tcp_ipv4.c 64.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Implementation of the Transmission Control Protocol(TCP).
 *
 *		IPv4 specific functions
 *
 *
 *		code split from:
 *		linux/ipv4/tcp.c
 *		linux/ipv4/tcp_input.c
 *		linux/ipv4/tcp_output.c
 *
 *		See tcp.c for author information
 *
 *	This program is free software; you can redistribute it and/or
 *      modify it under the terms of the GNU General Public License
 *      as published by the Free Software Foundation; either version
 *      2 of the License, or (at your option) any later version.
 */

/*
 * Changes:
 *		David S. Miller	:	New socket lookup architecture.
 *					This code is dedicated to John Dyson.
 *		David S. Miller :	Change semantics of established hash,
 *					half is devoted to TIME_WAIT sockets
 *					and the rest go in the other half.
 *		Andi Kleen :		Add support for syncookies and fixed
 *					some bugs: ip options weren't passed to
 *					the TCP layer, missed a check for an
 *					ACK bit.
 *		Andi Kleen :		Implemented fast path mtu discovery.
 *	     				Fixed many serious bugs in the
37
 *					request_sock handling and moved
L
Linus Torvalds 已提交
38 39
 *					most of it into the af independent code.
 *					Added tail drop and some other bugfixes.
S
Stephen Hemminger 已提交
40
 *					Added new listen semantics.
L
Linus Torvalds 已提交
41 42 43 44 45 46 47 48 49 50 51 52 53
 *		Mike McLagan	:	Routing by source
 *	Juan Jose Ciarlante:		ip_dynaddr bits
 *		Andi Kleen:		various fixes.
 *	Vitaly E. Lavrov	:	Transparent proxy revived after year
 *					coma.
 *	Andi Kleen		:	Fix new listen.
 *	Andi Kleen		:	Fix accept error reporting.
 *	YOSHIFUJI Hideaki @USAGI and:	Support IPV6_V6ONLY socket option, which
 *	Alexey Kuznetsov		allow both IPv4 and IPv6 sockets to bind
 *					a single port at the same time.
 */


H
Herbert Xu 已提交
54
#include <linux/bottom_half.h>
L
Linus Torvalds 已提交
55 56 57 58 59 60 61 62 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 <net/net_namespace.h>
L
Linus Torvalds 已提交
65
#include <net/icmp.h>
66
#include <net/inet_hashtables.h>
L
Linus Torvalds 已提交
67
#include <net/tcp.h>
68
#include <net/transp_v6.h>
L
Linus Torvalds 已提交
69 70
#include <net/ipv6.h>
#include <net/inet_common.h>
71
#include <net/timewait_sock.h>
L
Linus Torvalds 已提交
72
#include <net/xfrm.h>
C
Chris Leech 已提交
73
#include <net/netdma.h>
L
Linus Torvalds 已提交
74 75 76 77 78 79 80

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

81 82 83
#include <linux/crypto.h>
#include <linux/scatterlist.h>

84 85
int sysctl_tcp_tw_reuse __read_mostly;
int sysctl_tcp_low_latency __read_mostly;
L
Linus Torvalds 已提交
86 87


88
#ifdef CONFIG_TCP_MD5SIG
89 90
static struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk,
						   __be32 addr);
91 92
static int tcp_v4_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
			       __be32 daddr, __be32 saddr, struct tcphdr *th);
93 94 95 96 97 98
#else
static inline
struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk, __be32 addr)
{
	return NULL;
}
99 100
#endif

101
struct inet_hashinfo tcp_hashinfo;
L
Linus Torvalds 已提交
102

103
static inline __u32 tcp_v4_init_sequence(struct sk_buff *skb)
L
Linus Torvalds 已提交
104
{
105 106
	return secure_tcp_sequence_number(ip_hdr(skb)->daddr,
					  ip_hdr(skb)->saddr,
107 108
					  tcp_hdr(skb)->dest,
					  tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
109 110
}

111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
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 &&
129
			     get_seconds() - tcptw->tw_ts_recent_stamp > 1))) {
130 131 132 133 134 135 136 137 138 139 140 141 142 143
		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
144 145 146 147 148 149 150
/* This will initiate an outgoing connection. */
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
{
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
	struct sockaddr_in *usin = (struct sockaddr_in *)uaddr;
	struct rtable *rt;
151
	__be32 daddr, nexthop;
L
Linus Torvalds 已提交
152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167
	int tmp;
	int err;

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

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

	nexthop = daddr = usin->sin_addr.s_addr;
	if (inet->opt && inet->opt->srr) {
		if (!daddr)
			return -EINVAL;
		nexthop = inet->opt->faddr;
	}

E
Eric Dumazet 已提交
168
	tmp = ip_route_connect(&rt, nexthop, inet->inet_saddr,
L
Linus Torvalds 已提交
169 170
			       RT_CONN_FLAGS(sk), sk->sk_bound_dev_if,
			       IPPROTO_TCP,
E
Eric Dumazet 已提交
171
			       inet->inet_sport, usin->sin_port, sk, 1);
172 173
	if (tmp < 0) {
		if (tmp == -ENETUNREACH)
174
			IP_INC_STATS_BH(sock_net(sk), IPSTATS_MIB_OUTNOROUTES);
L
Linus Torvalds 已提交
175
		return tmp;
176
	}
L
Linus Torvalds 已提交
177 178 179 180 181 182 183 184 185

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

	if (!inet->opt || !inet->opt->srr)
		daddr = rt->rt_dst;

E
Eric Dumazet 已提交
186 187 188
	if (!inet->inet_saddr)
		inet->inet_saddr = rt->rt_src;
	inet->inet_rcv_saddr = inet->inet_saddr;
L
Linus Torvalds 已提交
189

E
Eric Dumazet 已提交
190
	if (tp->rx_opt.ts_recent_stamp && inet->inet_daddr != daddr) {
L
Linus Torvalds 已提交
191 192 193 194 195 196
		/* Reset inherited state */
		tp->rx_opt.ts_recent	   = 0;
		tp->rx_opt.ts_recent_stamp = 0;
		tp->write_seq		   = 0;
	}

197
	if (tcp_death_row.sysctl_tw_recycle &&
L
Linus Torvalds 已提交
198 199
	    !tp->rx_opt.ts_recent_stamp && rt->rt_dst == daddr) {
		struct inet_peer *peer = rt_get_peer(rt);
200 201 202 203 204
		/*
		 * VJ's idea. We save last timestamp seen from
		 * the destination in peer table, when entering state
		 * TIME-WAIT * and initialize rx_opt.ts_recent from it,
		 * when trying new connection.
L
Linus Torvalds 已提交
205
		 */
206
		if (peer != NULL &&
E
Eric Dumazet 已提交
207
		    (u32)get_seconds() - peer->tcp_ts_stamp <= TCP_PAWS_MSL) {
L
Linus Torvalds 已提交
208 209 210 211 212
			tp->rx_opt.ts_recent_stamp = peer->tcp_ts_stamp;
			tp->rx_opt.ts_recent = peer->tcp_ts;
		}
	}

E
Eric Dumazet 已提交
213 214
	inet->inet_dport = usin->sin_port;
	inet->inet_daddr = daddr;
L
Linus Torvalds 已提交
215

216
	inet_csk(sk)->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
217
	if (inet->opt)
218
		inet_csk(sk)->icsk_ext_hdr_len = inet->opt->optlen;
L
Linus Torvalds 已提交
219

220
	tp->rx_opt.mss_clamp = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
221 222 223 224 225 226 227

	/* 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);
228
	err = inet_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
229 230 231
	if (err)
		goto failure;

232
	err = ip_route_newports(&rt, IPPROTO_TCP,
E
Eric Dumazet 已提交
233
				inet->inet_sport, inet->inet_dport, sk);
L
Linus Torvalds 已提交
234 235 236 237
	if (err)
		goto failure;

	/* OK, now commit destination to socket.  */
238
	sk->sk_gso_type = SKB_GSO_TCPV4;
239
	sk_setup_caps(sk, &rt->u.dst);
L
Linus Torvalds 已提交
240 241

	if (!tp->write_seq)
E
Eric Dumazet 已提交
242 243 244
		tp->write_seq = secure_tcp_sequence_number(inet->inet_saddr,
							   inet->inet_daddr,
							   inet->inet_sport,
L
Linus Torvalds 已提交
245 246
							   usin->sin_port);

E
Eric Dumazet 已提交
247
	inet->inet_id = tp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
248 249 250 251 252 253 254 255 256

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

	return 0;

failure:
257 258 259 260
	/*
	 * This unhashes the socket and releases the local port,
	 * if necessary.
	 */
L
Linus Torvalds 已提交
261 262 263
	tcp_set_state(sk, TCP_CLOSE);
	ip_rt_put(rt);
	sk->sk_route_caps = 0;
E
Eric Dumazet 已提交
264
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
265 266 267 268 269 270
	return err;
}

/*
 * This routine does path mtu discovery as defined in RFC1191.
 */
S
Stephen Hemminger 已提交
271
static void do_pmtu_discovery(struct sock *sk, struct iphdr *iph, u32 mtu)
L
Linus Torvalds 已提交
272 273 274 275 276 277 278 279 280 281 282 283 284 285
{
	struct dst_entry *dst;
	struct inet_sock *inet = inet_sk(sk);

	/* We are not interested in TCP_LISTEN and open_requests (SYN-ACKs
	 * send out by Linux are always <576bytes so they should go through
	 * unfragmented).
	 */
	if (sk->sk_state == TCP_LISTEN)
		return;

