tcp_ipv4.c 66.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Implementation of the Transmission Control Protocol(TCP).
 *
 *		IPv4 specific functions
 *
 *
 *		code split from:
 *		linux/ipv4/tcp.c
 *		linux/ipv4/tcp_input.c
 *		linux/ipv4/tcp_output.c
 *
 *		See tcp.c for author information
 *
 *	This program is free software; you can redistribute it and/or
 *      modify it under the terms of the GNU General Public License
 *      as published by the Free Software Foundation; either version
 *      2 of the License, or (at your option) any later version.
 */

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


H
Herbert Xu 已提交
54
#include <linux/bottom_half.h>
L
Linus Torvalds 已提交
55 56 57 58 59 60 61 62
#include <linux/types.h>
#include <linux/fcntl.h>
#include <linux/module.h>
#include <linux/random.h>
#include <linux/cache.h>
#include <linux/jhash.h>
#include <linux/init.h>
#include <linux/times.h>
63
#include <linux/slab.h>
L
Linus Torvalds 已提交
64

65
#include <net/net_namespace.h>
L
Linus Torvalds 已提交
66
#include <net/icmp.h>
67
#include <net/inet_hashtables.h>
L
Linus Torvalds 已提交
68
#include <net/tcp.h>
69
#include <net/transp_v6.h>
L
Linus Torvalds 已提交
70 71
#include <net/ipv6.h>
#include <net/inet_common.h>
72
#include <net/timewait_sock.h>
L
Linus Torvalds 已提交
73
#include <net/xfrm.h>
C
Chris Leech 已提交
74
#include <net/netdma.h>
L
Linus Torvalds 已提交
75 76 77 78 79 80 81

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

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

85 86
int sysctl_tcp_tw_reuse __read_mostly;
int sysctl_tcp_low_latency __read_mostly;
L
Linus Torvalds 已提交
87 88


89
#ifdef CONFIG_TCP_MD5SIG
90 91
static struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk,
						   __be32 addr);
92 93
static int tcp_v4_md5_hash_hdr(char *md5_hash, struct tcp_md5sig_key *key,
			       __be32 daddr, __be32 saddr, struct tcphdr *th);
94 95 96 97 98 99
#else
static inline
struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk, __be32 addr)
{
	return NULL;
}
100 101
#endif

102
struct inet_hashinfo tcp_hashinfo;
L
Linus Torvalds 已提交
103

104
static inline __u32 tcp_v4_init_sequence(struct sk_buff *skb)
L
Linus Torvalds 已提交
105
{
106 107
	return secure_tcp_sequence_number(ip_hdr(skb)->daddr,
					  ip_hdr(skb)->saddr,
108 109
					  tcp_hdr(skb)->dest,
					  tcp_hdr(skb)->source);
L
Linus Torvalds 已提交
110 111
}

112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129
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 &&
130
			     get_seconds() - tcptw->tw_ts_recent_stamp > 1))) {
131 132 133 134 135 136 137 138 139 140 141 142 143 144
		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
145 146 147 148 149 150 151
/* 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;
152
	__be32 daddr, nexthop;
L
Linus Torvalds 已提交
153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
	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
169
	tmp = ip_route_connect(&rt, nexthop, inet->inet_saddr,
L
Linus Torvalds 已提交
170 171
			       RT_CONN_FLAGS(sk), sk->sk_bound_dev_if,
			       IPPROTO_TCP,
E
Eric Dumazet 已提交
172
			       inet->inet_sport, usin->sin_port, sk, 1);
173 174
	if (tmp < 0) {
		if (tmp == -ENETUNREACH)
175
			IP_INC_STATS_BH(sock_net(sk), IPSTATS_MIB_OUTNOROUTES);
L
Linus Torvalds 已提交
176
		return tmp;
177
	}
L
Linus Torvalds 已提交
178 179 180 181 182 183 184 185 186

	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
187 188 189
	if (!inet->inet_saddr)
		inet->inet_saddr = rt->rt_src;
	inet->inet_rcv_saddr = inet->inet_saddr;
L
Linus Torvalds 已提交
190

E
Eric Dumazet 已提交
191
	if (tp->rx_opt.ts_recent_stamp && inet->inet_daddr != daddr) {
L
Linus Torvalds 已提交
192 193 194 195 196 197
		/* Reset inherited state */
		tp->rx_opt.ts_recent	   = 0;
		tp->rx_opt.ts_recent_stamp = 0;
		tp->write_seq		   = 0;
	}

198
	if (tcp_death_row.sysctl_tw_recycle &&
L
Linus Torvalds 已提交
199 200
	    !tp->rx_opt.ts_recent_stamp && rt->rt_dst == daddr) {
		struct inet_peer *peer = rt_get_peer(rt);
201 202 203 204 205
		/*
		 * 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
206
		 */
207 208 209 210 211 212
		if (peer) {
			inet_peer_refcheck(peer);
			if ((u32)get_seconds() - peer->tcp_ts_stamp <= TCP_PAWS_MSL) {
				tp->rx_opt.ts_recent_stamp = peer->tcp_ts_stamp;
				tp->rx_opt.ts_recent = peer->tcp_ts;
			}
L
Linus Torvalds 已提交
213 214 215
		}
	}

E
Eric Dumazet 已提交
216 217
	inet->inet_dport = usin->sin_port;
	inet->inet_daddr = daddr;
L
Linus Torvalds 已提交
218

219
	inet_csk(sk)->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
220
	if (inet->opt)
221
		inet_csk(sk)->icsk_ext_hdr_len = inet->opt->optlen;
L
Linus Torvalds 已提交
222

223
	tp->rx_opt.mss_clamp = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
224 225 226 227 228 229 230

	/* 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);
231
	err = inet_hash_connect(&tcp_death_row, sk);
L
Linus Torvalds 已提交
232 233 234
	if (err)
		goto failure;

235
	err = ip_route_newports(&rt, IPPROTO_TCP,
E
Eric Dumazet 已提交
236
				inet->inet_sport, inet->inet_dport, sk);
L
Linus Torvalds 已提交
237 238 239 240
	if (err)
		goto failure;

	/* OK, now commit destination to socket.  */
241
	sk->sk_gso_type = SKB_GSO_TCPV4;
242
	sk_setup_caps(sk, &rt->dst);
L
Linus Torvalds 已提交
243 244

	if (!tp->write_seq)
E
Eric Dumazet 已提交
245 246 247
		tp->write_seq = secure_tcp_sequence_number(inet->inet_saddr,
							   inet->inet_daddr,
							   inet->inet_sport,
L
Linus Torvalds 已提交
248 249
							   usin->sin_port);

E
Eric Dumazet 已提交
250
	inet->inet_id = tp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
251 252 253 254 255 256 257 258 259

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

	return 0;

failure:
260 261 262 263
	/*
	 * This unhashes the socket and releases the local port,
	 * if necessary.
	 */
L
Linus Torvalds 已提交
264 265 266
	tcp_set_state(sk, TCP_CLOSE);
	ip_rt_put(rt);
	sk->sk_route_caps = 0;
E
Eric Dumazet 已提交
267
	inet->inet_dport = 0;
L
Linus Torvalds 已提交
268 269 270 271 272 273
	return err;
}

/*
 * This routine does path mtu discovery as defined in RFC1191.
 */
S
Stephen Hemminger 已提交
274
static void do_pmtu_discovery(struct sock *sk, struct iphdr *iph, u32 mtu)
L
Linus Torvalds 已提交
275 276 277 278 279 280 281 282 283 284 285 286 287 288
{
	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.
289
	 * There is a small race when the user changes this flag in the
L
Linus Torvalds 已提交
290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
	 * 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 &&
306
	    inet_csk(sk)->icsk_pmtu_cookie > mtu) {
L
Linus Torvalds 已提交
307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333
		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.
 *
 */

334
void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
L
Linus Torvalds 已提交
335
{
336 337
	struct iphdr *iph = (struct iphdr *)icmp_skb->data;
	struct tcphdr *th = (struct tcphdr *)(icmp_skb->data + (iph->ihl << 2));
338
	struct inet_connection_sock *icsk;
L
Linus Torvalds 已提交
339 340
	struct tcp_sock *tp;
	struct inet_sock *inet;
341 342
	const int type = icmp_hdr(icmp_skb)->type;
	const int code = icmp_hdr(icmp_skb)->code;
L
Linus Torvalds 已提交
343
	struct sock *sk;
344
	struct sk_buff *skb;
L
Linus Torvalds 已提交
345
	__u32 seq;
346
	__u32 remaining;
L
Linus Torvalds 已提交
347
	int err;
348
	struct net *net = dev_net(icmp_skb->dev);
L
Linus Torvalds 已提交
349

350
	if (icmp_skb->len < (iph->ihl << 2) + 8) {
351
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
352 353 354
		return;
	}

355
	sk = inet_lookup(net, &tcp_hashinfo, iph->daddr, th->dest,
356
			iph->saddr, th->source, inet_iif(icmp_skb));
L
Linus Torvalds 已提交
357
	if (!sk) {
358
		ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
L
Linus Torvalds 已提交
359 360 361
		return;
	}
	if (sk->sk_state == TCP_TIME_WAIT) {
362
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
363 364 365 366 367 368 369 370
		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))
371
		NET_INC_STATS_BH(net, LINUX_MIB_LOCKDROPPEDICMPS);
L
Linus Torvalds 已提交
372 373 374 375

