transport.c 21.0 KB
Newer Older
1
/* SCTP kernel implementation
L
Linus Torvalds 已提交
2 3 4 5 6 7
 * Copyright (c) 1999-2000 Cisco, Inc.
 * Copyright (c) 1999-2001 Motorola, Inc.
 * Copyright (c) 2001-2003 International Business Machines Corp.
 * Copyright (c) 2001 Intel Corp.
 * Copyright (c) 2001 La Monte H.P. Yarroll
 *
8
 * This file is part of the SCTP kernel implementation
L
Linus Torvalds 已提交
9 10 11 12 13
 *
 * This module provides the abstraction for an SCTP tranport representing
 * a remote transport address.  For local transport addresses, we just use
 * union sctp_addr.
 *
14
 * This SCTP implementation is free software;
L
Linus Torvalds 已提交
15 16 17 18 19
 * 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, or (at your option)
 * any later version.
 *
20
 * This SCTP implementation is distributed in the hope that it
L
Linus Torvalds 已提交
21 22 23 24 25 26
 * will be useful, but WITHOUT ANY WARRANTY; without even the implied
 *                 ************************
 * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
 * See the GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
27 28
 * along with GNU CC; see the file COPYING.  If not, see
 * <http://www.gnu.org/licenses/>.
L
Linus Torvalds 已提交
29 30 31
 *
 * Please send any bug reports or fixes you make to the
 * email address(es):
32
 *    lksctp developers <linux-sctp@vger.kernel.org>
L
Linus Torvalds 已提交
33 34 35 36 37 38 39 40 41 42 43
 *
 * Written or modified by:
 *    La Monte H.P. Yarroll <piggy@acm.org>
 *    Karl Knutson          <karl@athena.chicago.il.us>
 *    Jon Grimm             <jgrimm@us.ibm.com>
 *    Xingang Guo           <xingang.guo@intel.com>
 *    Hui Huang             <hui.huang@nokia.com>
 *    Sridhar Samudrala	    <sri@us.ibm.com>
 *    Ardelle Fan	    <ardelle.fan@intel.com>
 */

44 45
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

46
#include <linux/slab.h>
L
Linus Torvalds 已提交
47
#include <linux/types.h>
48
#include <linux/random.h>
L
Linus Torvalds 已提交
49 50 51 52 53 54
#include <net/sctp/sctp.h>
#include <net/sctp/sm.h>

/* 1st Level Abstractions.  */

/* Initialize a new transport from provided memory.  */
55 56
static struct sctp_transport *sctp_transport_init(struct net *net,
						  struct sctp_transport *peer,
L
Linus Torvalds 已提交
57
						  const union sctp_addr *addr,
A
Al Viro 已提交
58
						  gfp_t gfp)
L
Linus Torvalds 已提交
59 60 61 62 63 64
{
	/* Copy in the address.  */
	peer->ipaddr = *addr;
	peer->af_specific = sctp_get_af_specific(addr->sa.sa_family);
	memset(&peer->saddr, 0, sizeof(union sctp_addr));

65 66
	peer->sack_generation = 0;

L
Linus Torvalds 已提交
67 68 69 70 71 72
	/* From 6.3.1 RTO Calculation:
	 *
	 * C1) Until an RTT measurement has been made for a packet sent to the
	 * given destination transport address, set RTO to the protocol
	 * parameter 'RTO.Initial'.
	 */
73
	peer->rto = msecs_to_jiffies(net->sctp.rto_initial);
L
Linus Torvalds 已提交
74

T
Thomas Gleixner 已提交
75
	peer->last_time_heard = 0;
L
Linus Torvalds 已提交
76 77
	peer->last_time_ecne_reduced = jiffies;

78 79 80
	peer->param_flags = SPP_HB_DISABLE |
			    SPP_PMTUD_ENABLE |
			    SPP_SACKDELAY_ENABLE;
L
Linus Torvalds 已提交
81 82

