
	    (req == NULL || seq != tcp_rsk(req)->snt_isn)) {
		/* For a Fast Open socket, allow seq to be snt_isn. */
386
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
387 388 389 390
		goto out;
	}

	switch (type) {
391 392 393
	case ICMP_REDIRECT:
		do_redirect(icmp_skb, sk);
		goto out;
L
Linus Torvalds 已提交
394 395 396 397 398 399 400 401 402 403 404
	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) */
405 406 407 408 409 410 411
			/* We are not interested in TCP_LISTEN and open_requests
			 * (SYN-ACKs send out by Linux are always <576bytes so
			 * they should go through unfragmented).
			 */
			if (sk->sk_state == TCP_LISTEN)
				goto out;

412
			tp->mtu_info = info;
413
			if (!sock_owned_by_user(sk)) {
414
				tcp_v4_mtu_reduced(sk);
415 416 417 418
			} else {
				if (!test_and_set_bit(TCP_MTU_REDUCED_DEFERRED, &tp->tsq_flags))
					sock_hold(sk);
			}
L
Linus Torvalds 已提交
419 420 421 422
			goto out;
		}

		err = icmp_err_convert[code].errno;
423 424 425 426 427 428 429 430
		/* 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;

431 432
		/* XXX (TFO) - revisit the following logic for TFO */

433 434 435
		if (sock_owned_by_user(sk))
			break;

436
		icsk->icsk_backoff--;
437 438
		inet_csk(sk)->icsk_rto = (tp->srtt ? __tcp_set_rto(tp) :
			TCP_TIMEOUT_INIT) << icsk->icsk_backoff;
439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455
		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
456 457 458 459 460 461 462 463
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

464 465 466 467 468 469 470 471
	/* XXX (TFO) - if it's a TFO socket and has been accepted, rather
	 * than following the TCP_SYN_RECV case and closing the socket,
	 * we ignore the ICMP error and keep trying like a fully established
	 * socket. Is this the right thing to do?
	 */
	if (req && req->sk == NULL)
		goto out;

L
Linus Torvalds 已提交
472
	switch (sk->sk_state) {
473
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
474 475 476 477
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

478 479
		req = inet_csk_search_req(sk, &prev, th->dest,
					  iph->daddr, iph->saddr);
L
Linus Torvalds 已提交
480 481 482 483 484 485
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		   an established socket here.
		 */
486
		WARN_ON(req->sk);
L
Linus Torvalds 已提交
487

488
		if (seq != tcp_rsk(req)->snt_isn) {
489
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
490 491 492 493 494 495 496 497 498
			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().
		 */
499
		inet_csk_reqsk_queue_drop(sk, req, prev);
500
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
501 502 503 504
		goto out;

	case TCP_SYN_SENT:
	case TCP_SYN_RECV:  /* Cannot happen.
505 506
			       It can f.e. if SYNs crossed,
			       or Fast Open.
L
Linus Torvalds 已提交
507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548
			     */
		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);
}

549
void __tcp_v4_send_check(struct sk_buff *skb, __be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
550
{
551
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
552

553
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
554
		th->check = ~tcp_v4_check(skb->len, saddr, daddr, 0);
555
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
556
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
557
	} else {
558
		th->check = tcp_v4_check(skb->len, saddr, daddr,
559
					 csum_partial(th,
L
Linus Torvalds 已提交
560 561 562 563 564
						      th->doff << 2,
						      skb->csum));
	}
}

565
/* This routine computes an IPv4 TCP checksum. */
566
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb)
567
{
568
	const struct inet_sock *inet = inet_sk(sk);
569 570 571

	__tcp_v4_send_check(skb, inet->inet_saddr, inet->inet_daddr);
}
E
Eric Dumazet 已提交
572
EXPORT_SYMBOL(tcp_v4_send_check);
573

L
Linus Torvalds 已提交
574 575 576 577 578 579 580 581 582 583 584 585 586
/*
 *	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.
 */

587
static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
588
{
589
	const struct tcphdr *th = tcp_hdr(skb);
590 591 592
	struct {
		struct tcphdr th;
#ifdef CONFIG_TCP_MD5SIG
593
		__be32 opt[(TCPOLEN_MD5SIG_ALIGNED >> 2)];
594 595
#endif
	} rep;
L
Linus Torvalds 已提交
596
	struct ip_reply_arg arg;
597 598
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
599 600 601 602
	const __u8 *hash_location = NULL;
	unsigned char newhash[16];
	int genhash;
	struct sock *sk1 = NULL;
603
#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
633 634 635 636 637 638 639 640 641 642
	hash_location = tcp_parse_md5sig_option(th);
	if (!sk && hash_location) {
		/*
		 * active side is lost. Try to find listening socket through
		 * source port, and then find md5 key through listening socket.
		 * we are not loose security here:
		 * Incoming packet is checked with md5 hash with finding key,
		 * no RST generated if md5 hash doesn't match.
		 */
		sk1 = __inet_lookup_listener(dev_net(skb_dst(skb)->dev),
643 644
					     &tcp_hashinfo, ip_hdr(skb)->saddr,
					     th->source, ip_hdr(skb)->daddr,
645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663
					     ntohs(th->source), inet_iif(skb));
		/* don't send rst if it can't find key */
		if (!sk1)
			return;
		rcu_read_lock();
		key = tcp_md5_do_lookup(sk1, (union tcp_md5_addr *)
					&ip_hdr(skb)->saddr, AF_INET);
		if (!key)
			goto release_sk1;

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

664 665 666 667 668 669 670 671 672
	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;

673
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
674 675
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
676 677
	}
#endif
678 679
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
680
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
681
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
682
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
683
	/* When socket is gone, all binding information is lost.
A
Alexey Kuznetsov 已提交
684 685
	 * routing might fail in this case. No choice here, if we choose to force
	 * input interface, we will misroute in case of asymmetric route.
686
	 */
A
Alexey Kuznetsov 已提交
687 688
	if (sk)
		arg.bound_dev_if = sk->sk_bound_dev_if;
L
Linus Torvalds 已提交
689

E
Eric Dumazet 已提交
690
	net = dev_net(skb_dst(skb)->dev);
691
	arg.tos = ip_hdr(skb)->tos;
692
	ip_send_unicast_reply(net, skb, ip_hdr(skb)->saddr,
693
			      ip_hdr(skb)->daddr, &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
694

695 696
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
697 698 699 700 701 702 703 704

#ifdef CONFIG_TCP_MD5SIG
release_sk1:
	if (sk1) {
		rcu_read_unlock();
		sock_put(sk1);
	}
#endif
L
Linus Torvalds 已提交
705 706 707 708 709 710
}

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

711
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
712
			    u32 win, u32 tsval, u32 tsecr, int oif,
713
			    struct tcp_md5sig_key *key,
714
			    int reply_flags, u8 tos)
L
Linus Torvalds 已提交
715
{
716
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
717 718
	struct {
		struct tcphdr th;
719
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
720
#ifdef CONFIG_TCP_MD5SIG
721
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
722 723
#endif
			];
L
Linus Torvalds 已提交
724 725
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
726
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
727 728

	memset(&rep.th, 0, sizeof(struct tcphdr));
729
	memset(&arg, 0, sizeof(arg));
L
Linus Torvalds 已提交
730 731 732

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
733
	if (tsecr) {
734 735 736
		rep.opt[0] = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				   (TCPOPT_TIMESTAMP << 8) |
				   TCPOLEN_TIMESTAMP);
737 738
		rep.opt[1] = htonl(tsval);
		rep.opt[2] = htonl(tsecr);
739
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
740 741 742 743 744 745 746 747 748 749 750
	}

	/* 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);

751 752
#ifdef CONFIG_TCP_MD5SIG
	if (key) {
753
		int offset = (tsecr) ? 3 : 0;
754 755 756 757 758 759 760 761

		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;

762
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[offset],
763 764
				    key, ip_hdr(skb)->saddr,
				    ip_hdr(skb)->daddr, &rep.th);
765 766
	}
#endif
767
	arg.flags = reply_flags;
768 769
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
L
Linus Torvalds 已提交
770 771
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
772 773
	if (oif)
		arg.bound_dev_if = oif;
774
	arg.tos = tos;
775
	ip_send_unicast_reply(net, skb, ip_hdr(skb)->saddr,
776
			      ip_hdr(skb)->daddr, &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
777

778
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
779 780 781 782
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
783
	struct inet_timewait_sock *tw = inet_twsk(sk);
784
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
785

786
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
787
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
788
			tcp_time_stamp + tcptw->tw_ts_offset,
789 790
			tcptw->tw_ts_recent,
			tw->tw_bound_dev_if,
791
			tcp_twsk_md5_key(tcptw),
792 793
			tw->tw_transparent ? IP_REPLY_ARG_NOSRCCHECK : 0,
			tw->tw_tos
794
			);
L
Linus Torvalds 已提交
795