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

376 377 378 379 380
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
		goto out;
	}

381
	icsk = inet_csk(sk);
L
Linus Torvalds 已提交
382 383 384 385
	tp = tcp_sk(sk);
	seq = ntohl(th->seq);
	if (sk->sk_state != TCP_LISTEN &&
	    !between(seq, tp->snd_una, tp->snd_nxt)) {
386
		NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407
		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;
408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440
		/* 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
441 442 443 444 445 446 447 448 449
		break;
	case ICMP_TIME_EXCEEDED:
		err = EHOSTUNREACH;
		break;
	default:
		goto out;
	}

	switch (sk->sk_state) {
450
		struct request_sock *req, **prev;
L
Linus Torvalds 已提交
451 452 453 454
	case TCP_LISTEN:
		if (sock_owned_by_user(sk))
			goto out;

455 456
		req = inet_csk_search_req(sk, &prev, th->dest,
					  iph->daddr, iph->saddr);
L
Linus Torvalds 已提交
457 458 459 460 461 462
		if (!req)
			goto out;

		/* ICMPs are not backlogged, hence we cannot get
		   an established socket here.
		 */
463
		WARN_ON(req->sk);
L
Linus Torvalds 已提交
464

465
		if (seq != tcp_rsk(req)->snt_isn) {
466
			NET_INC_STATS_BH(net, LINUX_MIB_OUTOFWINDOWICMPS);
L
Linus Torvalds 已提交
467 468 469 470 471 472 473 474 475
			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().
		 */
476
		inet_csk_reqsk_queue_drop(sk, req, prev);
L
Linus Torvalds 已提交
477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
		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);
}

524 525
static void __tcp_v4_send_check(struct sk_buff *skb,
				__be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
526
{
527
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
528

529
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
530
		th->check = ~tcp_v4_check(skb->len, saddr, daddr, 0);
531
		skb->csum_start = skb_transport_header(skb) - skb->head;
A
Al Viro 已提交
532
		skb->csum_offset = offsetof(struct tcphdr, check);
L
Linus Torvalds 已提交
533
	} else {
534
		th->check = tcp_v4_check(skb->len, saddr, daddr,
535
					 csum_partial(th,
L
Linus Torvalds 已提交
536 537 538 539 540
						      th->doff << 2,
						      skb->csum));
	}
}

541
/* This routine computes an IPv4 TCP checksum. */
542
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb)
543 544 545 546 547 548
{
	struct inet_sock *inet = inet_sk(sk);

	__tcp_v4_send_check(skb, inet->inet_saddr, inet->inet_daddr);
}

549 550
int tcp_v4_gso_send_check(struct sk_buff *skb)
{
551
	const struct iphdr *iph;
552 553 554 555 556
	struct tcphdr *th;

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

557
	iph = ip_hdr(skb);
558
	th = tcp_hdr(skb);
559 560

	th->check = 0;
561
	skb->ip_summed = CHECKSUM_PARTIAL;
562
	__tcp_v4_send_check(skb, iph->saddr, iph->daddr);
563 564 565
	return 0;
}

L
Linus Torvalds 已提交
566 567 568 569 570 571 572 573 574 575 576 577 578
/*
 *	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.
 */

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

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

E
Eric Dumazet 已提交
598
	if (skb_rtable(skb)->rt_type != RTN_LOCAL)
L
Linus Torvalds 已提交
599 600 601
		return;

	/* Swap the send and the receive. */
602 603 604 605 606
	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
607 608

	if (th->ack) {
609
		rep.th.seq = th->ack_seq;
L
Linus Torvalds 已提交
610
	} else {
611 612 613
		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
614 615
	}

616
	memset(&arg, 0, sizeof(arg));
617 618 619 620
	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);

#ifdef CONFIG_TCP_MD5SIG
621
	key = sk ? tcp_v4_md5_do_lookup(sk, ip_hdr(skb)->daddr) : NULL;
622 623 624 625 626 627 628 629 630
	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;

631
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[1],
632 633
				     key, ip_hdr(skb)->saddr,
				     ip_hdr(skb)->daddr, &rep.th);
634 635
	}
#endif
636 637
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
638
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
L
Linus Torvalds 已提交
639
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
640
	arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
L
Linus Torvalds 已提交
641

E
Eric Dumazet 已提交
642
	net = dev_net(skb_dst(skb)->dev);
643
	ip_send_reply(net->ipv4.tcp_sock, skb,
644
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
645

646 647
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
	TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
L
Linus Torvalds 已提交
648 649 650 651 652 653
}

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

654 655
static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
			    u32 win, u32 ts, int oif,
656 657
			    struct tcp_md5sig_key *key,
			    int reply_flags)
L
Linus Torvalds 已提交
658
{
659
	struct tcphdr *th = tcp_hdr(skb);
L
Linus Torvalds 已提交
660 661
	struct {
		struct tcphdr th;
662
		__be32 opt[(TCPOLEN_TSTAMP_ALIGNED >> 2)
663
#ifdef CONFIG_TCP_MD5SIG
664
			   + (TCPOLEN_MD5SIG_ALIGNED >> 2)
665 666
#endif
			];
L
Linus Torvalds 已提交
667 668
	} rep;
	struct ip_reply_arg arg;
E
Eric Dumazet 已提交
669
	struct net *net = dev_net(skb_dst(skb)->dev);
L
Linus Torvalds 已提交
670 671

	memset(&rep.th, 0, sizeof(struct tcphdr));
672
	memset(&arg, 0, sizeof(arg));
L
Linus Torvalds 已提交
673 674 675 676

	arg.iov[0].iov_base = (unsigned char *)&rep;
	arg.iov[0].iov_len  = sizeof(rep.th);
	if (ts) {
677 678 679 680 681
		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);
682
		arg.iov[0].iov_len += TCPOLEN_TSTAMP_ALIGNED;
L
Linus Torvalds 已提交
683 684 685 686 687 688 689 690 691 692 693
	}

	/* 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);

694 695 696 697 698 699 700 701 702 703 704
#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;

705
		tcp_v4_md5_hash_hdr((__u8 *) &rep.opt[offset],
706 707
				    key, ip_hdr(skb)->saddr,
				    ip_hdr(skb)->daddr, &rep.th);
708 709
	}
#endif
710
	arg.flags = reply_flags;
711 712
	arg.csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
				      ip_hdr(skb)->saddr, /* XXX */
L
Linus Torvalds 已提交
713 714
				      arg.iov[0].iov_len, IPPROTO_TCP, 0);
	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
715 716
	if (oif)
		arg.bound_dev_if = oif;
L
Linus Torvalds 已提交
717

718
	ip_send_reply(net->ipv4.tcp_sock, skb,
719
		      &arg, arg.iov[0].iov_len);
L
Linus Torvalds 已提交
720

721
	TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
L
Linus Torvalds 已提交
722 723 724 725
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
726
	struct inet_timewait_sock *tw = inet_twsk(sk);
727
	struct tcp_timewait_sock *tcptw = tcp_twsk(sk);
L
Linus Torvalds 已提交
728

729
	tcp_v4_send_ack(skb, tcptw->tw_snd_nxt, tcptw->tw_rcv_nxt,
730
			tcptw->tw_rcv_wnd >> tw->tw_rcv_wscale,
731 732
			tcptw->tw_ts_recent,
			tw->tw_bound_dev_if,
733 734
			tcp_twsk_md5_key(tcptw),
			tw->tw_transparent ? IP_REPLY_ARG_NOSRCCHECK : 0
735
			);
L
Linus Torvalds 已提交
736

737
	inet_twsk_put(tw);
L
Linus Torvalds 已提交
738 739
}

740
static void tcp_v4_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
741
				  struct request_sock *req)
L
Linus Torvalds 已提交
742
{
743
	tcp_v4_send_ack(skb, tcp_rsk(req)->snt_isn + 1,
744
			tcp_rsk(req)->rcv_isn + 1, req->rcv_wnd,
745 746
			req->ts_recent,
			0,
747 748
			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
749 750 751
}

/*
752
 *	Send a SYN-ACK after having received a SYN.
753
 *	This still operates on a request_sock only, not on a big
L
Linus Torvalds 已提交
754 755
 *	socket.
 */
756 757 758
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
759
{
760
	const struct inet_request_sock *ireq = inet_rsk(req);
L
Linus Torvalds 已提交
761 762 763 764
	int err = -1;
	struct sk_buff * skb;

	/* First, grab a route. */
765
	if (!dst && (dst = inet_csk_route_req(sk, req)) == NULL)
766
		return -1;
L
Linus Torvalds 已提交
767

768
	skb = tcp_make_synack(sk, dst, req, rvp);
L
Linus Torvalds 已提交
769 770

	if (skb) {
771
		__tcp_v4_send_check(skb, ireq->loc_addr, ireq->rmt_addr);
L
Linus Torvalds 已提交
772

