tcp.h 15.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Definitions for the TCP protocol.
 *
 * Version:	@(#)tcp.h	1.0.2	04/28/93
 *
 * Author:	Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
 *
 *		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.
 */
#ifndef _LINUX_TCP_H
#define _LINUX_TCP_H


#include <linux/skbuff.h>
22
#include <linux/win_minmax.h>
L
Linus Torvalds 已提交
23
#include <net/sock.h>
24
#include <net/inet_connection_sock.h>
25
#include <net/inet_timewait_sock.h>
26
#include <uapi/linux/tcp.h>
L
Linus Torvalds 已提交
27

28 29
static inline struct tcphdr *tcp_hdr(const struct sk_buff *skb)
{
30
	return (struct tcphdr *)skb_transport_header(skb);
31 32
}

C
Craig Gallek 已提交
33 34 35 36 37
static inline unsigned int __tcp_hdrlen(const struct tcphdr *th)
{
	return th->doff * 4;
}

38 39
static inline unsigned int tcp_hdrlen(const struct sk_buff *skb)
{
C
Craig Gallek 已提交
40
	return __tcp_hdrlen(tcp_hdr(skb));
41 42
}

43 44 45 46 47 48 49 50 51 52
static inline struct tcphdr *inner_tcp_hdr(const struct sk_buff *skb)
{
	return (struct tcphdr *)skb_inner_transport_header(skb);
}

static inline unsigned int inner_tcp_hdrlen(const struct sk_buff *skb)
{
	return inner_tcp_hdr(skb)->doff * 4;
}

53 54
static inline unsigned int tcp_optlen(const struct sk_buff *skb)
{
55
	return (tcp_hdr(skb)->doff - 5) * 4;
56 57
}

Y
Yuchung Cheng 已提交
58 59 60
/* TCP Fast Open */
#define TCP_FASTOPEN_COOKIE_MIN	4	/* Min Fast Open Cookie size in bytes */
#define TCP_FASTOPEN_COOKIE_MAX	16	/* Max Fast Open Cookie size in bytes */
61
#define TCP_FASTOPEN_COOKIE_SIZE 8	/* the size employed by this impl. */
Y
Yuchung Cheng 已提交
62 63 64

/* TCP Fast Open Cookie as stored in memory */
struct tcp_fastopen_cookie {
65 66 67 68 69 70
	union {
		u8	val[TCP_FASTOPEN_COOKIE_MAX];
#if IS_ENABLED(CONFIG_IPV6)
		struct in6_addr addr;
#endif
	};
Y
Yuchung Cheng 已提交
71
	s8	len;
72
	bool	exp;	/* In RFC6994 experimental option format */
Y
Yuchung Cheng 已提交
73 74
};

L
Linus Torvalds 已提交
75
/* This defines a selective acknowledgement block. */
76 77 78 79 80
struct tcp_sack_block_wire {
	__be32	start_seq;
	__be32	end_seq;
};

L
Linus Torvalds 已提交
81
struct tcp_sack_block {
82 83
	u32	start_seq;
	u32	end_seq;
L
Linus Torvalds 已提交
84 85
};

86 87 88 89
/*These are used to set the sack_ok field in struct tcp_options_received */
#define TCP_SACK_SEEN     (1 << 0)   /*1 = peer is SACK capable, */
#define TCP_DSACK_SEEN    (1 << 2)   /*1 = DSACK was received from peer*/

L
Linus Torvalds 已提交
90 91 92
struct tcp_options_received {
/*	PAWS/RTTM data	*/
	long	ts_recent_stamp;/* Time we stored ts_recent (for aging) */
93 94 95 96
	u32	ts_recent;	/* Time stamp to echo next		*/
	u32	rcv_tsval;	/* Time stamp value             	*/
	u32	rcv_tsecr;	/* Time stamp echo reply        	*/
	u16 	saw_tstamp : 1,	/* Saw TIMESTAMP on last packet		*/
L
Linus Torvalds 已提交
97 98 99
		tstamp_ok : 1,	/* TIMESTAMP seen on SYN packet		*/
		dsack : 1,	/* D-SACK is scheduled			*/
		wscale_ok : 1,	/* Wscale seen on SYN packet		*/
100 101
		sack_ok : 3,	/* SACK seen on SYN packet		*/
		smc_ok : 1,	/* SMC seen on SYN packet		*/
L
Linus Torvalds 已提交
102 103
		snd_wscale : 4,	/* Window scaling received from sender	*/
		rcv_wscale : 4;	/* Window scaling to send to receiver	*/
104
	u8	num_sacks;	/* Number of SACK blocks		*/
105
	u16	user_mss;	/* mss requested by user in ioctl	*/
106
	u16	mss_clamp;	/* Maximal mss, negotiated at connection setup */
L
Linus Torvalds 已提交
107 108
};