	/* Initialize the default path max_retrans.  */
83 84
	peer->pathmaxrxt  = net->sctp.max_retrans_path;
	peer->pf_retrans  = net->sctp.pf_retrans;
L
Linus Torvalds 已提交
85 86 87 88 89

	INIT_LIST_HEAD(&peer->transmitted);
	INIT_LIST_HEAD(&peer->send_ready);
	INIT_LIST_HEAD(&peer->transports);

90
	setup_timer(&peer->T3_rtx_timer, sctp_generate_t3_rtx_event,
X
Xin Long 已提交
91
		    (unsigned long)peer);
92
	setup_timer(&peer->hb_timer, sctp_generate_heartbeat_event,
X
Xin Long 已提交
93 94 95
		    (unsigned long)peer);
	setup_timer(&peer->reconf_timer, sctp_generate_reconf_event,
		    (unsigned long)peer);
96 97
	setup_timer(&peer->proto_unreach_timer,
		    sctp_generate_proto_unreach_event, (unsigned long)peer);
L
Linus Torvalds 已提交
98

99 100 101
	/* Initialize the 64-bit random nonce sent with heartbeat. */
	get_random_bytes(&peer->hb_nonce, sizeof(peer->hb_nonce));

102
	refcount_set(&peer->refcnt, 1);
L
Linus Torvalds 已提交
103 104 105 106 107

	return peer;
}

/* Allocate and initialize a new transport.  */
108 109
struct sctp_transport *sctp_transport_new(struct net *net,
					  const union sctp_addr *addr,
A
Al Viro 已提交
110
					  gfp_t gfp)
L
Linus Torvalds 已提交
111
{
112
	struct sctp_transport *transport;
L
Linus Torvalds 已提交
113

114
	transport = kzalloc(sizeof(*transport), gfp);
L
Linus Torvalds 已提交
115 116 117
	if (!transport)
		goto fail;

118
	if (!sctp_transport_init(net, transport, addr, gfp))
L
Linus Torvalds 已提交
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
		goto fail_init;

	SCTP_DBG_OBJCNT_INC(transport);

	return transport;

fail_init:
	kfree(transport);

fail:
	return NULL;
}

/* This transport is no longer needed.  Free up if possible, or
 * delay until it last reference count.
 */
void sctp_transport_free(struct sctp_transport *transport)
{
	/* Try to delete the heartbeat timer.  */
	if (del_timer(&transport->hb_timer))
		sctp_transport_put(transport);

	/* Delete the T3_rtx timer if it's active.
	 * There is no point in not doing this now and letting
	 * structure hang around in memory since we know
	 * the tranport is going away.
	 */
146
	if (del_timer(&transport->T3_rtx_timer))
L
Linus Torvalds 已提交
147 148
		sctp_transport_put(transport);

X
Xin Long 已提交
149 150 151
	if (del_timer(&transport->reconf_timer))
		sctp_transport_put(transport);

152
	/* Delete the ICMP proto unreachable timer if it's active. */
153
	if (del_timer(&transport->proto_unreach_timer))
154
		sctp_association_put(transport->asoc);
L
Linus Torvalds 已提交
155 156 157 158

	sctp_transport_put(transport);
}

159
static void sctp_transport_destroy_rcu(struct rcu_head *head)
L
Linus Torvalds 已提交
160
{
161
	struct sctp_transport *transport;
L
Linus Torvalds 已提交
162

163
	transport = container_of(head, struct sctp_transport, rcu);
L
Linus Torvalds 已提交
164 165 166 167 168 169

	dst_release(transport->dst);
	kfree(transport);
	SCTP_DBG_OBJCNT_DEC(transport);
}

170 171 172 173 174
/* Destroy the transport data structure.
 * Assumes there are no more users of this structure.
 */
static void sctp_transport_destroy(struct sctp_transport *transport)
{
175
	if (unlikely(refcount_read(&transport->refcnt))) {
176 177 178
		WARN(1, "Attempt to destroy undead transport %p!\n", transport);
		return;
	}
179