773 774 775
		err = ip_build_and_send_pkt(skb, sk, ireq->loc_addr,
					    ireq->rmt_addr,
					    ireq->opt);
776
		err = net_xmit_eval(err);
L
Linus Torvalds 已提交
777 778 779 780 781 782
	}

	dst_release(dst);
	return err;
}

783
static int tcp_v4_rtx_synack(struct sock *sk, struct request_sock *req,
784
			      struct request_values *rvp)
785
{
786 787
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
	return tcp_v4_send_synack(sk, NULL, req, rvp);
788 789
}

L
Linus Torvalds 已提交
790
/*
791
 *	IPv4 request_sock destructor.
L
Linus Torvalds 已提交
792
 */
793
static void tcp_v4_reqsk_destructor(struct request_sock *req)
L
Linus Torvalds 已提交
794
{
J
Jesper Juhl 已提交
795
	kfree(inet_rsk(req)->opt);
L
Linus Torvalds 已提交
796 797
}

798
static void syn_flood_warning(const struct sk_buff *skb)
L
Linus Torvalds 已提交
799
{
800
	const char *msg;
L
Linus Torvalds 已提交
801

802 803 804 805
#ifdef CONFIG_SYN_COOKIES
	if (sysctl_tcp_syncookies)
		msg = "Sending cookies";
	else
806
#endif
807 808 809 810 811
		msg = "Dropping request";

	pr_info("TCP: Possible SYN flooding on port %d. %s.\n",
				ntohs(tcp_hdr(skb)->dest), msg);
}
L
Linus Torvalds 已提交
812 813

/*
814
 * Save and compile IPv4 options into the request_sock if needed.
L
Linus Torvalds 已提交
815
 */
S
Stephen Hemminger 已提交
816 817
static struct ip_options *tcp_v4_save_options(struct sock *sk,
					      struct sk_buff *skb)
L
Linus Torvalds 已提交
818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834
{
	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;
}

835 836 837 838 839 840 841 842
#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.  */
843 844
static struct tcp_md5sig_key *
			tcp_v4_md5_do_lookup(struct sock *sk, __be32 addr)
845 846 847 848 849 850 851 852
{
	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)
853
			return &tp->md5sig_info->keys4[i].base;
854 855 856 857 858 859 860
	}
	return NULL;
}

struct tcp_md5sig_key *tcp_v4_md5_lookup(struct sock *sk,
					 struct sock *addr_sk)
{
E
Eric Dumazet 已提交
861
	return tcp_v4_md5_do_lookup(sk, inet_sk(addr_sk)->inet_daddr);
862 863 864 865
}

EXPORT_SYMBOL(tcp_v4_md5_lookup);

A
Adrian Bunk 已提交
866 867
static struct tcp_md5sig_key *tcp_v4_reqsk_md5_lookup(struct sock *sk,
						      struct request_sock *req)
868 869 870 871 872 873 874 875 876
{
	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 */
877
	struct tcp_md5sig_key *key;
878 879 880
	struct tcp_sock *tp = tcp_sk(sk);
	struct tcp4_md5sig_key *keys;

881
	key = tcp_v4_md5_do_lookup(sk, addr);
882 883
	if (key) {
		/* Pre-existing entry - just update that one. */
884 885 886
		kfree(key->key);
		key->key = newkey;
		key->keylen = newkeylen;
887
	} else {
888 889
		struct tcp_md5sig_info *md5sig;

890
		if (!tp->md5sig_info) {
891 892
			tp->md5sig_info = kzalloc(sizeof(*tp->md5sig_info),
						  GFP_ATOMIC);
893 894 895 896
			if (!tp->md5sig_info) {
				kfree(newkey);
				return -ENOMEM;
			}
E
Eric Dumazet 已提交
897
			sk_nocaps_add(sk, NETIF_F_GSO_MASK);
898
		}
899
		if (tcp_alloc_md5sig_pool(sk) == NULL) {
900 901 902
			kfree(newkey);
			return -ENOMEM;
		}
903 904 905 906
		md5sig = tp->md5sig_info;

		if (md5sig->alloced4 == md5sig->entries4) {
			keys = kmalloc((sizeof(*keys) *
907
					(md5sig->entries4 + 1)), GFP_ATOMIC);
908 909 910 911 912 913
			if (!keys) {
				kfree(newkey);
				tcp_free_md5sig_pool();
				return -ENOMEM;
			}

914 915 916
			if (md5sig->entries4)
				memcpy(keys, md5sig->keys4,
				       sizeof(*keys) * md5sig->entries4);
917 918

			/* Free old key list, and reference new one */
919
			kfree(md5sig->keys4);
920 921
			md5sig->keys4 = keys;
			md5sig->alloced4++;
922
		}
923
		md5sig->entries4++;
924 925 926
		md5sig->keys4[md5sig->entries4 - 1].addr        = addr;
		md5sig->keys4[md5sig->entries4 - 1].base.key    = newkey;
		md5sig->keys4[md5sig->entries4 - 1].base.keylen = newkeylen;
927 928 929 930 931 932 933 934 935
	}
	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
936
	return tcp_v4_md5_do_add(sk, inet_sk(addr_sk)->inet_daddr,
937 938 939 940 941 942 943 944 945 946 947
				 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 */
948
			kfree(tp->md5sig_info->keys4[i].base.key);
949 950 951 952 953
			tp->md5sig_info->entries4--;

			if (tp->md5sig_info->entries4 == 0) {
				kfree(tp->md5sig_info->keys4);
				tp->md5sig_info->keys4 = NULL;
954
				tp->md5sig_info->alloced4 = 0;
955
			} else if (tp->md5sig_info->entries4 != i) {
956
				/* Need to do some manipulation */
957 958 959 960
				memmove(&tp->md5sig_info->keys4[i],
					&tp->md5sig_info->keys4[i+1],
					(tp->md5sig_info->entries4 - i) *
					 sizeof(struct tcp4_md5sig_key));
961 962 963 964 965 966 967 968 969 970
			}
			tcp_free_md5sig_pool();
			return 0;
		}
	}
	return -ENOENT;
}

EXPORT_SYMBOL(tcp_v4_md5_do_del);

971
static void tcp_v4_clear_md5_list(struct sock *sk)
972 973 974 975 976 977 978 979 980 981
{
	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++)
982
			kfree(tp->md5sig_info->keys4[i].base.key);
983 984 985 986 987 988 989 990 991 992
		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;
	}
}

993 994
static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval,
				 int optlen)
995 996 997 998 999 1000 1001 1002
{
	struct tcp_md5sig cmd;
	struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.tcpm_addr;
	u8 *newkey;

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

1003
	if (copy_from_user(&cmd, optval, sizeof(cmd)))
1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019
		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);
1020
		struct tcp_md5sig_info *p;
1021

1022
		p = kzalloc(sizeof(*p), sk->sk_allocation);
1023 1024 1025 1026
		if (!p)
			return -EINVAL;

		tp->md5sig_info = p;
E
Eric Dumazet 已提交
1027
		sk_nocaps_add(sk, NETIF_F_GSO_MASK);
1028 1029
	}

1030
	newkey = kmemdup(cmd.tcpm_key, cmd.tcpm_keylen, sk->sk_allocation);
1031 1032 1033 1034 1035 1036
	if (!newkey)
		return -ENOMEM;
	return tcp_v4_md5_do_add(sk, sin->sin_addr.s_addr,
				 newkey, cmd.tcpm_keylen);
}

1037 1038
static int tcp_v4_md5_hash_pseudoheader(struct tcp_md5sig_pool *hp,
					__be32 daddr, __be32 saddr, int nbytes)
1039 1040
{
	struct tcp4_pseudohdr *bp;
1041
	struct scatterlist sg;
1042 1043 1044 1045

	bp = &hp->md5_blk.ip4;

	/*
1046
	 * 1. the TCP pseudo-header (in the order: source IP address,
1047 1048 1049 1050 1051 1052
	 * destination IP address, zero-padded protocol number, and
	 * segment length)
	 */
	bp->saddr = saddr;
	bp->daddr = daddr;
	bp->pad = 0;
1053
	bp->protocol = IPPROTO_TCP;
1054
	bp->len = cpu_to_be16(nbytes);
1055

1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079
	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))
1080 1081 1082 1083
		goto clear_hash;

	tcp_put_md5sig_pool();
	return 0;
1084

1085 1086 1087 1088
clear_hash:
	tcp_put_md5sig_pool();
clear_hash_noput:
	memset(md5_hash, 0, 16);
1089
	return 1;
1090 1091
}

1092 1093 1094
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)
1095
{
1096 1097 1098
	struct tcp_md5sig_pool *hp;
	struct hash_desc *desc;
	struct tcphdr *th = tcp_hdr(skb);
1099 1100 1101
	__be32 saddr, daddr;

	if (sk) {
E
Eric Dumazet 已提交
1102 1103
		saddr = inet_sk(sk)->inet_saddr;
		daddr = inet_sk(sk)->inet_daddr;
1104 1105 1106
	} else if (req) {
		saddr = inet_rsk(req)->loc_addr;
		daddr = inet_rsk(req)->rmt_addr;
1107
	} else {
1108 1109 1110
		const struct iphdr *iph = ip_hdr(skb);
		saddr = iph->saddr;
		daddr = iph->daddr;
1111
	}
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139

