output.c 13.4 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 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
/*
 *  net/dccp/output.c
 * 
 *  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/config.h>
#include <linux/dccp.h>
#include <linux/skbuff.h>

#include <net/sock.h>

#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);
}

/*
 * 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.
 */
int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb)
{
	if (likely(skb != NULL)) {
		const struct inet_sock *inet = inet_sk(sk);
		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 */
		const int dccp_header_size = sizeof(*dh) +
					     sizeof(struct dccp_hdr_ext) +
43
					  dccp_packet_hdr_len(dcb->dccpd_type);
44 45 46 47 48 49 50 51 52 53 54 55 56 57
		int err, set_ack = 1;
		u64 ackno = dp->dccps_gsr;

		dccp_inc_seqno(&dp->dccps_gss);

		switch (dcb->dccpd_type) {
		case DCCP_PKT_DATA:
			set_ack = 0;
			break;
		case DCCP_PKT_SYNC:
		case DCCP_PKT_SYNCACK:
			ackno = dcb->dccpd_seq;
			break;
		}
58 59 60

		dcb->dccpd_seq = dp->dccps_gss;
		dccp_insert_options(sk, skb);
61 62 63
		
		skb->h.raw = skb_push(skb, dccp_header_size);
		dh = dccp_hdr(skb);
64 65 66
		/*
		 * Data packets are not cloned as they are never retransmitted
		 */
67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86
		if (skb_cloned(skb))
			skb_set_owner_w(skb, sk);

		/* Build DCCP header and checksum it. */
		memset(dh, 0, dccp_header_size);
		dh->dccph_type	= dcb->dccpd_type;
		dh->dccph_sport	= inet->sport;
		dh->dccph_dport	= inet->dport;
		dh->dccph_doff	= (dccp_header_size + dcb->dccpd_opt_len) / 4;
		dh->dccph_ccval	= dcb->dccpd_ccval;
		/* XXX For now we're using only 48 bits sequence numbers */
		dh->dccph_x	= 1;

		dp->dccps_awh = dp->dccps_gss;
		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:
87 88
			dccp_hdr_request(skb)->dccph_req_service =
							dcb->dccpd_service;
89 90
			break;
		case DCCP_PKT_RESET:
91 92
			dccp_hdr_reset(skb)->dccph_reset_code =
							dcb->dccpd_reset_code;
93 94 95
			break;
		}

96 97
		dh->dccph_checksum = dccp_v4_checksum(skb, inet->saddr,
						      inet->daddr);
98

99
		if (set_ack)
100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
			dccp_event_ack_sent(sk);

		DCCP_INC_STATS(DCCP_MIB_OUTSEGS);

		err = ip_queue_xmit(skb, 0);
		if (err <= 0)
			return err;

		/* NET_XMIT_CN is special. It does not guarantee,
		 * that this packet is lost. It tells that device
		 * is about to start to drop packets or already
		 * drops some packets of the same priority and
		 * invokes us to send less aggressively.
		 */
		return err == NET_XMIT_CN ? 0 : err;
	}
	return -ENOBUFS;
}

unsigned int dccp_sync_mss(struct sock *sk, u32 pmtu)
{
	struct dccp_sock *dp = dccp_sk(sk);
	int mss_now;

	/*
125 126 127 128
	 * FIXME: we really should be using the af_specific thing to support
	 * 	  IPv6.
	 * mss_now = pmtu - tp->af_specific->net_header_len -
	 * 	     sizeof(struct dccp_hdr) - sizeof(struct dccp_hdr_ext);
129
	 */
130 131
	mss_now = pmtu - sizeof(struct iphdr) - sizeof(struct dccp_hdr) -
		  sizeof(struct dccp_hdr_ext);
132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152

	/* Now subtract optional transport overhead */
	mss_now -= dp->dccps_ext_header_len;

	/*
	 * FIXME: this should come from the CCID infrastructure, where, say,
	 * TFRC will say it wants TIMESTAMPS, ELAPSED time, etc, for now lets
	 * put a rough estimate for NDP + TIMESTAMP + TIMESTAMP_ECHO + ELAPSED
	 * TIME + TFRC_OPT_LOSS_EVENT_RATE + TFRC_OPT_RECEIVE_RATE + padding to
	 * make it a multiple of 4
	 */

	mss_now -= ((5 + 6 + 10 + 6 + 6 + 6 + 3) / 4) * 4;