796
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
797 798
}

799
static void tcp_v4_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
800
				  struct request_sock *req)
L
Linus Torvalds 已提交
801
{
802 803 804 805 806 807
	/* sk->sk_state == TCP_LISTEN -> for regular TCP_SYN_RECV
	 * sk->sk_state == TCP_SYN_RECV -> for Fast Open.
	 */
	tcp_v4_send_ack(skb, (sk->sk_state == TCP_LISTEN) ?
			tcp_rsk(req)->snt_isn + 1 : tcp_sk(sk)->snd_nxt,
			tcp_rsk(req)->rcv_nxt, req->rcv_wnd,
808
			tcp_time_stamp,
809 810
			req->ts_recent,
			0,
E
Eric Dumazet 已提交
811 812
			tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&ip_hdr(skb)->daddr,
					  AF_INET),
813 814
			inet_rsk(req)->no_srccheck ? IP_REPLY_ARG_NOSRCCHECK : 0,
			ip_hdr(skb)->tos);
L
Linus Torvalds 已提交
815 816 817
}

/*
818
 *	Send a SYN-ACK after having received a SYN.
819
 *	This still operates on a request_sock only, not on a big
L
Linus Torvalds 已提交
820 821
 *	socket.
 */
822 823
static int tcp_v4_send_synack(struct sock *sk, struct dst_entry *dst,
			      struct request_sock *req,
824 825
			      u16 queue_mapping,
			      bool nocache)
L
Linus Torvalds 已提交
826
{
827
	const struct inet_request_sock *ireq = inet_rsk(req);
828
	struct flowi4 fl4;
L
Linus Torvalds 已提交
829 830 831 832
	int err = -1;
	struct sk_buff * skb;

	/* First, grab a route. */
833
	if (!dst && (dst = inet_csk_route_req(sk, &fl4, req)) == NULL)
834
		return -1;
L
Linus Torvalds 已提交
835

C
Christoph Paasch 已提交
836
	skb = tcp_make_synack(sk, dst, req, NULL);
L
Linus Torvalds 已提交
837 838

	if (skb) {
839
		__tcp_v4_send_check(skb, ireq->loc_addr, ireq->rmt_addr);
L
Linus Torvalds 已提交
840

841
		skb_set_queue_mapping(skb, queue_mapping);
842 843 844
		err = ip_build_and_send_pkt(skb, sk, ireq->loc_addr,
					    ireq->rmt_addr,
					    ireq->opt);
845
		err = net_xmit_eval(err);
846 847
		if (!tcp_rsk(req)->snt_synack && !err)
			tcp_rsk(req)->snt_synack = tcp_time_stamp;
L
Linus Torvalds 已提交
848 849 850 851 852
	}

	return err;
}

C
Christoph Paasch 已提交
853
static int tcp_v4_rtx_synack(struct sock *sk, struct request_sock *req)
854
{
C
Christoph Paasch 已提交
855
	int res = tcp_v4_send_synack(sk, NULL, req, 0, false);
856 857 858 859

	if (!res)
		TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
	return res;
860 861
}

L
Linus Torvalds 已提交
862
/*
863
 *	IPv4 request_sock destructor.
L
Linus Torvalds 已提交
864
 */
865
static void tcp_v4_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
866
{
J
Jesper Juhl 已提交
867
	kfree(inet_rsk(req)->opt);
L
Linus Torvalds 已提交
868 869
}

870
/*
E
Eric Dumazet 已提交
871
 * Return true if a syncookie should be sent
872
 */
E
Eric Dumazet 已提交
873
bool tcp_syn_flood_action(struct sock *sk,
874 875
			 const struct sk_buff *skb,
			 const char *proto)
L
Linus Torvalds 已提交
876
{
877
	const char *msg = "Dropping request";
E
Eric Dumazet 已提交
878
	bool want_cookie = false;
879 880 881
	struct listen_sock *lopt;


L
Linus Torvalds 已提交
882

883
#ifdef CONFIG_SYN_COOKIES
884
	if (sysctl_tcp_syncookies) {
885
		msg = "Sending cookies";
E
Eric Dumazet 已提交
886
		want_cookie = true;
887 888
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDOCOOKIES);
	} else
889
#endif
890 891 892 893 894
		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;
895
		pr_info("%s: Possible SYN flooding on port %d. %s.  Check SNMP counters.\n",
896 897 898
			proto, ntohs(tcp_hdr(skb)->dest), msg);
	}
	return want_cookie;
899
}
900
EXPORT_SYMBOL(tcp_syn_flood_action);
L
Linus Torvalds 已提交
901 902

/*
903
 * Save and compile IPv4 options into the request_sock if needed.
L
Linus Torvalds 已提交
904
 */
905
static struct ip_options_rcu *tcp_v4_save_options(struct sk_buff *skb)
L
Linus Torvalds 已提交
906
{
907 908
	const struct ip_options *opt = &(IPCB(skb)->opt);
	struct ip_options_rcu *dopt = NULL;
L
Linus Torvalds 已提交
909 910

	if (opt && opt->optlen) {
911 912
		int opt_size = sizeof(*dopt) + opt->optlen;

L
Linus Torvalds 已提交
913 914
		dopt = kmalloc(opt_size, GFP_ATOMIC);
		if (dopt) {
915
			if (ip_options_echo(&dopt->opt, skb)) {
L
Linus Torvalds 已提交
916 917 918 919 920 921 922 923
				kfree(dopt);
				dopt = NULL;
			}
		}
	}
	return dopt;
}

924 925 926 927 928 929 930 931
#ifdef CONFIG_TCP_MD5SIG
/*
 * RFC2385 MD5 checksumming requires a mapping of
 * IP address->MD5 Key.
 * We need to maintain these in the sk structure.
 */

/* Find the Key structure for an address.  */
E
Eric Dumazet 已提交
932 933 934
struct tcp_md5sig_key *tcp_md5_do_lookup(struct sock *sk,
					 const union tcp_md5_addr *addr,
					 int family)
935 936
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
937 938
	struct tcp_md5sig_key *key;
	unsigned int size = sizeof(struct in_addr);
939
	struct tcp_md5sig_info *md5sig;
940

941 942
	/* caller either holds rcu_read_lock() or socket lock */
	md5sig = rcu_dereference_check(tp->md5sig_info,
943 944
				       sock_owned_by_user(sk) ||
				       lockdep_is_held(&sk->sk_lock.slock));
945
	if (!md5sig)
946
		return NULL;
E
Eric Dumazet 已提交
947 948 949 950
#if IS_ENABLED(CONFIG_IPV6)
	if (family == AF_INET6)
		size = sizeof(struct in6_addr);
#endif
951
	hlist_for_each_entry_rcu(key, &md5sig->head, node) {
E
Eric Dumazet 已提交
952 953 954 955
		if (key->family != family)
			continue;
		if (!memcmp(&key->addr, addr, size))
			return key;
956 957 958
	}
	return NULL;
}
E
Eric Dumazet 已提交
959
EXPORT_SYMBOL(tcp_md5_do_lookup);
960 961 962 963

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
964 965 966 967
	union tcp_md5_addr *addr;

	addr = (union tcp_md5_addr *)&inet_sk(addr_sk)->inet_daddr;
	return tcp_md5_do_lookup(sk, addr, AF_INET);
968 969 970
}
EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
971 972
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
973
{
E
Eric Dumazet 已提交
974 975 976 977
	union tcp_md5_addr *addr;

	addr = (union tcp_md5_addr *)&inet_rsk(req)->rmt_addr;
	return tcp_md5_do_lookup(sk, addr, AF_INET);
978 979 980
}

/* This can be called on a newly created socket, from other files */
E
Eric Dumazet 已提交
981 982
int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
		   int family, const u8 *newkey, u8 newkeylen, gfp_t gfp)
983 984
{
	/* Add Key to the list */
985
	struct tcp_md5sig_key *key;
986
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
987
	struct tcp_md5sig_info *md5sig;
988

989
	key = tcp_md5_do_lookup(sk, addr, family);
990 991
	if (key) {
		/* Pre-existing entry - just update that one. */
E
Eric Dumazet 已提交
992
		memcpy(key->key, newkey, newkeylen);
993
		key->keylen = newkeylen;
E
Eric Dumazet 已提交
994 995
		return 0;
	}
996

997 998
	md5sig = rcu_dereference_protected(tp->md5sig_info,
					   sock_owned_by_user(sk));
E
Eric Dumazet 已提交
999 1000 1001
	if (!md5sig) {
		md5sig = kmalloc(sizeof(*md5sig), gfp);
		if (!md5sig)
1002 1003
			return -ENOMEM;

E
Eric Dumazet 已提交
1004 1005
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
		INIT_HLIST_HEAD(&md5sig->head);
1006
		rcu_assign_pointer(tp->md5sig_info, md5sig);
E
Eric Dumazet 已提交
1007
	}
1008