	/* We don't check in the destentry if pmtu discovery is forbidden
	 * on this route. We just assume that no packet_to_big packets
	 * are send back when pmtu discovery is not active.
286
	 * There is a small race when the user changes this flag in the
L
Linus Torvalds 已提交
287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302
	 * route, but I think that's acceptable.
	 */
	if ((dst = __sk_dst_check(sk, 0)) == NULL)
		return;

	dst->ops->update_pmtu(dst, mtu);

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

	mtu = dst_mtu(dst);

	if (inet->pmtudisc != IP_PMTUDISC_DONT &&
303
	    inet_csk(sk)->icsk_pmtu_cookie > mtu) {
L
Linus Torvalds 已提交
304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
		tcp_sync_mss(sk, mtu);

		/* Resend the TCP packet because it's
		 * clear that the old packet has been
		 * dropped. This is the new "fast" path mtu
		 * discovery.
		 */
		tcp_simple_retransmit(sk);
	} /* else let the usual retransmit timer handle it */
}

/*
 * This routine is called by the ICMP module when it gets some
 * sort of error condition.  If err < 0 then the socket should
 * be closed and the error returned to the user.  If err > 0
 * it's just the icmp type << 8 | icmp code.  After adjustment
 * header points to the first 8 bytes of the tcp header.  We need
 * to find the appropriate port.
 *
 * The locking strategy used here is very "optimistic". When
 * someone else accesses the socket the ICMP is just dropped
 * and for some paths there is no check at all.
 * A more general error queue to queue errors for later handling
 * is probably better.
 *
 */

331
void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
L
Linus Torvalds 已提交
332
{
333 334
	struct iphdr *iph = (struct iphdr *)icmp_skb->data;
	struct tcphdr *th = (struct tcphdr *)(icmp_skb->data + (iph->ihl << 2));
335
	struct inet_connection_sock *icsk;
L
Linus Torvalds 已提交
336 337
	struct tcp_sock *tp;
	struct inet_sock *inet;
338 339
	const int type = icmp_hdr(icmp_skb)->type;
	const int code = icmp_hdr(icmp_skb)->code;
L
Linus Torvalds 已提交
340
	struct sock *sk;
341
	struct sk_buff *skb;
L
Linus Torvalds 已提交
342
	__u32 seq;
343
	__u32 remaining;
L
Linus Torvalds 已提交
344
	int err;
345
	struct net *net = dev_net(icmp_skb->dev);
L
Linus Torvalds 已提交
346

347
	if (icmp_skb->len < (iph->ihl << 2) + 8) {
348
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
349 350 351
		return;
	}

352
	sk = inet_lookup(net, &tcp_hashinfo, iph->daddr, th->dest,
353
			iph->saddr, th->source, inet_iif(icmp_skb));
L
Linus Torvalds 已提交
354
	if (!sk) {
355
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
356 357 358
		return;
	}
	if (sk->sk_state == TCP_TIME_WAIT) {
359
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
360 361 362 363 364 365 366 367
		return;
	}

	bh_lock_sock(sk);
	/* If too many ICMPs get dropped on busy
	 * servers this needs to be solved differently.
	 */
	if (sock_owned_by_user(sk))
368
		NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
L
Linus Torvalds 已提交
369 370 371 372

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

373
	icsk = inet_csk(sk);
L
Linus Torvalds 已提交
374 375 376 377
	tp = tcp_sk(sk);
	seq = ntohl(th->seq);
	if (sk->sk_state != TCP_LISTEN &&
	    !between(seq, tp->snd_una, tp->snd_nxt)) {
378
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
		goto out;
	}

	switch (type) {
	case ICMP_SOURCE_QUENCH:
		/* Just silently ignore these. */
		goto out;
	case ICMP_PARAMETERPROB:
		err = EPROTO;
		break;
	case ICMP_DEST_UNREACH:
		if (code > NR_ICMP_UNREACH)
			goto out;

		if (code == ICMP_FRAG_NEEDED) { /* PMTU discovery (RFC1191) */
			if (!sock_owned_by_user(sk))
				do_pmtu_discovery(sk, iph, info);
			goto out;
		}

		err = icmp_err_convert[code].errno;
400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432
		/* 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;

		icsk->icsk_backoff--;
		inet_csk(sk)->icsk_rto = __tcp_set_rto(tp) <<
					 icsk->icsk_backoff;
		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 if (sock_owned_by_user(sk)) {
			/* RTO revert clocked out retransmission,
			 * but socket is locked. Will defer. */
			inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
						  HZ/20, TCP_RTO_MAX);
		} else {
			/* RTO revert clocked out retransmission.
			 * Will retransmit now */
			tcp_retransmit_timer(sk);
		}

L
Linus Torvalds 已提交
433 434 435 436 437 438 439 440 441
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

	switch (sk->sk_state) {
442
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
443 444 445 446
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

447 448
		req = inet_csk_search_req(sk, &prev, th->dest,
					  iph->daddr, iph->saddr);
L
Linus Torvalds 已提交
449 450 451 452 453 454
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		   an established socket here.
		 */
455
		WARN_ON(req->sk);
L
Linus Torvalds 已提交
456

457
		if (seq != tcp_rsk(req)->snt_isn) {
458
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
459 460 461 462 463 464 465 466 467
			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().
		 */
468
		inet_csk_reqsk_queue_drop(sk, req, prev);
L
Linus Torvalds 已提交
469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516
		goto out;

	case TCP_SYN_SENT:
	case TCP_SYN_RECV:  /* Cannot happen.
			       It can f.e. if SYNs crossed.
			     */
		if (!sock_owned_by_user(sk)) {
			sk->sk_err = err;

			sk->sk_error_report(sk);

			tcp_done(sk);
		} else {
			sk->sk_err_soft = err;
		}
		goto out;
	}

	/* If we've already connected we will keep trying
	 * until we time out, or the user gives up.
	 *
	 * rfc1122 4.2.3.9 allows to consider as hard errors
	 * only PROTO_UNREACH and PORT_UNREACH (well, FRAG_FAILED too,
	 * but it is obsoleted by pmtu discovery).
	 *
	 * Note, that in modern internet, where routing is unreliable
	 * and in each dark corner broken firewalls sit, sending random
	 * errors ordered by their masters even this two messages finally lose
	 * their original sense (even Linux sends invalid PORT_UNREACHs)
	 *
	 * Now we are in compliance with RFCs.
	 *							--ANK (980905)
	 */

	inet = inet_sk(sk);
	if (!sock_owned_by_user(sk) && inet->recverr) {
		sk->sk_err = err;
		sk->sk_error_report(sk);
	} else	{ /* Only an error on timeout */
		sk->sk_err_soft = err;
	}

out:
	bh_unlock_sock(sk);
	sock_put(sk);
}

/* This routine computes an IPv4 TCP checksum. */
517
void tcp_v4_send_check(struct sock *sk, int len, struct sk_buff *skb)
L
Linus Torvalds 已提交
518 519
{
	struct inet_sock *inet = inet_sk(sk);
520
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
521

522
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
E
Eric Dumazet 已提交
523 524
		th->check = ~tcp_v4_check(len, inet->inet_saddr,
					  inet->inet_daddr, 0);
525
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
526
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
527
	} else {
E
Eric Dumazet 已提交
528 529
		th->check = tcp_v4_check(len, inet->inet_saddr,
					 inet->inet_daddr,
530
					 csum_partial(th,
L
Linus Torvalds 已提交
531 532 533 534 535
						      th->doff << 2,
						      skb->csum));
	}
}

536 537
int tcp_v4_gso_send_check(struct sk_buff *skb)
{
538
	const struct iphdr *iph;
539 540 541 542 543
	struct tcphdr *th;

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

544
	iph = ip_hdr(skb);
545
	th = tcp_hdr(skb);
546 547

	th->check = 0;
548
	th->check = ~tcp_v4_check(skb->len, iph->saddr, iph->daddr, 0);
549
	skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
550
	skb->csum_offset = offsetof(struct tcphdr, check);
551
	skb->ip_summed = CHECKSUM_PARTIAL;
552 553 554
	return 0;
}

L
Linus Torvalds 已提交
555 556 557 558 559 560 561 562 563 564 565 566 567
/*
 *	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.
 */

568
static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
L
Linus Torvalds 已提交
569
{
570
	struct tcphdr *th = tcp_hdr(skb);
571 572 573
	struct {
		struct tcphdr th;
#ifdef CONFIG_TCP_MD5SIG
574
		__be32 opt[(TCPOLEN_MD5SIG_ALIGNED >> 2)];
575 576
#endif
	} rep;
L
Linus Torvalds 已提交
577
	struct ip_reply_arg arg;
578 579 580
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
581
	struct net *net;
L
Linus Torvalds 已提交
582 583 584 585 586

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

E
Eric Dumazet 已提交
587
	if (skb_rtable(skb)->rt_type != RTN_LOCAL)
L
Linus Torvalds 已提交
588 589 590
		return;

	/* Swap the send and the receive. */
591 592 593 594 595
	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
596 597

	if (th->ack) {
598
		rep.th.seq = th->ack_seq;
L
Linus Torvalds 已提交
599
	} else {
600 601 602
		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
603 604
	}

605
	memset(&arg, 0, sizeof(arg));
606 607 608 609
	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);

#ifdef CONFIG_TCP_MD5SIG
610
	key = sk ? tcp_v4_md5_do_lookup(sk, ip_hdr(skb)->daddr) : NULL;
611 612 613 614 615 616 617 618 619
	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;

