transport.c 21.4 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 91 92 93 94
	timer_setup(&peer->T3_rtx_timer, sctp_generate_t3_rtx_event, 0);
	timer_setup(&peer->hb_timer, sctp_generate_heartbeat_event, 0);
	timer_setup(&peer->reconf_timer, sctp_generate_reconf_event, 0);
	timer_setup(&peer->proto_unreach_timer,
		    sctp_generate_proto_unreach_event, 0);
L
Linus Torvalds 已提交
95

96 97 98
	/* Initialize the 64-bit random nonce sent with heartbeat. */
	get_random_bytes(&peer->hb_nonce, sizeof(peer->hb_nonce));

99
	refcount_set(&peer->refcnt, 1);
L
Linus Torvalds 已提交
100 101 102 103 104

	return peer;
}

/* Allocate and initialize a new transport.  */
105 106
struct sctp_transport *sctp_transport_new(struct net *net,
					  const union sctp_addr *addr,
A
Al Viro 已提交
107
					  gfp_t gfp)
L
Linus Torvalds 已提交
108
{
109
	struct sctp_transport *transport;
L
Linus Torvalds 已提交
110

111
	transport = kzalloc(sizeof(*transport), gfp);
L
Linus Torvalds 已提交
112 113 114
	if (!transport)
		goto fail;

115
	if (!sctp_transport_init(net, transport, addr, gfp))
L
Linus Torvalds 已提交
116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142
		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.
	 */
143
	if (del_timer(&transport->T3_rtx_timer))
L
Linus Torvalds 已提交
144 145
		sctp_transport_put(transport);

X
Xin Long 已提交
146 147 148
	if (del_timer(&transport->reconf_timer))
		sctp_transport_put(transport);

149
	/* Delete the ICMP proto unreachable timer if it's active. */
150
	if (del_timer(&transport->proto_unreach_timer))
151
		sctp_association_put(transport->asoc);
L
Linus Torvalds 已提交
152 153 154 155

	sctp_transport_put(transport);
}

156
static void sctp_transport_destroy_rcu(struct rcu_head *head)
L
Linus Torvalds 已提交
157
{
158
	struct sctp_transport *transport;
L
Linus Torvalds 已提交
159

160
	transport = container_of(head, struct sctp_transport, rcu);
L
Linus Torvalds 已提交
161 162 163 164 165 166

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

167 168 169 170 171
/* Destroy the transport data structure.
 * Assumes there are no more users of this structure.
 */
static void sctp_transport_destroy(struct sctp_transport *transport)
{
172
	if (unlikely(refcount_read(&transport->refcnt))) {
173 174 175
		WARN(1, "Attempt to destroy undead transport %p!\n", transport);
		return;
	}
176

177 178 179 180
	sctp_packet_free(&transport->packet);

	if (transport->asoc)
		sctp_association_put(transport->asoc);
181 182

	call_rcu(&transport->rcu, sctp_transport_destroy_rcu);
183 184
}

L
Linus Torvalds 已提交
185 186 187
/* 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.
 */
188
void sctp_transport_reset_t3_rtx(struct sctp_transport *transport)
L
Linus Torvalds 已提交
189 190 191 192 193 194 195 196 197
{
	/* 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.
	 */

198
	if (!timer_pending(&transport->T3_rtx_timer))
L
Linus Torvalds 已提交
199 200 201
		if (!mod_timer(&transport->T3_rtx_timer,
			       jiffies + transport->rto))
			sctp_transport_hold(transport);
202 203 204 205 206
}

void sctp_transport_reset_hb_timer(struct sctp_transport *transport)
{
	unsigned long expires;
L
Linus Torvalds 已提交
207 208

	/* When a data chunk is sent, reset the heartbeat interval.  */
209
	expires = jiffies + sctp_transport_timeout(transport);
210 211
	if ((time_before(transport->hb_timer.expires, expires) ||
	     !timer_pending(&transport->hb_timer)) &&
212 213 214
	    !mod_timer(&transport->hb_timer,
		       expires + prandom_u32_max(transport->rto)))
		sctp_transport_hold(transport);
L
Linus Torvalds 已提交
215 216
}

X
Xin Long 已提交
217 218 219 220 221 222 223 224
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
225 226 227 228 229 230 231 232 233 234 235 236
/* 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. */
237
void sctp_transport_pmtu(struct sctp_transport *transport, struct sock *sk)
L
Linus Torvalds 已提交
238
{
239
	/* If we don't have a fresh route, look one up */
240
	if (!transport->dst || transport->dst->obsolete) {
241
		sctp_transport_dst_release(transport);
242
		transport->af_specific->get_dst(transport, &transport->saddr,
243
						&transport->fl, sk);
244
	}
L
Linus Torvalds 已提交
245

246 247 248 249 250 251 252 253 254
	if (transport->param_flags & SPP_PMTUD_DISABLE) {
		struct sctp_association *asoc = transport->asoc;

		if (!transport->pathmtu && asoc && asoc->pathmtu)
			transport->pathmtu = asoc->pathmtu;
		if (transport->pathmtu)
			return;
	}