1009
	key = sock_kmalloc(sk, sizeof(*key), gfp);
E
Eric Dumazet 已提交
1010 1011
	if (!key)
		return -ENOMEM;
1012
	if (!tcp_alloc_md5sig_pool()) {
1013
		sock_kfree_s(sk, key, sizeof(*key));
E
Eric Dumazet 已提交
1014
		return -ENOMEM;
1015
	}
E
Eric Dumazet 已提交
1016 1017 1018 1019 1020 1021 1022 1023

	memcpy(key->key, newkey, newkeylen);
	key->keylen = newkeylen;
	key->family = family;
	memcpy(&key->addr, addr,
	       (family == AF_INET6) ? sizeof(struct in6_addr) :
				      sizeof(struct in_addr));
	hlist_add_head_rcu(&key->node, &md5sig->head);
1024 1025
	return 0;
}
E
Eric Dumazet 已提交
1026
EXPORT_SYMBOL(tcp_md5_do_add);
1027

E
Eric Dumazet 已提交
1028
int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr, int family)
1029
{
E
Eric Dumazet 已提交
1030 1031
	struct tcp_md5sig_key *key;

1032
	key = tcp_md5_do_lookup(sk, addr, family);
E
Eric Dumazet 已提交
1033 1034 1035
	if (!key)
		return -ENOENT;
	hlist_del_rcu(&key->node);
1036
	atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1037 1038
	kfree_rcu(key, rcu);
	return 0;
1039
}
E
Eric Dumazet 已提交
1040
EXPORT_SYMBOL(tcp_md5_do_del);
1041

1042
static void tcp_clear_md5_list(struct sock *sk)
1043 1044
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
1045
	struct tcp_md5sig_key *key;
1046
	struct hlist_node *n;
1047
	struct tcp_md5sig_info *md5sig;
1048

1049 1050
	md5sig = rcu_dereference_protected(tp->md5sig_info, 1);

1051
	hlist_for_each_entry_safe(key, n, &md5sig->head, node) {
E
Eric Dumazet 已提交
1052
		hlist_del_rcu(&key->node);
1053
		atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1054
		kfree_rcu(key, rcu);
1055 1056 1057
	}
}

1058 1059
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
1060 1061 1062 1063 1064 1065 1066
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;

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

1067
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
1068 1069 1070 1071 1072
		return -EFAULT;

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

1073
	if (!cmd.tcpm_key || !cmd.tcpm_keylen)
E
Eric Dumazet 已提交
1074 1075
		return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin->sin_addr.s_addr,
				      AF_INET);
1076 1077 1078 1079

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

E
Eric Dumazet 已提交
1080 1081 1082
	return tcp_md5_do_add(sk, (union tcp_md5_addr *)&sin->sin_addr.s_addr,
			      AF_INET, cmd.tcpm_key, cmd.tcpm_keylen,
			      GFP_KERNEL);
1083 1084
}

1085 1086
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1087 1088
{
	struct tcp4_pseudohdr *bp;
1089
	struct scatterlist sg;
1090 1091 1092 1093

	bp = &hp->md5_blk.ip4;

	/*
1094
	 * 1. the TCP pseudo-header (in the order: source IP address,
1095 1096 1097 1098 1099 1100
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1101
	bp->protocol = IPPROTO_TCP;
1102
	bp->len = cpu_to_be16(nbytes);
1103

1104 1105 1106 1107
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}

E
Eric Dumazet 已提交
1108
static int tcp_v4_md5_hash_hdr(char *md5_hash, const struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1109
			       __be32 daddr, __be32 saddr, const struct tcphdr *th)
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127
{
	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))
1128 1129 1130 1131
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1132

1133 1134 1135 1136
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1137
	return 1;
1138 1139
}

1140
int tcp_v4_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1141 1142
			const struct sock *sk, const struct request_sock *req,
			const struct sk_buff *skb)
1143
{
1144 1145
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
E
Eric Dumazet 已提交
1146
	const struct tcphdr *th = tcp_hdr(skb);
1147 1148 1149
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1150 1151
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1152 1153 1154
	} else if (req) {
		saddr = inet_rsk(req)->loc_addr;
		daddr = inet_rsk(req)->rmt_addr;
1155
	} else {
1156 1157 1158
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
1159
	}
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187

	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;
1188
}
1189
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1190

E
Eric Dumazet 已提交
1191
static bool tcp_v4_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
1192 1193 1194 1195 1196 1197 1198 1199 1200
{
	/*
	 * 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.
	 */
1201
	const __u8 *hash_location = NULL;
1202
	struct tcp_md5sig_key *hash_expected;
1203
	const struct iphdr *iph = ip_hdr(skb);
1204
	const struct tcphdr *th = tcp_hdr(skb);
1205 1206 1207
	int genhash;
	unsigned char newhash[16];

E
Eric Dumazet 已提交
1208 1209
	hash_expected = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&iph->saddr,
					  AF_INET);
1210
	hash_location = tcp_parse_md5sig_option(th);
1211 1212 1213

	/* We've parsed the options - do we have a hash? */
	if (!hash_expected && !hash_location)
E
Eric Dumazet 已提交
1214
		return false;
1215 1216

	if (hash_expected && !hash_location) {
1217
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
E
Eric Dumazet 已提交
1218
		return true;
1219 1220 1221
	}

	if (!hash_expected && hash_location) {
1222
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
E
Eric Dumazet 已提交
1223
		return true;
1224 1225 1226 1227 1228
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1229 1230 1231
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1232 1233

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
1234 1235 1236 1237 1238
		net_info_ratelimited("MD5 Hash failed for (%pI4, %d)->(%pI4, %d)%s\n",
				     &iph->saddr, ntohs(th->source),
				     &iph->daddr, ntohs(th->dest),
				     genhash ? " tcp_v4_calc_md5_hash failed"
				     : "");
E
Eric Dumazet 已提交
1239
		return true;
1240
	}
E
Eric Dumazet 已提交
1241
	return false;
1242 1243 1244 1245
}

#endif

1246
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1247
	.family		=	PF_INET,
1248
	.obj_size	=	sizeof(struct tcp_request_sock),
1249
	.rtx_syn_ack	=	tcp_v4_rtx_synack,
1250 1251
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1252
	.send_reset	=	tcp_v4_send_reset,
1253
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
1254 1255
};

1256
#ifdef CONFIG_TCP_MD5SIG
1257
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1258
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1259
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1260
};
1261
#endif
1262

1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
static bool tcp_fastopen_check(struct sock *sk, struct sk_buff *skb,
			       struct request_sock *req,
			       struct tcp_fastopen_cookie *foc,
			       struct tcp_fastopen_cookie *valid_foc)
{
	bool skip_cookie = false;
	struct fastopen_queue *fastopenq;

	if (likely(!fastopen_cookie_present(foc))) {
		/* See include/net/tcp.h for the meaning of these knobs */
		if ((sysctl_tcp_fastopen & TFO_SERVER_ALWAYS) ||
		    ((sysctl_tcp_fastopen & TFO_SERVER_COOKIE_NOT_REQD) &&
		    (TCP_SKB_CB(skb)->end_seq != TCP_SKB_CB(skb)->seq + 1)))
			skip_cookie = true; /* no cookie to validate */
		else
			return false;
	}
	fastopenq = inet_csk(sk)->icsk_accept_queue.fastopenq;
	/* A FO option is present; bump the counter. */
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPFASTOPENPASSIVE);

	/* Make sure the listener has enabled fastopen, and we don't
	 * exceed the max # of pending TFO requests allowed before trying
	 * to validating the cookie in order to avoid burning CPU cycles
	 * unnecessarily.
	 *
	 * XXX (TFO) - The implication of checking the max_qlen before
	 * processing a cookie request is that clients can't differentiate
	 * between qlen overflow causing Fast Open to be disabled
	 * temporarily vs a server not supporting Fast Open at all.
	 */
	if ((sysctl_tcp_fastopen & TFO_SERVER_ENABLE) == 0 ||
	    fastopenq == NULL || fastopenq->max_qlen == 0)
		return false;

	if (fastopenq->qlen >= fastopenq->max_qlen) {
		struct request_sock *req1;
		spin_lock(&fastopenq->lock);
		req1 = fastopenq->rskq_rst_head;
		if ((req1 == NULL) || time_after(req1->expires, jiffies)) {
			spin_unlock(&fastopenq->lock);
			NET_INC_STATS_BH(sock_net(sk),
			    LINUX_MIB_TCPFASTOPENLISTENOVERFLOW);
			/* Avoid bumping LINUX_MIB_TCPFASTOPENPASSIVEFAIL*/
			foc->len = -1;
			return false;
		}
		fastopenq->rskq_rst_head = req1->dl_next;
		fastopenq->qlen--;
		spin_unlock(&fastopenq->lock);
		reqsk_free(req1);
	}
	if (skip_cookie) {
		tcp_rsk(req)->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
		return true;
	}
	if (foc->len == TCP_FASTOPEN_COOKIE_SIZE) {
		if ((sysctl_tcp_fastopen & TFO_SERVER_COOKIE_NOT_CHKED) == 0) {
			tcp_fastopen_cookie_gen(ip_hdr(skb)->saddr, valid_foc);
			if ((valid_foc->len != TCP_FASTOPEN_COOKIE_SIZE) ||
			    memcmp(&foc->val[0], &valid_foc->val[0],
			    TCP_FASTOPEN_COOKIE_SIZE) != 0)
				return false;
			valid_foc->len = -1;
		}
		/* Acknowledge the data received from the peer. */
		tcp_rsk(req)->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
		return true;
	} else if (foc->len == 0) { /* Client requesting a cookie */
		tcp_fastopen_cookie_gen(ip_hdr(skb)->saddr, valid_foc);
		NET_INC_STATS_BH(sock_net(sk),
		    LINUX_MIB_TCPFASTOPENCOOKIEREQD);
	} else {
		/* Client sent a cookie with wrong size. Treat it
		 * the same as invalid and return a valid one.
		 */
		tcp_fastopen_cookie_gen(ip_hdr(skb)->saddr, valid_foc);
	}
	return false;
}

static int tcp_v4_conn_req_fastopen(struct sock *sk,
				    struct sk_buff *skb,
				    struct sk_buff *skb_synack,
C
Christoph Paasch 已提交
1347
				    struct request_sock *req)
1348 1349 1350 1351 1352
{
	struct tcp_sock *tp = tcp_sk(sk);
	struct request_sock_queue *queue = &inet_csk(sk)->icsk_accept_queue;
	const struct inet_request_sock *ireq = inet_rsk(req);
	struct sock *child;
1353
	int err;
1354