109 110
static inline void tcp_clear_options(struct tcp_options_received *rx_opt)
{
111 112
	rx_opt->tstamp_ok = rx_opt->sack_ok = 0;
	rx_opt->wscale_ok = rx_opt->snd_wscale = 0;
113 114 115
#if IS_ENABLED(CONFIG_SMC)
	rx_opt->smc_ok = 0;
#endif
116 117
}

118
/* This is the max number of SACKS that we'll generate and process. It's safe
119
 * to increase this, although since:
120 121 122 123
 *   size = TCPOLEN_SACK_BASE_ALIGNED (4) + n * TCPOLEN_SACK_PERBLOCK (8)
 * only four options will fit in a standard TCP header */
#define TCP_NUM_SACKS 4

124 125
struct tcp_request_sock_ops;

126
struct tcp_request_sock {
127
	struct inet_request_sock 	req;
128
	const struct tcp_request_sock_ops *af_specific;
129
	u64				snt_synack; /* first SYNACK sent time */
130
	bool				tfo_listener;
131
	u32				txhash;
132 133
	u32				rcv_isn;
	u32				snt_isn;
134
	u32				ts_off;
135
	u32				last_oow_ack_time; /* last SYNACK */
136 137 138 139
	u32				rcv_nxt; /* the ack # by SYNACK. For
						  * FastOpen it's the seq#
						  * after data-in-SYN.
						  */
140 141
};

142
static inline struct tcp_request_sock *tcp_rsk(const struct request_sock *req)
143 144 145 146
{
	return (struct tcp_request_sock *)req;
}

L
Linus Torvalds 已提交
147
struct tcp_sock {
148 149
	/* inet_connection_sock has to be the first member of tcp_sock */
	struct inet_connection_sock	inet_conn;
150
	u16	tcp_header_len;	/* Bytes of tcp header to send		*/
E
Eric Dumazet 已提交
151
	u16	gso_segs;	/* Max number of segs per GSO packet	*/
L
Linus Torvalds 已提交
152

153 154 155 156 157 158
/*
 *	Header prediction flags
 *	0x5?10 << 16 + snd_wnd in net byte order
 */
	__be32	pred_flags;

L
Linus Torvalds 已提交
159 160 161 162 163
/*
 *	RFC793 variables by their proper names. This means you can
 *	read the code and the spec side by side (and laugh ...)
 *	See RFC793 and RFC1122. The RFC writes these in capitals.
 */
164 165 166 167
	u64	bytes_received;	/* RFC4898 tcpEStatsAppHCThruOctetsReceived
				 * sum(delta(rcv_nxt)), or how many bytes
				 * were acked.
				 */
168 169 170
	u32	segs_in;	/* RFC4898 tcpEStatsPerfSegsIn
				 * total number of segments in.
				 */
171 172 173
	u32	data_segs_in;	/* RFC4898 tcpEStatsPerfDataSegsIn
				 * total number of data segments in.
				 */
174
 	u32	rcv_nxt;	/* What we want to receive next 	*/
175 176
	u32	copied_seq;	/* Head of yet unread data		*/
	u32	rcv_wup;	/* rcv_nxt on last window update sent	*/
177
 	u32	snd_nxt;	/* Next sequence we send		*/
178 179 180
	u32	segs_out;	/* RFC4898 tcpEStatsPerfSegsOut
				 * The total number of segments sent.
				 */
181 182 183
	u32	data_segs_out;	/* RFC4898 tcpEStatsPerfDataSegsOut
				 * total number of data segments sent.
				 */
184 185 186 187
	u64	bytes_acked;	/* RFC4898 tcpEStatsAppHCThruOctetsAcked
				 * sum(delta(snd_una)), or how many bytes
				 * were acked.
				 */
188 189 190 191
 	u32	snd_una;	/* First byte we want an ack for	*/
 	u32	snd_sml;	/* Last byte of the most recently transmitted small packet */
	u32	rcv_tstamp;	/* timestamp of last received ACK (for keepalives) */
	u32	lsndtime;	/* timestamp of last sent data packet (for restart window) */
192
	u32	last_oow_ack_time;  /* timestamp of last out-of-window ACK */
L
Linus Torvalds 已提交
193

