output.c 20.2 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 31
/* enqueue @skb on sk_send_head for retransmission, return clone to send now */
static struct sk_buff *dccp_skb_entail(struct sock *sk, struct sk_buff *skb)
32 33 34 35
{
	skb_set_owner_w(skb, sk);
	WARN_ON(sk->sk_send_head);
	sk->sk_send_head = skb;
36
	return skb_clone(sk->sk_send_head, gfp_any());
37 38
}

39 40 41 42 43 44
/*
 * 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.
 */
45
static int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb)
46 47
{
	if (likely(skb != NULL)) {
48
		struct inet_sock *inet = inet_sk(sk);
49
		const struct inet_connection_sock *icsk = inet_csk(sk);
50 51 52 53
		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 */
54
		const u32 dccp_header_size = sizeof(*dh) +
55
					     sizeof(struct dccp_hdr_ext) +
56
					  dccp_packet_hdr_len(dcb->dccpd_type);
57 58
		int err, set_ack = 1;
		u64 ackno = dp->dccps_gsr;
59 60 61 62 63
		/*
		 * 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);
64 65 66 67

		switch (dcb->dccpd_type) {
		case DCCP_PKT_DATA:
			set_ack = 0;
68 69
			/* fall through */
		case DCCP_PKT_DATAACK:
70
		case DCCP_PKT_RESET:
71
			break;
72

73 74
		case DCCP_PKT_REQUEST:
			set_ack = 0;
75 76 77
			/* Use ISS on the first (non-retransmitted) Request. */
			if (icsk->icsk_retransmits == 0)
				dcb->dccpd_seq = dp->dccps_iss;
78 79
			/* fall through */

80 81
		case DCCP_PKT_SYNC:
		case DCCP_PKT_SYNCACK:
G
Gerrit Renker 已提交
82
			ackno = dcb->dccpd_ack_seq;
83 84 85
			/* fall through */
		default:
			/*
86 87 88 89
			 * 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.
90 91 92
			 */
			WARN_ON(skb->sk);
			skb_set_owner_w(skb, sk);
93 94
			break;
		}
95

96 97 98 99
		if (dccp_insert_options(sk, skb)) {
			kfree_skb(skb);
			return -EPROTO;
		}
100

101

102
		/* Build DCCP header and checksum it. */
103
		dh = dccp_zeroed_hdr(skb, dccp_header_size);
104
		dh->dccph_type	= dcb->dccpd_type;
E
Eric Dumazet 已提交
105 106
		dh->dccph_sport	= inet->inet_sport;
		dh->dccph_dport	= inet->inet_dport;
107 108
		dh->dccph_doff	= (dccp_header_size + dcb->dccpd_opt_len) / 4;
		dh->dccph_ccval	= dcb->dccpd_ccval;
109
		dh->dccph_cscov = dp->dccps_pcslen;
110 111 112
		/* XXX For now we're using only 48 bits sequence numbers */
		dh->dccph_x	= 1;

113
		dccp_update_gss(sk, dcb->dccpd_seq);
114 115 116 117 118 119
		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:
120
			dccp_hdr_request(skb)->dccph_req_service =
121
							dp->dccps_service;
122 123 124 125 126
			/*
			 * Limit Ack window to ISS <= P.ackno <= GSS, so that
			 * only Responses to Requests we sent are considered.
			 */
			dp->dccps_awl = dp->dccps_iss;
127 128
			break;
		case DCCP_PKT_RESET:
129 130
			dccp_hdr_reset(skb)->dccph_reset_code =
							dcb->dccpd_reset_code;
131 132 133
			break;
		}

134
		icsk->icsk_af_ops->send_check(sk, skb);
135

136
		if (set_ack)
137 138 139 140
			dccp_event_ack_sent(sk);

		DCCP_INC_STATS(DCCP_MIB_OUTSEGS);

141
		err = icsk->icsk_af_ops->queue_xmit(sk, skb, &inet->cork.fl);
142
		return net_xmit_eval(err);
143 144 145 146
	}
	return -ENOBUFS;
}