180 181 182 183
	sctp_packet_free(&transport->packet);

	if (transport->asoc)
		sctp_association_put(transport->asoc);
184 185

	call_rcu(&transport->rcu, sctp_transport_destroy_rcu);
186 187
}

L
Linus Torvalds 已提交
188 189 190
/* Start T3_rtx timer if it is not already running and update the heartbeat
 * timer.  This routine is called every time a DATA chunk is sent.
 */
191
void sctp_transport_reset_t3_rtx(struct sctp_transport *transport)
L
Linus Torvalds 已提交
192 193 194 195 196 197 198 199 200
{
	/* RFC 2960 6.3.2 Retransmission Timer Rules
	 *
	 * R1) Every time a DATA chunk is sent to any address(including a
	 * retransmission), if the T3-rtx timer of that address is not running
	 * start it running so that it will expire after the RTO of that
	 * address.
	 */

201
	if (!timer_pending(&transport->T3_rtx_timer))
L
Linus Torvalds 已提交
202 203 204
		if (!mod_timer(&transport->T3_rtx_timer,
			       jiffies + transport->rto))
			sctp_transport_hold(transport);
205 206 207 208 209
}

void sctp_transport_reset_hb_timer(struct sctp_transport *transport)
{
	unsigned long expires;
L
Linus Torvalds 已提交
210 211

	/* When a data chunk is sent, reset the heartbeat interval.  */
212 213 214 215 216
	expires = jiffies + sctp_transport_timeout(transport);
	if (time_before(transport->hb_timer.expires, expires) &&
	    !mod_timer(&transport->hb_timer,
		       expires + prandom_u32_max(transport->rto)))
		sctp_transport_hold(transport);
L
Linus Torvalds 已提交
217 218
}

X
Xin Long 已提交
219 220 221 222 223 224 225 226
void sctp_transport_reset_reconf_timer(struct sctp_transport *transport)
{
	if (!timer_pending(&transport->reconf_timer))
		if (!mod_timer(&transport->reconf_timer,
			       jiffies + transport->rto))
			sctp_transport_hold(transport);
}

L
Linus Torvalds 已提交
227 228 229 230 231 232 233 234 235 236 237 238
/* This transport has been assigned to an association.
 * Initialize fields from the association or from the sock itself.
 * Register the reference count in the association.
 */
void sctp_transport_set_owner(struct sctp_transport *transport,
			      struct sctp_association *asoc)
{
	transport->asoc = asoc;
	sctp_association_hold(asoc);
}

/* Initialize the pmtu of a transport. */
239
void sctp_transport_pmtu(struct sctp_transport *transport, struct sock *sk)
L
Linus Torvalds 已提交
240
{
241
	/* If we don't have a fresh route, look one up */
242
	if (!transport->dst || transport->dst->obsolete) {
243
		sctp_transport_dst_release(transport);
244
		transport->af_specific->get_dst(transport, &transport->saddr,
245
						&transport->fl, sk);
246
	}
L
Linus Torvalds 已提交
247

248
	if (transport->dst) {
249
		transport->pathmtu = SCTP_TRUNC4(dst_mtu(transport->dst));
L
Linus Torvalds 已提交
250
	} else
251
		transport->pathmtu = SCTP_DEFAULT_MAXSEGMENT;
L
Linus Torvalds 已提交
252 253
}

254
void sctp_transport_update_pmtu(struct sctp_transport *t, u32 pmtu)
255
{
256
	struct dst_entry *dst = sctp_transport_dst_check(t);
257 258

	if (unlikely(pmtu < SCTP_DEFAULT_MINSEGMENT)) {
259
		pr_warn("%s: Reported pmtu %d too low, using default minimum of %d\n",
260
			__func__, pmtu, SCTP_DEFAULT_MINSEGMENT);
261 262 263 264 265 266 267 268
		/* Use default minimum segment size and disable
		 * pmtu discovery on this transport.
		 */
		t->pathmtu = SCTP_DEFAULT_MINSEGMENT;
	} else {
		t->pathmtu = pmtu;
	}

