ackvec.c 13.9 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 *  net/dccp/ackvec.c
 *
 *  An implementation of the DCCP protocol
 *  Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
 *
 *      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; version 2 of the License;
 */

#include "ackvec.h"
#include "dccp.h"

#include <linux/dccp.h>
16 17 18
#include <linux/init.h>
#include <linux/errno.h>
#include <linux/kernel.h>
19
#include <linux/skbuff.h>
20
#include <linux/slab.h>
21 22 23

#include <net/sock.h>

24
static kmem_cache_t *dccp_ackvec_slab;
25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
static kmem_cache_t *dccp_ackvec_record_slab;

static struct dccp_ackvec_record *dccp_ackvec_record_new(void)
{
	struct dccp_ackvec_record *avr =
			kmem_cache_alloc(dccp_ackvec_record_slab, GFP_ATOMIC);

	if (avr != NULL)
		INIT_LIST_HEAD(&avr->dccpavr_node);

	return avr;
}

static void dccp_ackvec_record_delete(struct dccp_ackvec_record *avr)
{
	if (unlikely(avr == NULL))
		return;
	/* Check if deleting a linked record */
	WARN_ON(!list_empty(&avr->dccpavr_node));
	kmem_cache_free(dccp_ackvec_record_slab, avr);
}

static void dccp_ackvec_insert_avr(struct dccp_ackvec *av,
				   struct dccp_ackvec_record *avr)
{
	/*
	 * AVRs are sorted by seqno. Since we are sending them in order, we
	 * just add the AVR at the head of the list.
	 * -sorbo.
	 */
	if (!list_empty(&av->dccpav_records)) {
		const struct dccp_ackvec_record *head =
					list_entry(av->dccpav_records.next,
						   struct dccp_ackvec_record,
						   dccpavr_node);
		BUG_ON(before48(avr->dccpavr_ack_seqno,
				head->dccpavr_ack_seqno));
	}

	list_add(&avr->dccpavr_node, &av->dccpav_records);
}
66

67 68 69
int dccp_insert_option_ackvec(struct sock *sk, struct sk_buff *skb)
{
	struct dccp_sock *dp = dccp_sk(sk);
70 71 72 73
#ifdef CONFIG_IP_DCCP_DEBUG
	const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ?
				"CLIENT tx: " : "server tx: ";
#endif
74 75 76 77 78
	struct dccp_ackvec *av = dp->dccps_hc_rx_ackvec;
	int len = av->dccpav_vec_len + 2;
	struct timeval now;
	u32 elapsed_time;
	unsigned char *to, *from;
79 80 81 82 83
	struct dccp_ackvec_record *avr;

	if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN)
		return -1;

84 85 86
	dccp_timestamp(sk, &now);
	elapsed_time = timeval_delta(&now, &av->dccpav_time) / 10;

87 88 89 90 91 92 93
	if (elapsed_time != 0 &&
	    dccp_insert_option_elapsed_time(sk, skb, elapsed_time))
		return -1;

	avr = dccp_ackvec_record_new();
	if (avr == NULL)
		return -1;
94 95 96 97 98 99 100 101 102 103 104

	DCCP_SKB_CB(skb)->dccpd_opt_len += len;

	to    = skb_push(skb, len);
	*to++ = DCCPO_ACK_VECTOR_0;
	*to++ = len;

	len  = av->dccpav_vec_len;
	from = av->dccpav_buf + av->dccpav_buf_head;

