output.c 18.3 KB
Newer Older
1 2
/*
 *  net/dccp/output.c
3
 *
4 5 6 7 8 9 10 11 12 13
 *  An implementation of the DCCP protocol
 *  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.
 */

#include <linux/dccp.h>
14
#include <linux/kernel.h>
15
#include <linux/skbuff.h>
16
#include <linux/slab.h>
17

18
#include <net/inet_sock.h>
19 20
#include <net/sock.h>

21
#include "ackvec.h"
22 23 24 25 26 27 28 29
#include "ccid.h"
#include "dccp.h"

static inline void dccp_event_ack_sent(struct sock *sk)
{
	inet_csk_clear_xmit_timer(sk, ICSK_TIME_DACK);
}

30
static void dccp_skb_entail(struct sock *sk, struct sk_buff *skb)
31 32 33 34 35 36
{
	skb_set_owner_w(skb, sk);
	WARN_ON(sk->sk_send_head);
	sk->sk_send_head = skb;
}

37 38 39 40 41 42
/*
 * All SKB's seen here are completely headerless. It is our
 * job to build the DCCP header, and pass the packet down to
 * IP so it can do the same plus pass the packet off to the
 * device.
 */
43
static int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb)
44 45 46
{
	if (likely(skb != NULL)) {
		const struct inet_sock *inet = inet_sk(sk);
47
		const struct inet_connection_sock *icsk = inet_csk(sk);
48 49 50 51
		struct dccp_sock *dp = dccp_sk(sk);
		struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
		struct dccp_hdr *dh;
		/* XXX For now we're using only 48 bits sequence numbers */
52
		const u32 dccp_header_size = sizeof(*dh) +
53
					     sizeof(struct dccp_hdr_ext) +
54
					  dccp_packet_hdr_len(dcb->dccpd_type);
55 56
		int err, set_ack = 1;
		u64 ackno = dp->dccps_gsr;
57 58 59 60 61
		/*
		 * Increment GSS here already in case the option code needs it.
		 * Update GSS for real only if option processing below succeeds.
		 */
		dcb->dccpd_seq = ADD48(dp->dccps_gss, 1);
62 63 64 65

		switch (dcb->dccpd_type) {
		case DCCP_PKT_DATA:
			set_ack = 0;
66 67
			/* fall through */
		case DCCP_PKT_DATAACK:
68
		case DCCP_PKT_RESET:
69
			break;
70

71 72
		case DCCP_PKT_REQUEST:
			set_ack = 0;
73 74 75
			/* Use ISS on the first (non-retransmitted) Request. */
			if (icsk->icsk_retransmits == 0)
				dcb->dccpd_seq = dp->dccps_iss;
76 77
			/* fall through */

78 79
		case DCCP_PKT_SYNC:
		case DCCP_PKT_SYNCACK:
G
Gerrit Renker 已提交
80
			ackno = dcb->dccpd_ack_seq;
81 82 83
			/* fall through */
		default:
			/*
84 85 86 87
			 * Set owner/destructor: some skbs are allocated via
			 * alloc_skb (e.g. when retransmission may happen).
			 * Only Data, DataAck, and Reset packets should come
			 * through here with skb->sk set.
88 89 90
			 */
			WARN_ON(skb->sk);
			skb_set_owner_w(skb, sk);
91 92
			break;
		}
93

94 95 96 97
		if (dccp_insert_options(sk, skb)) {
			kfree_skb(skb);
			return -EPROTO;
		}
98

99

100
		/* Build DCCP header and checksum it. */
101
		dh = dccp_zeroed_hdr(skb, dccp_header_size);
102
		dh->dccph_type	= dcb->dccpd_type;
E
Eric Dumazet 已提交
103 104
		dh->dccph_sport	= inet->inet_sport;
		dh->dccph_dport	= inet->inet_dport;
105 106
		dh->dccph_doff	= (dccp_header_size + dcb->dccpd_opt_len) / 4;
		dh->dccph_ccval	= dcb->dccpd_ccval;
107
		dh->dccph_cscov = dp->dccps_pcslen;
108 109 110
		/* XXX For now we're using only 48 bits sequence numbers */
		dh->dccph_x	= 1;