620
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
621 622
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
623 624
	}
#endif
625 626
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
627
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
628
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
629
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
L
Linus Torvalds 已提交
630

E
Eric Dumazet 已提交
631
	net = dev_net(skb_dst(skb)->dev);
632
	ip_send_reply(net->ipv4.tcp_sock, skb,
633
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
634

635 636
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
L
Linus Torvalds 已提交
637 638 639 640 641 642
}

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

643 644
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
			    u32 win, u32 ts, int oif,
645 646
			    struct tcp_md5sig_key *key,
			    int reply_flags)
L
Linus Torvalds 已提交
647
{
648
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
649 650
	struct {
		struct tcphdr th;
651
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
652
#ifdef CONFIG_TCP_MD5SIG
653
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
654 655
#endif
			];
L
Linus Torvalds 已提交
656 657
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
658
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
659 660

	memset(&rep.th, 0, sizeof(struct tcphdr));
661
	memset(&arg, 0, sizeof(arg));
L
Linus Torvalds 已提交
662 663 664 665

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
	if (ts) {
666 667 668 669 670
		rep.opt[0] = htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) |
				   (TCPOPT_TIMESTAMP << 8) |
				   TCPOLEN_TIMESTAMP);
		rep.opt[1] = htonl(tcp_time_stamp);
		rep.opt[2] = htonl(ts);
671
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
672 673 674 675 676 677 678 679 680 681 682
	}

	/* 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);

683 684 685 686 687 688 689 690 691 692 693
#ifdef CONFIG_TCP_MD5SIG
	if (key) {
		int offset = (ts) ? 3 : 0;

		rep.opt[offset++] = htonl((TCPOPT_NOP << 24) |
					  (TCPOPT_NOP << 16) |
					  (TCPOPT_MD5SIG << 8) |
					  TCPOLEN_MD5SIG);
		arg.iov[0].iov_len += TCPOLEN_MD5SIG_ALIGNED;
		rep.th.doff = arg.iov[0].iov_len/4;

694
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[offset],
695 696
				    key, ip_hdr(skb)->saddr,
				    ip_hdr(skb)->daddr, &rep.th);
697 698
	}
#endif
699
	arg.flags = reply_flags;
700 701
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
L
Linus Torvalds 已提交
702 703
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
704 705
	if (oif)
		arg.bound_dev_if = oif;
L
Linus Torvalds 已提交
706

707
	ip_send_reply(net->ipv4.tcp_sock, skb,
708
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
709

710
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
711 712 713 714
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
715
	struct inet_timewait_sock *tw = inet_twsk(sk);
716
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
717

718
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
719
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
720 721
			tcptw->tw_ts_recent,
			tw->tw_bound_dev_if,
722 723
			tcp_twsk_md5_key(tcptw),
			tw->tw_transparent ? IP_REPLY_ARG_NOSRCCHECK : 0
724
			);
L
Linus Torvalds 已提交
725

726
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
727 728
}

729
static void tcp_v4_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
730
				  struct request_sock *req)
L
Linus Torvalds 已提交
731
{
732
	tcp_v4_send_ack(skb, tcp_rsk(req)->snt_isn + 1,
733
			tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd,
734 735
			req->ts_recent,
			0,
736 737
			tcp_v4_md5_do_lookup(sk, ip_hdr(skb)->daddr),
			inet_rsk(req)->no_srccheck ? IP_REPLY_ARG_NOSRCCHECK : 0);
L
Linus Torvalds 已提交
738 739 740
}

/*
741
 *	Send a SYN-ACK after having received a SYN.
742
 *	This still operates on a request_sock only, not on a big
L
Linus Torvalds 已提交
743 744
 *	socket.
 */
745 746 747
static int __tcp_v4_send_synack(struct sock *sk, struct dst_entry *dst,
				struct request_sock *req,
				struct request_values *rvp)
L
Linus Torvalds 已提交
748
{
749
	const struct inet_request_sock *ireq = inet_rsk(req);
L
Linus Torvalds 已提交
750 751 752 753
	int err = -1;
	struct sk_buff * skb;

	/* First, grab a route. */
754
	if (!dst && (dst = inet_csk_route_req(sk, req)) == NULL)
755
		return -1;
L
Linus Torvalds 已提交
756

757
	skb = tcp_make_synack(sk, dst, req, rvp);
L
Linus Torvalds 已提交
758 759

	if (skb) {
760
		struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
761

762
		th->check = tcp_v4_check(skb->len,
763 764
					 ireq->loc_addr,
					 ireq->rmt_addr,
765
					 csum_partial(th, skb->len,
L
Linus Torvalds 已提交
766 767
						      skb->csum));

768 769 770
		err = ip_build_and_send_pkt(skb, sk, ireq->loc_addr,
					    ireq->rmt_addr,
					    ireq->opt);
771
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
772 773 774 775 776 777
	}

	dst_release(dst);
	return err;
}

778 779
static int tcp_v4_send_synack(struct sock *sk, struct request_sock *req,
			      struct request_values *rvp)
780
{
781
	return __tcp_v4_send_synack(sk, NULL, req, rvp);
782 783
}

L
Linus Torvalds 已提交
784
/*
785
 *	IPv4 request_sock destructor.
L
Linus Torvalds 已提交
786
 */
787
static void tcp_v4_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
788
{
J
Jesper Juhl 已提交
789
	kfree(inet_rsk(req)->opt);
L
Linus Torvalds 已提交
790 791
}

792
#ifdef CONFIG_SYN_COOKIES
S
Stephen Hemminger 已提交
793
static void syn_flood_warning(struct sk_buff *skb)
L
Linus Torvalds 已提交
794 795 796 797 798 799 800
{
	static unsigned long warntime;

	if (time_after(jiffies, (warntime + HZ * 60))) {
		warntime = jiffies;
		printk(KERN_INFO
		       "possible SYN flooding on port %d. Sending cookies.\n",
801
		       ntohs(tcp_hdr(skb)->dest));
L
Linus Torvalds 已提交
802 803
	}
}
804
#endif
L
Linus Torvalds 已提交
805 806

/*
807
 * Save and compile IPv4 options into the request_sock if needed.
L
Linus Torvalds 已提交
808
 */
S
Stephen Hemminger 已提交
809 810
static struct ip_options *tcp_v4_save_options(struct sock *sk,
					      struct sk_buff *skb)
L
Linus Torvalds 已提交
811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827
{
	struct ip_options *opt = &(IPCB(skb)->opt);
	struct ip_options *dopt = NULL;

	if (opt && opt->optlen) {
		int opt_size = optlength(opt);
		dopt = kmalloc(opt_size, GFP_ATOMIC);
		if (dopt) {
			if (ip_options_echo(dopt, skb)) {
				kfree(dopt);
				dopt = NULL;
			}
		}
	}
	return dopt;
}

828 829 830 831 832 833 834 835
#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.  */
836 837
static struct tcp_md5sig_key *
			tcp_v4_md5_do_lookup(struct sock *sk, __be32 addr)
838 839 840 841 842 843 844 845
{
	struct tcp_sock *tp = tcp_sk(sk);
	int i;

	if (!tp->md5sig_info || !tp->md5sig_info->entries4)
		return NULL;
	for (i = 0; i < tp->md5sig_info->entries4; i++) {
		if (tp->md5sig_info->keys4[i].addr == addr)
846
			return &tp->md5sig_info->keys4[i].base;
847 848 849 850 851 852 853
	}
	return NULL;
}

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
854
	return tcp_v4_md5_do_lookup(sk, inet_sk(addr_sk)->inet_daddr);
855 856 857 858
}

EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
859 860
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
861 862 863 864 865 866 867 868 869
{
	return tcp_v4_md5_do_lookup(sk, inet_rsk(req)->rmt_addr);
}

/* This can be called on a newly created socket, from other files */
int tcp_v4_md5_do_add(struct sock *sk, __be32 addr,
		      u8 *newkey, u8 newkeylen)
{
	/* Add Key to the list */
870
	struct tcp_md5sig_key *key;
871 872 873
	struct tcp_sock *tp = tcp_sk(sk);
	struct tcp4_md5sig_key *keys;

874
	key = tcp_v4_md5_do_lookup(sk, addr);
875 876
	if (key) {
		/* Pre-existing entry - just update that one. */
877 878 879
		kfree(key->key);
		key->key = newkey;
		key->keylen = newkeylen;
880
	} else {
881 882
		struct tcp_md5sig_info *md5sig;

883
		if (!tp->md5sig_info) {
884 885
			tp->md5sig_info = kzalloc(sizeof(*tp->md5sig_info),
						  GFP_ATOMIC);
886 887 888 889
			if (!tp->md5sig_info) {
				kfree(newkey);
				return -ENOMEM;
			}
890
			sk->sk_route_caps &= ~NETIF_F_GSO_MASK;
891
		}
892
		if (tcp_alloc_md5sig_pool(sk) == NULL) {
893 894 895
			kfree(newkey);
			return -ENOMEM;
		}
896 897 898 899
		md5sig = tp->md5sig_info;

		if (md5sig->alloced4 == md5sig->entries4) {
			keys = kmalloc((sizeof(*keys) *
900
					(md5sig->entries4 + 1)), GFP_ATOMIC);
901 902 903 904 905 906
			if (!keys) {
				kfree(newkey);
				tcp_free_md5sig_pool();
				return -ENOMEM;
			}