	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;
1140 1141
}

1142
EXPORT_SYMBOL(tcp_v4_md5_hash_skb);
1143

1144
static int tcp_v4_inbound_md5_hash(struct sock *sk, struct sk_buff *skb)
1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155
{
	/*
	 * 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;
1156
	const struct iphdr *iph = ip_hdr(skb);
1157
	struct tcphdr *th = tcp_hdr(skb);
1158 1159 1160 1161
	int genhash;
	unsigned char newhash[16];

	hash_expected = tcp_v4_md5_do_lookup(sk, iph->saddr);
1162
	hash_location = tcp_parse_md5sig_option(th);
1163 1164 1165 1166 1167 1168

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

	if (hash_expected && !hash_location) {
1169
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5NOTFOUND);
1170 1171 1172 1173
		return 1;
	}

	if (!hash_expected && hash_location) {
1174
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPMD5UNEXPECTED);
1175 1176 1177 1178 1179 1180
		return 1;
	}

	/* Okay, so this is hash_expected and hash_location -
	 * so we need to calculate the checksum.
	 */
1181 1182 1183
	genhash = tcp_v4_md5_hash_skb(newhash,
				      hash_expected,
				      NULL, NULL, skb);
1184 1185 1186

	if (genhash || memcmp(hash_location, newhash, 16) != 0) {
		if (net_ratelimit()) {
1187 1188 1189
			printk(KERN_INFO "MD5 Hash failed for (%pI4, %d)->(%pI4, %d)%s\n",
			       &iph->saddr, ntohs(th->source),
			       &iph->daddr, ntohs(th->dest),
1190 1191 1192 1193 1194 1195 1196 1197 1198
			       genhash ? " tcp_v4_calc_md5_hash failed" : "");
		}
		return 1;
	}
	return 0;
}

#endif

1199
struct request_sock_ops tcp_request_sock_ops __read_mostly = {
L
Linus Torvalds 已提交
1200
	.family		=	PF_INET,
1201
	.obj_size	=	sizeof(struct tcp_request_sock),
1202
	.rtx_syn_ack	=	tcp_v4_rtx_synack,
1203 1204
	.send_ack	=	tcp_v4_reqsk_send_ack,
	.destructor	=	tcp_v4_reqsk_destructor,
L
Linus Torvalds 已提交
1205
	.send_reset	=	tcp_v4_send_reset,
1206
	.syn_ack_timeout = 	tcp_syn_ack_timeout,
L
Linus Torvalds 已提交
1207 1208
};

1209
#ifdef CONFIG_TCP_MD5SIG
1210
static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
1211
	.md5_lookup	=	tcp_v4_reqsk_md5_lookup,
1212
	.calc_md5_hash	=	tcp_v4_md5_hash_skb,
1213
};
1214
#endif
1215

1216 1217 1218
static struct timewait_sock_ops tcp_timewait_sock_ops = {
	.twsk_obj_size	= sizeof(struct tcp_timewait_sock),
	.twsk_unique	= tcp_twsk_unique,
1219
	.twsk_destructor= tcp_twsk_destructor,
1220 1221
};

L
Linus Torvalds 已提交
1222 1223
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
{
1224
	struct tcp_extend_values tmp_ext;
L
Linus Torvalds 已提交
1225
	struct tcp_options_received tmp_opt;
1226
	u8 *hash_location;
1227
	struct request_sock *req;
1228
	struct inet_request_sock *ireq;
1229
	struct tcp_sock *tp = tcp_sk(sk);
1230
	struct dst_entry *dst = NULL;
1231 1232
	__be32 saddr = ip_hdr(skb)->saddr;
	__be32 daddr = ip_hdr(skb)->daddr;
L
Linus Torvalds 已提交
1233 1234 1235 1236 1237 1238 1239 1240
	__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
1241
	if (skb_rtable(skb)->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
L
Linus Torvalds 已提交
1242 1243 1244 1245 1246 1247
		goto drop;

	/* TW buckets are converted to open requests without
	 * limitations, they conserve resources and peer is
	 * evidently real one.
	 */
1248
	if (inet_csk_reqsk_queue_is_full(sk) && !isn) {
1249 1250
		if (net_ratelimit())
			syn_flood_warning(skb);
L
Linus Torvalds 已提交
1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263
#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.
	 */
1264
	if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
L
Linus Torvalds 已提交
1265 1266
		goto drop;

1267
	req = inet_reqsk_alloc(&tcp_request_sock_ops);
L
Linus Torvalds 已提交
1268 1269 1270
	if (!req)
		goto drop;

1271 1272 1273 1274
#ifdef CONFIG_TCP_MD5SIG
	tcp_rsk(req)->af_specific = &tcp_request_sock_ipv4_ops;
#endif

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

	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 */
1294 1295
		*mess++ ^= (__force u32)daddr;
		*mess++ ^= (__force u32)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 1323 1324 1325 1326 1327
	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);

1328
	if (security_inet_conn_request(sk, skb, req))
1329
		goto drop_and_free;
1330

L
Linus Torvalds 已提交
1331
	if (!want_cookie)
1332
		TCP_ECN_create_request(req, tcp_hdr(skb));
L
Linus Torvalds 已提交
1333 1334 1335

	if (want_cookie) {
#ifdef CONFIG_SYN_COOKIES
1336
		req->cookie_ts = tmp_opt.tstamp_ok;
L
Linus Torvalds 已提交
1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351
#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 &&
1352
		    tcp_death_row.sysctl_tw_recycle &&
1353
		    (dst = inet_csk_route_req(sk, req)) != NULL &&
L
Linus Torvalds 已提交
1354 1355
		    (peer = rt_get_peer((struct rtable *)dst)) != NULL &&
		    peer->v4daddr == saddr) {
1356
			inet_peer_refcheck(peer);
E
Eric Dumazet 已提交
1357
			if ((u32)get_seconds() - peer->tcp_ts_stamp < TCP_PAWS_MSL &&
L
Linus Torvalds 已提交
1358 1359
			    (s32)(peer->tcp_ts - req->ts_recent) >
							TCP_PAWS_WINDOW) {
1360
				NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1361
				goto drop_and_release;
L
Linus Torvalds 已提交
1362 1363 1364 1365
			}
		}
		/* Kill the following clause, if you dislike this way. */
		else if (!sysctl_tcp_syncookies &&
1366
			 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
L
Linus Torvalds 已提交
1367 1368 1369 1370 1371 1372 1373 1374 1375 1376
			  (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.
			 */
1377 1378
			LIMIT_NETDEBUG(KERN_DEBUG "TCP: drop open request from %pI4/%u\n",
				       &saddr, ntohs(tcp_hdr(skb)->source));
1379
			goto drop_and_release;
L
Linus Torvalds 已提交
1380 1381
		}

1382
		isn = tcp_v4_init_sequence(skb);
L
Linus Torvalds 已提交
1383
	}
1384
	tcp_rsk(req)->snt_isn = isn;
L
Linus Torvalds 已提交
1385

1386 1387
	if (tcp_v4_send_synack(sk, dst, req,
			       (struct request_values *)&tmp_ext) ||
1388
	    want_cookie)
L
Linus Torvalds 已提交
1389 1390
		goto drop_and_free;

1391
	inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
L
Linus Torvalds 已提交
1392 1393
	return 0;

1394 1395
drop_and_release:
	dst_release(dst);
L
Linus Torvalds 已提交
1396
drop_and_free:
1397
	reqsk_free(req);
L
Linus Torvalds 已提交
1398 1399 1400 1401 1402 1403 1404 1405 1406 1407
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,
1408
				  struct request_sock *req,
L
Linus Torvalds 已提交
1409 1410
				  struct dst_entry *dst)
{
1411
	struct inet_request_sock *ireq;
L
Linus Torvalds 已提交
1412 1413 1414
	struct inet_sock *newinet;
	struct tcp_sock *newtp;
	struct sock *newsk;
1415 1416 1417
#ifdef CONFIG_TCP_MD5SIG
	struct tcp_md5sig_key *key;
#endif
L
Linus Torvalds 已提交
1418 1419 1420 1421

	if (sk_acceptq_is_full(sk))
		goto exit_overflow;

1422
	if (!dst && (dst = inet_csk_route_req(sk, req)) == NULL)
L
Linus Torvalds 已提交
1423 1424 1425 1426 1427 1428
		goto exit;

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

1429
	newsk->sk_gso_type = SKB_GSO_TCPV4;
1430
	sk_setup_caps(newsk, dst);
L
Linus Torvalds 已提交
1431 1432 1433

	newtp		      = tcp_sk(newsk);
	newinet		      = inet_sk(newsk);
1434
	ireq		      = inet_rsk(req);
E
Eric Dumazet 已提交
1435 1436 1437
	newinet->inet_daddr   = ireq->rmt_addr;
	newinet->inet_rcv_saddr = ireq->loc_addr;
	newinet->inet_saddr	      = ireq->loc_addr;
1438 1439
	newinet->opt	      = ireq->opt;
	ireq->opt	      = NULL;
1440
	newinet->mc_index     = inet_iif(skb);
1441
	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
1442
	inet_csk(newsk)->icsk_ext_hdr_len = 0;
L
Linus Torvalds 已提交
1443
	if (newinet->opt)
1444
		inet_csk(newsk)->icsk_ext_hdr_len = newinet->opt->optlen;
E
Eric Dumazet 已提交
1445
	newinet->inet_id = newtp->write_seq ^ jiffies;
L
Linus Torvalds 已提交
1446