111
		dccp_update_gss(sk, dcb->dccpd_seq);
112 113 114 115 116 117
		dccp_hdr_set_seq(dh, dp->dccps_gss);
		if (set_ack)
			dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), ackno);

		switch (dcb->dccpd_type) {
		case DCCP_PKT_REQUEST:
118
			dccp_hdr_request(skb)->dccph_req_service =
119
							dp->dccps_service;
120 121 122 123 124
			/*
			 * Limit Ack window to ISS <= P.ackno <= GSS, so that
			 * only Responses to Requests we sent are considered.
			 */
			dp->dccps_awl = dp->dccps_iss;
125 126
			break;
		case DCCP_PKT_RESET:
127 128
			dccp_hdr_reset(skb)->dccph_reset_code =
							dcb->dccpd_reset_code;
129 130 131
			break;
		}

132
		icsk->icsk_af_ops->send_check(sk, skb);
133

134
		if (set_ack)
135 136 137 138
			dccp_event_ack_sent(sk);

		DCCP_INC_STATS(DCCP_MIB_OUTSEGS);

139
		err = icsk->icsk_af_ops->queue_xmit(skb);
140
		return net_xmit_eval(err);
141 142 143 144
	}
	return -ENOBUFS;
}

G
Gerrit Renker 已提交
145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
/**
 * dccp_determine_ccmps  -  Find out about CCID-specfic packet-size limits
 * We only consider the HC-sender CCID for setting the CCMPS (RFC 4340, 14.),
 * since the RX CCID is restricted to feedback packets (Acks), which are small
 * in comparison with the data traffic. A value of 0 means "no current CCMPS".
 */
static u32 dccp_determine_ccmps(const struct dccp_sock *dp)
{
	const struct ccid *tx_ccid = dp->dccps_hc_tx_ccid;

	if (tx_ccid == NULL || tx_ccid->ccid_ops == NULL)
		return 0;
	return tx_ccid->ccid_ops->ccid_ccmps;
}

160 161
unsigned int dccp_sync_mss(struct sock *sk, u32 pmtu)
{
162
	struct inet_connection_sock *icsk = inet_csk(sk);
163
	struct dccp_sock *dp = dccp_sk(sk);
G
Gerrit Renker 已提交
164
	u32 ccmps = dccp_determine_ccmps(dp);
165
	u32 cur_mps = ccmps ? min(pmtu, ccmps) : pmtu;
166

G
Gerrit Renker 已提交
167 168 169
	/* Account for header lengths and IPv4/v6 option overhead */
	cur_mps -= (icsk->icsk_af_ops->net_header_len + icsk->icsk_ext_hdr_len +
		    sizeof(struct dccp_hdr) + sizeof(struct dccp_hdr_ext));
170 171

	/*
172 173 174 175 176 177 178 179 180 181 182
	 * Leave enough headroom for common DCCP header options.
	 * This only considers options which may appear on DCCP-Data packets, as
	 * per table 3 in RFC 4340, 5.8. When running out of space for other
	 * options (eg. Ack Vector which can take up to 255 bytes), it is better
	 * to schedule a separate Ack. Thus we leave headroom for the following:
	 *  - 1 byte for Slow Receiver (11.6)
	 *  - 6 bytes for Timestamp (13.1)
	 *  - 10 bytes for Timestamp Echo (13.3)
	 *  - 8 bytes for NDP count (7.7, when activated)
	 *  - 6 bytes for Data Checksum (9.3)
	 *  - %DCCPAV_MIN_OPTLEN bytes for Ack Vector size (11.4, when enabled)
183
	 */
184 185
	cur_mps -= roundup(1 + 6 + 10 + dp->dccps_send_ndp_count * 8 + 6 +
			   (dp->dccps_hc_rx_ackvec ? DCCPAV_MIN_OPTLEN : 0), 4);
186 187

	/* And store cached results */
188
	icsk->icsk_pmtu_cookie = pmtu;
G
Gerrit Renker 已提交
189
	dp->dccps_mss_cache = cur_mps;
190