269
	if (dst) {
270
		dst->ops->update_pmtu(dst, t->asoc->base.sk, NULL, pmtu);
271 272
		dst = sctp_transport_dst_check(t);
	}
273 274 275

	if (!dst)
		t->af_specific->get_dst(t, &t->saddr, &t->fl, t->asoc->base.sk);
276 277
}

L
Linus Torvalds 已提交
278 279 280 281 282 283 284 285 286
/* Caches the dst entry and source address for a transport's destination
 * address.
 */
void sctp_transport_route(struct sctp_transport *transport,
			  union sctp_addr *saddr, struct sctp_sock *opt)
{
	struct sctp_association *asoc = transport->asoc;
	struct sctp_af *af = transport->af_specific;

287
	af->get_dst(transport, saddr, &transport->fl, sctp_opt2sk(opt));
L
Linus Torvalds 已提交
288 289 290 291

	if (saddr)
		memcpy(&transport->saddr, saddr, sizeof(union sctp_addr));
	else
292
		af->get_saddr(opt, transport, &transport->fl);
L
Linus Torvalds 已提交
293

294 295 296
	if ((transport->param_flags & SPP_PMTUD_DISABLE) && transport->pathmtu) {
		return;
	}
297
	if (transport->dst) {
298
		transport->pathmtu = SCTP_TRUNC4(dst_mtu(transport->dst));
L
Linus Torvalds 已提交
299 300 301

		/* Initialize sk->sk_rcv_saddr, if the transport is the
		 * association's active path for getsockname().
302
		 */
303 304
		if (asoc && (!asoc->peer.primary_path ||
				(transport == asoc->peer.active_path)))
305 306
			opt->pf->to_sk_saddr(&transport->saddr,
					     asoc->base.sk);
L
Linus Torvalds 已提交
307
	} else
308
		transport->pathmtu = SCTP_DEFAULT_MAXSEGMENT;
L
Linus Torvalds 已提交
309 310 311
}

/* Hold a reference to a transport.  */
312
int sctp_transport_hold(struct sctp_transport *transport)
L
Linus Torvalds 已提交
313
{
314
	return refcount_inc_not_zero(&transport->refcnt);
L
Linus Torvalds 已提交
315 316 317 318 319 320 321
}

/* Release a reference to a transport and clean up
 * if there are no more references.
 */
void sctp_transport_put(struct sctp_transport *transport)
{
322
	if (refcount_dec_and_test(&transport->refcnt))
L
Linus Torvalds 已提交
323 324 325 326 327 328
		sctp_transport_destroy(transport);
}

/* Update transport's RTO based on the newly calculated RTT. */
void sctp_transport_update_rto(struct sctp_transport *tp, __u32 rtt)
{
329 330 331
	if (unlikely(!tp->rto_pending))
		/* We should not be doing any RTO updates unless rto_pending is set.  */
		pr_debug("%s: rto_pending not set on transport %p!\n", __func__, tp);
L
Linus Torvalds 已提交
332 333

	if (tp->rttvar || tp->srtt) {
334
		struct net *net = sock_net(tp->asoc->base.sk);
L
Linus Torvalds 已提交
335 336 337 338 339 340 341 342 343 344 345
		/* 6.3.1 C3) When a new RTT measurement R' is made, set
		 * RTTVAR <- (1 - RTO.Beta) * RTTVAR + RTO.Beta * |SRTT - R'|
		 * SRTT <- (1 - RTO.Alpha) * SRTT + RTO.Alpha * R'
		 */

		/* Note:  The above algorithm has been rewritten to
		 * express rto_beta and rto_alpha as inverse powers
		 * of two.
		 * For example, assuming the default value of RTO.Alpha of
		 * 1/8, rto_alpha would be expressed as 3.
		 */
346
		tp->rttvar = tp->rttvar - (tp->rttvar >> net->sctp.rto_beta)
A
Andrew Morton 已提交
347
			+ (((__u32)abs((__s64)tp->srtt - (__s64)rtt)) >> net->sctp.rto_beta);
348 349
		tp->srtt = tp->srtt - (tp->srtt >> net->sctp.rto_alpha)
			+ (rtt >> net->sctp.rto_alpha);
L
Linus Torvalds 已提交
350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378
	} else {
		/* 6.3.1 C2) When the first RTT measurement R is made, set
		 * SRTT <- R, RTTVAR <- R/2.
		 */
		tp->srtt = rtt;
		tp->rttvar = rtt >> 1;
	}