J
John Heffner 已提交
1447
	tcp_mtup_init(newsk);
L
Linus Torvalds 已提交
1448 1449
	tcp_sync_mss(newsk, dst_mtu(dst));
	newtp->advmss = dst_metric(dst, RTAX_ADVMSS);
1450 1451 1452 1453
	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
1454 1455
	tcp_initialize_rcv_mss(newsk);

1456 1457
#ifdef CONFIG_TCP_MD5SIG
	/* Copy over the MD5 key from the original socket */
E
Eric Dumazet 已提交
1458 1459
	key = tcp_v4_md5_do_lookup(sk, newinet->inet_daddr);
	if (key != NULL) {
1460 1461 1462 1463 1464 1465
		/*
		 * 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.
		 */
1466 1467
		char *newkey = kmemdup(key->key, key->keylen, GFP_ATOMIC);
		if (newkey != NULL)
E
Eric Dumazet 已提交
1468
			tcp_v4_md5_do_add(newsk, newinet->inet_daddr,
1469
					  newkey, key->keylen);
E
Eric Dumazet 已提交
1470
		sk_nocaps_add(newsk, NETIF_F_GSO_MASK);
1471 1472 1473
	}
#endif

1474
	__inet_hash_nolisten(newsk, NULL);
1475
	__inet_inherit_port(sk, newsk);
L
Linus Torvalds 已提交
1476 1477 1478 1479

	return newsk;

exit_overflow:
1480
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
L
Linus Torvalds 已提交
1481
exit:
1482
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
L
Linus Torvalds 已提交
1483 1484 1485 1486 1487 1488
	dst_release(dst);
	return NULL;
}

static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
{
1489
	struct tcphdr *th = tcp_hdr(skb);
1490
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1491
	struct sock *nsk;
1492
	struct request_sock **prev;
L
Linus Torvalds 已提交
1493
	/* Find possible connection requests. */
1494 1495
	struct request_sock *req = inet_csk_search_req(sk, &prev, th->source,
						       iph->saddr, iph->daddr);
L
Linus Torvalds 已提交
1496 1497 1498
	if (req)
		return tcp_check_req(sk, skb, req, prev);

1499
	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
1500
			th->source, iph->daddr, th->dest, inet_iif(skb));
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505 1506

	if (nsk) {
		if (nsk->sk_state != TCP_TIME_WAIT) {
			bh_lock_sock(nsk);
			return nsk;
		}
1507
		inet_twsk_put(inet_twsk(nsk));
L
Linus Torvalds 已提交
1508 1509 1510 1511
		return NULL;
	}

#ifdef CONFIG_SYN_COOKIES
1512
	if (!th->syn)
L
Linus Torvalds 已提交
1513 1514 1515 1516 1517
		sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
#endif
	return sk;
}

1518
static __sum16 tcp_v4_checksum_init(struct sk_buff *skb)
L
Linus Torvalds 已提交
1519
{
1520 1521
	const struct iphdr *iph = ip_hdr(skb);

1522
	if (skb->ip_summed == CHECKSUM_COMPLETE) {
1523 1524
		if (!tcp_v4_check(skb->len, iph->saddr,
				  iph->daddr, skb->csum)) {
1525
			skb->ip_summed = CHECKSUM_UNNECESSARY;
L
Linus Torvalds 已提交
1526
			return 0;
1527
		}
L
Linus Torvalds 已提交
1528
	}
1529

1530
	skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
1531 1532
				       skb->len, IPPROTO_TCP, 0);

L
Linus Torvalds 已提交
1533
	if (skb->len <= 76) {
1534
		return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549
	}
	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)
{
1550 1551 1552 1553 1554 1555 1556 1557
	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
	 */
1558
	if (tcp_v4_inbound_md5_hash(sk, skb))
1559 1560 1561
		goto discard;
#endif

L
Linus Torvalds 已提交
1562
	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
1563
		sock_rps_save_rxhash(sk, skb->rxhash);
L
Linus Torvalds 已提交
1564
		TCP_CHECK_TIMER(sk);
1565
		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
1566
			rsk = sk;
L
Linus Torvalds 已提交
1567
			goto reset;
1568
		}
L
Linus Torvalds 已提交
1569 1570 1571 1572
		TCP_CHECK_TIMER(sk);
		return 0;
	}

1573
	if (skb->len < tcp_hdrlen(skb) || tcp_checksum_complete(skb))
L
Linus Torvalds 已提交
1574 1575 1576 1577 1578 1579 1580 1581
		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) {
1582 1583
			if (tcp_child_process(sk, nsk, skb)) {
				rsk = nsk;
L
Linus Torvalds 已提交
1584
				goto reset;
1585
			}
L
Linus Torvalds 已提交
1586 1587
			return 0;
		}
1588 1589 1590
	} else
		sock_rps_save_rxhash(sk, skb->rxhash);

L
Linus Torvalds 已提交
1591 1592

	TCP_CHECK_TIMER(sk);
1593
	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
1594
		rsk = sk;
L
Linus Torvalds 已提交
1595
		goto reset;
1596
	}
L
Linus Torvalds 已提交
1597 1598 1599 1600
	TCP_CHECK_TIMER(sk);
	return 0;

reset:
1601
	tcp_v4_send_reset(rsk, skb);
L
Linus Torvalds 已提交
1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
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:
1612
	TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1613 1614 1615 1616 1617 1618 1619 1620 1621
	goto discard;
}

/*
 *	From tcp_input.c
 */

int tcp_v4_rcv(struct sk_buff *skb)
{
1622
	const struct iphdr *iph;
L
Linus Torvalds 已提交
1623 1624 1625
	struct tcphdr *th;
	struct sock *sk;
	int ret;
1626
	struct net *net = dev_net(skb->dev);
L
Linus Torvalds 已提交
1627 1628 1629 1630 1631

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

	/* Count it even if it's bad */
1632
	TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
L
Linus Torvalds 已提交
1633 1634 1635 1636

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

1637
	th = tcp_hdr(skb);
L
Linus Torvalds 已提交
1638 1639 1640 1641 1642 1643 1644 1645

	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
1646
	 * provided case of th->doff==0 is eliminated.
L
Linus Torvalds 已提交
1647
	 * So, we defer the checks. */
1648
	if (!skb_csum_unnecessary(skb) && tcp_v4_checksum_init(skb))
L
Linus Torvalds 已提交
1649 1650
		goto bad_packet;

1651
	th = tcp_hdr(skb);
1652
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1653 1654 1655 1656 1657
	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;
1658
	TCP_SKB_CB(skb)->flags	 = iph->tos;
L
Linus Torvalds 已提交
1659 1660
	TCP_SKB_CB(skb)->sacked	 = 0;

1661
	sk = __inet_lookup_skb(&tcp_hashinfo, skb, th->source, th->dest);
L
Linus Torvalds 已提交
1662 1663 1664
	if (!sk)
		goto no_tcp_socket;

E
Eric Dumazet 已提交
1665 1666 1667 1668
process:
	if (sk->sk_state == TCP_TIME_WAIT)
		goto do_time_wait;

1669 1670
	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1671
		goto discard_and_relse;
1672
	}
1673

L
Linus Torvalds 已提交
1674 1675
	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
		goto discard_and_relse;
1676
	nf_reset(skb);
L
Linus Torvalds 已提交
1677

1678
	if (sk_filter(sk, skb))
L
Linus Torvalds 已提交
1679 1680 1681 1682
		goto discard_and_relse;

	skb->dev = NULL;

1683
	bh_lock_sock_nested(sk);
L
Linus Torvalds 已提交
1684 1685
	ret = 0;
	if (!sock_owned_by_user(sk)) {
C
Chris Leech 已提交
1686 1687 1688
#ifdef CONFIG_NET_DMA
		struct tcp_sock *tp = tcp_sk(sk);
		if (!tp->ucopy.dma_chan && tp->ucopy.pinned_list)
1689
			tp->ucopy.dma_chan = dma_find_channel(DMA_MEMCPY);
C
Chris Leech 已提交
1690
		if (tp->ucopy.dma_chan)
L
Linus Torvalds 已提交
1691
			ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1692 1693 1694 1695
		else
#endif
		{
			if (!tcp_prequeue(sk, skb))
S
Shan Wei 已提交
1696
				ret = tcp_v4_do_rcv(sk, skb);
C
Chris Leech 已提交
1697
		}
1698
	} else if (unlikely(sk_add_backlog(sk, skb))) {
Z
Zhu Yi 已提交
1699
		bh_unlock_sock(sk);
1700
		NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
Z
Zhu Yi 已提交
1701 1702
		goto discard_and_relse;
	}
L
Linus Torvalds 已提交
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714
	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:
1715
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
L
Linus Torvalds 已提交
1716
	} else {
1717
		tcp_v4_send_reset(NULL, skb);
L
Linus Torvalds 已提交
1718 1719 1720 1721 1722
	}

discard_it:
	/* Discard frame. */
	kfree_skb(skb);
1723
	return 0;
L
Linus Torvalds 已提交
1724 1725 1726 1727 1728 1729 1730

discard_and_relse:
	sock_put(sk);
	goto discard_it;

do_time_wait:
	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
1731
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1732 1733 1734 1735
		goto discard_it;
	}

	if (skb->len < (th->doff << 2) || tcp_checksum_complete(skb)) {
1736
		TCP_INC_STATS_BH(net, TCP_MIB_INERRS);
1737
		inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1738 1739
		goto discard_it;
	}
1740
	switch (tcp_timewait_state_process(inet_twsk(sk), skb, th)) {
L
Linus Torvalds 已提交
1741
	case TCP_TW_SYN: {
1742
		struct sock *sk2 = inet_lookup_listener(dev_net(skb->dev),
1743
							&tcp_hashinfo,
1744
							iph->daddr, th->dest,
1745
							inet_iif(skb));
L
Linus Torvalds 已提交
1746
		if (sk2) {
1747 1748
			inet_twsk_deschedule(inet_twsk(sk), &tcp_death_row);
			inet_twsk_put(inet_twsk(sk));
L
Linus Torvalds 已提交
1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
			sk = sk2;
			goto process;
		}
		/* Fall through to ACK */
	}
	case TCP_TW_ACK:
		tcp_v4_timewait_ack(sk, skb);
		break;
	case TCP_TW_RST:
		goto no_tcp_socket;
	case TCP_TW_SUCCESS:;
	}
	goto discard_it;
}

