tcp_ipv4.c 63.2 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>
75
#include <net/secure_seq.h>
G
Glauber Costa 已提交
76
#include <net/tcp_memcontrol.h>
77
#include <net/busy_poll.h>
L
Linus Torvalds 已提交
78 79 80 81 82 83 84

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

85 86 87
#include <linux/crypto.h>
#include <linux/scatterlist.h>

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

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

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

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

108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125
int tcp_twsk_unique(struct sock *sk, struct sock *sktw, void *twp)
{
	const struct tcp_timewait_sock *tcptw = tcp_twsk(sktw);
	struct tcp_sock *tp = tcp_sk(sk);

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

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

	   If TW bucket has been already destroyed we fall back to VJ's scheme
	   and use initial timestamp retrieved from peer table.
	 */
	if (tcptw->tw_ts_recent_stamp &&
	    (twp == NULL || (sysctl_tcp_tw_reuse &&
126
			     get_seconds() - tcptw->tw_ts_recent_stamp > 1))) {
127 128 129 130 131 132 133 134 135 136 137 138 139
		tp->write_seq = tcptw->tw_snd_nxt + 65535 + 2;
		if (tp->write_seq == 0)
			tp->write_seq = 1;
		tp->rx_opt.ts_recent	   = tcptw->tw_ts_recent;
		tp->rx_opt.ts_recent_stamp = tcptw->tw_ts_recent_stamp;
		sock_hold(sktw);
		return 1;
	}

	return 0;
}
EXPORT_SYMBOL_GPL(tcp_twsk_unique);

L
Linus Torvalds 已提交
140 141 142
/* This will initiate an outgoing connection. */
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
{
143
	struct sockaddr_in *usin = (struct sockaddr_in *)uaddr;
L
Linus Torvalds 已提交
144 145
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
146
	__be16 orig_sport, orig_dport;
147
	__be32 daddr, nexthop;
148
	struct flowi4 *fl4;
149
	struct rtable *rt;
L
Linus Torvalds 已提交
150
	int err;
151
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
152 153 154 155 156 157 158 159

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

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

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

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

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

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

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

E
Eric Dumazet 已提交
194
	if (tp->rx_opt.ts_recent_stamp && inet->inet_daddr != daddr) {
L
Linus Torvalds 已提交
195 196 197
		/* Reset inherited state */
		tp->rx_opt.ts_recent	   = 0;
		tp->rx_opt.ts_recent_stamp = 0;
P
Pavel Emelyanov 已提交
198 199
		if (likely(!tp->repair))
			tp->write_seq	   = 0;
L
Linus Torvalds 已提交
200 201
	}

202
	if (tcp_death_row.sysctl_tw_recycle &&
203 204
	    !tp->rx_opt.ts_recent_stamp && fl4->daddr == daddr)
		tcp_fetch_timewait_stamp(sk, &rt->dst);
L
Linus Torvalds 已提交
205

E
Eric Dumazet 已提交
206
	inet->inet_dport = usin->sin_port;
207
	sk_daddr_set(sk, daddr);
L
Linus Torvalds 已提交
208

209
	inet_csk(sk)->icsk_ext_hdr_len = 0;
210 211
	if (inet_opt)
		inet_csk(sk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
L
Linus Torvalds 已提交
212

213
	tp->rx_opt.mss_clamp = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
214 215 216 217 218 219 220

	/* 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);
221
	err = inet_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
222 223 224
	if (err)
		goto failure;

225 226
	inet_set_txhash(sk);

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
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
	    ip_sk_accept_pmtu(sk) &&
292
	    inet_csk(sk)->icsk_pmtu_cookie > mtu) {
L
Linus Torvalds 已提交
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
EXPORT_SYMBOL(tcp_v4_mtu_reduced);
L
Linus Torvalds 已提交
304

305 306 307 308
static void do_redirect(struct sk_buff *skb, struct sock *sk)
{
	struct dst_entry *dst = __sk_dst_check(sk, 0);

309
	if (dst)
310
		dst->ops->redirect(dst, sk, skb);
311 312
}

L
Linus Torvalds 已提交
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328
/*
 * 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.
 *
 */

329
void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
L
Linus Torvalds 已提交
330
{
331
	const struct iphdr *iph = (const struct iphdr *)icmp_skb->data;
332
	struct tcphdr *th = (struct tcphdr *)(icmp_skb->data + (iph->ihl << 2));
333
	struct inet_connection_sock *icsk;
L
Linus Torvalds 已提交
334 335
	struct tcp_sock *tp;
	struct inet_sock *inet;
336 337
	const int type = icmp_hdr(icmp_skb)->type;
	const int code = icmp_hdr(icmp_skb)->code;
L
Linus Torvalds 已提交
338
	struct sock *sk;
339
	struct sk_buff *skb;
340 341
	struct request_sock *fastopen;
	__u32 seq, snd_una;
342
	__u32 remaining;
L
Linus Torvalds 已提交
343
	int err;
344
	struct net *net = dev_net(icmp_skb->dev);
L
Linus Torvalds 已提交
345

346
	sk = inet_lookup(net, &tcp_hashinfo, iph->daddr, th->dest,
347
			iph->saddr, th->source, inet_iif(icmp_skb));
L
Linus Torvalds 已提交
348
	if (!sk) {
349
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
350 351 352
		return;
	}
	if (sk->sk_state == TCP_TIME_WAIT) {
353
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
354 355 356 357 358 359
		return;
	}

	bh_lock_sock(sk);
	/* If too many ICMPs get dropped on busy
	 * servers this needs to be solved differently.
360 361
	 * 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
362
	 */
363 364 365 366
	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
367 368 369
	if (sk->sk_state == TCP_CLOSE)
		goto out;

370 371 372 373 374
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto out;
	}

375
	icsk = inet_csk(sk);
L
Linus Torvalds 已提交
376 377
	tp = tcp_sk(sk);
	seq = ntohl(th->seq);
378 379 380
	/* XXX (TFO) - tp->snd_una should be ISN (tcp_create_openreq_child() */
	fastopen = tp->fastopen_rsk;
	snd_una = fastopen ? tcp_rsk(fastopen)->snt_isn : tp->snd_una;
L
Linus Torvalds 已提交
381
	if (sk->sk_state != TCP_LISTEN &&
382
	    !between(seq, snd_una, tp->snd_nxt)) {
383
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
384 385 386 387
		goto out;
	}

	switch (type) {
388 389 390
	case ICMP_REDIRECT:
		do_redirect(icmp_skb, sk);
		goto out;
L
Linus Torvalds 已提交
391 392 393 394 395 396 397 398 399 400 401
	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) */
402 403 404 405 406 407 408
			/* 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;

409
			tp->mtu_info = info;
410
			if (!sock_owned_by_user(sk)) {
411
				tcp_v4_mtu_reduced(sk);
412 413 414 415
			} else {
				if (!test_and_set_bit(TCP_MTU_REDUCED_DEFERRED, &tp->tsq_flags))
					sock_hold(sk);
			}
L
Linus Torvalds 已提交
416 417 418 419
			goto out;
		}

		err = icmp_err_convert[code].errno;
420 421 422 423 424
		/* 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 ||
425
		    !icsk->icsk_backoff || fastopen)
426 427
			break;

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

431
		icsk->icsk_backoff--;
432 433 434
		icsk->icsk_rto = tp->srtt_us ? __tcp_set_rto(tp) :
					       TCP_TIMEOUT_INIT;
		icsk->icsk_rto = inet_csk_rto_backoff(icsk, TCP_RTO_MAX);
435 436 437 438

		skb = tcp_write_queue_head(sk);
		BUG_ON(!skb);