	/* And store cached results */
	dp->dccps_pmtu_cookie = pmtu;
	dp->dccps_mss_cache = mss_now;

	return mss_now;
}

153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205
/**
 * dccp_wait_for_ccid - Wait for ccid to tell us we can send a packet
 * @sk: socket to wait for
 * @timeo: for how long
 */
static int dccp_wait_for_ccid(struct sock *sk, struct sk_buff *skb,
			      long *timeo)
{
	struct dccp_sock *dp = dccp_sk(sk);
	DEFINE_WAIT(wait);
	long delay;
	int rc;

	while (1) {
		prepare_to_wait(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);

		if (sk->sk_err || (sk->sk_shutdown & SEND_SHUTDOWN))
			goto do_error;
		if (!*timeo)
			goto do_nonblock;
		if (signal_pending(current))
			goto do_interrupted;

		rc = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb,
					    skb->len);
		if (rc <= 0)
			break;
		delay = msecs_to_jiffies(rc);
		if (delay > *timeo || delay < 0)
			goto do_nonblock;

		sk->sk_write_pending++;
		release_sock(sk);
		*timeo -= schedule_timeout(delay);
		lock_sock(sk);
		sk->sk_write_pending--;
	}
out:
	finish_wait(sk->sk_sleep, &wait);
	return rc;

do_error:
	rc = -EPIPE;
	goto out;
do_nonblock:
	rc = -EAGAIN;
	goto out;
do_interrupted:
	rc = sock_intr_errno(*timeo);
	goto out;
}

int dccp_write_xmit(struct sock *sk, struct sk_buff *skb, long *timeo)
206 207
{
	const struct dccp_sock *dp = dccp_sk(sk);
208 209 210 211 212
	int err = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb,
					 skb->len);

	if (err > 0)
		err = dccp_wait_for_ccid(sk, skb, timeo);
213 214 215 216

	if (err == 0) {
		const struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts;
		struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
217
		const int len = skb->len;
218 219 220 221 222 223 224 225 226 227 228 229 230

		if (sk->sk_state == DCCP_PARTOPEN) {
			/* See 8.1.5.  Handshake Completion */
			inet_csk_schedule_ack(sk);
			inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
						  inet_csk(sk)->icsk_rto,
						  DCCP_RTO_MAX);
			dcb->dccpd_type = DCCP_PKT_DATAACK;
			/*
			 * FIXME: we really should have a
			 * dccps_ack_pending or use icsk.
			 */
		} else if (inet_csk_ack_scheduled(sk) ||
231
			   dp->dccps_timestamp_echo != 0 ||
232 233 234 235 236 237 238 239 240 241 242 243 244 245
			   (dp->dccps_options.dccpo_send_ack_vector &&
			    ap->dccpap_buf_ackno != DCCP_MAX_SEQNO + 1 &&
			    ap->dccpap_ack_seqno == DCCP_MAX_SEQNO + 1))
			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);
	}

	return err;
}

246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285
int dccp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
{
	if (inet_sk_rebuild_header(sk) != 0)
		return -EHOSTUNREACH; /* Routing failure or similar. */

	return dccp_transmit_skb(sk, (skb_cloned(skb) ?
				      pskb_copy(skb, GFP_ATOMIC):
				      skb_clone(skb, GFP_ATOMIC)));
}

struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst,
				   struct request_sock *req)
{
	struct dccp_hdr *dh;
	const int dccp_header_size = sizeof(struct dccp_hdr) +
				     sizeof(struct dccp_hdr_ext) +
				     sizeof(struct dccp_hdr_response);
	struct sk_buff *skb = sock_wmalloc(sk, MAX_HEADER + DCCP_MAX_OPT_LEN +
					       dccp_header_size, 1,
					   GFP_ATOMIC);
	if (skb == NULL)
		return NULL;