255 256 257
	if (transport->dst)
		transport->pathmtu = sctp_dst_mtu(transport->dst);
	else
258
		transport->pathmtu = SCTP_DEFAULT_MAXSEGMENT;
L
Linus Torvalds 已提交
259 260
}

261
bool sctp_transport_update_pmtu(struct sctp_transport *t, u32 pmtu)
262
{
263
	struct dst_entry *dst = sctp_transport_dst_check(t);
264
	struct sock *sk = t->asoc->base.sk;
265
	bool change = true;
266 267

	if (unlikely(pmtu < SCTP_DEFAULT_MINSEGMENT)) {
268 269 270 271
		pr_warn_ratelimited("%s: Reported pmtu %d too low, using default minimum of %d\n",
				    __func__, pmtu, SCTP_DEFAULT_MINSEGMENT);
		/* Use default minimum segment instead */
		pmtu = SCTP_DEFAULT_MINSEGMENT;
272
	}
273
	pmtu = SCTP_TRUNC4(pmtu);
274

275
	if (dst) {
276 277 278 279 280 281 282 283
		struct sctp_pf *pf = sctp_get_pf_specific(dst->ops->family);
		union sctp_addr addr;

		pf->af->from_sk(&addr, sk);
		pf->to_sk_daddr(&t->ipaddr, sk);
		dst->ops->update_pmtu(dst, sk, NULL, pmtu);
		pf->to_sk_daddr(&addr, sk);

284 285
		dst = sctp_transport_dst_check(t);
	}
286

287
	if (!dst) {
288
		t->af_specific->get_dst(t, &t->saddr, &t->fl, sk);
289 290 291 292 293
		dst = t->dst;
	}

	if (dst) {
		/* Re-fetch, as under layers may have a higher minimum size */
294
		pmtu = sctp_dst_mtu(dst);
295 296 297 298 299
		change = t->pathmtu != pmtu;
	}
	t->pathmtu = pmtu;

	return change;
300 301
}

L
Linus Torvalds 已提交
302 303 304 305 306 307 308 309 310
/* 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;

311
	sctp_transport_dst_release(transport);
312
	af->get_dst(transport, saddr, &transport->fl, sctp_opt2sk(opt));
L
Linus Torvalds 已提交
313 314 315 316

	if (saddr)
		memcpy(&transport->saddr, saddr, sizeof(union sctp_addr));
	else
317
		af->get_saddr(opt, transport, &transport->fl);
L
Linus Torvalds 已提交
318

319 320 321 322 323 324 325 326
	sctp_transport_pmtu(transport, sctp_opt2sk(opt));

	/* Initialize sk->sk_rcv_saddr, if the transport is the
	 * association's active path for getsockname().
	 */
	if (transport->dst && asoc &&
	    (!asoc->peer.primary_path || transport == asoc->peer.active_path))
		opt->pf->to_sk_saddr(&transport->saddr, asoc->base.sk);
L
Linus Torvalds 已提交
327 328 329
}

/* Hold a reference to a transport.  */
330
int sctp_transport_hold(struct sctp_transport *transport)
L
Linus Torvalds 已提交
331
{
332
	return refcount_inc_not_zero(&transport->refcnt);
L
Linus Torvalds 已提交
333 334 335 336 337 338 339
}

/* Release a reference to a transport and clean up
 * if there are no more references.
 */
void sctp_transport_put(struct sctp_transport *transport)
{
340
	if (refcount_dec_and_test(&transport->refcnt))
L
Linus Torvalds 已提交
341 342 343 344 345 346
		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)
{
347 348 349
	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
350 351

	if (tp->rttvar || tp->srtt) {
352
		struct net *net = sock_net(tp->asoc->base.sk);
L
Linus Torvalds 已提交
353 354 355 356 357 358 359 360 361 362 363
		/* 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.
		 */
364
		tp->rttvar = tp->rttvar - (tp->rttvar >> net->sctp.rto_beta)
A
Andrew Morton 已提交
365
			+ (((__u32)abs((__s64)tp->srtt - (__s64)rtt)) >> net->sctp.rto_beta);
366 367
		tp->srtt = tp->srtt - (tp->srtt >> net->sctp.rto_alpha)
			+ (rtt >> net->sctp.rto_alpha);
L
Linus Torvalds 已提交
368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
	} 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;

397
	sctp_max_rto(tp->asoc, tp);
L
Linus Torvalds 已提交
398 399 400 401 402 403 404
	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;

405 406
	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
407 408 409 410 411 412 413 414
}

/* 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)
{
415
	struct sctp_association *asoc = transport->asoc;
L
Linus Torvalds 已提交
416 417 418 419 420
	__u32 cwnd, ssthresh, flight_size, pba, pmtu;

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

421
	/* See if we need to exit Fast Recovery first */
422 423 424
	if (asoc->fast_recovery &&
	    TSN_lte(asoc->fast_recovery_exit, sack_ctsn))
		asoc->fast_recovery = 0;
425