G
Gerrit Renker 已提交
147
/**
L
Lucas De Marchi 已提交
148
 * dccp_determine_ccmps  -  Find out about CCID-specific packet-size limits
G
Gerrit Renker 已提交
149 150 151 152 153 154 155 156 157 158 159 160 161
 * 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;
}

162 163
unsigned int dccp_sync_mss(struct sock *sk, u32 pmtu)
{
164
	struct inet_connection_sock *icsk = inet_csk(sk);
165
	struct dccp_sock *dp = dccp_sk(sk);
G
Gerrit Renker 已提交
166
	u32 ccmps = dccp_determine_ccmps(dp);
167
	u32 cur_mps = ccmps ? min(pmtu, ccmps) : pmtu;
168

G
Gerrit Renker 已提交
169 170 171
	/* 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));
172 173

	/*
174 175 176 177 178 179 180 181 182 183 184
	 * 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)
185
	 */
186 187
	cur_mps -= roundup(1 + 6 + 10 + dp->dccps_send_ndp_count * 8 + 6 +
			   (dp->dccps_hc_rx_ackvec ? DCCPAV_MIN_OPTLEN : 0), 4);
188 189

	/* And store cached results */
190
	icsk->icsk_pmtu_cookie = pmtu;
G
Gerrit Renker 已提交
191
	dp->dccps_mss_cache = cur_mps;
192

G
Gerrit Renker 已提交
193
	return cur_mps;
194 195
}

196 197
EXPORT_SYMBOL_GPL(dccp_sync_mss);

198 199
void dccp_write_space(struct sock *sk)
{
200
	struct socket_wq *wq;
201

202 203 204 205
	rcu_read_lock();
	wq = rcu_dereference(sk->sk_wq);
	if (wq_has_sleeper(wq))
		wake_up_interruptible(&wq->wait);
206 207
	/* Should agree with poll, otherwise some programs break */
	if (sock_writeable(sk))
208
		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
209

210
	rcu_read_unlock();
211 212
}

213
/**
214
 * dccp_wait_for_ccid  -  Await CCID send permission
G
Gerrit Renker 已提交
215
 * @sk:    socket to wait for
216
 * @delay: timeout in jiffies
217
 *
218
 * This is used by CCIDs which need to delay the send time in process context.
219
 */
220
static int dccp_wait_for_ccid(struct sock *sk, unsigned long delay)
221 222
{
	DEFINE_WAIT(wait);
223
	long remaining;
224

225 226 227
	prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
	sk->sk_write_pending++;
	release_sock(sk);
G
Gerrit Renker 已提交
228

229
	remaining = schedule_timeout(delay);
230

231 232
	lock_sock(sk);
	sk->sk_write_pending--;
E
Eric Dumazet 已提交
233
	finish_wait(sk_sleep(sk), &wait);
234 235 236 237

	if (signal_pending(current) || sk->sk_err)
		return -1;
	return remaining;
238 239
}

240 241 242 243 244 245 246 247
/**
 * dccp_xmit_packet  -  Send data packet under control of CCID
 * Transmits next-queued payload and informs CCID to account for the packet.
 */
static void dccp_xmit_packet(struct sock *sk)
{
	int err, len;
	struct dccp_sock *dp = dccp_sk(sk);
248
	struct sk_buff *skb = dccp_qpolicy_pop(sk);
249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288

	if (unlikely(skb == NULL))
		return;
	len = skb->len;

	if (sk->sk_state == DCCP_PARTOPEN) {
		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);
		}

		inet_csk_schedule_ack(sk);
		inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
					      inet_csk(sk)->icsk_rto,
					      DCCP_RTO_MAX);
		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_DATAACK;
	} else if (dccp_ack_pending(sk)) {
		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_DATAACK;
	} else {
		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_DATA;
	}

	err = dccp_transmit_skb(sk, skb);
	if (err)
		dccp_pr_debug("transmit_skb() returned err=%d\n", err);
	/*
	 * Register this one as sent even if an error occurred. To the remote
	 * end a local packet drop is indistinguishable from network loss, i.e.
	 * any local drop will eventually be reported via receiver feedback.
	 */
	ccid_hc_tx_packet_sent(dp->dccps_hc_tx_ccid, sk, len);
289 290 291 292 293 294 295 296 297