G
Gerrit Renker 已提交
191
	return cur_mps;
192 193
}

194 195
EXPORT_SYMBOL_GPL(dccp_sync_mss);

196 197 198 199
void dccp_write_space(struct sock *sk)
{
	read_lock(&sk->sk_callback_lock);

200
	if (sk_has_sleeper(sk))
E
Eric Dumazet 已提交
201
		wake_up_interruptible(sk_sleep(sk));
202 203
	/* Should agree with poll, otherwise some programs break */
	if (sock_writeable(sk))
204
		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
205 206 207 208

	read_unlock(&sk->sk_callback_lock);
}

209 210
/**
 * dccp_wait_for_ccid - Wait for ccid to tell us we can send a packet
G
Gerrit Renker 已提交
211 212 213 214 215
 * @sk:    socket to wait for
 * @skb:   current skb to pass on for waiting
 * @delay: sleep timeout in milliseconds (> 0)
 * This function is called by default when the socket is closed, and
 * when a non-zero linger time is set on the socket. For consistency
216
 */
G
Gerrit Renker 已提交
217
static int dccp_wait_for_ccid(struct sock *sk, struct sk_buff *skb, int delay)
218 219 220
{
	struct dccp_sock *dp = dccp_sk(sk);
	DEFINE_WAIT(wait);
G
Gerrit Renker 已提交
221
	unsigned long jiffdelay;
222 223
	int rc;

G
Gerrit Renker 已提交
224 225 226 227
	do {
		dccp_pr_debug("delayed send by %d msec\n", delay);
		jiffdelay = msecs_to_jiffies(delay);

E
Eric Dumazet 已提交
228
		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
229

G
Gerrit Renker 已提交
230 231 232 233 234 235
		sk->sk_write_pending++;
		release_sock(sk);
		schedule_timeout(jiffdelay);
		lock_sock(sk);
		sk->sk_write_pending--;

I
Ian McDonald 已提交
236
		if (sk->sk_err)
237 238 239 240
			goto do_error;
		if (signal_pending(current))
			goto do_interrupted;

241
		rc = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb);
G
Gerrit Renker 已提交
242
	} while ((delay = rc) > 0);
243
out:
E
Eric Dumazet 已提交
244
	finish_wait(sk_sleep(sk), &wait);
245 246 247 248 249 250
	return rc;

do_error:
	rc = -EPIPE;
	goto out;
do_interrupted:
I
Ian McDonald 已提交
251
	rc = -EINTR;
252 253 254
	goto out;
}

I
Ian McDonald 已提交
255
void dccp_write_xmit(struct sock *sk, int block)
256
{
I
Ian McDonald 已提交
257 258 259 260
	struct dccp_sock *dp = dccp_sk(sk);
	struct sk_buff *skb;

	while ((skb = skb_peek(&sk->sk_write_queue))) {
261
		int err = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb);
262

I
Ian McDonald 已提交
263 264 265 266 267
		if (err > 0) {
			if (!block) {
				sk_reset_timer(sk, &dp->dccps_xmit_timer,
						msecs_to_jiffies(err)+jiffies);
				break;
I
Ian McDonald 已提交
268
			} else
G
Gerrit Renker 已提交
269
				err = dccp_wait_for_ccid(sk, skb, err);
270
			if (err && err != -EINTR)
271
				DCCP_BUG("err=%d after dccp_wait_for_ccid", err);
I
Ian McDonald 已提交
272
		}
273

I
Ian McDonald 已提交
274 275 276 277
		skb_dequeue(&sk->sk_write_queue);
		if (err == 0) {
			struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
			const int len = skb->len;
278

I
Ian McDonald 已提交
279
			if (sk->sk_state == DCCP_PARTOPEN) {
280 281 282 283 284 285 286 287 288 289 290 291 292 293
				const u32 cur_mps = dp->dccps_mss_cache - DCCP_FEATNEG_OVERHEAD;
				/*
				 * See 8.1.5 - Handshake Completion.
				 *
				 * For robustness we resend Confirm options until the client has
				 * entered OPEN. During the initial feature negotiation, the MPS
				 * is smaller than usual, reduced by the Change/Confirm options.
				 */
				if (!list_empty(&dp->dccps_featneg) && len > cur_mps) {
					DCCP_WARN("Payload too large (%d) for featneg.\n", len);
					dccp_send_ack(sk);
					dccp_feat_list_purge(&dp->dccps_featneg);
				}