907 908 909
			if (md5sig->entries4)
				memcpy(keys, md5sig->keys4,
				       sizeof(*keys) * md5sig->entries4);
910 911

			/* Free old key list, and reference new one */
912
			kfree(md5sig->keys4);
913 914
			md5sig->keys4 = keys;
			md5sig->alloced4++;
915
		}
916
		md5sig->entries4++;
917 918 919
		md5sig->keys4[md5sig->entries4 - 1].addr        = addr;
		md5sig->keys4[md5sig->entries4 - 1].base.key    = newkey;
		md5sig->keys4[md5sig->entries4 - 1].base.keylen = newkeylen;
920 921 922 923 924 925 926 927 928
	}
	return 0;
}

EXPORT_SYMBOL(tcp_v4_md5_do_add);

static int tcp_v4_md5_add_func(struct sock *sk, struct sock *addr_sk,
			       u8 *newkey, u8 newkeylen)
{
E
Eric Dumazet 已提交
929
	return tcp_v4_md5_do_add(sk, inet_sk(addr_sk)->inet_daddr,
930 931 932 933 934 935 936 937 938 939 940
				 newkey, newkeylen);
}

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

	for (i = 0; i < tp->md5sig_info->entries4; i++) {
		if (tp->md5sig_info->keys4[i].addr == addr) {
			/* Free the key */
941
			kfree(tp->md5sig_info->keys4[i].base.key);
942 943 944 945 946
			tp->md5sig_info->entries4--;

			if (tp->md5sig_info->entries4 == 0) {
				kfree(tp->md5sig_info->keys4);
				tp->md5sig_info->keys4 = NULL;
947
				tp->md5sig_info->alloced4 = 0;
948
			} else if (tp->md5sig_info->entries4 != i) {
949
				/* Need to do some manipulation */
950 951 952 953
				memmove(&tp->md5sig_info->keys4[i],
					&tp->md5sig_info->keys4[i+1],
					(tp->md5sig_info->entries4 - i) *
					 sizeof(struct tcp4_md5sig_key));
954 955 956 957 958 959 960 961 962 963
			}
			tcp_free_md5sig_pool();
			return 0;
		}
	}
	return -ENOENT;
}

EXPORT_SYMBOL(tcp_v4_md5_do_del);

964
static void tcp_v4_clear_md5_list(struct sock *sk)
965 966 967 968 969 970 971 972 973 974
{
	struct tcp_sock *tp = tcp_sk(sk);

	/* Free each key, then the set of key keys,
	 * the crypto element, and then decrement our
	 * hold on the last resort crypto.
	 */
	if (tp->md5sig_info->entries4) {
		int i;
		for (i = 0; i < tp->md5sig_info->entries4; i++)
975
			kfree(tp->md5sig_info->keys4[i].base.key);
976 977 978 979 980 981 982 983 984 985
		tp->md5sig_info->entries4 = 0;
		tcp_free_md5sig_pool();
	}
	if (tp->md5sig_info->keys4) {
		kfree(tp->md5sig_info->keys4);
		tp->md5sig_info->keys4 = NULL;
		tp->md5sig_info->alloced4  = 0;
	}
}

986 987
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
988 989 990 991 992 993 994 995
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;
	u8 *newkey;

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

996
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
		return -EFAULT;

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

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

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

	if (!tcp_sk(sk)->md5sig_info) {
		struct tcp_sock *tp = tcp_sk(sk);
1013
		struct tcp_md5sig_info *p;
1014

1015
		p = kzalloc(sizeof(*p), sk->sk_allocation);
1016 1017 1018 1019
		if (!p)
			return -EINVAL;

		tp->md5sig_info = p;
1020
		sk->sk_route_caps &= ~NETIF_F_GSO_MASK;
1021 1022
	}

1023
	newkey = kmemdup(cmd.tcpm_key, cmd.tcpm_keylen, sk->sk_allocation);
1024 1025 1026 1027 1028 1029
	if (!newkey)
		return -ENOMEM;
	return tcp_v4_md5_do_add(sk, sin->sin_addr.s_addr,
				 newkey, cmd.tcpm_keylen);
}

1030 1031
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1032 1033
{
	struct tcp4_pseudohdr *bp;
1034
	struct scatterlist sg;
1035 1036 1037 1038

	bp = &hp->md5_blk.ip4;

	/*
1039
	 * 1. the TCP pseudo-header (in the order: source IP address,
1040 1041 1042 1043 1044 1045
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1046
	bp->protocol = IPPROTO_TCP;
1047
	bp->len = cpu_to_be16(nbytes);
1048

1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
	sg_init_one(&sg, bp, sizeof(*bp));
	return crypto_hash_update(&hp->md5_desc, &sg, sizeof(*bp));
}

static int tcp_v4_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
			       __be32 daddr, __be32 saddr, struct tcphdr *th)
{
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;

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

	if (crypto_hash_init(desc))
		goto clear_hash;
	if (tcp_v4_md5_hash_pseudoheader(hp, daddr, saddr, th->doff << 2))
		goto clear_hash;
	if (tcp_md5_hash_header(hp, th))
		goto clear_hash;
	if (tcp_md5_hash_key(hp, key))
		goto clear_hash;
	if (crypto_hash_final(desc, md5_hash))
1073 1074 1075 1076
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1077

1078 1079 1080 1081
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1082
	return 1;
1083 1084
}

1085 1086 1087
int tcp_v4_md5_hash_skb(char *md5_hash, struct tcp_md5sig_key *key,
			struct sock *sk, struct request_sock *req,
			struct sk_buff *skb)
1088
{
1089 1090 1091
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
	struct tcphdr *th = tcp_hdr(skb);
1092 1093 1094
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1095 1096
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1097 1098 1099
	} else if (req) {
		saddr = inet_rsk(req)->loc_addr;
		daddr = inet_rsk(req)->rmt_addr;
1100
	} else {
1101 1102 1103
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
1104
	}
1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132

	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;
1133 1134
}

1135
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1136

1137
static int tcp_v4_inbound_md5_hash(struct sock *sk, struct sk_buff *skb)
1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148
{
	/*
	 * This gets called for each TCP segment that arrives
	 * so we want to be efficient.
	 * We have 3 drop cases:
	 * o No MD5 hash and one expected.
	 * o MD5 hash and we're not expecting one.
	 * o MD5 hash and its wrong.
	 */
	__u8 *hash_location = NULL;
	struct tcp_md5sig_key *hash_expected;
1149
	const struct iphdr *iph = ip_hdr(skb);
1150
	struct tcphdr *th = tcp_hdr(skb);
1151 1152 1153 1154
	int genhash;
	unsigned char newhash[16];

	hash_expected = tcp_v4_md5_do_lookup(sk, iph->saddr);
1155
	hash_location = tcp_parse_md5sig_option(th);
1156 1157 1158 1159 1160 1161

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

	if (hash_expected && !hash_location) {
1162
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
1163 1164 1165 1166
		return 1;
	}

	if (!hash_expected && hash_location) {
1167
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
1168 1169 1170 1171 1172 1173
		return 1;
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1174 1175 1176
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1177 1178 1179

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
		if (net_ratelimit()) {
1180 1181 1182
			printk(KERN_INFO "MD5 Hash failed for (%pI4, %d)->(%pI4, %d)%s\n",
			       &iph->saddr, ntohs(th->source),
			       &iph->daddr, ntohs(th->dest),
1183 1184 1185 1186 1187 1188 1189 1190 1191
			       genhash ? " tcp_v4_calc_md5_hash failed" : "");
		}
		return 1;
	}
	return 0;
}

#endif

1192
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1193
	.family		=	PF_INET,
1194
	.obj_size	=	sizeof(struct tcp_request_sock),
L
Linus Torvalds 已提交
1195
	.rtx_syn_ack	=	tcp_v4_send_synack,
1196 1197
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1198 1199 1200
	.send_reset	=	tcp_v4_send_reset,
};

1201
#ifdef CONFIG_TCP_MD5SIG
1202
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1203
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1204
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1205
};
1206
#endif
1207

1208 1209 1210
static struct timewait_sock_ops tcp_timewait_sock_ops = {
	.twsk_obj_size	= sizeof(struct tcp_timewait_sock),
	.twsk_unique	= tcp_twsk_unique,
1211
	.twsk_destructor= tcp_twsk_destructor,
1212 1213
};

L
Linus Torvalds 已提交
1214 1215
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
1216
	struct tcp_extend_values tmp_ext;
L
Linus Torvalds 已提交
1217
	struct tcp_options_received tmp_opt;
1218
	u8 *hash_location;
1219
	struct request_sock *req;
1220
	struct inet_request_sock *ireq;
1221
	struct tcp_sock *tp = tcp_sk(sk);
1222
	struct dst_entry *dst = NULL;
1223 1224
	__be32 saddr = ip_hdr(skb)->saddr;
	__be32 daddr = ip_hdr(skb)->daddr;
L
Linus Torvalds 已提交
1225 1226 1227 1228 1229 1230 1231 1232
	__u32 isn = TCP_SKB_CB(skb)->when;
#ifdef CONFIG_SYN_COOKIES
	int want_cookie = 0;
#else
#define want_cookie 0 /* Argh, why doesn't gcc optimize this :( */
#endif

	/* Never answer to SYNs send to broadcast or multicast */
E
Eric Dumazet 已提交
1233
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1234 1235 1236 1237 1238 1239
		goto drop;