	/*
	 * If the CCID needs to transfer additional header options out-of-band
	 * (e.g. Ack Vectors or feature-negotiation options), it activates this
	 * flag to schedule a Sync. The Sync will automatically incorporate all
	 * currently pending header options, thus clearing the backlog.
	 */
	if (dp->dccps_sync_scheduled)
		dccp_send_sync(sk, dp->dccps_gsr, DCCP_PKT_SYNC);
298 299
}

300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346
/**
 * dccp_flush_write_queue  -  Drain queue at end of connection
 * Since dccp_sendmsg queues packets without waiting for them to be sent, it may
 * happen that the TX queue is not empty at the end of a connection. We give the
 * HC-sender CCID a grace period of up to @time_budget jiffies. If this function
 * returns with a non-empty write queue, it will be purged later.
 */
void dccp_flush_write_queue(struct sock *sk, long *time_budget)
{
	struct dccp_sock *dp = dccp_sk(sk);
	struct sk_buff *skb;
	long delay, rc;

	while (*time_budget > 0 && (skb = skb_peek(&sk->sk_write_queue))) {
		rc = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb);

		switch (ccid_packet_dequeue_eval(rc)) {
		case CCID_PACKET_WILL_DEQUEUE_LATER:
			/*
			 * If the CCID determines when to send, the next sending
			 * time is unknown or the CCID may not even send again
			 * (e.g. remote host crashes or lost Ack packets).
			 */
			DCCP_WARN("CCID did not manage to send all packets\n");
			return;
		case CCID_PACKET_DELAY:
			delay = msecs_to_jiffies(rc);
			if (delay > *time_budget)
				return;
			rc = dccp_wait_for_ccid(sk, delay);
			if (rc < 0)
				return;
			*time_budget -= (delay - rc);
			/* check again if we can send now */
			break;
		case CCID_PACKET_SEND_AT_ONCE:
			dccp_xmit_packet(sk);
			break;
		case CCID_PACKET_ERR:
			skb_dequeue(&sk->sk_write_queue);
			kfree_skb(skb);
			dccp_pr_debug("packet discarded due to err=%ld\n", rc);
		}
	}
}

void dccp_write_xmit(struct sock *sk)
347
{
I
Ian McDonald 已提交
348 349 350
	struct dccp_sock *dp = dccp_sk(sk);
	struct sk_buff *skb;

351
	while ((skb = dccp_qpolicy_top(sk))) {
352
		int rc = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb);
353

354 355 356 357
		switch (ccid_packet_dequeue_eval(rc)) {
		case CCID_PACKET_WILL_DEQUEUE_LATER:
			return;
		case CCID_PACKET_DELAY:
358 359 360
			sk_reset_timer(sk, &dp->dccps_xmit_timer,
				       jiffies + msecs_to_jiffies(rc));
			return;
361 362 363 364
		case CCID_PACKET_SEND_AT_ONCE:
			dccp_xmit_packet(sk);
			break;
		case CCID_PACKET_ERR:
365
			dccp_qpolicy_drop(sk, skb);
366
			dccp_pr_debug("packet discarded due to err=%d\n", rc);
367
		}
I
Ian McDonald 已提交
368
	}
369 370
}

371 372 373 374 375 376 377 378 379 380
/**
 * 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)
381
{
382 383
	WARN_ON(sk->sk_send_head == NULL);

384
	if (inet_csk(sk)->icsk_af_ops->rebuild_header(sk) != 0)
385 386
		return -EHOSTUNREACH; /* Routing failure or similar. */