L
Linus Torvalds 已提交
426 427
	ssthresh = transport->ssthresh;
	pba = transport->partial_bytes_acked;
428
	pmtu = transport->asoc->pathmtu;
L
Linus Torvalds 已提交
429 430

	if (cwnd <= ssthresh) {
431 432 433 434 435 436 437 438 439 440 441 442 443
		/* 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
444
		 */
445
		if (asoc->fast_recovery)
446 447
			return;

448 449 450 451 452 453 454 455
		/* 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
456 457 458 459
		if (bytes_acked > pmtu)
			cwnd += pmtu;
		else
			cwnd += bytes_acked;
460 461 462 463 464

		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
465 466
	} else {
		/* RFC 2960 7.2.2 Whenever cwnd is greater than ssthresh,
467 468 469 470 471
		 * 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
472
		 *
473 474 475 476 477 478 479
		 * 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)
		 *
480 481 482 483 484 485 486
		 * 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
487 488
		 */
		pba += bytes_acked;
489 490 491
		if (pba > cwnd && flight_size < cwnd)
			pba = cwnd;
		if (pba >= cwnd && flight_size >= cwnd) {
492
			pba = pba - cwnd;
L
Linus Torvalds 已提交
493 494
			cwnd += pmtu;
		}
495 496 497 498 499 500

		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
501 502 503 504 505 506 507 508 509 510
	}

	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,
511
			       enum sctp_lower_cwnd reason)
L
Linus Torvalds 已提交
512
{
513 514
	struct sctp_association *asoc = transport->asoc;

L
Linus Torvalds 已提交
515 516 517 518 519 520 521 522 523 524
	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,
525 526
					  4*asoc->pathmtu);
		transport->cwnd = asoc->pathmtu;
527

528 529
		/* T3-rtx also clears fast recovery */
		asoc->fast_recovery = 0;
L
Linus Torvalds 已提交
530 531 532 533 534 535 536
		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.
537 538
		 *
		 * RFC 2960 7.2.3, sctpimpguide Upon detection of packet
L
Linus Torvalds 已提交
539 540 541 542 543 544
		 * 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
		 */
545
		if (asoc->fast_recovery)
546 547 548
			return;

		/* Mark Fast recovery */
549 550
		asoc->fast_recovery = 1;
		asoc->fast_recovery_exit = asoc->next_tsn - 1;
551

L
Linus Torvalds 已提交
552
		transport->ssthresh = max(transport->cwnd/2,
553
					  4*asoc->pathmtu);
L
Linus Torvalds 已提交
554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569
		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).
		 */
570 571
		if (time_after(jiffies, transport->last_time_ecne_reduced +
					transport->rtt)) {
L
Linus Torvalds 已提交
572
			transport->ssthresh = max(transport->cwnd/2,
573
						  4*asoc->pathmtu);
L
Linus Torvalds 已提交
574 575 576 577 578 579 580 581 582 583 584 585 586 587
			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.
		 */
588
		transport->cwnd = max(transport->cwnd/2,
589
					 4*asoc->pathmtu);
590 591
		/* RFC 4960 Errata 3.27.2: also adjust sshthresh */
		transport->ssthresh = transport->cwnd;
L
Linus Torvalds 已提交
592
		break;
593
	}
L
Linus Torvalds 已提交
594 595

	transport->partial_bytes_acked = 0;
596 597 598 599

	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
600 601
}

602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617
/* 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;

618
	if (t->burst_limited || asoc->max_burst == 0)
619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
		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
639
/* What is the next timeout value for this transport? */
640
unsigned long sctp_transport_timeout(struct sctp_transport *trans)
L
Linus Torvalds 已提交
641
{
642
	/* RTO + timer slack +/- 50% of RTO */
643
	unsigned long timeout = trans->rto >> 1;
644 645 646 647 648

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

649
	return max_t(unsigned long, timeout, HZ / 5);
L
Linus Torvalds 已提交
650
}
651 652 653 654 655 656 657 658 659 660 661 662

/* 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));
663
	t->burst_limited = 0;
664
	t->ssthresh = asoc->peer.i.a_rwnd;
665
	t->rto = asoc->rto_initial;
666
	sctp_max_rto(asoc, t);
667 668 669 670
	t->rtt = 0;
	t->srtt = 0;
	t->rttvar = 0;

671
	/* Reset these additional variables so that we have a clean slate. */
672 673 674 675
	t->partial_bytes_acked = 0;
	t->flight_size = 0;
	t->error_count = 0;
	t->rto_pending = 0;
676
	t->hb_sent = 0;
677 678 679 680 681 682 683

	/* 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;
}
684 685 686 687 688

/* Schedule retransmission on the given transport */
void sctp_transport_immediate_rtx(struct sctp_transport *t)
{
	/* Stop pending T3_rtx_timer */
689
	if (del_timer(&t->T3_rtx_timer))
690
		sctp_transport_put(t);
691

692 693 694 695 696 697
	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);
	}
}
698 699 700 701 702 703 704 705 706 707 708 709 710 711

/* 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;
}