	/* 6.3.1 G1) Whenever RTTVAR is computed, if RTTVAR = 0, then
	 * adjust RTTVAR <- G, where G is the CLOCK GRANULARITY.
	 */
	if (tp->rttvar == 0)
		tp->rttvar = SCTP_CLOCK_GRANULARITY;

	/* 6.3.1 C3) After the computation, update RTO <- SRTT + 4 * RTTVAR. */
	tp->rto = tp->srtt + (tp->rttvar << 2);

	/* 6.3.1 C6) Whenever RTO is computed, if it is less than RTO.Min
	 * seconds then it is rounded up to RTO.Min seconds.
	 */
	if (tp->rto < tp->asoc->rto_min)
		tp->rto = tp->asoc->rto_min;

	/* 6.3.1 C7) A maximum value may be placed on RTO provided it is
	 * at least RTO.max seconds.
	 */
	if (tp->rto > tp->asoc->rto_max)
		tp->rto = tp->asoc->rto_max;

379
	sctp_max_rto(tp->asoc, tp);
L
Linus Torvalds 已提交
380 381 382 383 384 385 386
	tp->rtt = rtt;

	/* Reset rto_pending so that a new RTT measurement is started when a
	 * new data chunk is sent.
	 */
	tp->rto_pending = 0;

387 388
	pr_debug("%s: transport:%p, rtt:%d, srtt:%d rttvar:%d, rto:%ld\n",
		 __func__, tp, rtt, tp->srtt, tp->rttvar, tp->rto);
L
Linus Torvalds 已提交
389 390 391 392 393 394 395 396
}

/* This routine updates the transport's cwnd and partial_bytes_acked
 * parameters based on the bytes acked in the received SACK.
 */
void sctp_transport_raise_cwnd(struct sctp_transport *transport,
			       __u32 sack_ctsn, __u32 bytes_acked)
{
397
	struct sctp_association *asoc = transport->asoc;
L
Linus Torvalds 已提交
398 399 400 401 402
	__u32 cwnd, ssthresh, flight_size, pba, pmtu;

	cwnd = transport->cwnd;
	flight_size = transport->flight_size;

403
	/* See if we need to exit Fast Recovery first */
404 405 406
	if (asoc->fast_recovery &&
	    TSN_lte(asoc->fast_recovery_exit, sack_ctsn))
		asoc->fast_recovery = 0;
407

L
Linus Torvalds 已提交
408 409
	ssthresh = transport->ssthresh;
	pba = transport->partial_bytes_acked;
410
	pmtu = transport->asoc->pathmtu;
L
Linus Torvalds 已提交
411 412

	if (cwnd <= ssthresh) {
413 414 415 416 417 418 419 420 421 422 423 424 425
		/* RFC 4960 7.2.1
		 * o  When cwnd is less than or equal to ssthresh, an SCTP
		 *    endpoint MUST use the slow-start algorithm to increase
		 *    cwnd only if the current congestion window is being fully
		 *    utilized, an incoming SACK advances the Cumulative TSN
		 *    Ack Point, and the data sender is not in Fast Recovery.
		 *    Only when these three conditions are met can the cwnd be
		 *    increased; otherwise, the cwnd MUST not be increased.
		 *    If these conditions are met, then cwnd MUST be increased
		 *    by, at most, the lesser of 1) the total size of the
		 *    previously outstanding DATA chunk(s) acknowledged, and
		 *    2) the destination's path MTU.  This upper bound protects
		 *    against the ACK-Splitting attack outlined in [SAVAGE99].
L
Linus Torvalds 已提交
426
		 */
427
		if (asoc->fast_recovery)
428 429
			return;