387 388 389 390
	/* 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));
391 392 393 394 395 396
}

struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst,
				   struct request_sock *req)
{
	struct dccp_hdr *dh;
397
	struct dccp_request_sock *dreq;
398
	const u32 dccp_header_size = sizeof(struct dccp_hdr) +
399 400
				     sizeof(struct dccp_hdr_ext) +
				     sizeof(struct dccp_hdr_response);
401
	struct sk_buff *skb = sock_wmalloc(sk, sk->sk_prot->max_header, 1,
402 403 404 405 406
					   GFP_ATOMIC);
	if (skb == NULL)
		return NULL;

	/* Reserve space for headers. */
407
	skb_reserve(skb, sk->sk_prot->max_header);
408

E
Eric Dumazet 已提交
409
	skb_dst_set(skb, dst_clone(dst));
410

411
	dreq = dccp_rsk(req);
412 413
	if (inet_rsk(req)->acked)	/* increase GSS upon retransmission */
		dccp_inc_seqno(&dreq->dreq_gss);
414
	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESPONSE;
415
	DCCP_SKB_CB(skb)->dccpd_seq  = dreq->dreq_gss;
416

417 418 419 420 421 422
	/* 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;
423

424
	/* Build and checksum header */
425
	dh = dccp_zeroed_hdr(skb, dccp_header_size);
426

E
Eric Dumazet 已提交
427
	dh->dccph_sport	= htons(inet_rsk(req)->ir_num);
428
	dh->dccph_dport	= inet_rsk(req)->ir_rmt_port;
429 430
	dh->dccph_doff	= (dccp_header_size +
			   DCCP_SKB_CB(skb)->dccpd_opt_len) / 4;
431 432
	dh->dccph_type	= DCCP_PKT_RESPONSE;
	dh->dccph_x	= 1;
433 434
	dccp_hdr_set_seq(dh, dreq->dreq_gss);
	dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dreq->dreq_gsr);
435
	dccp_hdr_response(skb)->dccph_resp_service = dreq->dreq_service;
436

437 438
	dccp_csum_outgoing(skb);

439 440
	/* We use `acked' to remember that a Response was already sent. */
	inet_rsk(req)->acked = 1;
441 442
	DCCP_INC_STATS(DCCP_MIB_OUTSEGS);
	return skb;
443 444 445
response_failed:
	kfree_skb(skb);
	return NULL;
446 447
}

448 449
EXPORT_SYMBOL_GPL(dccp_make_response);

450
/* answer offending packet in @rcv_skb with Reset from control socket @ctl */
451
struct sk_buff *dccp_ctl_make_reset(struct sock *sk, struct sk_buff *rcv_skb)
452 453 454 455 456 457 458 459 460
{
	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;

461
	skb = alloc_skb(sk->sk_prot->max_header, GFP_ATOMIC);
462 463 464
	if (skb == NULL)
		return NULL;

465
	skb_reserve(skb, sk->sk_prot->max_header);
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501

	/* 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);

502
/* send Reset on established socket, to close or abort the connection */
503 504
int dccp_send_reset(struct sock *sk, enum dccp_reset_codes code)
{
505
	struct sk_buff *skb;
506 507 508 509
	/*
	 * FIXME: what if rebuild_header fails?
	 * Should we be doing a rebuild_header here?
	 */
510
	int err = inet_csk(sk)->icsk_af_ops->rebuild_header(sk);
511

512 513 514 515 516 517 518 519 520 521 522
	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;
523

524
	return dccp_transmit_skb(sk, skb);
525 526
}

527 528 529
/*
 * Do all connect socket setups that can be done AF independent.
 */
530
int dccp_connect(struct sock *sk)
531
{
532
	struct sk_buff *skb;
533
	struct dccp_sock *dp = dccp_sk(sk);
534 535 536 537 538
	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);
539

540 541
	dccp_sync_mss(sk, dst_mtu(dst));

542 543 544 545
	/* do not connect if feature negotiation setup fails */
	if (dccp_feat_finalise_settings(dccp_sk(sk)))
		return -EPROTO;

546 547
	/* Initialise GAR as per 8.5; AWL/AWH are set in dccp_transmit_skb() */
	dp->dccps_gar = dp->dccps_iss;
548

549
	skb = alloc_skb(sk->sk_prot->max_header, sk->sk_allocation);
550 551 552 553
	if (unlikely(skb == NULL))
		return -ENOBUFS;