	/* Reserve space for headers. */
	skb_reserve(skb, MAX_HEADER + DCCP_MAX_OPT_LEN + dccp_header_size);

	skb->dst = dst_clone(dst);
	skb->csum = 0;

	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESPONSE;
	DCCP_SKB_CB(skb)->dccpd_seq  = dccp_rsk(req)->dreq_iss;
	dccp_insert_options(sk, skb);

	skb->h.raw = skb_push(skb, dccp_header_size);

	dh = dccp_hdr(skb);
	memset(dh, 0, dccp_header_size);

	dh->dccph_sport	= inet_sk(sk)->sport;
	dh->dccph_dport	= inet_rsk(req)->rmt_port;
286 287
	dh->dccph_doff	= (dccp_header_size +
			   DCCP_SKB_CB(skb)->dccpd_opt_len) / 4;
288 289 290 291 292
	dh->dccph_type	= DCCP_PKT_RESPONSE;
	dh->dccph_x	= 1;
	dccp_hdr_set_seq(dh, dccp_rsk(req)->dreq_iss);
	dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dccp_rsk(req)->dreq_isr);

293 294
	dh->dccph_checksum = dccp_v4_checksum(skb, inet_rsk(req)->loc_addr,
					      inet_rsk(req)->rmt_addr);
295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334

	DCCP_INC_STATS(DCCP_MIB_OUTSEGS);
	return skb;
}

struct sk_buff *dccp_make_reset(struct sock *sk, struct dst_entry *dst,
				const enum dccp_reset_codes code)
				   
{
	struct dccp_hdr *dh;
	struct dccp_sock *dp = dccp_sk(sk);
	const int dccp_header_size = sizeof(struct dccp_hdr) +
				     sizeof(struct dccp_hdr_ext) +
				     sizeof(struct dccp_hdr_reset);
	struct sk_buff *skb = sock_wmalloc(sk, MAX_HEADER + DCCP_MAX_OPT_LEN +
					       dccp_header_size, 1,
					   GFP_ATOMIC);
	if (skb == NULL)
		return NULL;

	/* Reserve space for headers. */
	skb_reserve(skb, MAX_HEADER + DCCP_MAX_OPT_LEN + dccp_header_size);

	skb->dst = dst_clone(dst);
	skb->csum = 0;

	dccp_inc_seqno(&dp->dccps_gss);

	DCCP_SKB_CB(skb)->dccpd_reset_code = code;
	DCCP_SKB_CB(skb)->dccpd_type	   = DCCP_PKT_RESET;
	DCCP_SKB_CB(skb)->dccpd_seq	   = dp->dccps_gss;
	dccp_insert_options(sk, skb);

	skb->h.raw = skb_push(skb, dccp_header_size);

	dh = dccp_hdr(skb);
	memset(dh, 0, dccp_header_size);

	dh->dccph_sport	= inet_sk(sk)->sport;
	dh->dccph_dport	= inet_sk(sk)->dport;
335 336
	dh->dccph_doff	= (dccp_header_size +
			   DCCP_SKB_CB(skb)->dccpd_opt_len) / 4;
337 338 339 340 341 342 343
	dh->dccph_type	= DCCP_PKT_RESET;
	dh->dccph_x	= 1;
	dccp_hdr_set_seq(dh, dp->dccps_gss);
	dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dp->dccps_gsr);

	dccp_hdr_reset(skb)->dccph_reset_code = code;

344 345
	dh->dccph_checksum = dccp_v4_checksum(skb, inet_sk(sk)->saddr,
					      inet_sk(sk)->daddr);
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398

	DCCP_INC_STATS(DCCP_MIB_OUTSEGS);
	return skb;
}

/*
 * Do all connect socket setups that can be done AF independent.
 */
static inline void dccp_connect_init(struct sock *sk)
{
	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);
	
	dccp_sync_mss(sk, dst_mtu(dst));

	/*
	 * FIXME: set dp->{dccps_swh,dccps_swl}, with
	 * something like dccp_inc_seq
	 */

	icsk->icsk_retransmits = 0;
}

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

	dccp_connect_init(sk);

	skb = alloc_skb(MAX_DCCP_HEADER + 15, sk->sk_allocation);
	if (unlikely(skb == NULL))
		return -ENOBUFS;

