ccid2.c 19.1 KB
Newer Older
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
/*
 *  net/dccp/ccids/ccid2.c
 *
 *  Copyright (c) 2005, 2006 Andrea Bittau <a.bittau@cs.ucl.ac.uk>
 *
 *  Changes to meet Linux coding standards, and DCCP infrastructure fixes.
 *
 *  Copyright (c) 2006 Arnaldo Carvalho de Melo <acme@conectiva.com.br>
 *
 *  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.
 *
 *  This program is distributed in the hope that it 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
 *  along with this program; if not, write to the Free Software
 *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

/*
26
 * This implementation should follow RFC 4341
27
 */
28
#include "../feat.h"
29 30 31 32 33
#include "../ccid.h"
#include "../dccp.h"
#include "ccid2.h"


34
#ifdef CONFIG_IP_DCCP_CCID2_DEBUG
35 36
static int ccid2_debug;
#define ccid2_pr_debug(format, a...)	DCCP_PR_DEBUG(ccid2_debug, format, ##a)
37
#else
38
#define ccid2_pr_debug(format, a...)
39 40
#endif

G
Gerrit Renker 已提交
41
static int ccid2_hc_tx_alloc_seq(struct ccid2_hc_tx_sock *hctx)
42 43 44 45 46
{
	struct ccid2_seq *seqp;
	int i;

	/* check if we have space to preserve the pointer to the buffer */
47
	if (hctx->seqbufc >= sizeof(hctx->seqbuf) / sizeof(struct ccid2_seq *))
48 49 50
		return -ENOMEM;

	/* allocate buffer and initialize linked list */
G
Gerrit Renker 已提交
51
	seqp = kmalloc(CCID2_SEQBUF_LEN * sizeof(struct ccid2_seq), gfp_any());
52 53 54
	if (seqp == NULL)
		return -ENOMEM;

G
Gerrit Renker 已提交
55
	for (i = 0; i < (CCID2_SEQBUF_LEN - 1); i++) {
56 57 58
		seqp[i].ccid2s_next = &seqp[i + 1];
		seqp[i + 1].ccid2s_prev = &seqp[i];
	}
G
Gerrit Renker 已提交
59 60
	seqp[CCID2_SEQBUF_LEN - 1].ccid2s_next = seqp;
	seqp->ccid2s_prev = &seqp[CCID2_SEQBUF_LEN - 1];
61 62

	/* This is the first allocation.  Initiate the head and tail.  */
63 64
	if (hctx->seqbufc == 0)
		hctx->seqh = hctx->seqt = seqp;
65 66
	else {
		/* link the existing list with the one we just created */
67 68
		hctx->seqh->ccid2s_next = seqp;
		seqp->ccid2s_prev = hctx->seqh;
69

70 71
		hctx->seqt->ccid2s_prev = &seqp[CCID2_SEQBUF_LEN - 1];
		seqp[CCID2_SEQBUF_LEN - 1].ccid2s_next = hctx->seqt;
72 73 74
	}

	/* store the original pointer to the buffer so we can free it */
75 76
	hctx->seqbuf[hctx->seqbufc] = seqp;
	hctx->seqbufc++;
77 78 79 80

	return 0;
}

81
static int ccid2_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb)
82
{
G
Gerrit Renker 已提交
83 84 85
	if (ccid2_cwnd_network_limited(ccid2_hc_tx_sk(sk)))
		return CCID_PACKET_WILL_DEQUEUE_LATER;
	return CCID_PACKET_SEND_AT_ONCE;
86 87
}

88
static void ccid2_change_l_ack_ratio(struct sock *sk, u32 val)
89 90
{
	struct dccp_sock *dp = dccp_sk(sk);
91
	u32 max_ratio = DIV_ROUND_UP(ccid2_hc_tx_sk(sk)->cwnd, 2);
92

93
	/*
94 95 96 97
	 * Ensure that Ack Ratio does not exceed ceil(cwnd/2), which is (2) from
	 * RFC 4341, 6.1.2. We ignore the statement that Ack Ratio 2 is always
	 * acceptable since this causes starvation/deadlock whenever cwnd < 2.
	 * The same problem arises when Ack Ratio is 0 (ie. Ack Ratio disabled).
98
	 */
99 100 101
	if (val == 0 || val > max_ratio) {
		DCCP_WARN("Limiting Ack Ratio (%u) to %u\n", val, max_ratio);
		val = max_ratio;
102
	}
103 104
	if (val > DCCPF_ACK_RATIO_MAX)
		val = DCCPF_ACK_RATIO_MAX;
105