	/* Reserve space for headers. */
554
	skb_reserve(skb, sk->sk_prot->max_header);
555 556 557

	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_REQUEST;

558
	dccp_transmit_skb(sk, dccp_skb_entail(sk, skb));
559 560 561
	DCCP_INC_STATS(DCCP_MIB_ACTIVEOPENS);

	/* Timer for repeating the REQUEST until an answer. */
562
	icsk->icsk_retransmits = 0;
563 564
	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
				  icsk->icsk_rto, DCCP_RTO_MAX);
565 566 567
	return 0;
}

568 569
EXPORT_SYMBOL_GPL(dccp_connect);

570 571 572 573
void dccp_send_ack(struct sock *sk)
{
	/* If we have been reset, we may not send again. */
	if (sk->sk_state != DCCP_CLOSED) {
574 575
		struct sk_buff *skb = alloc_skb(sk->sk_prot->max_header,
						GFP_ATOMIC);
576 577 578 579

		if (skb == NULL) {
			inet_csk_schedule_ack(sk);
			inet_csk(sk)->icsk_ack.ato = TCP_ATO_MIN;
580 581 582
			inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
						  TCP_DELACK_MAX,
						  DCCP_RTO_MAX);
583 584 585 586
			return;
		}

		/* Reserve space for headers */
587
		skb_reserve(skb, sk->sk_prot->max_header);
588 589 590 591 592 593 594
		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
595
#if 0
596
/* FIXME: Is this still necessary (11.3) - currently nowhere used by DCCP. */
597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625
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
626
#endif
627

G
Gerrit Renker 已提交
628
void dccp_send_sync(struct sock *sk, const u64 ackno,
629
		    const enum dccp_pkt_type pkt_type)
630 631 632 633 634 635
{
	/*
	 * We are not putting this on the write queue, so
	 * dccp_transmit_skb() will set the ownership to this
	 * sock.
	 */
636
	struct sk_buff *skb = alloc_skb(sk->sk_prot->max_header, GFP_ATOMIC);
637

G
Gerrit Renker 已提交
638
	if (skb == NULL) {
639
		/* FIXME: how to make sure the sync is sent? */
G
Gerrit Renker 已提交
640
		DCCP_CRIT("could not send %s", dccp_packet_name(pkt_type));
641
		return;
G
Gerrit Renker 已提交
642
	}
643 644

	/* Reserve space for headers and prepare control bits. */
645
	skb_reserve(skb, sk->sk_prot->max_header);
646
	DCCP_SKB_CB(skb)->dccpd_type = pkt_type;
G
Gerrit Renker 已提交
647
	DCCP_SKB_CB(skb)->dccpd_ack_seq = ackno;
648

649 650 651 652 653 654
	/*
	 * Clear the flag in case the Sync was scheduled for out-of-band data,
	 * such as carrying a long Ack Vector.
	 */
	dccp_sk(sk)->dccps_sync_scheduled = 0;

655 656 657
	dccp_transmit_skb(sk, skb);
}

658 659
EXPORT_SYMBOL_GPL(dccp_send_sync);

660 661 662 663
/*
 * 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.
664
 */
665
void dccp_send_close(struct sock *sk, const int active)
666 667 668
{
	struct dccp_sock *dp = dccp_sk(sk);
	struct sk_buff *skb;
A
Al Viro 已提交
669
	const gfp_t prio = active ? GFP_KERNEL : GFP_ATOMIC;
670

671 672 673
	skb = alloc_skb(sk->sk_prot->max_header, prio);
	if (skb == NULL)
		return;
674 675 676

	/* Reserve space for headers and prepare control bits. */
	skb_reserve(skb, sk->sk_prot->max_header);
677 678 679 680
	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;
681

682
	if (active) {
683
		skb = dccp_skb_entail(sk, skb);
684 685 686 687 688 689 690 691 692 693 694 695
		/*
		 * 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);
696 697
	}
	dccp_transmit_skb(sk, skb);
698
}