	/* Check if buf_head wraps */
105 106
	if ((int)av->dccpav_buf_head + len > DCCP_MAX_ACKVEC_LEN) {
		const u32 tailsize = DCCP_MAX_ACKVEC_LEN - av->dccpav_buf_head;
107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123

		memcpy(to, from, tailsize);
		to   += tailsize;
		len  -= tailsize;
		from = av->dccpav_buf;
	}

	memcpy(to, from, len);
	/*
	 *	From draft-ietf-dccp-spec-11.txt:
	 *
	 *	For each acknowledgement it sends, the HC-Receiver will add an
	 *	acknowledgement record.  ack_seqno will equal the HC-Receiver
	 *	sequence number it used for the ack packet; ack_ptr will equal
	 *	buf_head; ack_ackno will equal buf_ackno; and ack_nonce will
	 *	equal buf_nonce.
	 */
124 125 126 127 128 129 130
	avr->dccpavr_ack_seqno = DCCP_SKB_CB(skb)->dccpd_seq;
	avr->dccpavr_ack_ptr   = av->dccpav_buf_head;
	avr->dccpavr_ack_ackno = av->dccpav_buf_ackno;
	avr->dccpavr_ack_nonce = av->dccpav_buf_nonce;
	avr->dccpavr_sent_len  = av->dccpav_vec_len;

	dccp_ackvec_insert_avr(av, avr);
131 132 133

	dccp_pr_debug("%sACK Vector 0, len=%d, ack_seqno=%llu, "
		      "ack_ackno=%llu\n",
134 135 136 137
		      debug_prefix, avr->dccpavr_sent_len,
		      (unsigned long long)avr->dccpavr_ack_seqno,
		      (unsigned long long)avr->dccpavr_ack_ackno);
	return 0;
138 139
}

140
struct dccp_ackvec *dccp_ackvec_alloc(const gfp_t priority)
141
{
142
	struct dccp_ackvec *av = kmem_cache_alloc(dccp_ackvec_slab, priority);
143

144
	if (av != NULL) {
145
		av->dccpav_buf_head	= DCCP_MAX_ACKVEC_LEN - 1;
146
		av->dccpav_buf_ackno	= DCCP_MAX_SEQNO + 1;
147 148 149 150
		av->dccpav_buf_nonce = av->dccpav_buf_nonce = 0;
		av->dccpav_ack_ptr	= 0;
		av->dccpav_time.tv_sec	= 0;
		av->dccpav_time.tv_usec	= 0;
151
		av->dccpav_vec_len	= 0;
152
		INIT_LIST_HEAD(&av->dccpav_records);
153 154 155 156 157 158 159
	}

	return av;
}

void dccp_ackvec_free(struct dccp_ackvec *av)
{
160 161
	if (unlikely(av == NULL))
		return;
162 163 164 165 166 167 168 169 170 171 172

	if (!list_empty(&av->dccpav_records)) {
		struct dccp_ackvec_record *avr, *next;

		list_for_each_entry_safe(avr, next, &av->dccpav_records,
					 dccpavr_node) {
			list_del_init(&avr->dccpavr_node);
			dccp_ackvec_record_delete(avr);
		}
	}

173
	kmem_cache_free(dccp_ackvec_slab, av);
174 175 176
}

static inline u8 dccp_ackvec_state(const struct dccp_ackvec *av,
177
				   const u8 index)
178 179 180 181 182
{
	return av->dccpav_buf[index] & DCCP_ACKVEC_STATE_MASK;
}

static inline u8 dccp_ackvec_len(const struct dccp_ackvec *av,
183
				 const u8 index)
184 185 186 187 188 189 190 191 192 193 194
{
	return av->dccpav_buf[index] & DCCP_ACKVEC_LEN_MASK;
}

/*
 * If several packets are missing, the HC-Receiver may prefer to enter multiple
 * bytes with run length 0, rather than a single byte with a larger run length;
 * this simplifies table updates if one of the missing packets arrives.
 */
static inline int dccp_ackvec_set_buf_head_state(struct dccp_ackvec *av,
						 const unsigned int packets,
195
						 const unsigned char state)
196 197
{
	unsigned int gap;
K
Kris Katterjohn 已提交
198
	long new_head;
199