1355 1356
	req->num_retrans = 0;
	req->num_timeout = 0;
1357 1358 1359 1360 1361 1362 1363 1364 1365
	req->sk = NULL;

	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL);
	if (child == NULL) {
		NET_INC_STATS_BH(sock_net(sk),
				 LINUX_MIB_TCPFASTOPENPASSIVEFAIL);
		kfree_skb(skb_synack);
		return -1;
	}
1366 1367 1368 1369 1370
	err = ip_build_and_send_pkt(skb_synack, sk, ireq->loc_addr,
				    ireq->rmt_addr, ireq->opt);
	err = net_xmit_eval(err);
	if (!err)
		tcp_rsk(req)->snt_synack = tcp_time_stamp;
1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430
	/* XXX (TFO) - is it ok to ignore error and continue? */

	spin_lock(&queue->fastopenq->lock);
	queue->fastopenq->qlen++;
	spin_unlock(&queue->fastopenq->lock);

	/* Initialize the child socket. Have to fix some values to take
	 * into account the child is a Fast Open socket and is created
	 * only out of the bits carried in the SYN packet.
	 */
	tp = tcp_sk(child);

	tp->fastopen_rsk = req;
	/* Do a hold on the listner sk so that if the listener is being
	 * closed, the child that has been accepted can live on and still
	 * access listen_lock.
	 */
	sock_hold(sk);
	tcp_rsk(req)->listener = sk;

	/* RFC1323: The window in SYN & SYN/ACK segments is never
	 * scaled. So correct it appropriately.
	 */
	tp->snd_wnd = ntohs(tcp_hdr(skb)->window);

	/* Activate the retrans timer so that SYNACK can be retransmitted.
	 * The request socket is not added to the SYN table of the parent
	 * because it's been added to the accept queue directly.
	 */
	inet_csk_reset_xmit_timer(child, ICSK_TIME_RETRANS,
	    TCP_TIMEOUT_INIT, TCP_RTO_MAX);

	/* Add the child socket directly into the accept queue */
	inet_csk_reqsk_queue_add(sk, req, child);

	/* Now finish processing the fastopen child socket. */
	inet_csk(child)->icsk_af_ops->rebuild_header(child);
	tcp_init_congestion_control(child);
	tcp_mtup_init(child);
	tcp_init_buffer_space(child);
	tcp_init_metrics(child);

	/* Queue the data carried in the SYN packet. We need to first
	 * bump skb's refcnt because the caller will attempt to free it.
	 *
	 * XXX (TFO) - we honor a zero-payload TFO request for now.
	 * (Any reason not to?)
	 */
	if (TCP_SKB_CB(skb)->end_seq == TCP_SKB_CB(skb)->seq + 1) {
		/* Don't queue the skb if there is no payload in SYN.
		 * XXX (TFO) - How about SYN+FIN?
		 */
		tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
	} else {
		skb = skb_get(skb);
		skb_dst_drop(skb);
		__skb_pull(skb, tcp_hdr(skb)->doff * 4);
		skb_set_owner_r(skb, child);
		__skb_queue_tail(&child->sk_receive_queue, skb);
		tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
Y
Yuchung Cheng 已提交
1431
		tp->syn_data_acked = 1;
1432 1433 1434 1435 1436 1437 1438 1439
	}
	sk->sk_data_ready(sk, 0);
	bh_unlock_sock(child);
	sock_put(child);
	WARN_ON(req->sk == NULL);
	return 0;
}

L
Linus Torvalds 已提交
1440 1441 1442
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
	struct tcp_options_received tmp_opt;
1443
	struct request_sock *req;
1444
	struct inet_request_sock *ireq;
1445
	struct tcp_sock *tp = tcp_sk(sk);
1446
	struct dst_entry *dst = NULL;
1447 1448
	__be32 saddr = ip_hdr(skb)->saddr;
	__be32 daddr = ip_hdr(skb)->daddr;
L
Linus Torvalds 已提交
1449
	__u32 isn = TCP_SKB_CB(skb)->when;
E
Eric Dumazet 已提交
1450
	bool want_cookie = false;
1451 1452 1453 1454 1455
	struct flowi4 fl4;
	struct tcp_fastopen_cookie foc = { .len = -1 };
	struct tcp_fastopen_cookie valid_foc = { .len = -1 };
	struct sk_buff *skb_synack;
	int do_fastopen;
L
Linus Torvalds 已提交
1456 1457

	/* Never answer to SYNs send to broadcast or multicast */
E
Eric Dumazet 已提交
1458
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1459 1460 1461 1462 1463 1464
		goto drop;

	/* TW buckets are converted to open requests without
	 * limitations, they conserve resources and peer is
	 * evidently real one.
	 */
1465
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
1466 1467 1468
		want_cookie = tcp_syn_flood_action(sk, skb, "TCP");
		if (!want_cookie)
			goto drop;
L
Linus Torvalds 已提交
1469 1470 1471 1472 1473 1474 1475
	}

	/* 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.
	 */
1476 1477
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1) {
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
L
Linus Torvalds 已提交
1478
		goto drop;
1479
	}
L
Linus Torvalds 已提交
1480

1481
	req = inet_reqsk_alloc(&tcp_request_sock_ops);
L
Linus Torvalds 已提交
1482 1483 1484
	if (!req)
		goto drop;

1485 1486 1487 1488
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv4_ops;
#endif

L
Linus Torvalds 已提交
1489
	tcp_clear_options(&tmp_opt);
1490
	tmp_opt.mss_clamp = TCP_MSS_DEFAULT;
1491
	tmp_opt.user_mss  = tp->rx_opt.user_mss;
C
Christoph Paasch 已提交
1492
	tcp_parse_options(skb, &tmp_opt, 0, want_cookie ? NULL : &foc);
L
Linus Torvalds 已提交
1493

1494
	if (want_cookie && !tmp_opt.saw_tstamp)
L
Linus Torvalds 已提交
1495 1496 1497 1498 1499
		tcp_clear_options(&tmp_opt);

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

1500 1501 1502 1503
	ireq = inet_rsk(req);
	ireq->loc_addr = daddr;
	ireq->rmt_addr = saddr;
	ireq->no_srccheck = inet_sk(sk)->transparent;
1504
	ireq->opt = tcp_v4_save_options(skb);
1505

1506
	if (security_inet_conn_request(sk, skb, req))
1507
		goto drop_and_free;
1508

1509
	if (!want_cookie || tmp_opt.tstamp_ok)
1510
		TCP_ECN_create_request(req, skb, sock_net(sk));
L
Linus Torvalds 已提交
1511 1512 1513

	if (want_cookie) {
		isn = cookie_v4_init_sequence(sk, skb, &req->mss);
1514
		req->cookie_ts = tmp_opt.tstamp_ok;
L
Linus Torvalds 已提交
1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525
	} else if (!isn) {
		/* VJ's idea. We save last timestamp seen
		 * from the destination in peer table, when entering
		 * state TIME-WAIT, and check against it before
		 * accepting new connection request.
		 *
		 * If "isn" is not zero, this request hit alive
		 * timewait bucket, so that all the necessary checks
		 * are made in the function processing timewait state.
		 */
		if (tmp_opt.saw_tstamp &&
1526
		    tcp_death_row.sysctl_tw_recycle &&
1527
		    (dst = inet_csk_route_req(sk, &fl4, req)) != NULL &&
1528 1529
		    fl4.daddr == saddr) {
			if (!tcp_peer_is_proven(req, dst, true)) {
1530
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1531
				goto drop_and_release;
L
Linus Torvalds 已提交
1532 1533 1534 1535
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
1536
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
L
Linus Torvalds 已提交
1537
			  (sysctl_max_syn_backlog >> 2)) &&
1538
			 !tcp_peer_is_proven(req, dst, false)) {
L
Linus Torvalds 已提交
1539 1540 1541 1542 1543 1544 1545
			/* 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.
			 */
1546
			LIMIT_NETDEBUG(KERN_DEBUG pr_fmt("drop open request from %pI4/%u\n"),
1547
				       &saddr, ntohs(tcp_hdr(skb)->source));
1548
			goto drop_and_release;
L
Linus Torvalds 已提交
1549 1550
		}