430 431 432 433 434 435 436 437
		/* The appropriate cwnd increase algorithm is performed
		 * if, and only if the congestion window is being fully
		 * utilized.  Note that RFC4960 Errata 3.22 removed the
		 * other condition on ctsn moving.
		 */
		if (flight_size < cwnd)
			return;

L
Linus Torvalds 已提交
438 439 440 441
		if (bytes_acked > pmtu)
			cwnd += pmtu;
		else
			cwnd += bytes_acked;
442 443 444 445 446

		pr_debug("%s: slow start: transport:%p, bytes_acked:%d, "
			 "cwnd:%d, ssthresh:%d, flight_size:%d, pba:%d\n",
			 __func__, transport, bytes_acked, cwnd, ssthresh,
			 flight_size, pba);
L
Linus Torvalds 已提交
447 448
	} else {
		/* RFC 2960 7.2.2 Whenever cwnd is greater than ssthresh,
449 450 451 452 453
		 * upon each SACK arrival, increase partial_bytes_acked
		 * by the total number of bytes of all new chunks
		 * acknowledged in that SACK including chunks
		 * acknowledged by the new Cumulative TSN Ack and by Gap
		 * Ack Blocks. (updated by RFC4960 Errata 3.22)
L
Linus Torvalds 已提交
454
		 *
455 456 457 458 459 460 461
		 * When partial_bytes_acked is greater than cwnd and
		 * before the arrival of the SACK the sender had less
		 * bytes of data outstanding than cwnd (i.e., before
		 * arrival of the SACK, flightsize was less than cwnd),
		 * reset partial_bytes_acked to cwnd. (RFC 4960 Errata
		 * 3.26)
		 *
462 463 464 465 466 467 468
		 * When partial_bytes_acked is equal to or greater than
		 * cwnd and before the arrival of the SACK the sender
		 * had cwnd or more bytes of data outstanding (i.e.,
		 * before arrival of the SACK, flightsize was greater
		 * than or equal to cwnd), partial_bytes_acked is reset
		 * to (partial_bytes_acked - cwnd). Next, cwnd is
		 * increased by MTU. (RFC 4960 Errata 3.12)
L
Linus Torvalds 已提交
469 470
		 */
		pba += bytes_acked;
471 472 473
		if (pba > cwnd && flight_size < cwnd)
			pba = cwnd;
		if (pba >= cwnd && flight_size >= cwnd) {
474
			pba = pba - cwnd;
L
Linus Torvalds 已提交
475 476
			cwnd += pmtu;
		}
477 478 479 480 481 482

		pr_debug("%s: congestion avoidance: transport:%p, "
			 "bytes_acked:%d, cwnd:%d, ssthresh:%d, "
			 "flight_size:%d, pba:%d\n", __func__,
			 transport, bytes_acked, cwnd, ssthresh,
			 flight_size, pba);
L
Linus Torvalds 已提交
483 484 485 486 487 488 489 490 491 492
	}

	transport->cwnd = cwnd;
	transport->partial_bytes_acked = pba;
}

/* This routine is used to lower the transport's cwnd when congestion is
 * detected.
 */
void sctp_transport_lower_cwnd(struct sctp_transport *transport,
493
			       enum sctp_lower_cwnd reason)
L
Linus Torvalds 已提交
494
{
495 496
	struct sctp_association *asoc = transport->asoc;

L
Linus Torvalds 已提交
497 498 499 500 501 502 503 504 505 506
	switch (reason) {
	case SCTP_LOWER_CWND_T3_RTX:
		/* RFC 2960 Section 7.2.3, sctpimpguide
		 * When the T3-rtx timer expires on an address, SCTP should
		 * perform slow start by:
		 *      ssthresh = max(cwnd/2, 4*MTU)
		 *      cwnd = 1*MTU
		 *      partial_bytes_acked = 0
		 */
		transport->ssthresh = max(transport->cwnd/2,
507 508
					  4*asoc->pathmtu);
		transport->cwnd = asoc->pathmtu;
509