200
	if (av->dccpav_vec_len + packets > DCCP_MAX_ACKVEC_LEN)
201 202 203 204 205 206 207 208 209 210 211
		return -ENOBUFS;

	gap	 = packets - 1;
	new_head = av->dccpav_buf_head - packets;

	if (new_head < 0) {
		if (gap > 0) {
			memset(av->dccpav_buf, DCCP_ACKVEC_STATE_NOT_RECEIVED,
			       gap + new_head + 1);
			gap = -new_head;
		}
212
		new_head += DCCP_MAX_ACKVEC_LEN;
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282
	} 

	av->dccpav_buf_head = new_head;

	if (gap > 0)
		memset(av->dccpav_buf + av->dccpav_buf_head + 1,
		       DCCP_ACKVEC_STATE_NOT_RECEIVED, gap);

	av->dccpav_buf[av->dccpav_buf_head] = state;
	av->dccpav_vec_len += packets;
	return 0;
}

/*
 * Implements the draft-ietf-dccp-spec-11.txt Appendix A
 */
int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk,
		    const u64 ackno, const u8 state)
{
	/*
	 * Check at the right places if the buffer is full, if it is, tell the
	 * caller to start dropping packets till the HC-Sender acks our ACK
	 * vectors, when we will free up space in dccpav_buf.
	 *
	 * We may well decide to do buffer compression, etc, but for now lets
	 * just drop.
	 *
	 * From Appendix A:
	 *
	 *	Of course, the circular buffer may overflow, either when the
	 *	HC-Sender is sending data at a very high rate, when the
	 *	HC-Receiver's acknowledgements are not reaching the HC-Sender,
	 *	or when the HC-Sender is forgetting to acknowledge those acks
	 *	(so the HC-Receiver is unable to clean up old state). In this
	 *	case, the HC-Receiver should either compress the buffer (by
	 *	increasing run lengths when possible), transfer its state to
	 *	a larger buffer, or, as a last resort, drop all received
	 *	packets, without processing them whatsoever, until its buffer
	 *	shrinks again.
	 */

	/* See if this is the first ackno being inserted */
	if (av->dccpav_vec_len == 0) {
		av->dccpav_buf[av->dccpav_buf_head] = state;
		av->dccpav_vec_len = 1;
	} else if (after48(ackno, av->dccpav_buf_ackno)) {
		const u64 delta = dccp_delta_seqno(av->dccpav_buf_ackno,
						   ackno);

		/*
		 * Look if the state of this packet is the same as the
		 * previous ackno and if so if we can bump the head len.
		 */
		if (delta == 1 &&
		    dccp_ackvec_state(av, av->dccpav_buf_head) == state &&
		    (dccp_ackvec_len(av, av->dccpav_buf_head) <
		     DCCP_ACKVEC_LEN_MASK))
			av->dccpav_buf[av->dccpav_buf_head]++;
		else if (dccp_ackvec_set_buf_head_state(av, delta, state))
			return -ENOBUFS;
	} else {
		/*
		 * A.1.2.  Old Packets
		 *
		 *	When a packet with Sequence Number S arrives, and
		 *	S <= buf_ackno, the HC-Receiver will scan the table
		 *	for the byte corresponding to S. (Indexing structures
		 *	could reduce the complexity of this scan.)
		 */
		u64 delta = dccp_delta_seqno(ackno, av->dccpav_buf_ackno);
283
		u8 index = av->dccpav_buf_head;
284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304

		while (1) {
			const u8 len = dccp_ackvec_len(av, index);
			const u8 state = dccp_ackvec_state(av, index);
			/*
			 * valid packets not yet in dccpav_buf have a reserved
			 * entry, with a len equal to 0.
			 */
			if (state == DCCP_ACKVEC_STATE_NOT_RECEIVED &&
			    len == 0 && delta == 0) { /* Found our
							 reserved seat! */
				dccp_pr_debug("Found %llu reserved seat!\n",
					      (unsigned long long)ackno);
				av->dccpav_buf[index] = state;
				goto out;
			}
			/* len == 0 means one packet */
			if (delta < len + 1)
				goto out_duplicate;

			delta -= len + 1;
305
			if (++index == DCCP_MAX_ACKVEC_LEN)
306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349
				index = 0;
		}
	}

	av->dccpav_buf_ackno = ackno;
	dccp_timestamp(sk, &av->dccpav_time);
out:
	return 0;

out_duplicate:
	/* Duplicate packet */
	dccp_pr_debug("Received a dup or already considered lost "
		      "packet: %llu\n", (unsigned long long)ackno);
	return -EILSEQ;
}