194 195
	u32	tsoffset;	/* timestamp offset */

E
Eric Dumazet 已提交
196
	struct list_head tsq_node; /* anchor in tsq_tasklet.head list */
197
	struct list_head tsorted_sent_queue; /* time-sorted sent but un-SACKed skbs */
E
Eric Dumazet 已提交
198

199 200 201 202
	u32	snd_wl1;	/* Sequence for window update		*/
	u32	snd_wnd;	/* The window we expect to receive	*/
	u32	max_window;	/* Maximal window ever seen from peer	*/
	u32	mss_cache;	/* Cached effective mss, not including SACKS */
L
Linus Torvalds 已提交
203

204 205
	u32	window_clamp;	/* Maximal window to advertise		*/
	u32	rcv_ssthresh;	/* Current window clamp			*/
L
Linus Torvalds 已提交
206

207 208
	/* Information of the most recently (s)acked skb */
	struct tcp_rack {
209
		u64 mstamp; /* (Re)sent time of the skb */
210
		u32 rtt_us;  /* Associated RTT */
211
		u32 end_seq; /* Ending TCP sequence of the skb */
212 213 214 215 216 217 218
		u32 last_delivered; /* tp->delivered at last reo_wnd adj */
		u8 reo_wnd_steps;   /* Allowed reordering window */
#define TCP_RACK_RECOVERY_THRESH 16
		u8 reo_wnd_persist:5, /* No. of recovery since last adj */
		   dsack_seen:1, /* Whether DSACK seen after last adj */
		   advanced:1,	 /* mstamp advanced since last lost marking */
		   reord:1;	 /* reordering detected */
219
	} rack;
220
	u16	advmss;		/* Advertised MSS			*/
E
Eric Dumazet 已提交
221
	u8	compressed_ack;
222 223 224 225
	u32	chrono_start;	/* Start time in jiffies of a TCP chrono */
	u32	chrono_stat[3];	/* Time in jiffies for chrono_stat stats */
	u8	chrono_type:2,	/* current chronograph type */
		rate_app_limited:1,  /* rate_{delivered,interval_us} limited? */
W
Wei Wang 已提交
226
		fastopen_connect:1, /* FASTOPEN_CONNECT sockopt */
227
		fastopen_no_cookie:1, /* Allow send/recv SYN+data without a cookie */
228 229
		is_sack_reneg:1,    /* in recovery from loss with SACK reneg? */
		unused:2;
A
Andreas Petlund 已提交
230 231
	u8	nonagle     : 4,/* Disable Nagle algorithm?             */
		thin_lto    : 1,/* Use linear timeouts for thin streams */
232
		recvmsg_inq : 1,/* Indicate # of bytes in queue upon recvmsg */
Y
Yuchung Cheng 已提交
233 234
		repair      : 1,
		frto        : 1;/* F-RTO (RFC5682) activated in CA_Loss */
P
Pavel Emelyanov 已提交
235
	u8	repair_queue;
Y
Yuchung Cheng 已提交
236
	u8	syn_data:1,	/* SYN includes data */
Y
Yuchung Cheng 已提交
237
		syn_fastopen:1,	/* SYN includes Fast Open option */
238
		syn_fastopen_exp:1,/* SYN includes Fast Open exp. option */
239
		syn_fastopen_ch:1, /* Active TFO re-enabling probe */
240
		syn_data_acked:1,/* data in SYN is acked by SYN-ACK */
241
		save_syn:1,	/* Save headers of SYN packet */
242 243
		is_cwnd_limited:1,/* forward progress limited by snd_cwnd? */
		syn_smc:1;	/* SYN includes SMC */
N
Nandita Dukkipati 已提交
244
	u32	tlp_high_seq;	/* snd_nxt at the time of TLP retransmit. */
L
Linus Torvalds 已提交
245 246