	/* Reserve space for headers. */
	skb_reserve(skb, MAX_DCCP_HEADER);

	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_REQUEST;
	/* FIXME: set service to something meaningful, coming
	 * from userspace*/
	DCCP_SKB_CB(skb)->dccpd_service = 0;
	skb->csum = 0;
	skb_set_owner_w(skb, sk);

	BUG_TRAP(sk->sk_send_head == NULL);
	sk->sk_send_head = skb;
	dccp_transmit_skb(sk, skb_clone(skb, GFP_KERNEL));
	DCCP_INC_STATS(DCCP_MIB_ACTIVEOPENS);

	/* Timer for repeating the REQUEST until an answer. */
399 400
	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
				  icsk->icsk_rto, DCCP_RTO_MAX);
401 402 403 404 405 406 407 408 409 410 411 412
	return 0;
}

void dccp_send_ack(struct sock *sk)
{
	/* If we have been reset, we may not send again. */
	if (sk->sk_state != DCCP_CLOSED) {
		struct sk_buff *skb = alloc_skb(MAX_DCCP_HEADER, GFP_ATOMIC);

		if (skb == NULL) {
			inet_csk_schedule_ack(sk);
			inet_csk(sk)->icsk_ack.ato = TCP_ATO_MIN;
413 414 415
			inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
						  TCP_DELACK_MAX,
						  DCCP_RTO_MAX);
416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459
			return;
		}

		/* Reserve space for headers */
		skb_reserve(skb, MAX_DCCP_HEADER);
		skb->csum = 0;
		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_ACK;
		skb_set_owner_w(skb, sk);
		dccp_transmit_skb(sk, skb);
	}
}

EXPORT_SYMBOL_GPL(dccp_send_ack);

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);
}

460 461
void dccp_send_sync(struct sock *sk, const u64 seq,
		    const enum dccp_pkt_type pkt_type)
462 463 464 465 466 467 468 469 470 471 472 473 474 475 476
{
	/*
	 * We are not putting this on the write queue, so
	 * dccp_transmit_skb() will set the ownership to this
	 * sock.
	 */
	struct sk_buff *skb = alloc_skb(MAX_DCCP_HEADER, GFP_ATOMIC);

	if (skb == NULL)
		/* FIXME: how to make sure the sync is sent? */
		return;

	/* Reserve space for headers and prepare control bits. */
	skb_reserve(skb, MAX_DCCP_HEADER);
	skb->csum = 0;
477
	DCCP_SKB_CB(skb)->dccpd_type = pkt_type;
478 479 480 481 482 483
	DCCP_SKB_CB(skb)->dccpd_seq = seq;

	skb_set_owner_w(skb, sk);
	dccp_transmit_skb(sk, skb);
}

484 485 486 487
/*
 * 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.
488
 */
489
void dccp_send_close(struct sock *sk, const int active)
490 491 492
{
	struct dccp_sock *dp = dccp_sk(sk);
	struct sk_buff *skb;
493
	const unsigned int prio = active ? GFP_KERNEL : GFP_ATOMIC;
494

495 496 497
	skb = alloc_skb(sk->sk_prot->max_header, prio);
	if (skb == NULL)
		return;
498 499 500 501

	/* Reserve space for headers and prepare control bits. */
	skb_reserve(skb, sk->sk_prot->max_header);
	skb->csum = 0;
502 503
	DCCP_SKB_CB(skb)->dccpd_type = dp->dccps_role == DCCP_ROLE_CLIENT ?
					DCCP_PKT_CLOSE : DCCP_PKT_CLOSEREQ;
504 505

	skb_set_owner_w(skb, sk);
506 507 508 509 510 511
	if (active) {
		BUG_TRAP(sk->sk_send_head == NULL);
		sk->sk_send_head = skb;
		dccp_transmit_skb(sk, skb_clone(skb, prio));
	} else
		dccp_transmit_skb(sk, skb);
512 513 514 515

	ccid_hc_rx_exit(dp->dccps_hc_rx_ccid, sk);
	ccid_hc_tx_exit(dp->dccps_hc_tx_ccid, sk);
}