E
Eric Dumazet 已提交
439 440 441
		remaining = icsk->icsk_rto -
			    min(icsk->icsk_rto,
				tcp_time_stamp - tcp_skb_timestamp(skb));
442 443 444 445 446 447 448 449 450 451

		if (remaining) {
			inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
						  remaining, TCP_RTO_MAX);
		} else {
			/* RTO revert clocked out retransmission.
			 * Will retransmit now */
			tcp_retransmit_timer(sk);
		}

L
Linus Torvalds 已提交
452 453 454 455 456 457 458 459 460
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

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

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

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

476
		if (seq != tcp_rsk(req)->snt_isn) {
477
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
478 479 480 481 482 483 484 485 486
			goto out;
		}

		/*
		 * Still in SYN_RECV, just remove it silently.
		 * There is no good way to pass the error to the newly
		 * created socket, and POSIX does not want network
		 * errors returned from accept().
		 */
487
		inet_csk_reqsk_queue_drop(sk, req, prev);
488
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
489 490 491
		goto out;

	case TCP_SYN_SENT:
492 493 494 495 496 497 498
	case TCP_SYN_RECV:
		/* Only in fast or simultaneous open. If a fast open socket is
		 * is already accepted it is treated as a connected one below.
		 */
		if (fastopen && fastopen->sk == NULL)
			break;

L
Linus Torvalds 已提交
499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
		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);
}

540
void __tcp_v4_send_check(struct sk_buff *skb, __be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
541
{
542
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
543

544
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
545
		th->check = ~tcp_v4_check(skb->len, saddr, daddr, 0);
546
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
547
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
548
	} else {
549
		th->check = tcp_v4_check(skb->len, saddr, daddr,
550
					 csum_partial(th,
L
Linus Torvalds 已提交
551 552 553 554 555
						      th->doff << 2,
						      skb->csum));
	}
}

556
/* This routine computes an IPv4 TCP checksum. */
557
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb)
558
{
559
	const struct inet_sock *inet = inet_sk(sk);
560 561 562

	__tcp_v4_send_check(skb, inet->inet_saddr, inet->inet_daddr);
}
E
Eric Dumazet 已提交
563
EXPORT_SYMBOL(tcp_v4_send_check);
564

L
Linus Torvalds 已提交
565 566 567 568 569 570 571 572 573 574 575 576 577
/*
 *	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.
 */

578
static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
579
{
580
	const struct tcphdr *th = tcp_hdr(skb);
581 582 583
	struct {
		struct tcphdr th;
#ifdef CONFIG_TCP_MD5SIG
584
		__be32 opt[(TCPOLEN_MD5SIG_ALIGNED >> 2)];
585 586
#endif
	} rep;
L
Linus Torvalds 已提交
587
	struct ip_reply_arg arg;
588 589
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
590 591 592 593
	const __u8 *hash_location = NULL;
	unsigned char newhash[16];
	int genhash;
	struct sock *sk1 = NULL;
594
#endif
595
	struct net *net;
L
Linus Torvalds 已提交
596 597 598 599 600

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

601 602 603 604
	/* If sk not NULL, it means we did a successful lookup and incoming
	 * route had to be correct. prequeue might have dropped our dst.
	 */
	if (!sk && skb_rtable(skb)->rt_type != RTN_LOCAL)
L
Linus Torvalds 已提交
605 606 607
		return;

	/* Swap the send and the receive. */
608 609 610 611 612
	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
613 614

	if (th->ack) {
615
		rep.th.seq = th->ack_seq;
L
Linus Torvalds 已提交
616
	} else {
617 618 619
		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
620 621
	}

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

626
	net = sk ? sock_net(sk) : dev_net(skb_dst(skb)->dev);
627
#ifdef CONFIG_TCP_MD5SIG
628 629 630 631 632 633 634 635 636
	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.
		 */
637
		sk1 = __inet_lookup_listener(net,
638 639
					     &tcp_hashinfo, ip_hdr(skb)->saddr,
					     th->source, ip_hdr(skb)->daddr,
640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658
					     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;
	}

659 660 661 662 663 664 665 666 667
	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;

668
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
669 670
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
671 672
	}
#endif
673 674
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
675
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
676
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
677
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
678
	/* When socket is gone, all binding information is lost.
A
Alexey Kuznetsov 已提交
679 680
	 * routing might fail in this case. No choice here, if we choose to force
	 * input interface, we will misroute in case of asymmetric route.
681
	 */
A
Alexey Kuznetsov 已提交
682 683
	if (sk)
		arg.bound_dev_if = sk->sk_bound_dev_if;
L
Linus Torvalds 已提交
684

685
	arg.tos = ip_hdr(skb)->tos;
686 687
	ip_send_unicast_reply(*this_cpu_ptr(net->ipv4.tcp_sk),
			      skb, &TCP_SKB_CB(skb)->header.h4.opt,
688 689
			      ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
			      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
690

691 692
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
693 694 695 696 697 698 699 700

#ifdef CONFIG_TCP_MD5SIG
release_sk1:
	if (sk1) {
		rcu_read_unlock();
		sock_put(sk1);
	}
#endif
L
Linus Torvalds 已提交
701 702 703 704 705 706
}

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

707
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
708
			    u32 win, u32 tsval, u32 tsecr, int oif,
709
			    struct tcp_md5sig_key *key,
710
			    int reply_flags, u8 tos)
L
Linus Torvalds 已提交
711
{
712
	const struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
713 714
	struct {
		struct tcphdr th;
715
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
716
#ifdef CONFIG_TCP_MD5SIG
717
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
718 719
#endif
			];
L
Linus Torvalds 已提交
720 721
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
722
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
723 724

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

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
729
	if (tsecr) {
730 731 732
		rep.opt[0] = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				   (TCPOPT_TIMESTAMP << 8) |
				   TCPOLEN_TIMESTAMP);
733 734
		rep.opt[1] = htonl(tsval);
		rep.opt[2] = htonl(tsecr);
735
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
736 737 738 739 740 741 742 743 744 745 746
	}

	/* 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);

747 748
#ifdef CONFIG_TCP_MD5SIG
	if (key) {
749
		int offset = (tsecr) ? 3 : 0;
750 751 752 753 754 755 756 757

		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;

758
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[offset],
759 760
				    key, ip_hdr(skb)->saddr,
				    ip_hdr(skb)->daddr, &rep.th);
761 762
	}
#endif
763
	arg.flags = reply_flags;
764 765
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
L
Linus Torvalds 已提交
766 767
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
768 769
	if (oif)
		arg.bound_dev_if = oif;
770
	arg.tos = tos;
771 772
	ip_send_unicast_reply(*this_cpu_ptr(net->ipv4.tcp_sk),
			      skb, &TCP_SKB_CB(skb)->header.h4.opt,
773 774
			      ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
			      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
775

776
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
777 778 779 780
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
781
	struct inet_timewait_sock *tw = inet_twsk(sk);
782
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
783

784
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
785
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
786
			tcp_time_stamp + tcptw->tw_ts_offset,
787 788
			tcptw->tw_ts_recent,
			tw->tw_bound_dev_if,
789
			tcp_twsk_md5_key(tcptw),
790 791
			tw->tw_transparent ? IP_REPLY_ARG_NOSRCCHECK : 0,
			tw->tw_tos
792
			);
L
Linus Torvalds 已提交
793

794
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
795 796
}

797
static void tcp_v4_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
798
				  struct request_sock *req)
L
Linus Torvalds 已提交
799
{
800 801 802 803 804 805
	/* 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,
806
			tcp_time_stamp,
807 808
			req->ts_recent,
			0,
E
Eric Dumazet 已提交
809 810
			tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&ip_hdr(skb)->daddr,
					  AF_INET),
811 812
			inet_rsk(req)->no_srccheck ? IP_REPLY_ARG_NOSRCCHECK : 0,
			ip_hdr(skb)->tos);
L
Linus Torvalds 已提交
813 814 815
}