/* RTT measurement */
247
	u64	tcp_mstamp;	/* most recent packet received/sent */
248 249 250 251
	u32	srtt_us;	/* smoothed round trip time << 3 in usecs */
	u32	mdev_us;	/* medium deviation			*/
	u32	mdev_max_us;	/* maximal mdev for the last rtt period	*/
	u32	rttvar_us;	/* smoothed mdev_max			*/
252
	u32	rtt_seq;	/* sequence number to update rttvar	*/
253
	struct  minmax rtt_min;
254 255 256

	u32	packets_out;	/* Packets which are "in flight"	*/
	u32	retrans_out;	/* Retransmitted packets out		*/
257 258
	u32	max_packets_out;  /* max packets_out in last window */
	u32	max_packets_seq;  /* right edge of max_packets_out flight */
259 260 261

	u16	urg_data;	/* Saved octet of OOB data and control flags */
	u8	ecn_flags;	/* ECN status bits.			*/
262 263
	u8	keepalive_probes; /* num of allowed keep alive probes	*/
	u32	reordering;	/* Packet reordering metric.		*/
I
Ilpo Järvinen 已提交
264 265
	u32	snd_up;		/* Urgent pointer		*/

L
Linus Torvalds 已提交
266 267 268 269 270 271 272 273
/*
 *      Options received (usually on last packet, some only on SYN packets).
 */
	struct tcp_options_received rx_opt;

/*
 *	Slow start and congestion control (see also Nagle, and Karn & Partridge)
 */
274 275
 	u32	snd_ssthresh;	/* Slow start size threshold		*/
 	u32	snd_cwnd;	/* Sending congestion window		*/
I
Ilpo Järvinen 已提交
276
	u32	snd_cwnd_cnt;	/* Linear increase counter		*/
277
	u32	snd_cwnd_clamp; /* Do not allow snd_cwnd to grow above this */
278 279
	u32	snd_cwnd_used;
	u32	snd_cwnd_stamp;
280
	u32	prior_cwnd;	/* cwnd right before starting loss recovery */
281 282 283
	u32	prr_delivered;	/* Number of newly delivered packets to
				 * receiver in Recovery. */
	u32	prr_out;	/* Total number of pkts sent during Recovery. */
Y
Yuchung Cheng 已提交
284
	u32	delivered;	/* Total data packets delivered incl. rexmits */
285
	u32	delivered_ce;	/* Like the above but only ECE marked packets */
286
	u32	lost;		/* Total data packets lost incl. rexmits */
287
	u32	app_limited;	/* limited until "delivered" reaches this val */
288 289
	u64	first_tx_mstamp;  /* start of window send phase */
	u64	delivered_mstamp; /* time we reached "delivered" */
Y
Yuchung Cheng 已提交
290 291
	u32	rate_delivered;    /* saved rate sample: packets delivered */
	u32	rate_interval_us;  /* saved rate sample: time elapsed */
L
Linus Torvalds 已提交
292

293 294
 	u32	rcv_wnd;	/* Current receiver window		*/
	u32	write_seq;	/* Tail(+1) of data held in tcp send buffer */
295
	u32	notsent_lowat;	/* TCP_NOTSENT_LOWAT */
296
	u32	pushed_seq;	/* Last pushed seq, required to talk to windows */
297 298
	u32	lost_out;	/* Lost packets			*/
	u32	sacked_out;	/* SACK'd packets			*/
L
Linus Torvalds 已提交
299

300
	struct hrtimer	pacing_timer;
E
Eric Dumazet 已提交
301
	struct hrtimer	compressed_ack_timer;
302

303 304 305 306
	/* from STCP, retrans queue hinting */
	struct sk_buff* lost_skb_hint;
	struct sk_buff *retransmit_skb_hint;

307 308 309
	/* OOO segments go in this rbtree. Socket lock must be held. */
	struct rb_root	out_of_order_queue;
	struct sk_buff	*ooo_last_skb; /* cache rb_last(out_of_order_queue) */
310

311
	/* SACKs data, these 2 need to be together (see tcp_options_write) */
L
Linus Torvalds 已提交
312 313 314
	struct tcp_sack_block duplicate_sack[1]; /* D-SACK block */
	struct tcp_sack_block selective_acks[4]; /* The SACKS themselves*/