	/* TW buckets are converted to open requests without
	 * limitations, they conserve resources and peer is
	 * evidently real one.
	 */
1240
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
L
Linus Torvalds 已提交
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253
#ifdef CONFIG_SYN_COOKIES
		if (sysctl_tcp_syncookies) {
			want_cookie = 1;
		} else
#endif
		goto drop;
	}

	/* Accept backlog is full. If we have already queued enough
	 * of warm entries in syn queue, drop request. It is better than
	 * clogging syn queue with openreqs with exponentially increasing
	 * timeout.
	 */
1254
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1255 1256
		goto drop;

1257
	req = inet_reqsk_alloc(&tcp_request_sock_ops);
L
Linus Torvalds 已提交
1258 1259 1260
	if (!req)
		goto drop;

1261 1262 1263 1264
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv4_ops;
#endif

1265 1266 1267 1268 1269 1270 1271 1272 1273 1274
	ireq = inet_rsk(req);
	ireq->loc_addr = daddr;
	ireq->rmt_addr = saddr;
	ireq->no_srccheck = inet_sk(sk)->transparent;
	ireq->opt = tcp_v4_save_options(sk, skb);

	dst = inet_csk_route_req(sk, req);
	if(!dst)
		goto drop_and_free;

L
Linus Torvalds 已提交
1275
	tcp_clear_options(&tmp_opt);
1276
	tmp_opt.mss_clamp = TCP_MSS_DEFAULT;
1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
	tmp_opt.user_mss  = tp->rx_opt.user_mss;
	tcp_parse_options(skb, &tmp_opt, &hash_location, 0, dst);

	if (tmp_opt.cookie_plus > 0 &&
	    tmp_opt.saw_tstamp &&
	    !tp->rx_opt.cookie_out_never &&
	    (sysctl_tcp_cookie_size > 0 ||
	     (tp->cookie_values != NULL &&
	      tp->cookie_values->cookie_desired > 0))) {
		u8 *c;
		u32 *mess = &tmp_ext.cookie_bakery[COOKIE_DIGEST_WORDS];
		int l = tmp_opt.cookie_plus - TCPOLEN_COOKIE_BASE;

		if (tcp_cookie_generator(&tmp_ext.cookie_bakery[0]) != 0)
			goto drop_and_release;

		/* Secret recipe starts with IP addresses */
		*mess++ ^= daddr;
		*mess++ ^= saddr;
L
Linus Torvalds 已提交
1296

1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314
		/* plus variable length Initiator Cookie */
		c = (u8 *)mess;
		while (l-- > 0)
			*c++ ^= *hash_location++;

#ifdef CONFIG_SYN_COOKIES
		want_cookie = 0;	/* not our kind of cookie */
#endif
		tmp_ext.cookie_out_never = 0; /* false */
		tmp_ext.cookie_plus = tmp_opt.cookie_plus;
	} else if (!tp->rx_opt.cookie_in_always) {
		/* redundant indications, but ensure initialization. */
		tmp_ext.cookie_out_never = 1; /* true */
		tmp_ext.cookie_plus = 0;
	} else {
		goto drop_and_release;
	}
	tmp_ext.cookie_in_always = tp->rx_opt.cookie_in_always;
L
Linus Torvalds 已提交
1315

1316
	if (want_cookie && !tmp_opt.saw_tstamp)
L
Linus Torvalds 已提交
1317 1318 1319 1320 1321
		tcp_clear_options(&tmp_opt);

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

1322
	if (security_inet_conn_request(sk, skb, req))
1323
		goto drop_and_release;
1324

L
Linus Torvalds 已提交
1325
	if (!want_cookie)
1326
		TCP_ECN_create_request(req, tcp_hdr(skb));
L
Linus Torvalds 已提交
1327 1328 1329 1330

	if (want_cookie) {
#ifdef CONFIG_SYN_COOKIES
		syn_flood_warning(skb);
1331
		req->cookie_ts = tmp_opt.tstamp_ok;
L
Linus Torvalds 已提交
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
#endif
		isn = cookie_v4_init_sequence(sk, skb, &req->mss);
	} else if (!isn) {
		struct inet_peer *peer = NULL;

		/* 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 &&
1347
		    tcp_death_row.sysctl_tw_recycle &&
L
Linus Torvalds 已提交
1348 1349
		    (peer = rt_get_peer((struct rtable *)dst)) != NULL &&
		    peer->v4daddr == saddr) {
E
Eric Dumazet 已提交
1350
			if ((u32)get_seconds() - peer->tcp_ts_stamp < TCP_PAWS_MSL &&
L
Linus Torvalds 已提交
1351 1352
			    (s32)(peer->tcp_ts - req->ts_recent) >
							TCP_PAWS_WINDOW) {
1353
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1354
				goto drop_and_release;
L
Linus Torvalds 已提交
1355 1356 1357 1358
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
1359
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
L
Linus Torvalds 已提交
1360 1361 1362 1363 1364 1365 1366 1367 1368 1369
			  (sysctl_max_syn_backlog >> 2)) &&
			 (!peer || !peer->tcp_ts_stamp) &&
			 (!dst || !dst_metric(dst, RTAX_RTT))) {
			/* Without syncookies last quarter of
			 * backlog is filled with destinations,
			 * proven to be alive.
			 * It means that we continue to communicate
			 * to destinations, already remembered
			 * to the moment of synflood.
			 */
1370 1371
			LIMIT_NETDEBUG(KERN_DEBUG "TCP: drop open request from %pI4/%u\n",
				       &saddr, ntohs(tcp_hdr(skb)->source));
1372
			goto drop_and_release;
L
Linus Torvalds 已提交
1373 1374
		}

1375
		isn = tcp_v4_init_sequence(skb);
L
Linus Torvalds 已提交
1376
	}
1377
	tcp_rsk(req)->snt_isn = isn;
L
Linus Torvalds 已提交
1378

1379 1380 1381
	if (__tcp_v4_send_synack(sk, dst, req,
				 (struct request_values *)&tmp_ext) ||
	    want_cookie)
L
Linus Torvalds 已提交
1382 1383
		goto drop_and_free;

1384
	inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
L
Linus Torvalds 已提交
1385 1386
	return 0;

1387 1388
drop_and_release:
	dst_release(dst);
L
Linus Torvalds 已提交
1389
drop_and_free:
1390
	reqsk_free(req);
L
Linus Torvalds 已提交
1391 1392 1393 1394 1395 1396 1397 1398 1399 1400
drop:
	return 0;
}


/*
 * 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,
1401
				  struct request_sock *req,
L
Linus Torvalds 已提交
1402 1403
				  struct dst_entry *dst)
{
1404
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1405 1406 1407
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1408 1409 1410
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
L
Linus Torvalds 已提交
1411 1412 1413 1414

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

1415
	if (!dst && (dst = inet_csk_route_req(sk, req)) == NULL)
L
Linus Torvalds 已提交
1416 1417 1418 1419 1420 1421
		goto exit;

	newsk = tcp_create_openreq_child(sk, req, skb);
	if (!newsk)
		goto exit;

1422
	newsk->sk_gso_type = SKB_GSO_TCPV4;
1423
	sk_setup_caps(newsk, dst);
L
Linus Torvalds 已提交
1424 1425 1426

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1427
	ireq		      = inet_rsk(req);
E
Eric Dumazet 已提交
1428 1429 1430
	newinet->inet_daddr   = ireq->rmt_addr;
	newinet->inet_rcv_saddr = ireq->loc_addr;
	newinet->inet_saddr	      = ireq->loc_addr;
1431 1432
	newinet->opt	      = ireq->opt;
	ireq->opt	      = NULL;
1433
	newinet->mc_index     = inet_iif(skb);
1434
	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
1435
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
1436
	if (newinet->opt)
1437
		inet_csk(newsk)->icsk_ext_hdr_len = newinet->opt->optlen;
E
Eric Dumazet 已提交
1438
	newinet->inet_id = newtp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
1439

J
John Heffner 已提交
1440
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1441 1442
	tcp_sync_mss(newsk, dst_mtu(dst));
	newtp->advmss = dst_metric(dst, RTAX_ADVMSS);
1443 1444 1445 1446
	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
1447 1448
	tcp_initialize_rcv_mss(newsk);

1449 1450
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1451 1452
	key = tcp_v4_md5_do_lookup(sk, newinet->inet_daddr);
	if (key != NULL) {
1453 1454 1455 1456 1457 1458
		/*
		 * 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.
		 */
1459 1460
		char *newkey = kmemdup(key->key, key->keylen, GFP_ATOMIC);
		if (newkey != NULL)
E
Eric Dumazet 已提交
1461
			tcp_v4_md5_do_add(newsk, newinet->inet_daddr,
1462
					  newkey, key->keylen);
1463
		newsk->sk_route_caps &= ~NETIF_F_GSO_MASK;
1464 1465 1466
	}
#endif

1467 1468
	__inet_hash_nolisten(newsk);
	__inet_inherit_port(sk, newsk);
L
Linus Torvalds 已提交
1469 1470 1471 1472

	return newsk;

exit_overflow:
1473
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
L
Linus Torvalds 已提交
1474
exit:
1475
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1476 1477 1478 1479 1480 1481
	dst_release(dst);
	return NULL;
}

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1482
	struct tcphdr *th = tcp_hdr(skb);
1483
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1484
	struct sock *nsk;
1485
	struct request_sock **prev;
L
Linus Torvalds 已提交
1486
	/* Find possible connection requests. */
1487 1488
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1489 1490 1491
	if (req)
		return tcp_check_req(sk, skb, req, prev);