106 107 108
	if (val == dp->dccps_l_ack_ratio)
		return;

109
	ccid2_pr_debug("changing local ack ratio to %u\n", val);
110 111 112 113 114 115 116
	dp->dccps_l_ack_ratio = val;
}

static void ccid2_hc_tx_rto_expire(unsigned long data)
{
	struct sock *sk = (struct sock *)data;
	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
G
Gerrit Renker 已提交
117
	const bool sender_was_blocked = ccid2_cwnd_network_limited(hctx);
118 119 120

	bh_lock_sock(sk);
	if (sock_owned_by_user(sk)) {
121
		sk_reset_timer(sk, &hctx->rtotimer, jiffies + HZ / 5);
122 123 124 125 126 127
		goto out;
	}

	ccid2_pr_debug("RTO_EXPIRE\n");

	/* back-off timer */
128
	hctx->rto <<= 1;
129 130
	if (hctx->rto > DCCP_RTO_MAX)
		hctx->rto = DCCP_RTO_MAX;
131 132

	/* adjust pipe, cwnd etc */
133 134 135 136 137
	hctx->ssthresh = hctx->cwnd / 2;
	if (hctx->ssthresh < 2)
		hctx->ssthresh = 2;
	hctx->cwnd = 1;
	hctx->pipe = 0;
138 139

	/* clear state about stuff we sent */
140 141
	hctx->seqt = hctx->seqh;
	hctx->packets_acked = 0;
142 143

	/* clear ack ratio state. */
144 145
	hctx->rpseq    = 0;
	hctx->rpdupack = -1;
146
	ccid2_change_l_ack_ratio(sk, 1);
G
Gerrit Renker 已提交
147 148 149 150

	/* if we were blocked before, we may now send cwnd=1 packet */
	if (sender_was_blocked)
		tasklet_schedule(&dccp_sk(sk)->dccps_xmitlet);
151 152
	/* restart backed-off timer */
	sk_reset_timer(sk, &hctx->rtotimer, jiffies + hctx->rto);
153 154
out:
	bh_unlock_sock(sk);
155
	sock_put(sk);
156 157
}

158
static void ccid2_hc_tx_packet_sent(struct sock *sk, unsigned int len)
159 160 161
{
	struct dccp_sock *dp = dccp_sk(sk);
	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
162
	struct ccid2_seq *next;
163

164
	hctx->pipe++;
165

166 167 168
	hctx->seqh->ccid2s_seq   = dp->dccps_gss;
	hctx->seqh->ccid2s_acked = 0;
	hctx->seqh->ccid2s_sent  = jiffies;
169

170
	next = hctx->seqh->ccid2s_next;
171
	/* check if we need to alloc more space */
172
	if (next == hctx->seqt) {
G
Gerrit Renker 已提交
173 174 175 176 177
		if (ccid2_hc_tx_alloc_seq(hctx)) {
			DCCP_CRIT("packet history - out of memory!");
			/* FIXME: find a more graceful way to bail out */
			return;
		}
178 179
		next = hctx->seqh->ccid2s_next;
		BUG_ON(next == hctx->seqt);
180
	}
181
	hctx->seqh = next;
182