/*
816
 *	Send a SYN-ACK after having received a SYN.
817
 *	This still operates on a request_sock only, not on a big
L
Linus Torvalds 已提交
818 819
 *	socket.
 */
820
static int tcp_v4_send_synack(struct sock *sk, struct dst_entry *dst,
821
			      struct flowi *fl,
822
			      struct request_sock *req,
823 824
			      u16 queue_mapping,
			      struct tcp_fastopen_cookie *foc)
L
Linus Torvalds 已提交
825
{
826
	const struct inet_request_sock *ireq = inet_rsk(req);
827
	struct flowi4 fl4;
L
Linus Torvalds 已提交
828
	int err = -1;
829
	struct sk_buff *skb;
L
Linus Torvalds 已提交
830 831

	/* First, grab a route. */
832
	if (!dst && (dst = inet_csk_route_req(sk, &fl4, req)) == NULL)
833
		return -1;
L
Linus Torvalds 已提交
834

835
	skb = tcp_make_synack(sk, dst, req, foc);
L
Linus Torvalds 已提交
836 837

	if (skb) {
838
		__tcp_v4_send_check(skb, ireq->ir_loc_addr, ireq->ir_rmt_addr);
L
Linus Torvalds 已提交
839

840
		skb_set_queue_mapping(skb, queue_mapping);
841 842
		err = ip_build_and_send_pkt(skb, sk, ireq->ir_loc_addr,
					    ireq->ir_rmt_addr,
843
					    ireq->opt);
844
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
845 846 847 848 849 850
	}

	return err;
}

/*
851
 *	IPv4 request_sock destructor.
L
Linus Torvalds 已提交
852
 */
853
static void tcp_v4_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
854
{
J
Jesper Juhl 已提交
855
	kfree(inet_rsk(req)->opt);
L
Linus Torvalds 已提交
856 857
}

858
/*
E
Eric Dumazet 已提交
859
 * Return true if a syncookie should be sent
860
 */
E
Eric Dumazet 已提交
861
bool tcp_syn_flood_action(struct sock *sk,
862 863
			 const struct sk_buff *skb,
			 const char *proto)
L
Linus Torvalds 已提交
864
{
865
	const char *msg = "Dropping request";
E
Eric Dumazet 已提交
866
	bool want_cookie = false;
867 868
	struct listen_sock *lopt;

869
#ifdef CONFIG_SYN_COOKIES
870
	if (sysctl_tcp_syncookies) {
871
		msg = "Sending cookies";
E
Eric Dumazet 已提交
872
		want_cookie = true;
873 874
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDOCOOKIES);
	} else
875
#endif
876 877 878
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDROP);

	lopt = inet_csk(sk)->icsk_accept_queue.listen_opt;
879
	if (!lopt->synflood_warned && sysctl_tcp_syncookies != 2) {
880
		lopt->synflood_warned = 1;
881
		pr_info("%s: Possible SYN flooding on port %d. %s.  Check SNMP counters.\n",
882 883 884
			proto, ntohs(tcp_hdr(skb)->dest), msg);
	}
	return want_cookie;
885
}
886
EXPORT_SYMBOL(tcp_syn_flood_action);
L
Linus Torvalds 已提交
887

888 889 890 891 892 893 894 895
#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
896 897 898
struct tcp_md5sig_key *tcp_md5_do_lookup(struct sock *sk,
					 const union tcp_md5_addr *addr,
					 int family)
899 900
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
901 902
	struct tcp_md5sig_key *key;
	unsigned int size = sizeof(struct in_addr);
903
	struct tcp_md5sig_info *md5sig;
904

905 906
	/* caller either holds rcu_read_lock() or socket lock */
	md5sig = rcu_dereference_check(tp->md5sig_info,
907 908
				       sock_owned_by_user(sk) ||
				       lockdep_is_held(&sk->sk_lock.slock));
909
	if (!md5sig)
910
		return NULL;
E
Eric Dumazet 已提交
911 912 913 914
#if IS_ENABLED(CONFIG_IPV6)
	if (family == AF_INET6)
		size = sizeof(struct in6_addr);
#endif
915
	hlist_for_each_entry_rcu(key, &md5sig->head, node) {
E
Eric Dumazet 已提交
916 917 918 919
		if (key->family != family)
			continue;
		if (!memcmp(&key->addr, addr, size))
			return key;
920 921 922
	}
	return NULL;
}
E
Eric Dumazet 已提交
923
EXPORT_SYMBOL(tcp_md5_do_lookup);
924 925 926 927

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
928 929 930 931
	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);
932 933 934
}
EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
935 936
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
937
{
E
Eric Dumazet 已提交
938 939
	union tcp_md5_addr *addr;

940
	addr = (union tcp_md5_addr *)&inet_rsk(req)->ir_rmt_addr;
E
Eric Dumazet 已提交
941
	return tcp_md5_do_lookup(sk, addr, AF_INET);
942 943 944
}

/* This can be called on a newly created socket, from other files */
E
Eric Dumazet 已提交
945 946
int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
		   int family, const u8 *newkey, u8 newkeylen, gfp_t gfp)
947 948
{
	/* Add Key to the list */
949
	struct tcp_md5sig_key *key;
950
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
951
	struct tcp_md5sig_info *md5sig;
952

953
	key = tcp_md5_do_lookup(sk, addr, family);
954 955
	if (key) {
		/* Pre-existing entry - just update that one. */
E
Eric Dumazet 已提交
956
		memcpy(key->key, newkey, newkeylen);
957
		key->keylen = newkeylen;
E
Eric Dumazet 已提交
958 959
		return 0;
	}
960

961 962
	md5sig = rcu_dereference_protected(tp->md5sig_info,
					   sock_owned_by_user(sk));
E
Eric Dumazet 已提交
963 964 965
	if (!md5sig) {
		md5sig = kmalloc(sizeof(*md5sig), gfp);
		if (!md5sig)
966 967
			return -ENOMEM;

E
Eric Dumazet 已提交
968 969
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
		INIT_HLIST_HEAD(&md5sig->head);
970
		rcu_assign_pointer(tp->md5sig_info, md5sig);
E
Eric Dumazet 已提交
971
	}
972

973
	key = sock_kmalloc(sk, sizeof(*key), gfp);
E
Eric Dumazet 已提交
974 975
	if (!key)
		return -ENOMEM;
976
	if (!tcp_alloc_md5sig_pool()) {
977
		sock_kfree_s(sk, key, sizeof(*key));
E
Eric Dumazet 已提交
978
		return -ENOMEM;
979
	}
E
Eric Dumazet 已提交
980 981 982 983 984 985 986 987

	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);
988 989
	return 0;
}
E
Eric Dumazet 已提交
990
EXPORT_SYMBOL(tcp_md5_do_add);
991

E
Eric Dumazet 已提交
992
int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr, int family)
993
{
E
Eric Dumazet 已提交
994 995
	struct tcp_md5sig_key *key;

996
	key = tcp_md5_do_lookup(sk, addr, family);
E
Eric Dumazet 已提交
997 998 999
	if (!key)
		return -ENOENT;
	hlist_del_rcu(&key->node);
1000
	atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1001 1002
	kfree_rcu(key, rcu);
	return 0;
1003
}
E
Eric Dumazet 已提交
1004
EXPORT_SYMBOL(tcp_md5_do_del);
1005

1006
static void tcp_clear_md5_list(struct sock *sk)
1007 1008
{
	struct tcp_sock *tp = tcp_sk(sk);
E
Eric Dumazet 已提交
1009
	struct tcp_md5sig_key *key;
1010
	struct hlist_node *n;
1011
	struct tcp_md5sig_info *md5sig;
1012