1492
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1493
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1494 1495 1496 1497 1498 1499

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1500
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
	if (!th->rst && !th->syn && th->ack)
		sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
#endif
	return sk;
}

1511
static __sum16 tcp_v4_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1512
{
1513 1514
	const struct iphdr *iph = ip_hdr(skb);

1515
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
1516 1517
		if (!tcp_v4_check(skb->len, iph->saddr,
				  iph->daddr, skb->csum)) {
1518
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1519
			return 0;
1520
		}
L
Linus Torvalds 已提交
1521
	}
1522

1523
	skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
1524 1525
				       skb->len, IPPROTO_TCP, 0);

L
Linus Torvalds 已提交
1526
	if (skb->len <= 76) {
1527
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
	}
	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)
{
1543 1544 1545 1546 1547 1548 1549 1550
	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
	 */
1551
	if (tcp_v4_inbound_md5_hash(sk, skb))
1552 1553 1554
		goto discard;
#endif

L
Linus Torvalds 已提交
1555 1556
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
		TCP_CHECK_TIMER(sk);
1557
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
1558
			rsk = sk;
L
Linus Torvalds 已提交
1559
			goto reset;
1560
		}
L
Linus Torvalds 已提交
1561 1562 1563 1564
		TCP_CHECK_TIMER(sk);
		return 0;
	}

1565
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1566 1567 1568 1569 1570 1571 1572 1573
		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) {
1574 1575
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1576
				goto reset;
1577
			}
L
Linus Torvalds 已提交
1578 1579 1580 1581 1582
			return 0;
		}
	}

	TCP_CHECK_TIMER(sk);
1583
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1584
		rsk = sk;
L
Linus Torvalds 已提交
1585
		goto reset;
1586
	}
L
Linus Torvalds 已提交
1587 1588 1589 1590
	TCP_CHECK_TIMER(sk);
	return 0;

reset:
1591
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601
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:
1602
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1603 1604 1605 1606 1607 1608 1609 1610 1611
	goto discard;
}

/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1612
	const struct iphdr *iph;
L
Linus Torvalds 已提交
1613 1614 1615
	struct tcphdr *th;
	struct sock *sk;
	int ret;
1616
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1617 1618 1619 1620 1621

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

	/* Count it even if it's bad */
1622
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1623 1624 1625 1626

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

1627
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1628 1629 1630 1631 1632 1633 1634 1635

	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
1636
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1637
	 * So, we defer the checks. */
1638
	if (!skb_csum_unnecessary(skb) && tcp_v4_checksum_init(skb))
L
Linus Torvalds 已提交
1639 1640
		goto bad_packet;

1641
	th = tcp_hdr(skb);
1642
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1643 1644 1645 1646 1647
	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;
1648
	TCP_SKB_CB(skb)->flags	 = iph->tos;
L
Linus Torvalds 已提交
1649 1650
	TCP_SKB_CB(skb)->sacked	 = 0;

1651
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1652 1653 1654 1655 1656 1657 1658 1659 1660
	if (!sk)
		goto no_tcp_socket;

process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1661
	nf_reset(skb);
L
Linus Torvalds 已提交
1662

1663
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1664 1665 1666 1667
		goto discard_and_relse;

	skb->dev = NULL;

1668
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1669 1670
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1671 1672 1673
#ifdef CONFIG_NET_DMA
		struct tcp_sock *tp = tcp_sk(sk);
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1674
			tp->ucopy.dma_chan = dma_find_channel(DMA_MEMCPY);
C
Chris Leech 已提交
1675
		if (tp->ucopy.dma_chan)
L
Linus Torvalds 已提交
1676
			ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1677 1678 1679 1680
		else
#endif
		{
			if (!tcp_prequeue(sk, skb))
S
Shan Wei 已提交
1681
				ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1682
		}
L
Linus Torvalds 已提交
1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696
	} else
		sk_add_backlog(sk, skb);
	bh_unlock_sock(sk);

	sock_put(sk);

	return ret;

no_tcp_socket:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
		goto discard_it;

	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
bad_packet:
1697
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1698
	} else {
1699
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1700 1701 1702 1703 1704
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
1705
	return 0;
L
Linus Torvalds 已提交
1706 1707 1708 1709 1710 1711 1712

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1713
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1714 1715 1716 1717
		goto discard_it;
	}

	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
1718
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1719
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1720 1721
		goto discard_it;
	}
1722
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1723
	case TCP_TW_SYN: {
1724
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
1725
							&tcp_hashinfo,
1726
							iph->daddr, th->dest,
1727
							inet_iif(skb));
L
Linus Torvalds 已提交
1728
		if (sk2) {
1729 1730
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759
			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;
}

/* VJ's idea. Save last timestamp seen from this destination
 * and hold it at least for normal timewait interval to use for duplicate
 * segment detection in subsequent connections, before they enter synchronized
 * state.
 */

int tcp_v4_remember_stamp(struct sock *sk)
{
	struct inet_sock *inet = inet_sk(sk);
	struct tcp_sock *tp = tcp_sk(sk);
	struct rtable *rt = (struct rtable *)__sk_dst_get(sk);
	struct inet_peer *peer = NULL;
	int release_it = 0;

E
Eric Dumazet 已提交
1760 1761
	if (!rt || rt->rt_dst != inet->inet_daddr) {
		peer = inet_getpeer(inet->inet_daddr, 1);
L
Linus Torvalds 已提交
1762 1763 1764 1765 1766 1767 1768 1769 1770
		release_it = 1;
	} else {
		if (!rt->peer)
			rt_bind_peer(rt, 1);
		peer = rt->peer;
	}

	if (peer) {
		if ((s32)(peer->tcp_ts - tp->rx_opt.ts_recent) <= 0 ||
E
Eric Dumazet 已提交
1771 1772 1773
		    ((u32)get_seconds() - peer->tcp_ts_stamp > TCP_PAWS_MSL &&
		     peer->tcp_ts_stamp <= (u32)tp->rx_opt.ts_recent_stamp)) {
			peer->tcp_ts_stamp = (u32)tp->rx_opt.ts_recent_stamp;
L
Linus Torvalds 已提交
1774 1775 1776 1777 1778 1779 1780 1781 1782 1783
			peer->tcp_ts = tp->rx_opt.ts_recent;
		}
		if (release_it)
			inet_putpeer(peer);
		return 1;
	}

	return 0;
}

1784
int tcp_v4_tw_remember_stamp(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
1785
{
1786
	struct inet_peer *peer = inet_getpeer(tw->tw_daddr, 1);
L
Linus Torvalds 已提交
1787 1788

	if (peer) {
1789 1790 1791
		const struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);

		if ((s32)(peer->tcp_ts - tcptw->tw_ts_recent) <= 0 ||
E
Eric Dumazet 已提交
1792 1793 1794
		    ((u32)get_seconds() - peer->tcp_ts_stamp > TCP_PAWS_MSL &&
		     peer->tcp_ts_stamp <= (u32)tcptw->tw_ts_recent_stamp)) {
			peer->tcp_ts_stamp = (u32)tcptw->tw_ts_recent_stamp;
1795
			peer->tcp_ts	   = tcptw->tw_ts_recent;
L
Linus Torvalds 已提交
1796 1797 1798 1799 1800 1801 1802 1803
		}
		inet_putpeer(peer);
		return 1;
	}

	return 0;
}

1804
const struct inet_connection_sock_af_ops ipv4_specific = {
1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815
	.queue_xmit	   = ip_queue_xmit,
	.send_check	   = tcp_v4_send_check,
	.rebuild_header	   = inet_sk_rebuild_header,
	.conn_request	   = tcp_v4_conn_request,
	.syn_recv_sock	   = tcp_v4_syn_recv_sock,
	.remember_stamp	   = tcp_v4_remember_stamp,
	.net_header_len	   = sizeof(struct iphdr),
	.setsockopt	   = ip_setsockopt,
	.getsockopt	   = ip_getsockopt,
	.addr2sockaddr	   = inet_csk_addr2sockaddr,
	.sockaddr_len	   = sizeof(struct sockaddr_in),
1816
	.bind_conflict	   = inet_csk_bind_conflict,
1817
#ifdef CONFIG_COMPAT
1818 1819
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
1820
#endif
L
Linus Torvalds 已提交
1821 1822
};

1823
#ifdef CONFIG_TCP_MD5SIG
1824
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
1825
	.md5_lookup		= tcp_v4_md5_lookup,
1826
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
1827 1828 1829
	.md5_add		= tcp_v4_md5_add_func,
	.md5_parse		= tcp_v4_parse_md5_keys,
};
1830
#endif
1831

L
Linus Torvalds 已提交
1832 1833 1834 1835 1836
/* 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)
{
1837
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1838 1839 1840 1841 1842 1843
	struct tcp_sock *tp = tcp_sk(sk);

	skb_queue_head_init(&tp->out_of_order_queue);
	tcp_init_xmit_timers(sk);
	tcp_prequeue_init(tp);

1844
	icsk->icsk_rto = TCP_TIMEOUT_INIT;
L
Linus Torvalds 已提交
1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856
	tp->mdev = TCP_TIMEOUT_INIT;

	/* So many TCP implementations out there (incorrectly) count the
	 * initial SYN frame in their delayed-ACK and congestion control
	 * algorithms that we must have the following bandaid to talk
	 * efficiently to them.  -DaveM
	 */
	tp->snd_cwnd = 2;