/* 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
1778 1779
	if (!rt || rt->rt_dst != inet->inet_daddr) {
		peer = inet_getpeer(inet->inet_daddr, 1);
L
Linus Torvalds 已提交
1780 1781 1782 1783 1784 1785 1786 1787 1788
		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
1789 1790 1791
		    ((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
1792 1793 1794 1795 1796 1797 1798 1799 1800 1801
			peer->tcp_ts = tp->rx_opt.ts_recent;
		}
		if (release_it)
			inet_putpeer(peer);
		return 1;
	}

	return 0;
}

1802
int tcp_v4_tw_remember_stamp(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
1803
{
1804
	struct inet_peer *peer = inet_getpeer(tw->tw_daddr, 1);
L
Linus Torvalds 已提交
1805 1806

	if (peer) {
1807 1808 1809
		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
1810 1811 1812
		    ((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;
1813
			peer->tcp_ts	   = tcptw->tw_ts_recent;
L
Linus Torvalds 已提交
1814 1815 1816 1817 1818 1819 1820 1821
		}
		inet_putpeer(peer);
		return 1;
	}

	return 0;
}

1822
const struct inet_connection_sock_af_ops ipv4_specific = {
1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833
	.queue_xmit	   = ip_queue_xmit,
	.send_check	   = tcp_v4_send_check,
	.rebuild_header	   = inet_sk_rebuild_header,
	.conn_request	   = tcp_v4_conn_request,
	.syn_recv_sock	   = tcp_v4_syn_recv_sock,
	.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),
1834
	.bind_conflict	   = inet_csk_bind_conflict,
1835
#ifdef CONFIG_COMPAT
1836 1837
	.compat_setsockopt = compat_ip_setsockopt,
	.compat_getsockopt = compat_ip_getsockopt,
1838
#endif
L
Linus Torvalds 已提交
1839 1840
};

1841
#ifdef CONFIG_TCP_MD5SIG
1842
static const struct tcp_sock_af_ops tcp_sock_ipv4_specific = {
1843
	.md5_lookup		= tcp_v4_md5_lookup,
1844
	.calc_md5_hash		= tcp_v4_md5_hash_skb,
1845 1846 1847
	.md5_add		= tcp_v4_md5_add_func,
	.md5_parse		= tcp_v4_parse_md5_keys,
};
1848
#endif
1849

L
Linus Torvalds 已提交
1850 1851 1852 1853 1854
/* 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)
{
1855
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
1856 1857 1858 1859 1860 1861
	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);

1862
	icsk->icsk_rto = TCP_TIMEOUT_INIT;
L
Linus Torvalds 已提交
1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
	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
1875
	tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
L
Linus Torvalds 已提交
1876
	tp->snd_cwnd_clamp = ~0;
1877
	tp->mss_cache = TCP_MSS_DEFAULT;
L
Linus Torvalds 已提交
1878 1879

	tp->reordering = sysctl_tcp_reordering;
1880
	icsk->icsk_ca_ops = &tcp_init_congestion_ops;
L
Linus Torvalds 已提交
1881 1882 1883 1884 1885 1886

	sk->sk_state = TCP_CLOSE;

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

1887
	icsk->icsk_af_ops = &ipv4_specific;
1888
	icsk->icsk_sync_mss = tcp_sync_mss;
1889 1890 1891
#ifdef CONFIG_TCP_MD5SIG
	tp->af_specific = &tcp_sock_ipv4_specific;
#endif
L
Linus Torvalds 已提交
1892

1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905
	/* 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
1906 1907 1908
	sk->sk_sndbuf = sysctl_tcp_wmem[1];
	sk->sk_rcvbuf = sysctl_tcp_rmem[1];

H
Herbert Xu 已提交
1909
	local_bh_disable();
1910
	percpu_counter_inc(&tcp_sockets_allocated);
H
Herbert Xu 已提交
1911
	local_bh_enable();
L
Linus Torvalds 已提交
1912 1913 1914 1915

	return 0;
}

1916
void tcp_v4_destroy_sock(struct sock *sk)
L
Linus Torvalds 已提交
1917 1918 1919 1920 1921
{
	struct tcp_sock *tp = tcp_sk(sk);

	tcp_clear_xmit_timers(sk);

1922
	tcp_cleanup_congestion_control(sk);
1923

L
Linus Torvalds 已提交
1924
	/* Cleanup up the write buffer. */
1925
	tcp_write_queue_purge(sk);
L
Linus Torvalds 已提交
1926 1927

	/* Cleans up our, hopefully empty, out_of_order_queue. */
1928
	__skb_queue_purge(&tp->out_of_order_queue);
L
Linus Torvalds 已提交
1929

1930 1931 1932 1933 1934 1935 1936 1937 1938
#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
1939 1940
#ifdef CONFIG_NET_DMA
	/* Cleans up our sk_async_wait_queue */
1941
	__skb_queue_purge(&sk->sk_async_wait_queue);
C
Chris Leech 已提交
1942 1943
#endif

L
Linus Torvalds 已提交
1944 1945 1946 1947
	/* Clean prequeue, it must be empty really */
	__skb_queue_purge(&tp->ucopy.prequeue);

	/* Clean up a referenced TCP bind bucket. */
1948
	if (inet_csk(sk)->icsk_bind_hash)
1949
		inet_put_port(sk);
L
Linus Torvalds 已提交
1950 1951 1952 1953 1954 1955 1956 1957 1958

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

1959 1960 1961 1962 1963 1964 1965
	/* TCP Cookie Transactions */
	if (tp->cookie_values != NULL) {
		kref_put(&tp->cookie_values->kref,
			 tcp_cookie_values_release);
		tp->cookie_values = NULL;
	}

1966
	percpu_counter_dec(&tcp_sockets_allocated);
L
Linus Torvalds 已提交
1967 1968 1969 1970 1971 1972 1973
}

EXPORT_SYMBOL(tcp_v4_destroy_sock);

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

1974
static inline struct inet_timewait_sock *tw_head(struct hlist_nulls_head *head)
L
Linus Torvalds 已提交
1975
{
1976
	return hlist_nulls_empty(head) ? NULL :
1977
		list_entry(head->first, struct inet_timewait_sock, tw_node);
L
Linus Torvalds 已提交
1978 1979
}

1980
static inline struct inet_timewait_sock *tw_next(struct inet_timewait_sock *tw)
L
Linus Torvalds 已提交
1981
{
1982 1983
	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
1984 1985
}

1986 1987 1988 1989 1990
/*
 * 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
1991 1992
static void *listening_get_next(struct seq_file *seq, void *cur)
{
1993
	struct inet_connection_sock *icsk;
1994
	struct hlist_nulls_node *node;
L
Linus Torvalds 已提交
1995
	struct sock *sk = cur;
1996
	struct inet_listen_hashbucket *ilb;
J
Jianjun Kong 已提交
1997
	struct tcp_iter_state *st = seq->private;
1998
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
1999 2000

	if (!sk) {
2001
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
2002
		spin_lock_bh(&ilb->lock);
2003
		sk = sk_nulls_head(&ilb->head);
2004
		st->offset = 0;
L
Linus Torvalds 已提交
2005 2006
		goto get_sk;
	}
2007
	ilb = &tcp_hashinfo.listening_hash[st->bucket];
L
Linus Torvalds 已提交
2008
	++st->num;
2009
	++st->offset;
L
Linus Torvalds 已提交
2010 2011

	if (st->state == TCP_SEQ_STATE_OPENREQ) {
2012
		struct request_sock *req = cur;
L
Linus Torvalds 已提交
2013