1013 1014
	md5sig = rcu_dereference_protected(tp->md5sig_info, 1);

1015
	hlist_for_each_entry_safe(key, n, &md5sig->head, node) {
E
Eric Dumazet 已提交
1016
		hlist_del_rcu(&key->node);
1017
		atomic_sub(sizeof(*key), &sk->sk_omem_alloc);
E
Eric Dumazet 已提交
1018
		kfree_rcu(key, rcu);
1019 1020 1021
	}
}

1022 1023
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
1024 1025 1026 1027 1028 1029 1030
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;

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

1031
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
1032 1033 1034 1035 1036
		return -EFAULT;

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

1037
	if (!cmd.tcpm_keylen)
E
Eric Dumazet 已提交
1038 1039
		return tcp_md5_do_del(sk, (union tcp_md5_addr *)&sin->sin_addr.s_addr,
				      AF_INET);
1040 1041 1042 1043

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

E
Eric Dumazet 已提交
1044 1045 1046
	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);
1047 1048
}

1049 1050
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1051 1052
{
	struct tcp4_pseudohdr *bp;
1053
	struct scatterlist sg;
1054 1055 1056 1057

	bp = &hp->md5_blk.ip4;

	/*
1058
	 * 1. the TCP pseudo-header (in the order: source IP address,
1059 1060 1061 1062 1063 1064
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1065
	bp->protocol = IPPROTO_TCP;
1066
	bp->len = cpu_to_be16(nbytes);
1067

1068 1069 1070 1071
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}

E
Eric Dumazet 已提交
1072
static int tcp_v4_md5_hash_hdr(char *md5_hash, const struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1073
			       __be32 daddr, __be32 saddr, const struct tcphdr *th)
1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
{
	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))
1092 1093 1094 1095
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1096

1097 1098 1099 1100
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1101
	return 1;
1102 1103
}

1104
int tcp_v4_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
E
Eric Dumazet 已提交
1105 1106
			const struct sock *sk, const struct request_sock *req,
			const struct sk_buff *skb)
1107
{
1108 1109
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
E
Eric Dumazet 已提交
1110
	const struct tcphdr *th = tcp_hdr(skb);
1111 1112 1113
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1114 1115
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1116
	} else if (req) {
1117 1118
		saddr = inet_rsk(req)->ir_loc_addr;
		daddr = inet_rsk(req)->ir_rmt_addr;
1119
	} else {
1120 1121 1122
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
1123
	}
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151

	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;
1152
}
1153
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1154

1155 1156
static bool __tcp_v4_inbound_md5_hash(struct sock *sk,
				      const struct sk_buff *skb)
1157 1158 1159 1160 1161 1162 1163 1164 1165
{
	/*
	 * 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.
	 */
1166
	const __u8 *hash_location = NULL;
1167
	struct tcp_md5sig_key *hash_expected;
1168
	const struct iphdr *iph = ip_hdr(skb);
1169
	const struct tcphdr *th = tcp_hdr(skb);
1170 1171 1172
	int genhash;
	unsigned char newhash[16];

E
Eric Dumazet 已提交
1173 1174
	hash_expected = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&iph->saddr,
					  AF_INET);
1175
	hash_location = tcp_parse_md5sig_option(th);
1176 1177 1178

	/* We've parsed the options - do we have a hash? */
	if (!hash_expected && !hash_location)
E
Eric Dumazet 已提交
1179
		return false;
1180 1181

	if (hash_expected && !hash_location) {
1182
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
E
Eric Dumazet 已提交
1183
		return true;
1184 1185 1186
	}

	if (!hash_expected && hash_location) {
1187
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
E
Eric Dumazet 已提交
1188
		return true;
1189 1190 1191 1192 1193
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1194 1195 1196
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1197 1198

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
1199 1200 1201 1202 1203
		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
1204
		return true;
1205
	}
E
Eric Dumazet 已提交
1206
	return false;
1207 1208
}

1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
static bool tcp_v4_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
{
	bool ret;

	rcu_read_lock();
	ret = __tcp_v4_inbound_md5_hash(sk, skb);
	rcu_read_unlock();

	return ret;
}

1220 1221
#endif

1222 1223 1224 1225 1226 1227 1228 1229 1230
static void tcp_v4_init_req(struct request_sock *req, struct sock *sk,
			    struct sk_buff *skb)
{
	struct inet_request_sock *ireq = inet_rsk(req);

	ireq->ir_loc_addr = ip_hdr(skb)->daddr;
	ireq->ir_rmt_addr = ip_hdr(skb)->saddr;
	ireq->no_srccheck = inet_sk(sk)->transparent;
	ireq->opt = tcp_v4_save_options(skb);
E
Eric Dumazet 已提交
1231
	ireq->ireq_family = AF_INET;
1232 1233
}

1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
static struct dst_entry *tcp_v4_route_req(struct sock *sk, struct flowi *fl,
					  const struct request_sock *req,
					  bool *strict)
{
	struct dst_entry *dst = inet_csk_route_req(sk, &fl->u.ip4, req);

	if (strict) {
		if (fl->u.ip4.daddr == inet_rsk(req)->ir_rmt_addr)
			*strict = true;
		else
			*strict = false;
	}

	return dst;
}

1250
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1251
	.family		=	PF_INET,
1252
	.obj_size	=	sizeof(struct tcp_request_sock),
1253
	.rtx_syn_ack	=	tcp_rtx_synack,
1254 1255
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1256
	.send_reset	=	tcp_v4_send_reset,
S
stephen hemminger 已提交
1257
	.syn_ack_timeout =	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
1258 1259
};

1260
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1261
	.mss_clamp	=	TCP_MSS_DEFAULT,
1262
#ifdef CONFIG_TCP_MD5SIG
1263
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1264
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1265
#endif
1266
	.init_req	=	tcp_v4_init_req,
1267 1268 1269
#ifdef CONFIG_SYN_COOKIES
	.cookie_init_seq =	cookie_v4_init_sequence,
#endif
1270
	.route_req	=	tcp_v4_route_req,
1271
	.init_seq	=	tcp_v4_init_sequence,
1272
	.send_synack	=	tcp_v4_send_synack,
1273
	.queue_hash_add =	inet_csk_reqsk_queue_hash_add,
1274
};
1275

L
Linus Torvalds 已提交
1276 1277 1278
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
	/* Never answer to SYNs send to broadcast or multicast */
E
Eric Dumazet 已提交
1279
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1280 1281
		goto drop;

O
Octavian Purdila 已提交
1282 1283
	return tcp_conn_request(&tcp_request_sock_ops,
				&tcp_request_sock_ipv4_ops, sk, skb);
L
Linus Torvalds 已提交
1284 1285

drop:
1286
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1287 1288
	return 0;
}
E
Eric Dumazet 已提交
1289
EXPORT_SYMBOL(tcp_v4_conn_request);
L
Linus Torvalds 已提交
1290 1291 1292 1293 1294 1295 1296


/*
 * 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,
1297
				  struct request_sock *req,
L
Linus Torvalds 已提交
1298 1299
				  struct dst_entry *dst)
{
1300
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1301 1302 1303
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1304 1305 1306
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
1307
	struct ip_options_rcu *inet_opt;
L
Linus Torvalds 已提交
1308 1309 1310 1311 1312 1313

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (!newsk)
1314
		goto exit_nonewsk;
L
Linus Torvalds 已提交
1315

1316
	newsk->sk_gso_type = SKB_GSO_TCPV4;
1317
	inet_sk_rx_dst_set(newsk, skb);
L
Linus Torvalds 已提交
1318 1319 1320