	/* See draft-stevens-tcpca-spec-01 for discussion of the
	 * initialization of these values.
	 */
I
Ilpo Järvinen 已提交
1857
	tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
L
Linus Torvalds 已提交
1858
	tp->snd_cwnd_clamp = ~0;
1859
	tp->mss_cache = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
1860 1861

	tp->reordering = sysctl_tcp_reordering;
1862
	icsk->icsk_ca_ops = &tcp_init_congestion_ops;
L
Linus Torvalds 已提交
1863 1864 1865 1866 1867 1868

	sk->sk_state = TCP_CLOSE;

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

1869
	icsk->icsk_af_ops = &ipv4_specific;
1870
	icsk->icsk_sync_mss = tcp_sync_mss;
1871 1872 1873
#ifdef CONFIG_TCP_MD5SIG
	tp->af_specific = &tcp_sock_ipv4_specific;
#endif
L
Linus Torvalds 已提交
1874

1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887
	/* TCP Cookie Transactions */
	if (sysctl_tcp_cookie_size > 0) {
		/* Default, cookies without s_data_payload. */
		tp->cookie_values =
			kzalloc(sizeof(*tp->cookie_values),
				sk->sk_allocation);
		if (tp->cookie_values != NULL)
			kref_init(&tp->cookie_values->kref);
	}
	/* Presumed zeroed, in order of appearance:
	 *	cookie_in_always, cookie_out_never,
	 *	s_data_constant, s_data_in, s_data_out
	 */
L
Linus Torvalds 已提交
1888 1889 1890
	sk->sk_sndbuf = sysctl_tcp_wmem[1];
	sk->sk_rcvbuf = sysctl_tcp_rmem[1];

H
Herbert Xu 已提交
1891
	local_bh_disable();
1892
	percpu_counter_inc(&tcp_sockets_allocated);
H
Herbert Xu 已提交
1893
	local_bh_enable();
L
Linus Torvalds 已提交
1894 1895 1896 1897

	return 0;
}

1898
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1899 1900 1901 1902 1903
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

1904
	tcp_cleanup_congestion_control(sk);
1905

L
Linus Torvalds 已提交
1906
	/* Cleanup up the write buffer. */
1907
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
1908 1909

	/* Cleans up our, hopefully empty, out_of_order_queue. */
1910
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
1911

1912 1913 1914 1915 1916 1917 1918 1919 1920
#ifdef CONFIG_TCP_MD5SIG
	/* Clean up the MD5 key list, if any */
	if (tp->md5sig_info) {
		tcp_v4_clear_md5_list(sk);
		kfree(tp->md5sig_info);
		tp->md5sig_info = NULL;
	}
#endif

C
Chris Leech 已提交
1921 1922
#ifdef CONFIG_NET_DMA
	/* Cleans up our sk_async_wait_queue */
1923
	__skb_queue_purge(&sk->sk_async_wait_queue);
C
Chris Leech 已提交
1924 1925
#endif

L
Linus Torvalds 已提交
1926 1927 1928 1929
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
1930
	if (inet_csk(sk)->icsk_bind_hash)
1931
		inet_put_port(sk);
L
Linus Torvalds 已提交
1932 1933 1934 1935 1936 1937 1938 1939 1940

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

1941 1942 1943 1944 1945 1946 1947
	/* TCP Cookie Transactions */
	if (tp->cookie_values != NULL) {
		kref_put(&tp->cookie_values->kref,
			 tcp_cookie_values_release);
		tp->cookie_values = NULL;
	}

1948
	percpu_counter_dec(&tcp_sockets_allocated);
L
Linus Torvalds 已提交
1949 1950 1951 1952 1953 1954 1955
}

EXPORT_SYMBOL(tcp_v4_destroy_sock);

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

1956
static inline struct inet_timewait_sock *tw_head(struct hlist_nulls_head *head)
L
Linus Torvalds 已提交
1957
{
1958
	return hlist_nulls_empty(head) ? NULL :
1959
		list_entry(head->first, struct inet_timewait_sock, tw_node);
L
Linus Torvalds 已提交
1960 1961
}

1962
static inline struct inet_timewait_sock *tw_next(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
1963
{
1964 1965
	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
1966 1967 1968 1969
}

static void *listening_get_next(struct seq_file *seq, void *cur)
{
1970
	struct inet_connection_sock *icsk;
1971
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
1972
	struct sock *sk = cur;
1973
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
1974
	struct tcp_iter_state *st = seq->private;
1975
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
1976 1977 1978

	if (!sk) {
		st->bucket = 0;
1979 1980
		ilb = &tcp_hashinfo.listening_hash[0];
		spin_lock_bh(&ilb->lock);
1981
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
1982 1983
		goto get_sk;
	}
1984
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
1985 1986 1987
	++st->num;

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
1988
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
1989

1990
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
1991 1992 1993
		req = req->dl_next;
		while (1) {
			while (req) {
1994
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
1995 1996 1997 1998 1999
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
2000
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
2001 2002
				break;
get_req:
2003
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
2004 2005 2006
		}
		sk	  = sk_next(st->syn_wait_sk);
		st->state = TCP_SEQ_STATE_LISTENING;
2007
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2008
	} else {
2009
		icsk = inet_csk(sk);
2010 2011
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
2012
			goto start_req;
2013
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2014 2015 2016
		sk = sk_next(sk);
	}
get_sk:
2017
	sk_nulls_for_each_from(sk, node) {
2018
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2019 2020 2021
			cur = sk;
			goto out;
		}
2022
		icsk = inet_csk(sk);
2023 2024
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
2025 2026 2027 2028 2029 2030 2031
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;
		}
2032
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2033
	}
2034
	spin_unlock_bh(&ilb->lock);
2035
	if (++st->bucket < INET_LHTABLE_SIZE) {
2036 2037
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
2038
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
	void *rc = listening_get_next(seq, NULL);

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

2057 2058
static inline int empty_bucket(struct tcp_iter_state *st)
{
2059 2060
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain) &&
		hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].twchain);
2061 2062
}

L
Linus Torvalds 已提交
2063 2064
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
2065
	struct tcp_iter_state *st = seq->private;
2066
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2067 2068
	void *rc = NULL;

2069
	for (st->bucket = 0; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
2070
		struct sock *sk;
2071
		struct hlist_nulls_node *node;
2072
		struct inet_timewait_sock *tw;
2073
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
2074

2075 2076 2077 2078
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

2079
		spin_lock_bh(lock);
2080
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
2081
			if (sk->sk_family != st->family ||
2082
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2083 2084 2085 2086 2087 2088
				continue;
			}
			rc = sk;
			goto out;
		}
		st->state = TCP_SEQ_STATE_TIME_WAIT;
2089
		inet_twsk_for_each(tw, node,
2090
				   &tcp_hashinfo.ehash[st->bucket].twchain) {
2091
			if (tw->tw_family != st->family ||
2092
			    !net_eq(twsk_net(tw), net)) {
L
Linus Torvalds 已提交
2093 2094 2095 2096 2097
				continue;
			}
			rc = tw;
			goto out;
		}
2098
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2099 2100 2101 2102 2103 2104 2105 2106 2107
		st->state = TCP_SEQ_STATE_ESTABLISHED;
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2108
	struct inet_timewait_sock *tw;
2109
	struct hlist_nulls_node *node;
J
Jianjun Kong 已提交
2110
	struct tcp_iter_state *st = seq->private;
2111
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2112 2113 2114 2115 2116 2117 2118

	++st->num;

	if (st->state == TCP_SEQ_STATE_TIME_WAIT) {
		tw = cur;
		tw = tw_next(tw);
get_tw:
2119
		while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
L
Linus Torvalds 已提交
2120 2121 2122 2123 2124 2125
			tw = tw_next(tw);
		}
		if (tw) {
			cur = tw;
			goto out;
		}
2126
		spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2127 2128
		st->state = TCP_SEQ_STATE_ESTABLISHED;

2129
		/* Look for next non empty bucket */
2130
		while (++st->bucket <= tcp_hashinfo.ehash_mask &&
2131 2132
				empty_bucket(st))
			;
2133
		if (st->bucket > tcp_hashinfo.ehash_mask)
2134 2135
			return NULL;

2136
		spin_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
2137
		sk = sk_nulls_head(&tcp_hashinfo.ehash[st->bucket].chain);
L
Linus Torvalds 已提交
2138
	} else
2139
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2140

2141
	sk_nulls_for_each_from(sk, node) {
2142
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
L
Linus Torvalds 已提交
2143 2144 2145 2146
			goto found;
	}

	st->state = TCP_SEQ_STATE_TIME_WAIT;
2147
	tw = tw_head(&tcp_hashinfo.ehash[st->bucket].twchain);
L
Linus Torvalds 已提交
2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161
	goto get_tw;
found:
	cur = sk;
out:
	return cur;
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc = established_get_first(seq);

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2162
	}
L
Linus Torvalds 已提交
2163 2164 2165 2166 2167 2168
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2169
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183

	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;
}

static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2184
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2185 2186 2187 2188 2189 2190 2191 2192
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
	return *pos ? tcp_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	void *rc = NULL;
J
Jianjun Kong 已提交
2193
	struct tcp_iter_state *st;
L
Linus Torvalds 已提交
2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221

	if (v == SEQ_START_TOKEN) {
		rc = tcp_get_idx(seq, 0);
		goto out;
	}
	st = seq->private;

	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;
			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;
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2222
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2223 2224 2225 2226