183
	ccid2_pr_debug("cwnd=%d pipe=%d\n", hctx->cwnd, hctx->pipe);
184

185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204
	/*
	 * FIXME: The code below is broken and the variables have been removed
	 * from the socket struct. The `ackloss' variable was always set to 0,
	 * and with arsent there are several problems:
	 *  (i) it doesn't just count the number of Acks, but all sent packets;
	 *  (ii) it is expressed in # of packets, not # of windows, so the
	 *  comparison below uses the wrong formula: Appendix A of RFC 4341
	 *  comes up with the number K = cwnd / (R^2 - R) of consecutive windows
	 *  of data with no lost or marked Ack packets. If arsent were the # of
	 *  consecutive Acks received without loss, then Ack Ratio needs to be
	 *  decreased by 1 when
	 *	      arsent >=  K * cwnd / R  =  cwnd^2 / (R^3 - R^2)
	 *  where cwnd / R is the number of Acks received per window of data
	 *  (cf. RFC 4341, App. A). The problems are that
	 *  - arsent counts other packets as well;
	 *  - the comparison uses a formula different from RFC 4341;
	 *  - computing a cubic/quadratic equation each time is too complicated.
	 *  Hence a different algorithm is needed.
	 */
#if 0
205
	/* Ack Ratio.  Need to maintain a concept of how many windows we sent */
206
	hctx->arsent++;
207
	/* We had an ack loss in this window... */
208 209 210 211
	if (hctx->ackloss) {
		if (hctx->arsent >= hctx->cwnd) {
			hctx->arsent  = 0;
			hctx->ackloss = 0;
212
		}
213 214
	} else {
		/* No acks lost up to now... */
215 216 217
		/* decrease ack ratio if enough packets were sent */
		if (dp->dccps_l_ack_ratio > 1) {
			/* XXX don't calculate denominator each time */
218 219
			int denom = dp->dccps_l_ack_ratio * dp->dccps_l_ack_ratio -
				    dp->dccps_l_ack_ratio;
220

221
			denom = hctx->cwnd * hctx->cwnd / denom;
222

223
			if (hctx->arsent >= denom) {
224
				ccid2_change_l_ack_ratio(sk, dp->dccps_l_ack_ratio - 1);
225
				hctx->arsent = 0;
226
			}
227 228
		} else {
			/* we can't increase ack ratio further [1] */
229
			hctx->arsent = 0; /* or maybe set it to cwnd*/
230 231
		}
	}
232
#endif
233 234

	/* setup RTO timer */
235
	if (!timer_pending(&hctx->rtotimer))
236
		sk_reset_timer(sk, &hctx->rtotimer, jiffies + hctx->rto);
237

238
#ifdef CONFIG_IP_DCCP_CCID2_DEBUG
239
	do {
240
		struct ccid2_seq *seqp = hctx->seqt;
241

242
		while (seqp != hctx->seqh) {
243
			ccid2_pr_debug("out seq=%llu acked=%d time=%lu\n",
244
				       (unsigned long long)seqp->ccid2s_seq,
R
Randy Dunlap 已提交
245
				       seqp->ccid2s_acked, seqp->ccid2s_sent);
246 247
			seqp = seqp->ccid2s_next;
		}
248
	} while (0);
249 250 251 252
	ccid2_pr_debug("=========\n");
#endif
}

253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333
/**
 * ccid2_rtt_estimator - Sample RTT and compute RTO using RFC2988 algorithm
 * This code is almost identical with TCP's tcp_rtt_estimator(), since
 * - it has a higher sampling frequency (recommended by RFC 1323),
 * - the RTO does not collapse into RTT due to RTTVAR going towards zero,
 * - it is simple (cf. more complex proposals such as Eifel timer or research
 *   which suggests that the gain should be set according to window size),
 * - in tests it was found to work well with CCID2 [gerrit].
 */
static void ccid2_rtt_estimator(struct sock *sk, const long mrtt)
{
	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
	long m = mrtt ? : 1;

	if (hctx->srtt == 0) {
		/* First measurement m */
		hctx->srtt = m << 3;
		hctx->mdev = m << 1;

		hctx->mdev_max = max(TCP_RTO_MIN, hctx->mdev);
		hctx->rttvar   = hctx->mdev_max;
		hctx->rtt_seq  = dccp_sk(sk)->dccps_gss;
	} else {
		/* Update scaled SRTT as SRTT += 1/8 * (m - SRTT) */
		m -= (hctx->srtt >> 3);
		hctx->srtt += m;

		/* Similarly, update scaled mdev with regard to |m| */
		if (m < 0) {
			m = -m;
			m -= (hctx->mdev >> 2);
			/*
			 * This neutralises RTO increase when RTT < SRTT - mdev
			 * (see P. Sarolahti, A. Kuznetsov,"Congestion Control
			 * in Linux TCP", USENIX 2002, pp. 49-62).
			 */
			if (m > 0)
				m >>= 3;
		} else {
			m -= (hctx->mdev >> 2);
		}
		hctx->mdev += m;

		if (hctx->mdev > hctx->mdev_max) {
			hctx->mdev_max = hctx->mdev;
			if (hctx->mdev_max > hctx->rttvar)
				hctx->rttvar = hctx->mdev_max;
		}