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1321
	ireq		      = inet_rsk(req);
1322 1323
	sk_daddr_set(newsk, ireq->ir_rmt_addr);
	sk_rcv_saddr_set(newsk, ireq->ir_loc_addr);
1324
	newinet->inet_saddr	      = ireq->ir_loc_addr;
1325 1326
	inet_opt	      = ireq->opt;
	rcu_assign_pointer(newinet->inet_opt, inet_opt);
1327
	ireq->opt	      = NULL;
1328
	newinet->mc_index     = inet_iif(skb);
1329
	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
1330
	newinet->rcv_tos      = ip_hdr(skb)->tos;
1331
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
1332
	inet_set_txhash(newsk);
1333 1334
	if (inet_opt)
		inet_csk(newsk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
E
Eric Dumazet 已提交
1335
	newinet->inet_id = newtp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
1336

E
Eric Dumazet 已提交
1337 1338 1339 1340 1341 1342 1343
	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() */
	}
1344 1345
	sk_setup_caps(newsk, dst);

1346 1347
	tcp_ca_openreq_child(newsk, dst);

L
Linus Torvalds 已提交
1348
	tcp_sync_mss(newsk, dst_mtu(dst));
1349
	newtp->advmss = dst_metric_advmss(dst);
1350 1351 1352 1353
	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
1354 1355
	tcp_initialize_rcv_mss(newsk);

1356 1357
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1358 1359
	key = tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&newinet->inet_daddr,
				AF_INET);
E
Eric Dumazet 已提交
1360
	if (key != NULL) {
1361 1362 1363 1364 1365 1366
		/*
		 * 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
1367 1368
		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
1369
		sk_nocaps_add(newsk, NETIF_F_GSO_MASK);
1370 1371 1372
	}
#endif

1373 1374
	if (__inet_inherit_port(sk, newsk) < 0)
		goto put_and_exit;
1375
	__inet_hash_nolisten(newsk, NULL);
L
Linus Torvalds 已提交
1376 1377 1378 1379

	return newsk;

exit_overflow:
1380
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
1381 1382
exit_nonewsk:
	dst_release(dst);
L
Linus Torvalds 已提交
1383
exit:
1384
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1385
	return NULL;
1386
put_and_exit:
1387 1388
	inet_csk_prepare_forced_close(newsk);
	tcp_done(newsk);
1389
	goto exit;
L
Linus Torvalds 已提交
1390
}
E
Eric Dumazet 已提交
1391
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
L
Linus Torvalds 已提交
1392 1393 1394

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1395
	struct tcphdr *th = tcp_hdr(skb);
1396
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1397
	struct sock *nsk;
1398
	struct request_sock **prev;
L
Linus Torvalds 已提交
1399
	/* Find possible connection requests. */
1400 1401
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1402
	if (req)
1403
		return tcp_check_req(sk, skb, req, prev, false);
L
Linus Torvalds 已提交
1404

1405
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1406
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1407 1408 1409 1410 1411 1412

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1413
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1414 1415 1416 1417
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
1418
	if (!th->syn)
C
Cong Wang 已提交
1419
		sk = cookie_v4_check(sk, skb);
L
Linus Torvalds 已提交
1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433
#endif
	return sk;
}

/* 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)
{
1434 1435
	struct sock *rsk;

L
Linus Torvalds 已提交
1436
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1437 1438
		struct dst_entry *dst = sk->sk_rx_dst;

1439
		sock_rps_save_rxhash(sk, skb);
1440
		sk_mark_napi_id(sk, skb);
1441
		if (dst) {
E
Eric Dumazet 已提交
1442 1443
			if (inet_sk(sk)->rx_dst_ifindex != skb->skb_iif ||
			    dst->ops->check(dst, 0) == NULL) {
1444 1445 1446 1447
				dst_release(dst);
				sk->sk_rx_dst = NULL;
			}
		}
1448
		tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len);
L
Linus Torvalds 已提交
1449 1450 1451
		return 0;
	}

1452
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1453 1454 1455 1456 1457 1458 1459 1460
		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) {
1461
			sock_rps_save_rxhash(nsk, skb);
1462
			sk_mark_napi_id(sk, skb);
1463 1464
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1465
				goto reset;
1466
			}
L
Linus Torvalds 已提交
1467 1468
			return 0;
		}
1469
	} else
1470
		sock_rps_save_rxhash(sk, skb);
1471

1472
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1473
		rsk = sk;
L
Linus Torvalds 已提交
1474
		goto reset;
1475
	}
L
Linus Torvalds 已提交
1476 1477 1478
	return 0;

reset:
1479
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1480 1481 1482 1483 1484 1485 1486 1487 1488 1489
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:
1490
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_CSUMERRORS);
1491
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1492 1493
	goto discard;
}
E
Eric Dumazet 已提交
1494
EXPORT_SYMBOL(tcp_v4_do_rcv);
L
Linus Torvalds 已提交
1495

1496
void tcp_v4_early_demux(struct sk_buff *skb)
D
David S. Miller 已提交
1497 1498 1499 1500 1501 1502
{
	const struct iphdr *iph;
	const struct tcphdr *th;
	struct sock *sk;

	if (skb->pkt_type != PACKET_HOST)
1503
		return;
D
David S. Miller 已提交
1504

1505
	if (!pskb_may_pull(skb, skb_transport_offset(skb) + sizeof(struct tcphdr)))
1506
		return;
D
David S. Miller 已提交
1507 1508

	iph = ip_hdr(skb);
1509
	th = tcp_hdr(skb);
D
David S. Miller 已提交
1510 1511

	if (th->doff < sizeof(struct tcphdr) / 4)
1512
		return;
D
David S. Miller 已提交
1513

1514
	sk = __inet_lookup_established(dev_net(skb->dev), &tcp_hashinfo,
D
David S. Miller 已提交
1515
				       iph->saddr, th->source,
1516
				       iph->daddr, ntohs(th->dest),
E
Eric Dumazet 已提交
1517
				       skb->skb_iif);
D
David S. Miller 已提交
1518 1519 1520
	if (sk) {
		skb->sk = sk;
		skb->destructor = sock_edemux;
1521
		if (sk_fullsock(sk)) {
D
David S. Miller 已提交
1522
			struct dst_entry *dst = sk->sk_rx_dst;
E
Eric Dumazet 已提交
1523

D
David S. Miller 已提交
1524 1525
			if (dst)
				dst = dst_check(dst, 0);
1526
			if (dst &&
E
Eric Dumazet 已提交
1527
			    inet_sk(sk)->rx_dst_ifindex == skb->skb_iif)
1528
				skb_dst_set_noref(skb, dst);
D
David S. Miller 已提交
1529 1530 1531 1532
		}
	}
}

E
Eric Dumazet 已提交
1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550
/* 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;

1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561
	/* Before escaping RCU protected region, we need to take care of skb
	 * dst. Prequeue is only enabled for established sockets.
	 * For such sockets, we might need the skb dst only to set sk->sk_rx_dst
	 * Instead of doing full sk_rx_dst validity here, let's perform
	 * an optimistic check.
	 */
	if (likely(sk->sk_rx_dst))
		skb_dst_drop(skb);
	else
		skb_dst_force(skb);

E
Eric Dumazet 已提交
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
	__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
1588 1589 1590 1591 1592 1593
/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1594
	const struct iphdr *iph;
1595
	const struct tcphdr *th;
L
Linus Torvalds 已提交
1596 1597
	struct sock *sk;
	int ret;
1598
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1599 1600 1601 1602 1603

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

	/* Count it even if it's bad */
1604
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1605 1606 1607 1608

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

1609
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1610 1611 1612 1613 1614 1615 1616 1617