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2227 2228
			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
2229 2230 2231
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2232
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2233 2234 2235 2236
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2237
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2238 2239 2240 2241 2242 2243 2244 2245
		break;
	}
}

static int tcp_seq_open(struct inode *inode, struct file *file)
{
	struct tcp_seq_afinfo *afinfo = PDE(inode)->data;
	struct tcp_iter_state *s;
2246
	int err;
L
Linus Torvalds 已提交
2247

2248 2249 2250 2251
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2252

2253
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2254
	s->family		= afinfo->family;
2255 2256 2257
	return 0;
}

2258
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2259 2260 2261 2262
{
	int rc = 0;
	struct proc_dir_entry *p;

2263 2264 2265 2266
	afinfo->seq_fops.open		= tcp_seq_open;
	afinfo->seq_fops.read		= seq_read;
	afinfo->seq_fops.llseek		= seq_lseek;
	afinfo->seq_fops.release	= seq_release_net;
2267

2268 2269 2270 2271
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2272 2273 2274
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
			     &afinfo->seq_fops, afinfo);
	if (!p)
L
Linus Torvalds 已提交
2275 2276 2277 2278
		rc = -ENOMEM;
	return rc;
}

2279
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2280
{
2281
	proc_net_remove(net, afinfo->name);
L
Linus Torvalds 已提交
2282 2283
}

2284
static void get_openreq4(struct sock *sk, struct request_sock *req,
2285
			 struct seq_file *f, int i, int uid, int *len)
L
Linus Torvalds 已提交
2286
{
2287
	const struct inet_request_sock *ireq = inet_rsk(req);
L
Linus Torvalds 已提交
2288 2289
	int ttd = req->expires - jiffies;

2290 2291
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %p%n",
L
Linus Torvalds 已提交
2292
		i,
2293
		ireq->loc_addr,
E
Eric Dumazet 已提交
2294
		ntohs(inet_sk(sk)->inet_sport),
2295 2296
		ireq->rmt_addr,
		ntohs(ireq->rmt_port),
L
Linus Torvalds 已提交
2297 2298 2299 2300 2301 2302 2303 2304 2305
		TCP_SYN_RECV,
		0, 0, /* could print option size, but that is af dependent. */
		1,    /* timers active (only the expire timer) */
		jiffies_to_clock_t(ttd),
		req->retrans,
		uid,
		0,  /* non standard timer */
		0, /* open_requests have no inode */
		atomic_read(&sk->sk_refcnt),
2306 2307
		req,
		len);
L
Linus Torvalds 已提交
2308 2309
}

2310
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2311 2312 2313
{
	int timer_active;
	unsigned long timer_expires;
2314 2315 2316
	struct tcp_sock *tp = tcp_sk(sk);
	const struct inet_connection_sock *icsk = inet_csk(sk);
	struct inet_sock *inet = inet_sk(sk);
E
Eric Dumazet 已提交
2317 2318 2319 2320
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
L
Linus Torvalds 已提交
2321

2322
	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
2323
		timer_active	= 1;
2324 2325
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2326
		timer_active	= 4;
2327
		timer_expires	= icsk->icsk_timeout;
2328
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2329
		timer_active	= 2;
2330
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2331 2332 2333 2334 2335
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2336
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
2337
			"%08X %5d %8d %lu %d %p %lu %lu %u %u %d%n",
2338
		i, src, srcp, dest, destp, sk->sk_state,
2339
		tp->write_seq - tp->snd_una,
2340
		sk->sk_state == TCP_LISTEN ? sk->sk_ack_backlog :
2341
					     (tp->rcv_nxt - tp->copied_seq),
L
Linus Torvalds 已提交
2342 2343
		timer_active,
		jiffies_to_clock_t(timer_expires - jiffies),
2344
		icsk->icsk_retransmits,
2345
		sock_i_uid(sk),
2346
		icsk->icsk_probes_out,
2347 2348
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2349 2350
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2351
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2352
		tp->snd_cwnd,
I
Ilpo Järvinen 已提交
2353
		tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh,
2354
		len);
L
Linus Torvalds 已提交
2355 2356
}

2357
static void get_timewait4_sock(struct inet_timewait_sock *tw,
2358
			       struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2359
{
2360
	__be32 dest, src;
L
Linus Torvalds 已提交
2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371
	__u16 destp, srcp;
	int ttd = tw->tw_ttd - jiffies;

	if (ttd < 0)
		ttd = 0;

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

2372 2373
	seq_printf(f, "%4d: %08X:%04X %08X:%04X"
		" %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %p%n",
L
Linus Torvalds 已提交
2374 2375
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
		3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
2376
		atomic_read(&tw->tw_refcnt), tw, len);
L
Linus Torvalds 已提交
2377 2378 2379 2380 2381 2382
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2383
	struct tcp_iter_state *st;
2384
	int len;
L
Linus Torvalds 已提交
2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397

	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:
2398
		get_tcp4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2399 2400
		break;
	case TCP_SEQ_STATE_OPENREQ:
2401
		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid, &len);
L
Linus Torvalds 已提交
2402 2403
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
2404
		get_timewait4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2405 2406
		break;
	}
2407
	seq_printf(seq, "%*s\n", TMPSZ - 1 - len, "");
L
Linus Torvalds 已提交
2408 2409 2410 2411 2412 2413 2414
out:
	return 0;
}

static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2415 2416 2417
	.seq_fops	= {
		.owner		= THIS_MODULE,
	},
2418 2419 2420
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2421 2422
};

2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437
static int tcp4_proc_init_net(struct net *net)
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

static void tcp4_proc_exit_net(struct net *net)
{
	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
2438 2439
int __init tcp4_proc_init(void)
{
2440
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2441 2442 2443 2444
}

void tcp4_proc_exit(void)
{
2445
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2446 2447 2448
}
#endif /* CONFIG_PROC_FS */

H
Herbert Xu 已提交
2449 2450
struct sk_buff **tcp4_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
H
Herbert Xu 已提交
2451
	struct iphdr *iph = skb_gro_network_header(skb);
H
Herbert Xu 已提交
2452 2453 2454

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
2455
		if (!tcp_v4_check(skb_gro_len(skb), iph->saddr, iph->daddr,
H
Herbert Xu 已提交
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
				  skb->csum)) {
			skb->ip_summed = CHECKSUM_UNNECESSARY;
			break;
		}

		/* fall through */
	case CHECKSUM_NONE:
		NAPI_GRO_CB(skb)->flush = 1;
		return NULL;
	}

	return tcp_gro_receive(head, skb);
}
EXPORT_SYMBOL(tcp4_gro_receive);

int tcp4_gro_complete(struct sk_buff *skb)
{
	struct iphdr *iph = ip_hdr(skb);
	struct tcphdr *th = tcp_hdr(skb);

	th->check = ~tcp_v4_check(skb->len - skb_transport_offset(skb),
				  iph->saddr, iph->daddr, 0);
	skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;

	return tcp_gro_complete(skb);
}
EXPORT_SYMBOL(tcp4_gro_complete);

L
Linus Torvalds 已提交
2484 2485 2486 2487 2488 2489
struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2490
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2491 2492 2493 2494 2495 2496 2497 2498
	.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,
	.backlog_rcv		= tcp_v4_do_rcv,
2499 2500 2501
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2502 2503
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
2504
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2505 2506 2507 2508 2509 2510 2511
	.memory_allocated	= &tcp_memory_allocated,
	.memory_pressure	= &tcp_memory_pressure,
	.sysctl_mem		= sysctl_tcp_mem,
	.sysctl_wmem		= sysctl_tcp_wmem,
	.sysctl_rmem		= sysctl_tcp_rmem,
	.max_header		= MAX_TCP_HEADER,
	.obj_size		= sizeof(struct tcp_sock),
2512
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2513
	.twsk_prot		= &tcp_timewait_sock_ops,
2514
	.rsk_prot		= &tcp_request_sock_ops,
2515
	.h.hashinfo		= &tcp_hashinfo,
2516 2517 2518 2519
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
L
Linus Torvalds 已提交
2520 2521
};

2522 2523 2524 2525 2526 2527 2528 2529 2530 2531

static int __net_init tcp_sk_init(struct net *net)
{
	return inet_ctl_sock_create(&net->ipv4.tcp_sock,
				    PF_INET, SOCK_RAW, IPPROTO_TCP, net);
}

static void __net_exit tcp_sk_exit(struct net *net)
{
	inet_ctl_sock_destroy(net->ipv4.tcp_sock);
E
Eric W. Biederman 已提交
2532 2533 2534 2535 2536
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2537 2538 2539
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2540 2541 2542
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2543 2544
};

2545
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2546
{
2547
	inet_hashinfo_init(&tcp_hashinfo);
2548
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567
		panic("Failed to create the TCP control socket.\n");
}

EXPORT_SYMBOL(ipv4_specific);
EXPORT_SYMBOL(tcp_hashinfo);
EXPORT_SYMBOL(tcp_prot);
EXPORT_SYMBOL(tcp_v4_conn_request);
EXPORT_SYMBOL(tcp_v4_connect);
EXPORT_SYMBOL(tcp_v4_do_rcv);
EXPORT_SYMBOL(tcp_v4_remember_stamp);
EXPORT_SYMBOL(tcp_v4_send_check);
EXPORT_SYMBOL(tcp_v4_syn_recv_sock);

#ifdef CONFIG_PROC_FS
EXPORT_SYMBOL(tcp_proc_register);
EXPORT_SYMBOL(tcp_proc_unregister);
#endif
EXPORT_SYMBOL(sysctl_tcp_low_latency);