1551
		isn = tcp_v4_init_sequence(skb);
L
Linus Torvalds 已提交
1552
	}
1553
	tcp_rsk(req)->snt_isn = isn;
L
Linus Torvalds 已提交
1554

1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589
	if (dst == NULL) {
		dst = inet_csk_route_req(sk, &fl4, req);
		if (dst == NULL)
			goto drop_and_free;
	}
	do_fastopen = tcp_fastopen_check(sk, skb, req, &foc, &valid_foc);

	/* We don't call tcp_v4_send_synack() directly because we need
	 * to make sure a child socket can be created successfully before
	 * sending back synack!
	 *
	 * XXX (TFO) - Ideally one would simply call tcp_v4_send_synack()
	 * (or better yet, call tcp_send_synack() in the child context
	 * directly, but will have to fix bunch of other code first)
	 * after syn_recv_sock() except one will need to first fix the
	 * latter to remove its dependency on the current implementation
	 * of tcp_v4_send_synack()->tcp_select_initial_window().
	 */
	skb_synack = tcp_make_synack(sk, dst, req,
	    fastopen_cookie_present(&valid_foc) ? &valid_foc : NULL);

	if (skb_synack) {
		__tcp_v4_send_check(skb_synack, ireq->loc_addr, ireq->rmt_addr);
		skb_set_queue_mapping(skb_synack, skb_get_queue_mapping(skb));
	} else
		goto drop_and_free;

	if (likely(!do_fastopen)) {
		int err;
		err = ip_build_and_send_pkt(skb_synack, sk, ireq->loc_addr,
		     ireq->rmt_addr, ireq->opt);
		err = net_xmit_eval(err);
		if (err || want_cookie)
			goto drop_and_free;

1590
		tcp_rsk(req)->snt_synack = tcp_time_stamp;
1591 1592 1593 1594 1595 1596
		tcp_rsk(req)->listener = NULL;
		/* Add the request_sock to the SYN table */
		inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
		if (fastopen_cookie_present(&foc) && foc.len != 0)
			NET_INC_STATS_BH(sock_net(sk),
			    LINUX_MIB_TCPFASTOPENPASSIVEFAIL);
C
Christoph Paasch 已提交
1597
	} else if (tcp_v4_conn_req_fastopen(sk, skb, skb_synack, req))
L
Linus Torvalds 已提交
1598 1599 1600 1601
		goto drop_and_free;

	return 0;

1602 1603
drop_and_release:
	dst_release(dst);
L
Linus Torvalds 已提交
1604
drop_and_free:
1605
	reqsk_free(req);
L
Linus Torvalds 已提交
1606
drop:
1607
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1608 1609
	return 0;
}
E
Eric Dumazet 已提交
1610
EXPORT_SYMBOL(tcp_v4_conn_request);
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616 1617


/*
 * 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,
1618
				  struct request_sock *req,
L
Linus Torvalds 已提交
1619 1620
				  struct dst_entry *dst)
{
1621
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1622 1623 1624
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1625 1626 1627
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
1628
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
1629 1630 1631 1632 1633 1634

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (!newsk)
1635
		goto exit_nonewsk;
L
Linus Torvalds 已提交
1636

1637
	newsk->sk_gso_type = SKB_GSO_TCPV4;
1638
	inet_sk_rx_dst_set(newsk, skb);
L
Linus Torvalds 已提交
1639 1640 1641

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1642
	ireq		      = inet_rsk(req);
E
Eric Dumazet 已提交
1643 1644 1645
	newinet->inet_daddr   = ireq->rmt_addr;
	newinet->inet_rcv_saddr = ireq->loc_addr;
	newinet->inet_saddr	      = ireq->loc_addr;
1646 1647
	inet_opt	      = ireq->opt;
	rcu_assign_pointer(newinet->inet_opt, inet_opt);
1648
	ireq->opt	      = NULL;
1649
	newinet->mc_index     = inet_iif(skb);
1650
	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
1651
	newinet->rcv_tos      = ip_hdr(skb)->tos;
1652
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
1653 1654
	if (inet_opt)
		inet_csk(newsk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
E
Eric Dumazet 已提交
1655
	newinet->inet_id = newtp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
1656

E
Eric Dumazet 已提交
1657 1658 1659 1660 1661 1662 1663
	if (!dst) {
		dst = inet_csk_route_child_sock(sk, newsk, req);
		if (!dst)
			goto put_and_exit;
	} else {
		/* syncookie case : see end of cookie_v4_check() */
	}
1664 1665
	sk_setup_caps(newsk, dst);

J
John Heffner 已提交
1666
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1667
	tcp_sync_mss(newsk, dst_mtu(dst));
1668
	newtp->advmss = dst_metric_advmss(dst);
1669 1670 1671 1672
	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
1673
	tcp_initialize_rcv_mss(newsk);
1674
	tcp_synack_rtt_meas(newsk, req);
1675
	newtp->total_retrans = req->num_retrans;
L
Linus Torvalds 已提交
1676

1677 1678
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1679 1680
	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&newinet->inet_daddr,
				AF_INET);
E
Eric Dumazet 已提交
1681
	if (key != NULL) {
1682 1683 1684 1685 1686 1687
		/*
		 * We're using one, so create a matching key
		 * on the newsk structure. If we fail to get
		 * memory, then we end up not copying the key
		 * across. Shucks.
		 */
E
Eric Dumazet 已提交
1688 1689
		tcp_md5_do_add(newsk, (union tcp_md5_addr *)&newinet->inet_daddr,
			       AF_INET, key->key, key->keylen, GFP_ATOMIC);
E
Eric Dumazet 已提交
1690
		sk_nocaps_add(newsk, NETIF_F_GSO_MASK);
1691 1692 1693
	}
#endif

1694 1695
	if (__inet_inherit_port(sk, newsk) < 0)
		goto put_and_exit;
1696
	__inet_hash_nolisten(newsk, NULL);
L
Linus Torvalds 已提交
1697 1698 1699 1700

	return newsk;

exit_overflow:
1701
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1702 1703
exit_nonewsk:
	dst_release(dst);
L
Linus Torvalds 已提交
1704
exit:
1705
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1706
	return NULL;
1707
put_and_exit:
1708 1709
	inet_csk_prepare_forced_close(newsk);
	tcp_done(newsk);
1710
	goto exit;
L
Linus Torvalds 已提交
1711
}
E
Eric Dumazet 已提交
1712
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
L
Linus Torvalds 已提交
1713 1714 1715

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1716
	struct tcphdr *th = tcp_hdr(skb);
1717
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1718
	struct sock *nsk;
1719
	struct request_sock **prev;
L
Linus Torvalds 已提交
1720
	/* Find possible connection requests. */
1721 1722
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1723
	if (req)
1724
		return tcp_check_req(sk, skb, req, prev, false);
L
Linus Torvalds 已提交
1725

1726
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1727
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1728 1729 1730 1731 1732 1733

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1734
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1735 1736 1737 1738
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
1739
	if (!th->syn)
L
Linus Torvalds 已提交
1740 1741 1742 1743 1744
		sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
#endif
	return sk;
}

1745
static __sum16 tcp_v4_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1746
{
1747 1748
	const struct iphdr *iph = ip_hdr(skb);

1749
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
1750 1751
		if (!tcp_v4_check(skb->len, iph->saddr,
				  iph->daddr, skb->csum)) {
1752
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1753
			return 0;
1754
		}
L
Linus Torvalds 已提交
1755
	}
1756

1757
	skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
1758 1759
				       skb->len, IPPROTO_TCP, 0);

L
Linus Torvalds 已提交
1760
	if (skb->len <= 76) {
1761
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
	}
	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)
{
1777 1778 1779 1780 1781 1782 1783 1784
	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
	 */
1785
	if (tcp_v4_inbound_md5_hash(sk, skb))
1786 1787 1788
		goto discard;
#endif

L
Linus Torvalds 已提交
1789
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1790 1791
		struct dst_entry *dst = sk->sk_rx_dst;

1792
		sock_rps_save_rxhash(sk, skb);
1793
		if (dst) {
E
Eric Dumazet 已提交
1794 1795
			if (inet_sk(sk)->rx_dst_ifindex != skb->skb_iif ||
			    dst->ops->check(dst, 0) == NULL) {
1796 1797 1798 1799
				dst_release(dst);
				sk->sk_rx_dst = NULL;
			}
		}
1800
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
1801
			rsk = sk;
L
Linus Torvalds 已提交
1802
			goto reset;
1803
		}
L
Linus Torvalds 已提交
1804 1805 1806
		return 0;
	}

1807
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1808 1809 1810 1811 1812 1813 1814 1815
		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) {
1816
			sock_rps_save_rxhash(nsk, skb);
1817 1818
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1819
				goto reset;
1820
			}
L
Linus Torvalds 已提交
1821 1822
			return 0;
		}
1823
	} else
1824
		sock_rps_save_rxhash(sk, skb);
1825