		/*
		 * Decay RTTVAR at most once per flight, exploiting that
		 *  1) pipe <= cwnd <= Sequence_Window = W  (RFC 4340, 7.5.2)
		 *  2) AWL = GSS-W+1 <= GAR <= GSS          (RFC 4340, 7.5.1)
		 * GAR is a useful bound for FlightSize = pipe, AWL is probably
		 * too low as it over-estimates pipe.
		 */
		if (after48(dccp_sk(sk)->dccps_gar, hctx->rtt_seq)) {
			if (hctx->mdev_max < hctx->rttvar)
				hctx->rttvar -= (hctx->rttvar -
						 hctx->mdev_max) >> 2;
			hctx->rtt_seq  = dccp_sk(sk)->dccps_gss;
			hctx->mdev_max = TCP_RTO_MIN;
		}
	}

	/*
	 * Set RTO from SRTT and RTTVAR
	 * Clock granularity is ignored since the minimum error for RTTVAR is
	 * clamped to 50msec (corresponding to HZ=20). This leads to a minimum
	 * RTO of 200msec. This agrees with TCP and RFC 4341, 5.: "Because DCCP
	 * does not retransmit data, DCCP does not require TCP's recommended
	 * minimum timeout of one second".
	 */
	hctx->rto = (hctx->srtt >> 3) + hctx->rttvar;

	if (hctx->rto > DCCP_RTO_MAX)
		hctx->rto = DCCP_RTO_MAX;
}

static void ccid2_new_ack(struct sock *sk, struct ccid2_seq *seqp,
			  unsigned int *maxincr)
334 335 336
{
	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);

337 338 339 340 341
	if (hctx->cwnd < hctx->ssthresh) {
		if (*maxincr > 0 && ++hctx->packets_acked == 2) {
			hctx->cwnd += 1;
			*maxincr   -= 1;
			hctx->packets_acked = 0;
342
		}
343 344 345
	} else if (++hctx->packets_acked >= hctx->cwnd) {
			hctx->cwnd += 1;
			hctx->packets_acked = 0;
346
	}
347 348 349 350 351 352 353 354 355
	/*
	 * FIXME: RTT is sampled several times per acknowledgment (for each
	 * entry in the Ack Vector), instead of once per Ack (as in TCP SACK).
	 * This causes the RTT to be over-estimated, since the older entries
	 * in the Ack Vector have earlier sending times.
	 * The cleanest solution is to not use the ccid2s_sent field at all
	 * and instead use DCCP timestamps - need to be resolved at some time.
	 */
	ccid2_rtt_estimator(sk, jiffies - seqp->ccid2s_sent);
356 357
}

358
static void ccid2_congestion_event(struct sock *sk, struct ccid2_seq *seqp)
359
{
360 361
	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);

362
	if (time_before(seqp->ccid2s_sent, hctx->last_cong)) {
363 364 365 366
		ccid2_pr_debug("Multiple losses in an RTT---treating as one\n");
		return;
	}

367
	hctx->last_cong = jiffies;
368

369 370
	hctx->cwnd     = hctx->cwnd / 2 ? : 1U;
	hctx->ssthresh = max(hctx->cwnd, 2U);
371 372

	/* Avoid spurious timeouts resulting from Ack Ratio > cwnd */
373 374
	if (dccp_sk(sk)->dccps_l_ack_ratio > hctx->cwnd)
		ccid2_change_l_ack_ratio(sk, hctx->cwnd);
375 376
}

377 378 379 380 381 382 383 384 385 386 387 388 389 390
static int ccid2_hc_tx_parse_options(struct sock *sk, u8 packet_type,
				     u8 option, u8 *optval, u8 optlen)
{
	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);

	switch (option) {
	case DCCPO_ACK_VECTOR_0:
	case DCCPO_ACK_VECTOR_1:
		return dccp_ackvec_parsed_add(&hctx->av_chunks, optval, optlen,
					      option - DCCPO_ACK_VECTOR_0);
	}
	return 0;
}