2014
		icsk = inet_csk(st->syn_wait_sk);
L
Linus Torvalds 已提交
2015 2016 2017
		req = req->dl_next;
		while (1) {
			while (req) {
2018
				if (req->rsk_ops->family == st->family) {
L
Linus Torvalds 已提交
2019 2020 2021 2022 2023
					cur = req;
					goto out;
				}
				req = req->dl_next;
			}
2024
			st->offset = 0;
2025
			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
L
Linus Torvalds 已提交
2026 2027
				break;
get_req:
2028
			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
L
Linus Torvalds 已提交
2029 2030 2031
		}
		sk	  = sk_next(st->syn_wait_sk);
		st->state = TCP_SEQ_STATE_LISTENING;
2032
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2033
	} else {
2034
		icsk = inet_csk(sk);
2035 2036
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue))
L
Linus Torvalds 已提交
2037
			goto start_req;
2038
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2039 2040 2041
		sk = sk_next(sk);
	}
get_sk:
2042
	sk_nulls_for_each_from(sk, node) {
2043
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2044 2045 2046
			cur = sk;
			goto out;
		}
2047
		icsk = inet_csk(sk);
2048 2049
		read_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
L
Linus Torvalds 已提交
2050 2051 2052 2053 2054 2055 2056
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;
		}
2057
		read_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
L
Linus Torvalds 已提交
2058
	}
2059
	spin_unlock_bh(&ilb->lock);
2060
	st->offset = 0;
2061
	if (++st->bucket < INET_LHTABLE_SIZE) {
2062 2063
		ilb = &tcp_hashinfo.listening_hash[st->bucket];
		spin_lock_bh(&ilb->lock);
2064
		sk = sk_nulls_head(&ilb->head);
L
Linus Torvalds 已提交
2065 2066 2067 2068 2069 2070 2071 2072 2073
		goto get_sk;
	}
	cur = NULL;
out:
	return cur;
}

static void *listening_get_idx(struct seq_file *seq, loff_t *pos)
{
2074 2075 2076 2077 2078 2079
	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
2080 2081 2082 2083 2084 2085 2086 2087

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

2088 2089
static inline int empty_bucket(struct tcp_iter_state *st)
{
2090 2091
	return hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].chain) &&
		hlist_nulls_empty(&tcp_hashinfo.ehash[st->bucket].twchain);
2092 2093
}

2094 2095 2096 2097
/*
 * 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
2098 2099
static void *established_get_first(struct seq_file *seq)
{
J
Jianjun Kong 已提交
2100
	struct tcp_iter_state *st = seq->private;
2101
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2102 2103
	void *rc = NULL;

2104 2105
	st->offset = 0;
	for (; st->bucket <= tcp_hashinfo.ehash_mask; ++st->bucket) {
L
Linus Torvalds 已提交
2106
		struct sock *sk;
2107
		struct hlist_nulls_node *node;
2108
		struct inet_timewait_sock *tw;
2109
		spinlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
L
Linus Torvalds 已提交
2110

2111 2112 2113 2114
		/* Lockless fast path for the common case of empty buckets */
		if (empty_bucket(st))
			continue;

2115
		spin_lock_bh(lock);
2116
		sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
2117
			if (sk->sk_family != st->family ||
2118
			    !net_eq(sock_net(sk), net)) {
L
Linus Torvalds 已提交
2119 2120 2121 2122 2123 2124
				continue;
			}
			rc = sk;
			goto out;
		}
		st->state = TCP_SEQ_STATE_TIME_WAIT;
2125
		inet_twsk_for_each(tw, node,
2126
				   &tcp_hashinfo.ehash[st->bucket].twchain) {
2127
			if (tw->tw_family != st->family ||
2128
			    !net_eq(twsk_net(tw), net)) {
L
Linus Torvalds 已提交
2129 2130 2131 2132 2133
				continue;
			}
			rc = tw;
			goto out;
		}
2134
		spin_unlock_bh(lock);
L
Linus Torvalds 已提交
2135 2136 2137 2138 2139 2140 2141 2142 2143
		st->state = TCP_SEQ_STATE_ESTABLISHED;
	}
out:
	return rc;
}

static void *established_get_next(struct seq_file *seq, void *cur)
{
	struct sock *sk = cur;
2144
	struct inet_timewait_sock *tw;
2145
	struct hlist_nulls_node *node;
J
Jianjun Kong 已提交
2146
	struct tcp_iter_state *st = seq->private;
2147
	struct net *net = seq_file_net(seq);
L
Linus Torvalds 已提交
2148 2149

	++st->num;
2150
	++st->offset;
L
Linus Torvalds 已提交
2151 2152 2153 2154 2155

	if (st->state == TCP_SEQ_STATE_TIME_WAIT) {
		tw = cur;
		tw = tw_next(tw);
get_tw:
2156
		while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
L
Linus Torvalds 已提交
2157 2158 2159 2160 2161 2162
			tw = tw_next(tw);
		}
		if (tw) {
			cur = tw;
			goto out;
		}
2163
		spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2164 2165
		st->state = TCP_SEQ_STATE_ESTABLISHED;

2166
		/* Look for next non empty bucket */
2167
		st->offset = 0;
2168
		while (++st->bucket <= tcp_hashinfo.ehash_mask &&
2169 2170
				empty_bucket(st))
			;
2171
		if (st->bucket > tcp_hashinfo.ehash_mask)
2172 2173
			return NULL;

2174
		spin_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
2175
		sk = sk_nulls_head(&tcp_hashinfo.ehash[st->bucket].chain);
L
Linus Torvalds 已提交
2176
	} else
2177
		sk = sk_nulls_next(sk);
L
Linus Torvalds 已提交
2178

2179
	sk_nulls_for_each_from(sk, node) {
2180
		if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
L
Linus Torvalds 已提交
2181 2182 2183 2184
			goto found;
	}

	st->state = TCP_SEQ_STATE_TIME_WAIT;
2185
	tw = tw_head(&tcp_hashinfo.ehash[st->bucket].twchain);
L
Linus Torvalds 已提交
2186 2187 2188 2189 2190 2191 2192 2193 2194
	goto get_tw;
found:
	cur = sk;
out:
	return cur;
}

static void *established_get_idx(struct seq_file *seq, loff_t pos)
{
2195 2196 2197 2198 2199
	struct tcp_iter_state *st = seq->private;
	void *rc;

	st->bucket = 0;
	rc = established_get_first(seq);
L
Linus Torvalds 已提交
2200 2201 2202 2203

	while (rc && pos) {
		rc = established_get_next(seq, rc);
		--pos;
2204
	}
L
Linus Torvalds 已提交
2205 2206 2207 2208 2209 2210
	return rc;
}

static void *tcp_get_idx(struct seq_file *seq, loff_t pos)
{
	void *rc;
J
Jianjun Kong 已提交
2211
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223

	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;
}

2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258
static void *tcp_seek_last_pos(struct seq_file *seq)
{
	struct tcp_iter_state *st = seq->private;
	int offset = st->offset;
	int orig_num = st->num;
	void *rc = NULL;

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
	case TCP_SEQ_STATE_LISTENING:
		if (st->bucket >= INET_LHTABLE_SIZE)
			break;
		st->state = TCP_SEQ_STATE_LISTENING;
		rc = listening_get_next(seq, NULL);
		while (offset-- && rc)
			rc = listening_get_next(seq, rc);
		if (rc)
			break;
		st->bucket = 0;
		/* Fallthrough */
	case TCP_SEQ_STATE_ESTABLISHED:
	case TCP_SEQ_STATE_TIME_WAIT:
		st->state = TCP_SEQ_STATE_ESTABLISHED;
		if (st->bucket > tcp_hashinfo.ehash_mask)
			break;
		rc = established_get_first(seq);
		while (offset-- && rc)
			rc = established_get_next(seq, rc);
	}

	st->num = orig_num;

	return rc;
}

L
Linus Torvalds 已提交
2259 2260
static void *tcp_seq_start(struct seq_file *seq, loff_t *pos)
{
J
Jianjun Kong 已提交
2261
	struct tcp_iter_state *st = seq->private;
2262 2263 2264 2265 2266 2267 2268 2269
	void *rc;

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

L
Linus Torvalds 已提交
2270 2271
	st->state = TCP_SEQ_STATE_LISTENING;
	st->num = 0;
2272 2273 2274 2275 2276 2277 2278
	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
2279 2280 2281 2282
}

static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
2283
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296
	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;
2297 2298
			st->bucket = 0;
			st->offset = 0;
L
Linus Torvalds 已提交
2299 2300 2301 2302 2303 2304 2305 2306 2307 2308
			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;
2309
	st->last_pos = *pos;
L
Linus Torvalds 已提交
2310 2311 2312 2313 2314
	return rc;
}

static void tcp_seq_stop(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2315
	struct tcp_iter_state *st = seq->private;
L
Linus Torvalds 已提交
2316 2317 2318 2319

	switch (st->state) {
	case TCP_SEQ_STATE_OPENREQ:
		if (v) {
2320 2321
			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
2322 2323 2324
		}
	case TCP_SEQ_STATE_LISTENING:
		if (v != SEQ_START_TOKEN)
2325
			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
L
Linus Torvalds 已提交
2326 2327 2328 2329
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
	case TCP_SEQ_STATE_ESTABLISHED:
		if (v)
2330
			spin_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
L
Linus Torvalds 已提交
2331 2332 2333 2334 2335 2336 2337 2338
		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;
2339
	int err;
L
Linus Torvalds 已提交
2340