510 511
		/* T3-rtx also clears fast recovery */
		asoc->fast_recovery = 0;
L
Linus Torvalds 已提交
512 513 514 515 516 517 518
		break;

	case SCTP_LOWER_CWND_FAST_RTX:
		/* RFC 2960 7.2.4 Adjust the ssthresh and cwnd of the
		 * destination address(es) to which the missing DATA chunks
		 * were last sent, according to the formula described in
		 * Section 7.2.3.
519 520
		 *
		 * RFC 2960 7.2.3, sctpimpguide Upon detection of packet
L
Linus Torvalds 已提交
521 522 523 524 525 526
		 * losses from SACK (see Section 7.2.4), An endpoint
		 * should do the following:
		 *      ssthresh = max(cwnd/2, 4*MTU)
		 *      cwnd = ssthresh
		 *      partial_bytes_acked = 0
		 */
527
		if (asoc->fast_recovery)
528 529 530
			return;

		/* Mark Fast recovery */
531 532
		asoc->fast_recovery = 1;
		asoc->fast_recovery_exit = asoc->next_tsn - 1;
533

L
Linus Torvalds 已提交
534
		transport->ssthresh = max(transport->cwnd/2,
535
					  4*asoc->pathmtu);
L
Linus Torvalds 已提交
536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551
		transport->cwnd = transport->ssthresh;
		break;

	case SCTP_LOWER_CWND_ECNE:
		/* RFC 2481 Section 6.1.2.
		 * If the sender receives an ECN-Echo ACK packet
		 * then the sender knows that congestion was encountered in the
		 * network on the path from the sender to the receiver. The
		 * indication of congestion should be treated just as a
		 * congestion loss in non-ECN Capable TCP. That is, the TCP
		 * source halves the congestion window "cwnd" and reduces the
		 * slow start threshold "ssthresh".
		 * A critical condition is that TCP does not react to
		 * congestion indications more than once every window of
		 * data (or more loosely more than once every round-trip time).
		 */
552 553
		if (time_after(jiffies, transport->last_time_ecne_reduced +
					transport->rtt)) {
L
Linus Torvalds 已提交
554
			transport->ssthresh = max(transport->cwnd/2,
555
						  4*asoc->pathmtu);
L
Linus Torvalds 已提交
556 557 558 559 560 561 562 563 564 565 566 567 568 569
			transport->cwnd = transport->ssthresh;
			transport->last_time_ecne_reduced = jiffies;
		}
		break;

	case SCTP_LOWER_CWND_INACTIVE:
		/* RFC 2960 Section 7.2.1, sctpimpguide
		 * When the endpoint does not transmit data on a given
		 * transport address, the cwnd of the transport address
		 * should be adjusted to max(cwnd/2, 4*MTU) per RTO.
		 * NOTE: Although the draft recommends that this check needs
		 * to be done every RTO interval, we do it every hearbeat
		 * interval.
		 */
570
		transport->cwnd = max(transport->cwnd/2,
571
					 4*asoc->pathmtu);
572 573
		/* RFC 4960 Errata 3.27.2: also adjust sshthresh */
		transport->ssthresh = transport->cwnd;
L
Linus Torvalds 已提交
574
		break;
575
	}
L
Linus Torvalds 已提交
576 577

	transport->partial_bytes_acked = 0;
578 579 580 581

	pr_debug("%s: transport:%p, reason:%d, cwnd:%d, ssthresh:%d\n",
		 __func__, transport, reason, transport->cwnd,
		 transport->ssthresh);
L
Linus Torvalds 已提交
582 583
}