391 392 393 394
static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
{
	struct dccp_sock *dp = dccp_sk(sk);
	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
G
Gerrit Renker 已提交
395
	const bool sender_was_blocked = ccid2_cwnd_network_limited(hctx);
396
	struct dccp_ackvec_parsed *avp;
397 398 399 400 401 402 403 404 405 406 407 408 409
	u64 ackno, seqno;
	struct ccid2_seq *seqp;
	int done = 0;
	unsigned int maxincr = 0;

	/* check reverse path congestion */
	seqno = DCCP_SKB_CB(skb)->dccpd_seq;

	/* XXX this whole "algorithm" is broken.  Need to fix it to keep track
	 * of the seqnos of the dupacks so that rpseq and rpdupack are correct
	 * -sorbo.
	 */
	/* need to bootstrap */
410 411 412
	if (hctx->rpdupack == -1) {
		hctx->rpdupack = 0;
		hctx->rpseq = seqno;
413
	} else {
414
		/* check if packet is consecutive */
415 416
		if (dccp_delta_seqno(hctx->rpseq, seqno) == 1)
			hctx->rpseq = seqno;
417
		/* it's a later packet */
418 419
		else if (after48(seqno, hctx->rpseq)) {
			hctx->rpdupack++;
420 421

			/* check if we got enough dupacks */
422 423 424
			if (hctx->rpdupack >= NUMDUPACK) {
				hctx->rpdupack = -1; /* XXX lame */
				hctx->rpseq = 0;
425

426
				ccid2_change_l_ack_ratio(sk, 2 * dp->dccps_l_ack_ratio);
427 428 429 430 431
			}
		}
	}

	/* check forward path congestion */
432
	if (dccp_packet_without_ack(skb))
433 434
		return;

435 436 437
	/* still didn't send out new data packets */
	if (hctx->seqh == hctx->seqt)
		goto done;
438 439

	ackno = DCCP_SKB_CB(skb)->dccpd_ack_seq;
440 441
	if (after48(ackno, hctx->high_ack))
		hctx->high_ack = ackno;
A
Andrea Bittau 已提交
442

443
	seqp = hctx->seqt;
A
Andrea Bittau 已提交
444 445
	while (before48(seqp->ccid2s_seq, ackno)) {
		seqp = seqp->ccid2s_next;
446 447
		if (seqp == hctx->seqh) {
			seqp = hctx->seqh->ccid2s_prev;
A
Andrea Bittau 已提交
448 449 450
			break;
		}
	}
451

452 453 454 455
	/*
	 * In slow-start, cwnd can increase up to a maximum of Ack Ratio/2
	 * packets per acknowledgement. Rounding up avoids that cwnd is not
	 * advanced when Ack Ratio is 1 and gives a slight edge otherwise.
456
	 */
457
	if (hctx->cwnd < hctx->ssthresh)
458
		maxincr = DIV_ROUND_UP(dp->dccps_l_ack_ratio, 2);
459 460

	/* go through all ack vectors */
461
	list_for_each_entry(avp, &hctx->av_chunks, node) {
462
		/* go through this ack vector */
463 464 465
		for (; avp->len--; avp->vec++) {
			u64 ackno_end_rl = SUB48(ackno,
						 dccp_ackvec_runlen(avp->vec));
466

467
			ccid2_pr_debug("ackvec %llu |%u,%u|\n",
R
Randy Dunlap 已提交
468
				       (unsigned long long)ackno,
469 470
				       dccp_ackvec_state(avp->vec) >> 6,
				       dccp_ackvec_runlen(avp->vec));
471 472 473 474 475
			/* if the seqno we are analyzing is larger than the
			 * current ackno, then move towards the tail of our
			 * seqnos.
			 */
			while (after48(seqp->ccid2s_seq, ackno)) {
476
				if (seqp == hctx->seqt) {
477 478 479 480 481 482 483 484 485 486 487 488
					done = 1;
					break;
				}
				seqp = seqp->ccid2s_prev;
			}
			if (done)
				break;