1826
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1827
		rsk = sk;
L
Linus Torvalds 已提交
1828
		goto reset;
1829
	}
L
Linus Torvalds 已提交
1830 1831 1832
	return 0;

reset:
1833
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1834 1835 1836 1837 1838 1839 1840 1841 1842 1843
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:
1844
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_CSUMERRORS);
1845
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1846 1847
	goto discard;
}
E
Eric Dumazet 已提交
1848
EXPORT_SYMBOL(tcp_v4_do_rcv);
L
Linus Torvalds 已提交
1849

1850
void tcp_v4_early_demux(struct sk_buff *skb)
D
David S. Miller 已提交
1851 1852 1853 1854 1855 1856
{
	const struct iphdr *iph;
	const struct tcphdr *th;
	struct sock *sk;

	if (skb->pkt_type != PACKET_HOST)
1857
		return;
D
David S. Miller 已提交
1858

1859
	if (!pskb_may_pull(skb, skb_transport_offset(skb) + sizeof(struct tcphdr)))
1860
		return;
D
David S. Miller 已提交
1861 1862

	iph = ip_hdr(skb);
1863
	th = tcp_hdr(skb);
D
David S. Miller 已提交
1864 1865

	if (th->doff < sizeof(struct tcphdr) / 4)
1866
		return;
D
David S. Miller 已提交
1867

1868
	sk = __inet_lookup_established(dev_net(skb->dev), &tcp_hashinfo,
D
David S. Miller 已提交
1869
				       iph->saddr, th->source,
1870
				       iph->daddr, ntohs(th->dest),
E
Eric Dumazet 已提交
1871
				       skb->skb_iif);
D
David S. Miller 已提交
1872 1873 1874 1875 1876
	if (sk) {
		skb->sk = sk;
		skb->destructor = sock_edemux;
		if (sk->sk_state != TCP_TIME_WAIT) {
			struct dst_entry *dst = sk->sk_rx_dst;
E
Eric Dumazet 已提交
1877

D
David S. Miller 已提交
1878 1879
			if (dst)
				dst = dst_check(dst, 0);
1880
			if (dst &&
E
Eric Dumazet 已提交
1881
			    inet_sk(sk)->rx_dst_ifindex == skb->skb_iif)
1882
				skb_dst_set_noref(skb, dst);
D
David S. Miller 已提交
1883 1884 1885 1886
		}
	}
}

E
Eric Dumazet 已提交
1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904
/* Packet is added to VJ-style prequeue for processing in process
 * context, if a reader task is waiting. Apparently, this exciting
 * idea (VJ's mail "Re: query about TCP header on tcp-ip" of 07 Sep 93)
 * failed somewhere. Latency? Burstiness? Well, at least now we will
 * see, why it failed. 8)8)				  --ANK
 *
 */
bool tcp_prequeue(struct sock *sk, struct sk_buff *skb)
{
	struct tcp_sock *tp = tcp_sk(sk);

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

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

1905
	skb_dst_force(skb);
E
Eric Dumazet 已提交
1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931
	__skb_queue_tail(&tp->ucopy.prequeue, skb);
	tp->ucopy.memory += skb->truesize;
	if (tp->ucopy.memory > sk->sk_rcvbuf) {
		struct sk_buff *skb1;

		BUG_ON(sock_owned_by_user(sk));

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

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

L
Linus Torvalds 已提交
1932 1933 1934 1935 1936 1937
/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1938
	const struct iphdr *iph;
1939
	const struct tcphdr *th;
L
Linus Torvalds 已提交
1940 1941
	struct sock *sk;
	int ret;
1942
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1943 1944 1945 1946 1947

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

	/* Count it even if it's bad */
1948
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1949 1950 1951 1952

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

1953
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1954 1955 1956 1957 1958 1959 1960 1961

	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
1962
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1963
	 * So, we defer the checks. */
1964
	if (!skb_csum_unnecessary(skb) && tcp_v4_checksum_init(skb))
1965
		goto csum_error;
L
Linus Torvalds 已提交
1966

1967
	th = tcp_hdr(skb);
1968
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1969 1970 1971 1972 1973
	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;
1974
	TCP_SKB_CB(skb)->ip_dsfield = ipv4_get_dsfield(iph);
L
Linus Torvalds 已提交
1975 1976
	TCP_SKB_CB(skb)->sacked	 = 0;

1977
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1978 1979 1980
	if (!sk)
		goto no_tcp_socket;

E
Eric Dumazet 已提交
1981 1982 1983 1984
process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1985 1986
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1987
		goto discard_and_relse;
1988
	}
1989

L
Linus Torvalds 已提交
1990 1991
	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1992
	nf_reset(skb);
L
Linus Torvalds 已提交
1993

1994
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1995 1996
		goto discard_and_relse;

1997
	sk_mark_ll(sk, skb);
L
Linus Torvalds 已提交
1998 1999
	skb->dev = NULL;

2000
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
2001 2002
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
2003 2004 2005
#ifdef CONFIG_NET_DMA
		struct tcp_sock *tp = tcp_sk(sk);
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
2006
			tp->ucopy.dma_chan = net_dma_find_channel();
C
Chris Leech 已提交
2007
		if (tp->ucopy.dma_chan)
L
Linus Torvalds 已提交
2008
			ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
2009 2010 2011 2012
		else
#endif
		{
			if (!tcp_prequeue(sk, skb))
S
Shan Wei 已提交
2013
				ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
2014
		}
2015 2016
	} else if (unlikely(sk_add_backlog(sk, skb,
					   sk->sk_rcvbuf + sk->sk_sndbuf))) {
Z
Zhu Yi 已提交
2017
		bh_unlock_sock(sk);
2018
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
2019 2020
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031
	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)) {
2032 2033
csum_error:
		TCP_INC_STATS_BH(net, TCP_MIB_CSUMERRORS);
L
Linus Torvalds 已提交
2034
bad_packet:
2035
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
2036
	} else {
2037
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
2038 2039 2040 2041 2042
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
2043
	return 0;
L
Linus Torvalds 已提交
2044 2045 2046 2047 2048 2049 2050

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
2051
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
2052 2053 2054
		goto discard_it;
	}

2055
	if (skb->len < (th->doff << 2)) {
2056
		inet_twsk_put(inet_twsk(sk));
2057 2058 2059 2060 2061
		goto bad_packet;
	}
	if (tcp_checksum_complete(skb)) {
		inet_twsk_put(inet_twsk(sk));
		goto csum_error;
L
Linus Torvalds 已提交
2062
	}
2063
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
2064
	case TCP_TW_SYN: {
2065
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
2066
							&tcp_hashinfo,
2067
							iph->saddr, th->source,
2068
							iph->daddr, th->dest,
2069
							inet_iif(skb));
L
Linus Torvalds 已提交
2070
		if (sk2) {
2071 2072
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087
			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;
}

2088 2089 2090 2091 2092
static struct timewait_sock_ops tcp_timewait_sock_ops = {
	.twsk_obj_size	= sizeof(struct tcp_timewait_sock),
	.twsk_unique	= tcp_twsk_unique,
	.twsk_destructor= tcp_twsk_destructor,
};
L
Linus Torvalds 已提交
2093

2094
void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
E
Eric Dumazet 已提交
2095 2096 2097 2098 2099 2100 2101
{
	struct dst_entry *dst = skb_dst(skb);

	dst_hold(dst);
	sk->sk_rx_dst = dst;
	inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
}
2102
EXPORT_SYMBOL(inet_sk_rx_dst_set);
E
Eric Dumazet 已提交
2103

2104
const struct inet_connection_sock_af_ops ipv4_specific = {
2105 2106 2107
	.queue_xmit	   = ip_queue_xmit,
	.send_check	   = tcp_v4_send_check,
	.rebuild_header	   = inet_sk_rebuild_header,
E
Eric Dumazet 已提交
2108
	.sk_rx_dst_set	   = inet_sk_rx_dst_set,
2109 2110 2111 2112 2113 2114 2115
	.conn_request	   = tcp_v4_conn_request,
	.syn_recv_sock	   = tcp_v4_syn_recv_sock,
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ip_setsockopt,
	.getsockopt	   = ip_getsockopt,
	.addr2sockaddr	   = inet_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in),
2116
	.bind_conflict	   = inet_csk_bind_conflict,
2117
#ifdef CONFIG_COMPAT
2118 2119
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
2120
#endif
L
Linus Torvalds 已提交
2121
};
E
Eric Dumazet 已提交
2122
EXPORT_SYMBOL(ipv4_specific);
L
Linus Torvalds 已提交
2123

2124
#ifdef CONFIG_TCP_MD5SIG
2125
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
2126
	.md5_lookup		= tcp_v4_md5_lookup,
2127
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
2128 2129
	.md5_parse		= tcp_v4_parse_md5_keys,
};
2130
#endif
2131

L
Linus Torvalds 已提交
2132 2133 2134 2135 2136
/* 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)
{
2137
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
2138