#ifdef CONFIG_IP_DCCP_DEBUG
void dccp_ackvector_print(const u64 ackno, const unsigned char *vector, int len)
{
	if (!dccp_debug)
		return;

	printk("ACK vector len=%d, ackno=%llu |", len,
	       (unsigned long long)ackno);

	while (len--) {
		const u8 state = (*vector & DCCP_ACKVEC_STATE_MASK) >> 6;
		const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;

		printk("%d,%d|", state, rl);
		++vector;
	}

	printk("\n");
}

void dccp_ackvec_print(const struct dccp_ackvec *av)
{
	dccp_ackvector_print(av->dccpav_buf_ackno,
			     av->dccpav_buf + av->dccpav_buf_head,
			     av->dccpav_vec_len);
}
#endif

350 351
static void dccp_ackvec_throw_record(struct dccp_ackvec *av,
				     struct dccp_ackvec_record *avr)
352
{
353 354
	struct dccp_ackvec_record *next;

355 356 357 358 359 360 361
	/* sort out vector length */
	if (av->dccpav_buf_head <= avr->dccpavr_ack_ptr)
		av->dccpav_vec_len = avr->dccpavr_ack_ptr - av->dccpav_buf_head;
	else
		av->dccpav_vec_len = DCCP_MAX_ACKVEC_LEN - 1
				     - av->dccpav_buf_head
				     + avr->dccpavr_ack_ptr;
362 363 364 365 366 367 368

	/* free records */
	list_for_each_entry_safe_from(avr, next, &av->dccpav_records,
				      dccpavr_node) {
		list_del_init(&avr->dccpavr_node);
		dccp_ackvec_record_delete(avr);
	}
369 370 371 372 373
}

void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av, struct sock *sk,
				 const u64 ackno)
{
374
	struct dccp_ackvec_record *avr;
375

376 377 378 379 380 381 382
	/*
	 * If we traverse backwards, it should be faster when we have large
	 * windows. We will be receiving ACKs for stuff we sent a while back
	 * -sorbo.
	 */
	list_for_each_entry_reverse(avr, &av->dccpav_records, dccpavr_node) {
		if (ackno == avr->dccpavr_ack_seqno) {
383
#ifdef CONFIG_IP_DCCP_DEBUG
384 385 386
			struct dccp_sock *dp = dccp_sk(sk);
			const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ?
						"CLIENT rx ack: " : "server rx ack: ";
387
#endif
388 389 390 391 392 393 394 395
			dccp_pr_debug("%sACK packet 0, len=%d, ack_seqno=%llu, "
				      "ack_ackno=%llu, ACKED!\n",
				      debug_prefix, 1,
				      (unsigned long long)avr->dccpavr_ack_seqno,
				      (unsigned long long)avr->dccpavr_ack_ackno);
			dccp_ackvec_throw_record(av, avr);
			break;
		}
396 397 398 399 400 401 402 403 404
	}
}

static void dccp_ackvec_check_rcv_ackvector(struct dccp_ackvec *av,
					    struct sock *sk, u64 ackno,
					    const unsigned char len,
					    const unsigned char *vector)
{
	unsigned char i;
405
	struct dccp_ackvec_record *avr;
406 407

	/* Check if we actually sent an ACK vector */
408
	if (list_empty(&av->dccpav_records))
409 410 411
		return;

	i = len;
412 413 414 415 416 417 418
	/*
	 * XXX
	 * I think it might be more efficient to work backwards. See comment on
	 * rcv_ackno. -sorbo.
	 */
	avr = list_entry(av->dccpav_records.next, struct dccp_ackvec_record,
			 dccpavr_node);
419 420 421 422 423 424 425
	while (i--) {
		const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;
		u64 ackno_end_rl;

		dccp_set_seqno(&ackno_end_rl, ackno - rl);