			/* check all seqnos in the range of the vector
			 * run length
			 */
			while (between48(seqp->ccid2s_seq,ackno_end_rl,ackno)) {
489
				const u8 state = dccp_ackvec_state(avp->vec);
490 491

				/* new packet received or marked */
492
				if (state != DCCPAV_NOT_RECEIVED &&
493
				    !seqp->ccid2s_acked) {
494
					if (state == DCCPAV_ECN_MARKED)
495
						ccid2_congestion_event(sk,
496
								       seqp);
497
					else
498 499 500 501 502
						ccid2_new_ack(sk, seqp,
							      &maxincr);

					seqp->ccid2s_acked = 1;
					ccid2_pr_debug("Got ack for %llu\n",
R
Randy Dunlap 已提交
503
						       (unsigned long long)seqp->ccid2s_seq);
504
					hctx->pipe--;
505
				}
506
				if (seqp == hctx->seqt) {
507 508 509
					done = 1;
					break;
				}
510
				seqp = seqp->ccid2s_prev;
511 512 513 514
			}
			if (done)
				break;

515
			ackno = SUB48(ackno_end_rl, 1);
516 517 518 519 520 521 522 523
		}
		if (done)
			break;
	}

	/* The state about what is acked should be correct now
	 * Check for NUMDUPACK
	 */
524 525
	seqp = hctx->seqt;
	while (before48(seqp->ccid2s_seq, hctx->high_ack)) {
A
Andrea Bittau 已提交
526
		seqp = seqp->ccid2s_next;
527 528
		if (seqp == hctx->seqh) {
			seqp = hctx->seqh->ccid2s_prev;
A
Andrea Bittau 已提交
529 530 531
			break;
		}
	}
532 533 534 535
	done = 0;
	while (1) {
		if (seqp->ccid2s_acked) {
			done++;
536
			if (done == NUMDUPACK)
537 538
				break;
		}
539
		if (seqp == hctx->seqt)
540 541 542 543 544 545 546
			break;
		seqp = seqp->ccid2s_prev;
	}

	/* If there are at least 3 acknowledgements, anything unacknowledged
	 * below the last sequence number is considered lost
	 */
547
	if (done == NUMDUPACK) {
548 549 550 551 552
		struct ccid2_seq *last_acked = seqp;

		/* check for lost packets */
		while (1) {
			if (!seqp->ccid2s_acked) {
553
				ccid2_pr_debug("Packet lost: %llu\n",
R
Randy Dunlap 已提交
554
					       (unsigned long long)seqp->ccid2s_seq);
555 556 557 558
				/* XXX need to traverse from tail -> head in
				 * order to detect multiple congestion events in
				 * one ack vector.
				 */
559
				ccid2_congestion_event(sk, seqp);
560
				hctx->pipe--;
561
			}
562
			if (seqp == hctx->seqt)
563 564 565 566
				break;
			seqp = seqp->ccid2s_prev;
		}

567
		hctx->seqt = last_acked;
568 569 570
	}

	/* trim acked packets in tail */
571 572
	while (hctx->seqt != hctx->seqh) {
		if (!hctx->seqt->ccid2s_acked)
573 574
			break;

575
		hctx->seqt = hctx->seqt->ccid2s_next;
576 577
	}

578 579 580 581
	/* restart RTO timer if not all outstanding data has been acked */
	if (hctx->pipe == 0)
		sk_stop_timer(sk, &hctx->rtotimer);
	else
582
		sk_reset_timer(sk, &hctx->rtotimer, jiffies + hctx->rto);
583
done:
G
Gerrit Renker 已提交
584 585 586
	/* check if incoming Acks allow pending packets to be sent */
	if (sender_was_blocked && !ccid2_cwnd_network_limited(hctx))
		tasklet_schedule(&dccp_sk(sk)->dccps_xmitlet);
587
	dccp_ackvec_parsed_cleanup(&hctx->av_chunks);
588 589
}

590
static int ccid2_hc_tx_init(struct ccid *ccid, struct sock *sk)
591
{
592
	struct ccid2_hc_tx_sock *hctx = ccid_priv(ccid);
593 594
	struct dccp_sock *dp = dccp_sk(sk);
	u32 max_ratio;
595