2341 2342 2343 2344
	err = seq_open_net(inode, file, &afinfo->seq_ops,
			  sizeof(struct tcp_iter_state));
	if (err < 0)
		return err;
2345

2346
	s = ((struct seq_file *)file->private_data)->private;
L
Linus Torvalds 已提交
2347
	s->family		= afinfo->family;
2348
	s->last_pos 		= 0;
2349 2350 2351
	return 0;
}

2352
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2353 2354 2355 2356
{
	int rc = 0;
	struct proc_dir_entry *p;

2357 2358 2359 2360
	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;
2361

2362 2363 2364 2365
	afinfo->seq_ops.start		= tcp_seq_start;
	afinfo->seq_ops.next		= tcp_seq_next;
	afinfo->seq_ops.stop		= tcp_seq_stop;

2366 2367 2368
	p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
			     &afinfo->seq_fops, afinfo);
	if (!p)
L
Linus Torvalds 已提交
2369 2370 2371 2372
		rc = -ENOMEM;
	return rc;
}

2373
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
L
Linus Torvalds 已提交
2374
{
2375
	proc_net_remove(net, afinfo->name);
L
Linus Torvalds 已提交
2376 2377
}

2378
static void get_openreq4(struct sock *sk, struct request_sock *req,
2379
			 struct seq_file *f, int i, int uid, int *len)
L
Linus Torvalds 已提交
2380
{
2381
	const struct inet_request_sock *ireq = inet_rsk(req);
L
Linus Torvalds 已提交
2382 2383
	int ttd = req->expires - jiffies;

2384 2385
	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
2386
		i,
2387
		ireq->loc_addr,
E
Eric Dumazet 已提交
2388
		ntohs(inet_sk(sk)->inet_sport),
2389 2390
		ireq->rmt_addr,
		ntohs(ireq->rmt_port),
L
Linus Torvalds 已提交
2391 2392 2393 2394 2395 2396 2397 2398 2399
		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),
2400 2401
		req,
		len);
L
Linus Torvalds 已提交
2402 2403
}

2404
static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2405 2406 2407
{
	int timer_active;
	unsigned long timer_expires;
2408 2409 2410
	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
2411 2412 2413 2414
	__be32 dest = inet->inet_daddr;
	__be32 src = inet->inet_rcv_saddr;
	__u16 destp = ntohs(inet->inet_dport);
	__u16 srcp = ntohs(inet->inet_sport);
2415
	int rx_queue;
L
Linus Torvalds 已提交
2416

2417
	if (icsk->icsk_pending == ICSK_TIME_RETRANS) {
L
Linus Torvalds 已提交
2418
		timer_active	= 1;
2419 2420
		timer_expires	= icsk->icsk_timeout;
	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
L
Linus Torvalds 已提交
2421
		timer_active	= 4;
2422
		timer_expires	= icsk->icsk_timeout;
2423
	} else if (timer_pending(&sk->sk_timer)) {
L
Linus Torvalds 已提交
2424
		timer_active	= 2;
2425
		timer_expires	= sk->sk_timer.expires;
L
Linus Torvalds 已提交
2426 2427 2428 2429 2430
	} else {
		timer_active	= 0;
		timer_expires = jiffies;
	}

2431 2432 2433 2434 2435 2436 2437 2438
	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);

2439
	seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
2440
			"%08X %5d %8d %lu %d %p %lu %lu %u %u %d%n",
2441
		i, src, srcp, dest, destp, sk->sk_state,
2442
		tp->write_seq - tp->snd_una,
2443
		rx_queue,
L
Linus Torvalds 已提交
2444 2445
		timer_active,
		jiffies_to_clock_t(timer_expires - jiffies),
2446
		icsk->icsk_retransmits,
2447
		sock_i_uid(sk),
2448
		icsk->icsk_probes_out,
2449 2450
		sock_i_ino(sk),
		atomic_read(&sk->sk_refcnt), sk,
2451 2452
		jiffies_to_clock_t(icsk->icsk_rto),
		jiffies_to_clock_t(icsk->icsk_ack.ato),
2453
		(icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
L
Linus Torvalds 已提交
2454
		tp->snd_cwnd,
I
Ilpo Järvinen 已提交
2455
		tcp_in_initial_slowstart(tp) ? -1 : tp->snd_ssthresh,
2456
		len);
L
Linus Torvalds 已提交
2457 2458
}

2459
static void get_timewait4_sock(struct inet_timewait_sock *tw,
2460
			       struct seq_file *f, int i, int *len)
L
Linus Torvalds 已提交
2461
{
2462
	__be32 dest, src;
L
Linus Torvalds 已提交
2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473
	__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);

2474 2475
	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
2476 2477
		i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
		3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
2478
		atomic_read(&tw->tw_refcnt), tw, len);
L
Linus Torvalds 已提交
2479 2480 2481 2482 2483 2484
}

#define TMPSZ 150

static int tcp4_seq_show(struct seq_file *seq, void *v)
{
J
Jianjun Kong 已提交
2485
	struct tcp_iter_state *st;
2486
	int len;
L
Linus Torvalds 已提交
2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499

	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:
2500
		get_tcp4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2501 2502
		break;
	case TCP_SEQ_STATE_OPENREQ:
2503
		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid, &len);
L
Linus Torvalds 已提交
2504 2505
		break;
	case TCP_SEQ_STATE_TIME_WAIT:
2506
		get_timewait4_sock(v, seq, st->num, &len);
L
Linus Torvalds 已提交
2507 2508
		break;
	}
2509
	seq_printf(seq, "%*s\n", TMPSZ - 1 - len, "");
L
Linus Torvalds 已提交
2510 2511 2512 2513 2514 2515 2516
out:
	return 0;
}

static struct tcp_seq_afinfo tcp4_seq_afinfo = {
	.name		= "tcp",
	.family		= AF_INET,
2517 2518 2519
	.seq_fops	= {
		.owner		= THIS_MODULE,
	},
2520 2521 2522
	.seq_ops	= {
		.show		= tcp4_seq_show,
	},
L
Linus Torvalds 已提交
2523 2524
};

2525
static int __net_init tcp4_proc_init_net(struct net *net)
2526 2527 2528 2529
{
	return tcp_proc_register(net, &tcp4_seq_afinfo);
}

2530
static void __net_exit tcp4_proc_exit_net(struct net *net)
2531 2532 2533 2534 2535 2536 2537 2538 2539
{
	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
2540 2541
int __init tcp4_proc_init(void)
{
2542
	return register_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2543 2544 2545 2546
}

void tcp4_proc_exit(void)
{
2547
	unregister_pernet_subsys(&tcp4_net_ops);
L
Linus Torvalds 已提交
2548 2549 2550
}
#endif /* CONFIG_PROC_FS */

H
Herbert Xu 已提交
2551 2552
struct sk_buff **tcp4_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
H
Herbert Xu 已提交
2553
	struct iphdr *iph = skb_gro_network_header(skb);
H
Herbert Xu 已提交
2554 2555 2556

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
2557
		if (!tcp_v4_check(skb_gro_len(skb), iph->saddr, iph->daddr,
H
Herbert Xu 已提交
2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585
				  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
2586 2587 2588 2589 2590 2591
struct proto tcp_prot = {
	.name			= "TCP",
	.owner			= THIS_MODULE,
	.close			= tcp_close,
	.connect		= tcp_v4_connect,
	.disconnect		= tcp_disconnect,
2592
	.accept			= inet_csk_accept,
L
Linus Torvalds 已提交
2593 2594 2595 2596 2597 2598 2599 2600
	.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,
2601 2602 2603
	.hash			= inet_hash,
	.unhash			= inet_unhash,
	.get_port		= inet_csk_get_port,
L
Linus Torvalds 已提交
2604 2605
	.enter_memory_pressure	= tcp_enter_memory_pressure,
	.sockets_allocated	= &tcp_sockets_allocated,
2606
	.orphan_count		= &tcp_orphan_count,
L
Linus Torvalds 已提交
2607 2608 2609 2610 2611 2612 2613
	.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),
2614
	.slab_flags		= SLAB_DESTROY_BY_RCU,
2615
	.twsk_prot		= &tcp_timewait_sock_ops,
2616
	.rsk_prot		= &tcp_request_sock_ops,
2617
	.h.hashinfo		= &tcp_hashinfo,
2618 2619 2620 2621
#ifdef CONFIG_COMPAT
	.compat_setsockopt	= compat_tcp_setsockopt,
	.compat_getsockopt	= compat_tcp_getsockopt,
#endif
L
Linus Torvalds 已提交
2622 2623
};

2624 2625 2626 2627 2628 2629 2630 2631 2632 2633

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
2634 2635 2636 2637 2638
}

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
	inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
2639 2640 2641
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
E
Eric W. Biederman 已提交
2642 2643 2644
       .init	   = tcp_sk_init,
       .exit	   = tcp_sk_exit,
       .exit_batch = tcp_sk_exit_batch,
2645 2646
};

2647
void __init tcp_v4_init(void)
L
Linus Torvalds 已提交
2648
{
2649
	inet_hashinfo_init(&tcp_hashinfo);
2650
	if (register_pernet_subsys(&tcp_sk_ops))
L
Linus Torvalds 已提交
2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669
		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);