	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
1618
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1619
	 * So, we defer the checks. */
1620 1621

	if (skb_checksum_init(skb, IPPROTO_TCP, inet_compute_pseudo))
1622
		goto csum_error;
L
Linus Torvalds 已提交
1623

1624
	th = tcp_hdr(skb);
1625
	iph = ip_hdr(skb);
1626 1627 1628 1629 1630 1631 1632
	/* This is tricky : We move IPCB at its correct location into TCP_SKB_CB()
	 * barrier() makes sure compiler wont play fool^Waliasing games.
	 */
	memmove(&TCP_SKB_CB(skb)->header.h4, IPCB(skb),
		sizeof(struct inet_skb_parm));
	barrier();

L
Linus Torvalds 已提交
1633 1634 1635 1636
	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);
1637
	TCP_SKB_CB(skb)->tcp_flags = tcp_flag_byte(th);
1638
	TCP_SKB_CB(skb)->tcp_tw_isn = 0;
1639
	TCP_SKB_CB(skb)->ip_dsfield = ipv4_get_dsfield(iph);
L
Linus Torvalds 已提交
1640 1641
	TCP_SKB_CB(skb)->sacked	 = 0;

1642
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1643 1644 1645
	if (!sk)
		goto no_tcp_socket;

E
Eric Dumazet 已提交
1646 1647 1648 1649
process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1650 1651
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1652
		goto discard_and_relse;
1653
	}
1654

L
Linus Torvalds 已提交
1655 1656
	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668

#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
	 */
	if (tcp_v4_inbound_md5_hash(sk, skb))
		goto discard_and_relse;
#endif

1669
	nf_reset(skb);
L
Linus Torvalds 已提交
1670

1671
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1672 1673
		goto discard_and_relse;

E
Eric Dumazet 已提交
1674
	sk_incoming_cpu_update(sk);
L
Linus Torvalds 已提交
1675 1676
	skb->dev = NULL;

1677
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1678 1679
	ret = 0;
	if (!sock_owned_by_user(sk)) {
D
Dan Williams 已提交
1680
		if (!tcp_prequeue(sk, skb))
L
Linus Torvalds 已提交
1681
			ret = tcp_v4_do_rcv(sk, skb);
1682 1683
	} else if (unlikely(sk_add_backlog(sk, skb,
					   sk->sk_rcvbuf + sk->sk_sndbuf))) {
Z
Zhu Yi 已提交
1684
		bh_unlock_sock(sk);
1685
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1686 1687
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698
	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)) {
1699 1700
csum_error:
		TCP_INC_STATS_BH(net, TCP_MIB_CSUMERRORS);
L
Linus Torvalds 已提交
1701
bad_packet:
1702
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1703
	} else {
1704
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1705 1706 1707 1708 1709
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
1710
	return 0;
L
Linus Torvalds 已提交
1711 1712 1713 1714 1715 1716 1717

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1718
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1719 1720 1721
		goto discard_it;
	}

1722
	if (skb->len < (th->doff << 2)) {
1723
		inet_twsk_put(inet_twsk(sk));
1724 1725 1726 1727 1728
		goto bad_packet;
	}
	if (tcp_checksum_complete(skb)) {
		inet_twsk_put(inet_twsk(sk));
		goto csum_error;
L
Linus Torvalds 已提交
1729
	}
1730
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1731
	case TCP_TW_SYN: {
1732
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
1733
							&tcp_hashinfo,
1734
							iph->saddr, th->source,
1735
							iph->daddr, th->dest,
1736
							inet_iif(skb));
L
Linus Torvalds 已提交
1737
		if (sk2) {
1738 1739
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754
			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;
}

1755 1756 1757 1758 1759
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
1760

1761
void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
E
Eric Dumazet 已提交
1762 1763 1764
{
	struct dst_entry *dst = skb_dst(skb);

1765 1766 1767 1768 1769
	if (dst) {
		dst_hold(dst);
		sk->sk_rx_dst = dst;
		inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
	}
E
Eric Dumazet 已提交
1770
}
1771
EXPORT_SYMBOL(inet_sk_rx_dst_set);
E
Eric Dumazet 已提交
1772

1773
const struct inet_connection_sock_af_ops ipv4_specific = {
1774 1775 1776
	.queue_xmit	   = ip_queue_xmit,
	.send_check	   = tcp_v4_send_check,
	.rebuild_header	   = inet_sk_rebuild_header,
E
Eric Dumazet 已提交
1777
	.sk_rx_dst_set	   = inet_sk_rx_dst_set,
1778 1779 1780 1781 1782 1783 1784
	.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),
1785
	.bind_conflict	   = inet_csk_bind_conflict,
1786
#ifdef CONFIG_COMPAT
1787 1788
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
1789
#endif
1790
	.mtu_reduced	   = tcp_v4_mtu_reduced,
L
Linus Torvalds 已提交
1791
};
E
Eric Dumazet 已提交
1792
EXPORT_SYMBOL(ipv4_specific);
L
Linus Torvalds 已提交
1793

1794
#ifdef CONFIG_TCP_MD5SIG
1795
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
1796
	.md5_lookup		= tcp_v4_md5_lookup,
1797
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
1798 1799
	.md5_parse		= tcp_v4_parse_md5_keys,
};
1800
#endif
1801

L
Linus Torvalds 已提交
1802 1803 1804 1805 1806
/* 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)
{
1807
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1808

1809
	tcp_init_sock(sk);
L
Linus Torvalds 已提交
1810

1811
	icsk->icsk_af_ops = &ipv4_specific;
1812

1813
#ifdef CONFIG_TCP_MD5SIG
1814
	tcp_sk(sk)->af_specific = &tcp_sock_ipv4_specific;
1815
#endif
L
Linus Torvalds 已提交
1816 1817 1818 1819

	return 0;
}

1820
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1821 1822 1823 1824 1825
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

1826
	tcp_cleanup_congestion_control(sk);
1827

L
Linus Torvalds 已提交
1828
	/* Cleanup up the write buffer. */
1829
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
1830 1831

	/* Cleans up our, hopefully empty, out_of_order_queue. */
1832
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
1833

1834 1835 1836
#ifdef CONFIG_TCP_MD5SIG
	/* Clean up the MD5 key list, if any */
	if (tp->md5sig_info) {
E
Eric Dumazet 已提交
1837
		tcp_clear_md5_list(sk);
1838
		kfree_rcu(tp->md5sig_info, rcu);
1839 1840 1841
		tp->md5sig_info = NULL;
	}
#endif
C
Chris Leech 已提交
1842

L
Linus Torvalds 已提交
1843 1844 1845 1846
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
1847
	if (inet_csk(sk)->icsk_bind_hash)
1848
		inet_put_port(sk);
L
Linus Torvalds 已提交
1849

1850
	BUG_ON(tp->fastopen_rsk != NULL);
1851

1852 1853 1854
	/* If socket is aborted during connect operation */
	tcp_free_fastopen_req(tp);

1855
	sk_sockets_allocated_dec(sk);
G
Glauber Costa 已提交
1856
	sock_release_memcg(sk);
L
Linus Torvalds 已提交
1857 1858 1859 1860 1861 1862
}
EXPORT_SYMBOL(tcp_v4_destroy_sock);

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

1863 1864 1865 1866 1867
/*
 * 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
1868 1869
static void *listening_get_next(struct seq_file *seq, void *cur)
{
1870
	struct inet_connection_sock *icsk;
1871
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
1872
	struct sock *sk = cur;
1873
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
1874
	struct tcp_iter_state *st = seq->private;
1875
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
1876 1877

	if (!sk) {
1878
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
1879
		spin_lock_bh(&ilb->lock);
1880
		sk = sk_nulls_head(&ilb->head);
1881
		st->offset = 0;
L
Linus Torvalds 已提交
1882 1883
		goto get_sk;
	}
1884
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
1885
	++st->num;
1886
	++st->offset;
L
Linus Torvalds 已提交
1887 1888