2139
	tcp_init_sock(sk);
L
Linus Torvalds 已提交
2140

2141
	icsk->icsk_af_ops = &ipv4_specific;
2142

2143
#ifdef CONFIG_TCP_MD5SIG
2144
	tcp_sk(sk)->af_specific = &tcp_sock_ipv4_specific;
2145
#endif
L
Linus Torvalds 已提交
2146 2147 2148 2149

	return 0;
}

2150
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
2151 2152 2153 2154 2155
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

2156
	tcp_cleanup_congestion_control(sk);
2157

L
Linus Torvalds 已提交
2158
	/* Cleanup up the write buffer. */
2159
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
2160 2161

	/* Cleans up our, hopefully empty, out_of_order_queue. */
2162
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
2163

2164 2165 2166
#ifdef CONFIG_TCP_MD5SIG
	/* Clean up the MD5 key list, if any */
	if (tp->md5sig_info) {
E
Eric Dumazet 已提交
2167
		tcp_clear_md5_list(sk);
2168
		kfree_rcu(tp->md5sig_info, rcu);
2169 2170 2171 2172
		tp->md5sig_info = NULL;
	}
#endif

C
Chris Leech 已提交
2173 2174
#ifdef CONFIG_NET_DMA
	/* Cleans up our sk_async_wait_queue */
2175
	__skb_queue_purge(&sk->sk_async_wait_queue);
C
Chris Leech 已提交
2176 2177
#endif

L
Linus Torvalds 已提交
2178 2179 2180 2181
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
2182
	if (inet_csk(sk)->icsk_bind_hash)
2183
		inet_put_port(sk);
L
Linus Torvalds 已提交
2184

2185
	BUG_ON(tp->fastopen_rsk != NULL);
2186

2187 2188 2189
	/* If socket is aborted during connect operation */
	tcp_free_fastopen_req(tp);

2190
	sk_sockets_allocated_dec(sk);
G
Glauber Costa 已提交
2191
	sock_release_memcg(sk);
L
Linus Torvalds 已提交
2192 2193 2194 2195 2196 2197
}
EXPORT_SYMBOL(tcp_v4_destroy_sock);

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

2198
static inline struct inet_timewait_sock *tw_head(struct hlist_nulls_head *head)
L
Linus Torvalds 已提交
2199
{
2200
	return hlist_nulls_empty(head) ? NULL :
2201
		list_entry(head->first, struct inet_timewait_sock, tw_node);
L
Linus Torvalds 已提交
2202 2203
}

2204
static inline struct inet_timewait_sock *tw_next(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
2205
{
2206 2207
	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
2208 2209
}

2210 2211 2212 2213 2214
/*
 * 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
2215 2216
static void *listening_get_next(struct seq_file *seq, void *cur)
{
2217
	struct inet_connection_sock *icsk;
2218
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
2219
	struct sock *sk = cur;
2220
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
2221
	struct tcp_iter_state *st = seq->private;
2222
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2223 2224

	if (!sk) {
2225
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
2226
		spin_lock_bh(&ilb->lock);
2227
		sk = sk_nulls_head(&ilb->head);
2228
		st->offset = 0;
L
Linus Torvalds 已提交
2229 2230
		goto get_sk;
	}
2231
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
2232
	++st->num;
2233
	++st->offset;
L
Linus Torvalds 已提交
2234 2235

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
2236
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
2237

2238
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
2239 2240 2241
		req = req->dl_next;
		while (1) {
			while (req) {
2242
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
2243 2244 2245 2246 2247
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
2248
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
2249 2250
				break;
get_req:
2251
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
2252
		}
E
Eric Dumazet 已提交
2253
		sk	  = sk_nulls_next(st->syn_wait_sk);
L
Linus Torvalds 已提交
2254
		st->state = TCP_SEQ_STATE_LISTENING;
2255
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2256
	} else {
2257
		icsk = inet_csk(sk);
2258 2259
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
2260
			goto start_req;
2261
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
E
Eric Dumazet 已提交
2262
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2263 2264
	}
get_sk:
2265
	sk_nulls_for_each_from(sk, node) {
2266 2267 2268
		if (!net_eq(sock_net(sk), net))
			continue;
		if (sk->sk_family == st->family) {
L
Linus Torvalds 已提交
2269 2270 2271
			cur = sk;
			goto out;
		}
2272
		icsk = inet_csk(sk);
2273 2274
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
2275 2276 2277 2278 2279 2280 2281
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;
		}
2282
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2283
	}
2284
	spin_unlock_bh(&ilb->lock);
2285
	st->offset = 0;
2286
	if (++st->bucket < INET_LHTABLE_SIZE) {
2287 2288
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
2289
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
2290 2291 2292 2293 2294 2295 2296 2297 2298
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
2299 2300 2301 2302 2303 2304
	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
2305 2306 2307 2308 2309 2310 2311 2312

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

E
Eric Dumazet 已提交
2313
static inline bool empty_bucket(struct tcp_iter_state *st)
2314
{
2315 2316
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain) &&
		hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].twchain);
2317 2318
}

2319 2320 2321 2322
/*
 * 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
2323 2324
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
2325
	struct tcp_iter_state *st = seq->private;
2326
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2327 2328
	void *rc = NULL;

2329 2330
	st->offset = 0;
	for (; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
2331
		struct sock *sk;
2332
		struct hlist_nulls_node *node;
2333
		struct inet_timewait_sock *tw;
2334
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
2335

2336 2337 2338 2339
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

2340
		spin_lock_bh(lock);
2341
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
2342
			if (sk->sk_family != st->family ||
2343
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2344 2345 2346 2347 2348 2349
				continue;
			}
			rc = sk;
			goto out;
		}
		st->state = TCP_SEQ_STATE_TIME_WAIT;
2350
		inet_twsk_for_each(tw, node,
2351
				   &tcp_hashinfo.ehash[st->bucket].twchain) {
2352
			if (tw->tw_family != st->family ||
2353
			    !net_eq(twsk_net(tw), net)) {
L
Linus Torvalds 已提交
2354 2355 2356 2357 2358
				continue;
			}
			rc = tw;
			goto out;
		}
2359
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2360 2361 2362 2363 2364 2365 2366 2367 2368
		st->state = TCP_SEQ_STATE_ESTABLISHED;
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2369
	struct inet_timewait_sock *tw;
2370
	struct hlist_nulls_node *node;
J
Jianjun Kong 已提交
2371
	struct tcp_iter_state *st = seq->private;
2372
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2373 2374

	++st->num;
2375
	++st->offset;
L
Linus Torvalds 已提交
2376 2377 2378 2379 2380

	if (st->state == TCP_SEQ_STATE_TIME_WAIT) {
		tw = cur;
		tw = tw_next(tw);
get_tw:
2381
		while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
L
Linus Torvalds 已提交
2382 2383 2384 2385 2386 2387
			tw = tw_next(tw);
		}
		if (tw) {
			cur = tw;
			goto out;
		}
2388
		spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2389 2390
		st->state = TCP_SEQ_STATE_ESTABLISHED;

2391
		/* Look for next non empty bucket */
2392
		st->offset = 0;
2393
		while (++st->bucket <= tcp_hashinfo.ehash_mask &&
2394 2395
				empty_bucket(st))
			;
2396
		if (st->bucket > tcp_hashinfo.ehash_mask)
2397 2398
			return NULL;

2399
		spin_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
2400
		sk = sk_nulls_head(&tcp_hashinfo.ehash[st->bucket].chain);
L
Linus Torvalds 已提交
2401
	} else
2402
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2403

2404
	sk_nulls_for_each_from(sk, node) {
2405
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
L
Linus Torvalds 已提交
2406 2407 2408 2409
			goto found;
	}

	st->state = TCP_SEQ_STATE_TIME_WAIT;
2410
	tw = tw_head(&tcp_hashinfo.ehash[st->bucket].twchain);
L
Linus Torvalds 已提交
2411 2412 2413 2414 2415 2416 2417 2418 2419
	goto get_tw;
found:
	cur = sk;
out:
	return cur;
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
2420 2421 2422 2423 2424
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	rc = established_get_first(seq);
L
Linus Torvalds 已提交
2425 2426 2427 2428

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2429
	}
L
Linus Torvalds 已提交
2430 2431 2432 2433 2434 2435
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2436
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448

	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;
}

2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483
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
2484 2485
static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2486
	struct tcp_iter_state *st = seq->private;
2487 2488 2489 2490 2491 2492 2493 2494
	void *rc;

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

L
Linus Torvalds 已提交
2495 2496
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
2497 2498 2499 2500 2501 2502 2503
	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
2504 2505 2506 2507
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2508
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521
	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;
2522 2523
			st->bucket = 0;
			st->offset = 0;
L
Linus Torvalds 已提交
2524 2525 2526 2527 2528 2529 2530 2531 2532 2533
			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;
2534
	st->last_pos = *pos;
L
Linus Torvalds 已提交
2535 2536 2537 2538 2539
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2540
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2541 2542 2543 2544