I
Ian McDonald 已提交
294 295
				inet_csk_schedule_ack(sk);
				inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
296 297
						  inet_csk(sk)->icsk_rto,
						  DCCP_RTO_MAX);
I
Ian McDonald 已提交
298 299 300 301 302 303 304 305
				dcb->dccpd_type = DCCP_PKT_DATAACK;
			} else if (dccp_ack_pending(sk))
				dcb->dccpd_type = DCCP_PKT_DATAACK;
			else
				dcb->dccpd_type = DCCP_PKT_DATA;

			err = dccp_transmit_skb(sk, skb);
			ccid_hc_tx_packet_sent(dp->dccps_hc_tx_ccid, sk, 0, len);
306 307 308
			if (err)
				DCCP_BUG("err=%d after ccid_hc_tx_packet_sent",
					 err);
309
		} else {
310
			dccp_pr_debug("packet discarded due to err=%d\n", err);
P
Patrick McHardy 已提交
311
			kfree_skb(skb);
312
		}
I
Ian McDonald 已提交
313
	}
314 315
}

316 317 318 319 320 321 322 323 324 325
/**
 * dccp_retransmit_skb  -  Retransmit Request, Close, or CloseReq packets
 * There are only four retransmittable packet types in DCCP:
 * - Request  in client-REQUEST  state (sec. 8.1.1),
 * - CloseReq in server-CLOSEREQ state (sec. 8.3),
 * - Close    in   node-CLOSING  state (sec. 8.3),
 * - Acks in client-PARTOPEN state (sec. 8.1.5, handled by dccp_delack_timer()).
 * This function expects sk->sk_send_head to contain the original skb.
 */
int dccp_retransmit_skb(struct sock *sk)
326
{
327 328
	WARN_ON(sk->sk_send_head == NULL);

329
	if (inet_csk(sk)->icsk_af_ops->rebuild_header(sk) != 0)
330 331
		return -EHOSTUNREACH; /* Routing failure or similar. */

332 333 334 335
	/* this count is used to distinguish original and retransmitted skb */
	inet_csk(sk)->icsk_retransmits++;

	return dccp_transmit_skb(sk, skb_clone(sk->sk_send_head, GFP_ATOMIC));
336 337 338 339 340 341
}

struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst,
				   struct request_sock *req)
{
	struct dccp_hdr *dh;
342
	struct dccp_request_sock *dreq;
343
	const u32 dccp_header_size = sizeof(struct dccp_hdr) +
344 345
				     sizeof(struct dccp_hdr_ext) +
				     sizeof(struct dccp_hdr_response);
346
	struct sk_buff *skb = sock_wmalloc(sk, sk->sk_prot->max_header, 1,
347 348 349 350 351
					   GFP_ATOMIC);
	if (skb == NULL)
		return NULL;

	/* Reserve space for headers. */
352
	skb_reserve(skb, sk->sk_prot->max_header);
353

E
Eric Dumazet 已提交
354
	skb_dst_set(skb, dst_clone(dst));
355

356
	dreq = dccp_rsk(req);
357 358
	if (inet_rsk(req)->acked)	/* increase ISS upon retransmission */
		dccp_inc_seqno(&dreq->dreq_iss);
359
	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESPONSE;
360
	DCCP_SKB_CB(skb)->dccpd_seq  = dreq->dreq_iss;
361

362 363 364 365 366 367
	/* Resolve feature dependencies resulting from choice of CCID */
	if (dccp_feat_server_ccid_dependencies(dreq))
		goto response_failed;

	if (dccp_insert_options_rsk(dreq, skb))
		goto response_failed;
368

369
	/* Build and checksum header */
370
	dh = dccp_zeroed_hdr(skb, dccp_header_size);
371