		/*
426 427
		 * If our AVR sequence number is greater than the ack, go
		 * forward in the AVR list until it is not so.
428
		 */
429 430 431 432 433 434 435 436 437
		list_for_each_entry_from(avr, &av->dccpav_records,
					 dccpavr_node) {
			if (!after48(avr->dccpavr_ack_seqno, ackno))
				goto found;
		}
		/* End of the dccpav_records list, not found, exit */
		break;
found:
		if (between48(avr->dccpavr_ack_seqno, ackno_end_rl, ackno)) {
438
			const u8 state = *vector & DCCP_ACKVEC_STATE_MASK;
439 440 441 442 443 444 445 446 447 448 449 450
			if (state != DCCP_ACKVEC_STATE_NOT_RECEIVED) {
#ifdef CONFIG_IP_DCCP_DEBUG
				struct dccp_sock *dp = dccp_sk(sk);
				const char *debug_prefix =
					dp->dccps_role == DCCP_ROLE_CLIENT ?
					"CLIENT rx ack: " : "server rx ack: ";
#endif
				dccp_pr_debug("%sACK vector 0, len=%d, "
					      "ack_seqno=%llu, ack_ackno=%llu, "
					      "ACKED!\n",
					      debug_prefix, len,
					      (unsigned long long)
451
					      avr->dccpavr_ack_seqno,
452
					      (unsigned long long)
453 454
					      avr->dccpavr_ack_ackno);
				dccp_ackvec_throw_record(av, avr);
455
				break;
456 457
			}
			/*
458 459
			 * If it wasn't received, continue scanning... we might
			 * find another one.
460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479
			 */
		}

		dccp_set_seqno(&ackno, ackno_end_rl - 1);
		++vector;
	}
}

int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb,
		      const u8 opt, const u8 *value, const u8 len)
{
	if (len > DCCP_MAX_ACKVEC_LEN)
		return -1;

	/* dccp_ackvector_print(DCCP_SKB_CB(skb)->dccpd_ack_seq, value, len); */
	dccp_ackvec_check_rcv_ackvector(dccp_sk(sk)->dccps_hc_rx_ackvec, sk,
					DCCP_SKB_CB(skb)->dccpd_ack_seq,
				        len, value);
	return 0;
}
480 481

static char dccp_ackvec_slab_msg[] __initdata =
482
	KERN_CRIT "DCCP: Unable to create ack vectors slab caches\n";
483 484 485 486 487 488

int __init dccp_ackvec_init(void)
{
	dccp_ackvec_slab = kmem_cache_create("dccp_ackvec",
					     sizeof(struct dccp_ackvec), 0,
					     SLAB_HWCACHE_ALIGN, NULL, NULL);
489 490 491 492 493 494 495 496 497
	if (dccp_ackvec_slab == NULL)
		goto out_err;

	dccp_ackvec_record_slab =
			kmem_cache_create("dccp_ackvec_record",
					  sizeof(struct dccp_ackvec_record),
					  0, SLAB_HWCACHE_ALIGN, NULL, NULL);
	if (dccp_ackvec_record_slab == NULL)
		goto out_destroy_slab;
498 499

	return 0;
500 501 502 503 504 505 506

out_destroy_slab:
	kmem_cache_destroy(dccp_ackvec_slab);
	dccp_ackvec_slab = NULL;
out_err:
	printk(dccp_ackvec_slab_msg);
	return -ENOBUFS;
507 508 509 510 511 512 513 514
}

void dccp_ackvec_exit(void)
{
	if (dccp_ackvec_slab != NULL) {
		kmem_cache_destroy(dccp_ackvec_slab);
		dccp_ackvec_slab = NULL;
	}
515 516 517 518
	if (dccp_ackvec_record_slab != NULL) {
		kmem_cache_destroy(dccp_ackvec_record_slab);
		dccp_ackvec_record_slab = NULL;
	}
519
}