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2545 2546
			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
2547 2548 2549
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2550
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2551 2552 2553 2554
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2555
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2556 2557 2558 2559
		break;
	}
}

2560
int tcp_seq_open(struct inode *inode, struct file *file)
L
Linus Torvalds 已提交
2561
{
A
Al Viro 已提交
2562
	struct tcp_seq_afinfo *afinfo = PDE_DATA(inode);
L
Linus Torvalds 已提交
2563
	struct tcp_iter_state *s;
2564
	int err;
L
Linus Torvalds 已提交
2565

2566 2567 2568 2569
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2570

2571
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2572
	s->family		= afinfo->family;
2573
	s->last_pos 		= 0;
2574 2575
	return 0;
}
2576
EXPORT_SYMBOL(tcp_seq_open);
2577

2578
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2579 2580 2581 2582
{
	int rc = 0;
	struct proc_dir_entry *p;

2583 2584 2585 2586
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2587
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
2588
			     afinfo->seq_fops, afinfo);
2589
	if (!p)
L
Linus Torvalds 已提交
2590 2591 2592
		rc = -ENOMEM;
	return rc;
}
E
Eric Dumazet 已提交
2593
EXPORT_SYMBOL(tcp_proc_register);
L
Linus Torvalds 已提交
2594

2595
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2596
{
2597
	remove_proc_entry(afinfo->name, net->proc_net);
L
Linus Torvalds 已提交
2598
}
E
Eric Dumazet 已提交
2599
EXPORT_SYMBOL(tcp_proc_unregister);
L
Linus Torvalds 已提交
2600

2601
static void get_openreq4(const struct sock *sk, const struct request_sock *req,
2602
			 struct seq_file *f, int i, kuid_t uid, int *len)
L
Linus Torvalds 已提交
2603
{
2604
	const struct inet_request_sock *ireq = inet_rsk(req);
2605
	long delta = req->expires - jiffies;
L
Linus Torvalds 已提交
2606

2607
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2608
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %pK%n",
L
Linus Torvalds 已提交
2609
		i,
2610
		ireq->loc_addr,
E
Eric Dumazet 已提交
2611
		ntohs(inet_sk(sk)->inet_sport),
2612 2613
		ireq->rmt_addr,
		ntohs(ireq->rmt_port),
L
Linus Torvalds 已提交
2614 2615 2616
		TCP_SYN_RECV,
		0, 0, /* could print option size, but that is af dependent. */
		1,    /* timers active (only the expire timer) */
2617
		jiffies_delta_to_clock_t(delta),
2618
		req->num_timeout,
2619
		from_kuid_munged(seq_user_ns(f), uid),
L
Linus Torvalds 已提交
2620 2621 2622
		0,  /* non standard timer */
		0, /* open_requests have no inode */
		atomic_read(&sk->sk_refcnt),
2623 2624
		req,
		len);
L
Linus Torvalds 已提交
2625 2626
}

2627
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2628 2629 2630
{
	int timer_active;
	unsigned long timer_expires;
2631
	const struct tcp_sock *tp = tcp_sk(sk);
2632
	const struct inet_connection_sock *icsk = inet_csk(sk);
2633
	const struct inet_sock *inet = inet_sk(sk);
2634
	struct fastopen_queue *fastopenq = icsk->icsk_accept_queue.fastopenq;
E
Eric Dumazet 已提交
2635 2636 2637 2638
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
2639
	int rx_queue;
L
Linus Torvalds 已提交
2640

N
Nandita Dukkipati 已提交
2641 2642 2643
	if (icsk->icsk_pending == ICSK_TIME_RETRANS ||
	    icsk->icsk_pending == ICSK_TIME_EARLY_RETRANS ||
	    icsk->icsk_pending == ICSK_TIME_LOSS_PROBE) {
L
Linus Torvalds 已提交
2644
		timer_active	= 1;
2645 2646
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2647
		timer_active	= 4;
2648
		timer_expires	= icsk->icsk_timeout;
2649
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2650
		timer_active	= 2;
2651
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2652 2653 2654 2655 2656
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2657 2658 2659 2660 2661 2662 2663 2664
	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);

2665
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
D
Dan Rosenberg 已提交
2666
			"%08X %5d %8d %lu %d %pK %lu %lu %u %u %d%n",
2667
		i, src, srcp, dest, destp, sk->sk_state,
2668
		tp->write_seq - tp->snd_una,
2669
		rx_queue,
L
Linus Torvalds 已提交
2670
		timer_active,
2671
		jiffies_delta_to_clock_t(timer_expires - jiffies),
2672
		icsk->icsk_retransmits,
2673
		from_kuid_munged(seq_user_ns(f), sock_i_uid(sk)),
2674
		icsk->icsk_probes_out,
2675 2676
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2677 2678
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2679
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2680
		tp->snd_cwnd,
2681 2682 2683
		sk->sk_state == TCP_LISTEN ?
		    (fastopenq ? fastopenq->max_qlen : 0) :
		    (tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh),
2684
		len);
L
Linus Torvalds 已提交
2685 2686
}

2687
static void get_timewait4_sock(const struct inet_timewait_sock *tw,
2688
			       struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2689
{
2690
	__be32 dest, src;
L
Linus Torvalds 已提交
2691
	__u16 destp, srcp;
2692
	long delta = tw->tw_ttd - jiffies;
L
Linus Torvalds 已提交
2693 2694 2695 2696 2697 2698

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

2699
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
D
Dan Rosenberg 已提交
2700
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK%n",
L
Linus Torvalds 已提交
2701
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
2702
		3, jiffies_delta_to_clock_t(delta), 0, 0, 0, 0,
2703
		atomic_read(&tw->tw_refcnt), tw, len);
L
Linus Torvalds 已提交
2704 2705 2706 2707 2708 2709
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2710
	struct tcp_iter_state *st;
2711
	int len;
L
Linus Torvalds 已提交
2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724

	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:
2725
		get_tcp4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2726 2727
		break;
	case TCP_SEQ_STATE_OPENREQ:
2728
		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid, &len);
L
Linus Torvalds 已提交
2729 2730
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
2731
		get_timewait4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2732 2733
		break;
	}
2734
	seq_printf(seq, "%*s\n", TMPSZ - 1 - len, "");
L
Linus Torvalds 已提交
2735 2736 2737 2738
out:
	return 0;
}

2739 2740 2741 2742 2743 2744 2745 2746
static const struct file_operations tcp_afinfo_seq_fops = {
	.owner   = THIS_MODULE,
	.open    = tcp_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
	.release = seq_release_net
};

L
Linus Torvalds 已提交
2747 2748 2749
static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2750
	.seq_fops	= &tcp_afinfo_seq_fops,
2751 2752 2753
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2754 2755
};

2756
static int __net_init tcp4_proc_init_net(struct net *net)
2757 2758 2759 2760
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

2761
static void __net_exit tcp4_proc_exit_net(struct net *net)
2762 2763 2764 2765 2766 2767 2768 2769 2770
{
	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
2771 2772
int __init tcp4_proc_init(void)
{
2773
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2774 2775 2776 2777
}

void tcp4_proc_exit(void)
{
2778
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2779 2780 2781 2782 2783 2784 2785 2786 2787
}
#endif /* CONFIG_PROC_FS */

struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2788
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2789 2790 2791 2792 2793 2794 2795
	.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,
2796 2797
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2798
	.backlog_rcv		= tcp_v4_do_rcv,
E
Eric Dumazet 已提交
2799
	.release_cb		= tcp_release_cb,
2800
	.mtu_reduced		= tcp_v4_mtu_reduced,
2801 2802 2803
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2804 2805
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
2806
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2807 2808 2809 2810 2811 2812
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp_sock),
2813
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2814
	.twsk_prot		= &tcp_timewait_sock_ops,
2815
	.rsk_prot		= &tcp_request_sock_ops,
2816
	.h.hashinfo		= &tcp_hashinfo,
2817
	.no_autobind		= true,
2818 2819 2820 2821
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
A
Andrew Morton 已提交
2822
#ifdef CONFIG_MEMCG_KMEM
G
Glauber Costa 已提交
2823 2824 2825 2826
	.init_cgroup		= tcp_init_cgroup,
	.destroy_cgroup		= tcp_destroy_cgroup,
	.proto_cgroup		= tcp_proto_cgroup,
#endif
L
Linus Torvalds 已提交
2827
};
E
Eric Dumazet 已提交
2828
EXPORT_SYMBOL(tcp_prot);
L
Linus Torvalds 已提交
2829

2830 2831
static int __net_init tcp_sk_init(struct net *net)
{
2832
	net->ipv4.sysctl_tcp_ecn = 2;
2833
	return 0;
2834 2835 2836 2837
}

static void __net_exit tcp_sk_exit(struct net *net)
{
E
Eric W. Biederman 已提交
2838 2839 2840 2841 2842
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2843 2844 2845
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2846 2847 2848
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2849 2850
};

2851
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2852
{
2853
	inet_hashinfo_init(&tcp_hashinfo);
2854
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2855 2856
		panic("Failed to create the TCP control socket.\n");
}