372
	dh->dccph_sport	= inet_rsk(req)->loc_port;
373
	dh->dccph_dport	= inet_rsk(req)->rmt_port;
374 375
	dh->dccph_doff	= (dccp_header_size +
			   DCCP_SKB_CB(skb)->dccpd_opt_len) / 4;
376 377
	dh->dccph_type	= DCCP_PKT_RESPONSE;
	dh->dccph_x	= 1;
378 379 380
	dccp_hdr_set_seq(dh, dreq->dreq_iss);
	dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dreq->dreq_isr);
	dccp_hdr_response(skb)->dccph_resp_service = dreq->dreq_service;
381

382 383
	dccp_csum_outgoing(skb);

384 385
	/* We use `acked' to remember that a Response was already sent. */
	inet_rsk(req)->acked = 1;
386 387
	DCCP_INC_STATS(DCCP_MIB_OUTSEGS);
	return skb;
388 389 390
response_failed:
	kfree_skb(skb);
	return NULL;
391 392
}

393 394
EXPORT_SYMBOL_GPL(dccp_make_response);

395
/* answer offending packet in @rcv_skb with Reset from control socket @ctl */
396
struct sk_buff *dccp_ctl_make_reset(struct sock *sk, struct sk_buff *rcv_skb)
397 398 399 400 401 402 403 404 405
{
	struct dccp_hdr *rxdh = dccp_hdr(rcv_skb), *dh;
	struct dccp_skb_cb *dcb = DCCP_SKB_CB(rcv_skb);
	const u32 dccp_hdr_reset_len = sizeof(struct dccp_hdr) +
				       sizeof(struct dccp_hdr_ext) +
				       sizeof(struct dccp_hdr_reset);
	struct dccp_hdr_reset *dhr;
	struct sk_buff *skb;

406
	skb = alloc_skb(sk->sk_prot->max_header, GFP_ATOMIC);
407 408 409
	if (skb == NULL)
		return NULL;

410
	skb_reserve(skb, sk->sk_prot->max_header);
411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446

	/* Swap the send and the receive. */
	dh = dccp_zeroed_hdr(skb, dccp_hdr_reset_len);
	dh->dccph_type	= DCCP_PKT_RESET;
	dh->dccph_sport	= rxdh->dccph_dport;
	dh->dccph_dport	= rxdh->dccph_sport;
	dh->dccph_doff	= dccp_hdr_reset_len / 4;
	dh->dccph_x	= 1;

	dhr = dccp_hdr_reset(skb);
	dhr->dccph_reset_code = dcb->dccpd_reset_code;

	switch (dcb->dccpd_reset_code) {
	case DCCP_RESET_CODE_PACKET_ERROR:
		dhr->dccph_reset_data[0] = rxdh->dccph_type;
		break;
	case DCCP_RESET_CODE_OPTION_ERROR:	/* fall through */
	case DCCP_RESET_CODE_MANDATORY_ERROR:
		memcpy(dhr->dccph_reset_data, dcb->dccpd_reset_data, 3);
		break;
	}
	/*
	 * From RFC 4340, 8.3.1:
	 *   If P.ackno exists, set R.seqno := P.ackno + 1.
	 *   Else set R.seqno := 0.
	 */
	if (dcb->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ)
		dccp_hdr_set_seq(dh, ADD48(dcb->dccpd_ack_seq, 1));
	dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dcb->dccpd_seq);

	dccp_csum_outgoing(skb);
	return skb;
}

EXPORT_SYMBOL_GPL(dccp_ctl_make_reset);

447
/* send Reset on established socket, to close or abort the connection */
448 449
int dccp_send_reset(struct sock *sk, enum dccp_reset_codes code)
{
450
	struct sk_buff *skb;
451 452 453 454
	/*
	 * FIXME: what if rebuild_header fails?
	 * Should we be doing a rebuild_header here?
	 */
455
	int err = inet_csk(sk)->icsk_af_ops->rebuild_header(sk);
456