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
1889
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
1890

1891
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
1892 1893 1894
		req = req->dl_next;
		while (1) {
			while (req) {
1895
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
1896 1897 1898 1899 1900
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
1901
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
1902 1903
				break;
get_req:
1904
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
1905
		}
E
Eric Dumazet 已提交
1906
		sk	  = sk_nulls_next(st->syn_wait_sk);
L
Linus Torvalds 已提交
1907
		st->state = TCP_SEQ_STATE_LISTENING;
1908
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
1909
	} else {
1910
		icsk = inet_csk(sk);
1911 1912
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
1913
			goto start_req;
1914
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
E
Eric Dumazet 已提交
1915
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
1916 1917
	}
get_sk:
1918
	sk_nulls_for_each_from(sk, node) {
1919 1920 1921
		if (!net_eq(sock_net(sk), net))
			continue;
		if (sk->sk_family == st->family) {
L
Linus Torvalds 已提交
1922 1923 1924
			cur = sk;
			goto out;
		}
1925
		icsk = inet_csk(sk);
1926 1927
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
1928 1929 1930 1931 1932 1933 1934
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;
		}
1935
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
1936
	}
1937
	spin_unlock_bh(&ilb->lock);
1938
	st->offset = 0;
1939
	if (++st->bucket < INET_LHTABLE_SIZE) {
1940 1941
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
1942
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
1943 1944 1945 1946 1947 1948 1949 1950 1951
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
1952 1953 1954 1955 1956 1957
	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
1958 1959 1960 1961 1962 1963 1964 1965

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

E
Eric Dumazet 已提交
1966
static inline bool empty_bucket(const struct tcp_iter_state *st)
1967
{
E
Eric Dumazet 已提交
1968
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain);
1969 1970
}

1971 1972 1973 1974
/*
 * 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
1975 1976
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
1977
	struct tcp_iter_state *st = seq->private;
1978
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
1979 1980
	void *rc = NULL;

1981 1982
	st->offset = 0;
	for (; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
1983
		struct sock *sk;
1984
		struct hlist_nulls_node *node;
1985
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
1986

1987 1988 1989 1990
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

1991
		spin_lock_bh(lock);
1992
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
1993
			if (sk->sk_family != st->family ||
1994
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
1995 1996 1997 1998 1999
				continue;
			}
			rc = sk;
			goto out;
		}
2000
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2001 2002 2003 2004 2005 2006 2007 2008
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2009
	struct hlist_nulls_node *node;
J
Jianjun Kong 已提交
2010
	struct tcp_iter_state *st = seq->private;
2011
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2012 2013

	++st->num;
2014
	++st->offset;
L
Linus Torvalds 已提交
2015

E
Eric Dumazet 已提交
2016
	sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2017

2018
	sk_nulls_for_each_from(sk, node) {
2019
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
E
Eric Dumazet 已提交
2020
			return sk;
L
Linus Torvalds 已提交
2021 2022
	}

E
Eric Dumazet 已提交
2023 2024 2025
	spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
	++st->bucket;
	return established_get_first(seq);
L
Linus Torvalds 已提交
2026 2027 2028 2029
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
2030 2031 2032 2033 2034
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	rc = established_get_first(seq);
L
Linus Torvalds 已提交
2035 2036 2037 2038

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2039
	}
L
Linus Torvalds 已提交
2040 2041 2042 2043 2044 2045
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2046
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058

	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;
}

2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077
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;
E
Eric Dumazet 已提交
2078
		st->state = TCP_SEQ_STATE_ESTABLISHED;
2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092
		/* Fallthrough */
	case 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
2093 2094
static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2095
	struct tcp_iter_state *st = seq->private;
2096 2097 2098 2099 2100 2101 2102 2103
	void *rc;

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

L
Linus Torvalds 已提交
2104 2105
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
2106 2107 2108 2109 2110 2111 2112
	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
2113 2114 2115 2116
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2117
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130
	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;
2131 2132
			st->bucket = 0;
			st->offset = 0;
L
Linus Torvalds 已提交
2133 2134 2135 2136 2137 2138 2139 2140 2141
			rc	  = established_get_first(seq);
		}
		break;
	case TCP_SEQ_STATE_ESTABLISHED:
		rc = established_get_next(seq, v);
		break;
	}
out:
	++*pos;
2142
	st->last_pos = *pos;
L
Linus Torvalds 已提交
2143 2144 2145 2146 2147
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2148
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2149 2150 2151 2152

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2153 2154
			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
2155 2156 2157
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2158
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2159 2160 2161
		break;
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2162
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2163 2164 2165 2166
		break;
	}
}

2167
int tcp_seq_open(struct inode *inode, struct file *file)
L
Linus Torvalds 已提交
2168
{
A
Al Viro 已提交
2169
	struct tcp_seq_afinfo *afinfo = PDE_DATA(inode);
L
Linus Torvalds 已提交
2170
	struct tcp_iter_state *s;
2171
	int err;
L
Linus Torvalds 已提交
2172

2173 2174 2175 2176
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2177

2178
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2179
	s->family		= afinfo->family;
S
stephen hemminger 已提交
2180
	s->last_pos		= 0;
2181 2182
	return 0;
}
2183
EXPORT_SYMBOL(tcp_seq_open);
2184

2185
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2186 2187 2188 2189
{
	int rc = 0;
	struct proc_dir_entry *p;

2190 2191 2192 2193
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2194
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
2195
			     afinfo->seq_fops, afinfo);
2196
	if (!p)
L
Linus Torvalds 已提交
2197 2198 2199
		rc = -ENOMEM;
	return rc;
}
E
Eric Dumazet 已提交
2200
EXPORT_SYMBOL(tcp_proc_register);
L
Linus Torvalds 已提交
2201

2202
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2203
{
2204
	remove_proc_entry(afinfo->name, net->proc_net);
L
Linus Torvalds 已提交
2205
}
E
Eric Dumazet 已提交
2206
EXPORT_SYMBOL(tcp_proc_unregister);
L
Linus Torvalds 已提交
2207

2208
static void get_openreq4(const struct request_sock *req,
2209
			 struct seq_file *f, int i, kuid_t uid)
L
Linus Torvalds 已提交
2210
{
2211
	const struct inet_request_sock *ireq = inet_rsk(req);
2212
	long delta = req->expires - jiffies;
L
Linus Torvalds 已提交
2213

2214
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
2215
		" %02X %08X:%08X %02X:%08lX %08X %5u %8d %u %d %pK",
L
Linus Torvalds 已提交
2216
		i,
2217
		ireq->ir_loc_addr,
2218
		ireq->ir_num,
2219 2220
		ireq->ir_rmt_addr,
		ntohs(ireq->ir_rmt_port),
L
Linus Torvalds 已提交
2221 2222 2223
		TCP_SYN_RECV,
		0, 0, /* could print option size, but that is af dependent. */
		1,    /* timers active (only the expire timer) */
2224
		jiffies_delta_to_clock_t(delta),
2225
		req->num_timeout,
2226
		from_kuid_munged(seq_user_ns(f), uid),
L
Linus Torvalds 已提交
2227 2228
		0,  /* non standard timer */
		0, /* open_requests have no inode */
2229
		0,
2230
		req);
L
Linus Torvalds 已提交
2231 2232
}

2233
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i)
L
Linus Torvalds 已提交
2234 2235 2236
{
	int timer_active;
	unsigned long timer_expires;
2237
	const struct tcp_sock *tp = tcp_sk(sk);
2238
	const struct inet_connection_sock *icsk = inet_csk(sk);
2239
	const struct inet_sock *inet = inet_sk(sk);
2240
	struct fastopen_queue *fastopenq = icsk->icsk_accept_queue.fastopenq;
E
Eric Dumazet 已提交
2241 2242 2243 2244
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
2245
	int rx_queue;
L
Linus Torvalds 已提交
2246