584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599
/* Apply Max.Burst limit to the congestion window:
 * sctpimpguide-05 2.14.2
 * D) When the time comes for the sender to
 * transmit new DATA chunks, the protocol parameter Max.Burst MUST
 * first be applied to limit how many new DATA chunks may be sent.
 * The limit is applied by adjusting cwnd as follows:
 * 	if ((flightsize+ Max.Burst * MTU) < cwnd)
 * 		cwnd = flightsize + Max.Burst * MTU
 */

void sctp_transport_burst_limited(struct sctp_transport *t)
{
	struct sctp_association *asoc = t->asoc;
	u32 old_cwnd = t->cwnd;
	u32 max_burst_bytes;

600
	if (t->burst_limited || asoc->max_burst == 0)
601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620
		return;

	max_burst_bytes = t->flight_size + (asoc->max_burst * asoc->pathmtu);
	if (max_burst_bytes < old_cwnd) {
		t->cwnd = max_burst_bytes;
		t->burst_limited = old_cwnd;
	}
}

/* Restore the old cwnd congestion window, after the burst had it's
 * desired effect.
 */
void sctp_transport_burst_reset(struct sctp_transport *t)
{
	if (t->burst_limited) {
		t->cwnd = t->burst_limited;
		t->burst_limited = 0;
	}
}

L
Linus Torvalds 已提交
621
/* What is the next timeout value for this transport? */
622
unsigned long sctp_transport_timeout(struct sctp_transport *trans)
L
Linus Torvalds 已提交
623
{
624
	/* RTO + timer slack +/- 50% of RTO */
625
	unsigned long timeout = trans->rto >> 1;
626 627 628 629 630

	if (trans->state != SCTP_UNCONFIRMED &&
	    trans->state != SCTP_PF)
		timeout += trans->hbinterval;

631
	return timeout;
L
Linus Torvalds 已提交
632
}
633 634 635 636 637 638 639 640 641 642 643 644

/* Reset transport variables to their initial values */
void sctp_transport_reset(struct sctp_transport *t)
{
	struct sctp_association *asoc = t->asoc;

	/* RFC 2960 (bis), Section 5.2.4
	 * All the congestion control parameters (e.g., cwnd, ssthresh)
	 * related to this peer MUST be reset to their initial values
	 * (see Section 6.2.1)
	 */
	t->cwnd = min(4*asoc->pathmtu, max_t(__u32, 2*asoc->pathmtu, 4380));
645
	t->burst_limited = 0;
646
	t->ssthresh = asoc->peer.i.a_rwnd;
647
	t->rto = asoc->rto_initial;
648
	sctp_max_rto(asoc, t);
649 650 651 652
	t->rtt = 0;
	t->srtt = 0;
	t->rttvar = 0;

653
	/* Reset these additional variables so that we have a clean slate. */
654 655 656 657
	t->partial_bytes_acked = 0;
	t->flight_size = 0;
	t->error_count = 0;
	t->rto_pending = 0;
658
	t->hb_sent = 0;
659 660 661 662 663 664 665

	/* Initialize the state information for SFR-CACC */
	t->cacc.changeover_active = 0;
	t->cacc.cycling_changeover = 0;
	t->cacc.next_tsn_at_change = 0;
	t->cacc.cacc_saw_newack = 0;
}
666 667 668 669 670

/* Schedule retransmission on the given transport */
void sctp_transport_immediate_rtx(struct sctp_transport *t)
{
	/* Stop pending T3_rtx_timer */
671
	if (del_timer(&t->T3_rtx_timer))
672
		sctp_transport_put(t);
673

674 675 676 677 678 679
	sctp_retransmit(&t->asoc->outqueue, t, SCTP_RTXR_T3_RTX);
	if (!timer_pending(&t->T3_rtx_timer)) {
		if (!mod_timer(&t->T3_rtx_timer, jiffies + t->rto))
			sctp_transport_hold(t);
	}
}
680 681 682 683 684 685 686 687 688 689 690 691 692 693

/* Drop dst */
void sctp_transport_dst_release(struct sctp_transport *t)
{
	dst_release(t->dst);
	t->dst = NULL;
	t->dst_pending_confirm = 0;
}

/* Schedule neighbour confirm */
void sctp_transport_dst_confirm(struct sctp_transport *t)
{
	t->dst_pending_confirm = 1;
}