457 458 459 460 461 462 463 464 465 466 467
	if (err != 0)
		return err;

	skb = sock_wmalloc(sk, sk->sk_prot->max_header, 1, GFP_ATOMIC);
	if (skb == NULL)
		return -ENOBUFS;

	/* Reserve space for headers and prepare control bits. */
	skb_reserve(skb, sk->sk_prot->max_header);
	DCCP_SKB_CB(skb)->dccpd_type	   = DCCP_PKT_RESET;
	DCCP_SKB_CB(skb)->dccpd_reset_code = code;
468

469
	return dccp_transmit_skb(sk, skb);
470 471
}

472 473 474 475 476
/*
 * Do all connect socket setups that can be done AF independent.
 */
static inline void dccp_connect_init(struct sock *sk)
{
477
	struct dccp_sock *dp = dccp_sk(sk);
478 479 480 481 482
	struct dst_entry *dst = __sk_dst_get(sk);
	struct inet_connection_sock *icsk = inet_csk(sk);

	sk->sk_err = 0;
	sock_reset_flag(sk, SOCK_DONE);
483

484 485
	dccp_sync_mss(sk, dst_mtu(dst));

486
	/* Initialise GAR as per 8.5; AWL/AWH are set in dccp_transmit_skb() */
487 488
	dp->dccps_gar = dp->dccps_iss;

489 490 491 492 493 494 495 496
	icsk->icsk_retransmits = 0;
}

int dccp_connect(struct sock *sk)
{
	struct sk_buff *skb;
	struct inet_connection_sock *icsk = inet_csk(sk);

497 498 499 500
	/* do not connect if feature negotiation setup fails */
	if (dccp_feat_finalise_settings(dccp_sk(sk)))
		return -EPROTO;

501 502
	dccp_connect_init(sk);

503
	skb = alloc_skb(sk->sk_prot->max_header, sk->sk_allocation);
504 505 506 507
	if (unlikely(skb == NULL))
		return -ENOBUFS;

	/* Reserve space for headers. */
508
	skb_reserve(skb, sk->sk_prot->max_header);
509 510 511

	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_REQUEST;

512
	dccp_skb_entail(sk, skb);
513 514 515 516
	dccp_transmit_skb(sk, skb_clone(skb, GFP_KERNEL));
	DCCP_INC_STATS(DCCP_MIB_ACTIVEOPENS);

	/* Timer for repeating the REQUEST until an answer. */
517 518
	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
				  icsk->icsk_rto, DCCP_RTO_MAX);
519 520 521
	return 0;
}

522 523
EXPORT_SYMBOL_GPL(dccp_connect);

524 525 526 527
void dccp_send_ack(struct sock *sk)
{
	/* If we have been reset, we may not send again. */
	if (sk->sk_state != DCCP_CLOSED) {
528 529
		struct sk_buff *skb = alloc_skb(sk->sk_prot->max_header,
						GFP_ATOMIC);
530 531 532 533

		if (skb == NULL) {
			inet_csk_schedule_ack(sk);
			inet_csk(sk)->icsk_ack.ato = TCP_ATO_MIN;
534 535 536
			inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
						  TCP_DELACK_MAX,
						  DCCP_RTO_MAX);
537 538 539 540
			return;
		}

		/* Reserve space for headers */
541
		skb_reserve(skb, sk->sk_prot->max_header);
542 543 544 545 546 547 548
		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_ACK;
		dccp_transmit_skb(sk, skb);
	}
}

EXPORT_SYMBOL_GPL(dccp_send_ack);

G
Gerrit Renker 已提交
549
#if 0
550
/* FIXME: Is this still necessary (11.3) - currently nowhere used by DCCP. */
551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579
void dccp_send_delayed_ack(struct sock *sk)
{
	struct inet_connection_sock *icsk = inet_csk(sk);
	/*
	 * FIXME: tune this timer. elapsed time fixes the skew, so no problem
	 * with using 2s, and active senders also piggyback the ACK into a
	 * DATAACK packet, so this is really for quiescent senders.
	 */
	unsigned long timeout = jiffies + 2 * HZ;