596
	/* RFC 4341, 5: initialise ssthresh to arbitrarily high (max) value */
597
	hctx->ssthresh = ~0U;
598

599 600
	/* Use larger initial windows (RFC 3390, rfc2581bis) */
	hctx->cwnd = rfc3390_bytes_to_packets(dp->dccps_mss_cache);
601 602

	/* Make sure that Ack Ratio is enabled and within bounds. */
603
	max_ratio = DIV_ROUND_UP(hctx->cwnd, 2);
604 605 606
	if (dp->dccps_l_ack_ratio == 0 || dp->dccps_l_ack_ratio > max_ratio)
		dp->dccps_l_ack_ratio = max_ratio;

607
	/* XXX init ~ to window size... */
G
Gerrit Renker 已提交
608
	if (ccid2_hc_tx_alloc_seq(hctx))
609
		return -ENOMEM;
610

611
	hctx->rto	= DCCP_TIMEOUT_INIT;
612 613 614
	hctx->rpdupack  = -1;
	hctx->last_cong = jiffies;
	setup_timer(&hctx->rtotimer, ccid2_hc_tx_rto_expire, (unsigned long)sk);
615
	INIT_LIST_HEAD(&hctx->av_chunks);
616 617 618 619 620
	return 0;
}

static void ccid2_hc_tx_exit(struct sock *sk)
{
621
	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
622
	int i;
623

624
	sk_stop_timer(sk, &hctx->rtotimer);
625

626 627 628
	for (i = 0; i < hctx->seqbufc; i++)
		kfree(hctx->seqbuf[i]);
	hctx->seqbufc = 0;
629 630 631 632 633 634 635 636 637 638
}

static void ccid2_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
{
	const struct dccp_sock *dp = dccp_sk(sk);
	struct ccid2_hc_rx_sock *hcrx = ccid2_hc_rx_sk(sk);

	switch (DCCP_SKB_CB(skb)->dccpd_type) {
	case DCCP_PKT_DATA:
	case DCCP_PKT_DATAACK:
639 640
		hcrx->data++;
		if (hcrx->data >= dp->dccps_r_ack_ratio) {
641
			dccp_send_ack(sk);
642
			hcrx->data = 0;
643 644 645 646 647
		}
		break;
	}
}

648
static struct ccid_operations ccid2 = {
649 650 651 652 653 654 655 656 657 658 659 660
	.ccid_id		  = DCCPC_CCID2,
	.ccid_name		  = "TCP-like",
	.ccid_owner		  = THIS_MODULE,
	.ccid_hc_tx_obj_size	  = sizeof(struct ccid2_hc_tx_sock),
	.ccid_hc_tx_init	  = ccid2_hc_tx_init,
	.ccid_hc_tx_exit	  = ccid2_hc_tx_exit,
	.ccid_hc_tx_send_packet	  = ccid2_hc_tx_send_packet,
	.ccid_hc_tx_packet_sent	  = ccid2_hc_tx_packet_sent,
	.ccid_hc_tx_parse_options = ccid2_hc_tx_parse_options,
	.ccid_hc_tx_packet_recv	  = ccid2_hc_tx_packet_recv,
	.ccid_hc_rx_obj_size	  = sizeof(struct ccid2_hc_rx_sock),
	.ccid_hc_rx_packet_recv	  = ccid2_hc_rx_packet_recv,
661 662
};

663
#ifdef CONFIG_IP_DCCP_CCID2_DEBUG
664
module_param(ccid2_debug, bool, 0644);
665
MODULE_PARM_DESC(ccid2_debug, "Enable debug messages");
666
#endif
667 668 669 670 671 672 673 674 675 676 677 678 679 680

static __init int ccid2_module_init(void)
{
	return ccid_register(&ccid2);
}
module_init(ccid2_module_init);

static __exit void ccid2_module_exit(void)
{
	ccid_unregister(&ccid2);
}
module_exit(ccid2_module_exit);

MODULE_AUTHOR("Andrea Bittau <a.bittau@cs.ucl.ac.uk>");
681
MODULE_DESCRIPTION("DCCP TCP-Like (CCID2) CCID");
682 683
MODULE_LICENSE("GPL");
MODULE_ALIAS("net-dccp-ccid-2");