N
Nandita Dukkipati 已提交
2247 2248 2249
	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
2250
		timer_active	= 1;
2251 2252
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2253
		timer_active	= 4;
2254
		timer_expires	= icsk->icsk_timeout;
2255
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2256
		timer_active	= 2;
2257
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2258 2259 2260 2261 2262
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2263 2264 2265 2266 2267 2268 2269 2270
	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);

2271
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
2272
			"%08X %5u %8d %lu %d %pK %lu %lu %u %u %d",
2273
		i, src, srcp, dest, destp, sk->sk_state,
2274
		tp->write_seq - tp->snd_una,
2275
		rx_queue,
L
Linus Torvalds 已提交
2276
		timer_active,
2277
		jiffies_delta_to_clock_t(timer_expires - jiffies),
2278
		icsk->icsk_retransmits,
2279
		from_kuid_munged(seq_user_ns(f), sock_i_uid(sk)),
2280
		icsk->icsk_probes_out,
2281 2282
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2283 2284
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2285
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2286
		tp->snd_cwnd,
2287 2288
		sk->sk_state == TCP_LISTEN ?
		    (fastopenq ? fastopenq->max_qlen : 0) :
2289
		    (tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh));
L
Linus Torvalds 已提交
2290 2291
}

2292
static void get_timewait4_sock(const struct inet_timewait_sock *tw,
2293
			       struct seq_file *f, int i)
L
Linus Torvalds 已提交
2294
{
2295
	__be32 dest, src;
L
Linus Torvalds 已提交
2296
	__u16 destp, srcp;
2297
	s32 delta = tw->tw_ttd - inet_tw_time_stamp();
L
Linus Torvalds 已提交
2298 2299 2300 2301 2302 2303

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

2304
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
2305
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %pK",
L
Linus Torvalds 已提交
2306
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
2307
		3, jiffies_delta_to_clock_t(delta), 0, 0, 0, 0,
2308
		atomic_read(&tw->tw_refcnt), tw);
L
Linus Torvalds 已提交
2309 2310 2311 2312 2313 2314
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2315
	struct tcp_iter_state *st;
E
Eric Dumazet 已提交
2316
	struct sock *sk = v;
L
Linus Torvalds 已提交
2317

2318
	seq_setwidth(seq, TMPSZ - 1);
L
Linus Torvalds 已提交
2319
	if (v == SEQ_START_TOKEN) {
2320
		seq_puts(seq, "  sl  local_address rem_address   st tx_queue "
L
Linus Torvalds 已提交
2321 2322 2323 2324 2325 2326 2327 2328 2329
			   "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:
E
Eric Dumazet 已提交
2330
		if (sk->sk_state == TCP_TIME_WAIT)
2331
			get_timewait4_sock(v, seq, st->num);
E
Eric Dumazet 已提交
2332
		else
2333
			get_tcp4_sock(v, seq, st->num);
L
Linus Torvalds 已提交
2334 2335
		break;
	case TCP_SEQ_STATE_OPENREQ:
2336
		get_openreq4(v, seq, st->num, st->uid);
L
Linus Torvalds 已提交
2337 2338 2339
		break;
	}
out:
2340
	seq_pad(seq, '\n');
L
Linus Torvalds 已提交
2341 2342 2343
	return 0;
}

2344 2345 2346 2347 2348 2349 2350 2351
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
2352 2353 2354
static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2355
	.seq_fops	= &tcp_afinfo_seq_fops,
2356 2357 2358
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2359 2360
};

2361
static int __net_init tcp4_proc_init_net(struct net *net)
2362 2363 2364 2365
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

2366
static void __net_exit tcp4_proc_exit_net(struct net *net)
2367 2368 2369 2370 2371 2372 2373 2374 2375
{
	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
2376 2377
int __init tcp4_proc_init(void)
{
2378
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2379 2380 2381 2382
}

void tcp4_proc_exit(void)
{
2383
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2384 2385 2386 2387 2388 2389 2390 2391 2392
}
#endif /* CONFIG_PROC_FS */

struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2393
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2394 2395 2396 2397 2398 2399 2400
	.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,
2401 2402
	.sendmsg		= tcp_sendmsg,
	.sendpage		= tcp_sendpage,
L
Linus Torvalds 已提交
2403
	.backlog_rcv		= tcp_v4_do_rcv,
E
Eric Dumazet 已提交
2404
	.release_cb		= tcp_release_cb,
2405 2406 2407
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2408
	.enter_memory_pressure	= tcp_enter_memory_pressure,
2409
	.stream_memory_free	= tcp_stream_memory_free,
L
Linus Torvalds 已提交
2410
	.sockets_allocated	= &tcp_sockets_allocated,
2411
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2412 2413
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
2414
	.sysctl_mem		= sysctl_tcp_mem,
L
Linus Torvalds 已提交
2415 2416 2417 2418
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp_sock),
2419
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2420
	.twsk_prot		= &tcp_timewait_sock_ops,
2421
	.rsk_prot		= &tcp_request_sock_ops,
2422
	.h.hashinfo		= &tcp_hashinfo,
2423
	.no_autobind		= true,
2424 2425 2426 2427
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
A
Andrew Morton 已提交
2428
#ifdef CONFIG_MEMCG_KMEM
G
Glauber Costa 已提交
2429 2430 2431 2432
	.init_cgroup		= tcp_init_cgroup,
	.destroy_cgroup		= tcp_destroy_cgroup,
	.proto_cgroup		= tcp_proto_cgroup,
#endif
L
Linus Torvalds 已提交
2433
};
E
Eric Dumazet 已提交
2434
EXPORT_SYMBOL(tcp_prot);
L
Linus Torvalds 已提交
2435

2436 2437 2438 2439 2440 2441 2442 2443 2444
static void __net_exit tcp_sk_exit(struct net *net)
{
	int cpu;

	for_each_possible_cpu(cpu)
		inet_ctl_sock_destroy(*per_cpu_ptr(net->ipv4.tcp_sk, cpu));
	free_percpu(net->ipv4.tcp_sk);
}

2445 2446
static int __net_init tcp_sk_init(struct net *net)
{
2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461
	int res, cpu;

	net->ipv4.tcp_sk = alloc_percpu(struct sock *);
	if (!net->ipv4.tcp_sk)
		return -ENOMEM;

	for_each_possible_cpu(cpu) {
		struct sock *sk;

		res = inet_ctl_sock_create(&sk, PF_INET, SOCK_RAW,
					   IPPROTO_TCP, net);
		if (res)
			goto fail;
		*per_cpu_ptr(net->ipv4.tcp_sk, cpu) = sk;
	}
2462
	net->ipv4.sysctl_tcp_ecn = 2;
F
Fan Du 已提交
2463
	net->ipv4.sysctl_tcp_base_mss = TCP_BASE_MSS;
2464
	net->ipv4.sysctl_tcp_probe_threshold = TCP_PROBE_THRESHOLD;
2465
	net->ipv4.sysctl_tcp_probe_interval = TCP_PROBE_INTERVAL;
2466
	return 0;
2467

2468 2469 2470 2471
fail:
	tcp_sk_exit(net);

	return res;
E
Eric W. Biederman 已提交
2472 2473 2474 2475 2476
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2477 2478 2479
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2480 2481 2482
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2483 2484
};

2485
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2486
{
2487
	inet_hashinfo_init(&tcp_hashinfo);
2488
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2489 2490
		panic("Failed to create the TCP control socket.\n");
}