	/* Use new timeout only if there wasn't a older one earlier. */
	if (icsk->icsk_ack.pending & ICSK_ACK_TIMER) {
		/* If delack timer was blocked or is about to expire,
		 * send ACK now.
		 *
		 * FIXME: check the "about to expire" part
		 */
		if (icsk->icsk_ack.blocked) {
			dccp_send_ack(sk);
			return;
		}

		if (!time_before(timeout, icsk->icsk_ack.timeout))
			timeout = icsk->icsk_ack.timeout;
	}
	icsk->icsk_ack.pending |= ICSK_ACK_SCHED | ICSK_ACK_TIMER;
	icsk->icsk_ack.timeout = timeout;
	sk_reset_timer(sk, &icsk->icsk_delack_timer, timeout);
}
G
Gerrit Renker 已提交
580
#endif
581

G
Gerrit Renker 已提交
582
void dccp_send_sync(struct sock *sk, const u64 ackno,
583
		    const enum dccp_pkt_type pkt_type)
584 585 586 587 588 589
{
	/*
	 * We are not putting this on the write queue, so
	 * dccp_transmit_skb() will set the ownership to this
	 * sock.
	 */
590
	struct sk_buff *skb = alloc_skb(sk->sk_prot->max_header, GFP_ATOMIC);
591

G
Gerrit Renker 已提交
592
	if (skb == NULL) {
593
		/* FIXME: how to make sure the sync is sent? */
G
Gerrit Renker 已提交
594
		DCCP_CRIT("could not send %s", dccp_packet_name(pkt_type));
595
		return;
G
Gerrit Renker 已提交
596
	}
597 598

	/* Reserve space for headers and prepare control bits. */
599
	skb_reserve(skb, sk->sk_prot->max_header);
600
	DCCP_SKB_CB(skb)->dccpd_type = pkt_type;
G
Gerrit Renker 已提交
601
	DCCP_SKB_CB(skb)->dccpd_ack_seq = ackno;
602 603 604 605

	dccp_transmit_skb(sk, skb);
}

606 607
EXPORT_SYMBOL_GPL(dccp_send_sync);

608 609 610 611
/*
 * Send a DCCP_PKT_CLOSE/CLOSEREQ. The caller locks the socket for us. This
 * cannot be allowed to fail queueing a DCCP_PKT_CLOSE/CLOSEREQ frame under
 * any circumstances.
612
 */
613
void dccp_send_close(struct sock *sk, const int active)
614 615 616
{
	struct dccp_sock *dp = dccp_sk(sk);
	struct sk_buff *skb;
A
Al Viro 已提交
617
	const gfp_t prio = active ? GFP_KERNEL : GFP_ATOMIC;
618

619 620 621
	skb = alloc_skb(sk->sk_prot->max_header, prio);
	if (skb == NULL)
		return;
622 623 624

	/* Reserve space for headers and prepare control bits. */
	skb_reserve(skb, sk->sk_prot->max_header);
625 626 627 628
	if (dp->dccps_role == DCCP_ROLE_SERVER && !dp->dccps_server_timewait)
		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_CLOSEREQ;
	else
		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_CLOSE;
629

630
	if (active) {
I
Ian McDonald 已提交
631
		dccp_write_xmit(sk, 1);
632
		dccp_skb_entail(sk, skb);
633
		dccp_transmit_skb(sk, skb_clone(skb, prio));
634 635 636 637 638 639 640 641 642 643 644 645
		/*
		 * Retransmission timer for active-close: RFC 4340, 8.3 requires
		 * to retransmit the Close/CloseReq until the CLOSING/CLOSEREQ
		 * state can be left. The initial timeout is 2 RTTs.
		 * Since RTT measurement is done by the CCIDs, there is no easy
		 * way to get an RTT sample. The fallback RTT from RFC 4340, 3.4
		 * is too low (200ms); we use a high value to avoid unnecessary
		 * retransmissions when the link RTT is > 0.2 seconds.
		 * FIXME: Let main module sample RTTs and use that instead.
		 */
		inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
					  DCCP_TIMEOUT_INIT, DCCP_RTO_MAX);
646 647
	} else
		dccp_transmit_skb(sk, skb);
648
}