315
	struct tcp_sack_block recv_sack_cache[4];
316

317 318
	struct sk_buff *highest_sack;   /* skb just after the highest
					 * skb with SACKed bit set
319 320 321
					 * (validity guaranteed only if
					 * sacked_out > 0)
					 */
322

323 324
	int     lost_cnt_hint;

I
Ilpo Järvinen 已提交
325
	u32	prior_ssthresh; /* ssthresh saved at recovery start	*/
326
	u32	high_seq;	/* snd_nxt at onset of congestion	*/
L
Linus Torvalds 已提交
327

328
	u32	retrans_stamp;	/* Timestamp of the last retransmit,
L
Linus Torvalds 已提交
329 330
				 * also used in SYN-SENT to remember stamp of
				 * the first SYN. */
Y
Yuchung Cheng 已提交
331
	u32	undo_marker;	/* snd_una upon a new recovery episode. */
L
Linus Torvalds 已提交
332
	int	undo_retrans;	/* number of undoable retransmissions. */
333 334
	u32	total_retrans;	/* Total retransmits for entire connection */

335
	u32	urg_seq;	/* Seq of received urgent pointer */
L
Linus Torvalds 已提交
336 337 338
	unsigned int		keepalive_time;	  /* time before keep alive takes place */
	unsigned int		keepalive_intvl;  /* time interval between keep alive probes */

339
	int			linger2;
L
Linus Torvalds 已提交
340

341 342 343 344 345 346 347 348 349 350 351

/* Sock_ops bpf program related variables */
#ifdef CONFIG_BPF
	u8	bpf_sock_ops_cb_flags;  /* Control calling BPF programs
					 * values defined in uapi/linux/tcp.h
					 */
#define BPF_SOCK_OPS_TEST_FLAG(TP, ARG) (TP->bpf_sock_ops_cb_flags & ARG)
#else
#define BPF_SOCK_OPS_TEST_FLAG(TP, ARG) 0
#endif

L
Linus Torvalds 已提交
352
/* Receiver side RTT estimation */
353
	u32 rcv_rtt_last_tsecr;
L
Linus Torvalds 已提交
354
	struct {
355 356 357
		u32	rtt_us;
		u32	seq;
		u64	time;
L
Linus Torvalds 已提交
358 359 360 361
	} rcv_rtt_est;

/* Receiver queue space */
	struct {
362
		u32	space;
363 364
		u32	seq;
		u64	time;
L
Linus Torvalds 已提交
365
	} rcvq_space;
366 367 368

/* TCP-specific MTU probe information. */
	struct {
369 370
		u32		  probe_seq_start;
		u32		  probe_seq_end;
371
	} mtu_probe;
372 373 374
	u32	mtu_info; /* We received an ICMP_FRAG_NEEDED / ICMPV6_PKT_TOOBIG
			   * while socket was owned by user.
			   */
375 376 377

#ifdef CONFIG_TCP_MD5SIG
/* TCP AF-Specific parts; only used by MD5 Signature support so far */
378
	const struct tcp_sock_af_ops	*af_specific;
379

380
/* TCP MD5 Signature Option information */
381
	struct tcp_md5sig_info	__rcu *md5sig_info;
382
#endif
383

384 385 386 387 388 389
/* TCP fastopen related information */
	struct tcp_fastopen_request *fastopen_req;
	/* fastopen_rsk points to request_sock that resulted in this big
	 * socket. Used to retransmit SYNACKs etc.
	 */
	struct request_sock *fastopen_rsk;
390
	u32	*saved_syn;
L
Linus Torvalds 已提交
391 392
};

E
Eric Dumazet 已提交
393
enum tsq_enum {
E
Eric Dumazet 已提交
394 395
	TSQ_THROTTLED,
	TSQ_QUEUED,
396 397 398
	TCP_TSQ_DEFERRED,	   /* tcp_tasklet_func() found socket was owned */
	TCP_WRITE_TIMER_DEFERRED,  /* tcp_write_timer() found socket was owned */
	TCP_DELACK_TIMER_DEFERRED, /* tcp_delack_timer() found socket was owned */
399 400 401
	TCP_MTU_REDUCED_DEFERRED,  /* tcp_v{4|6}_err() could not call
				    * tcp_v{4|6}_mtu_reduced()
				    */
E
Eric Dumazet 已提交
402 403
};

E
Eric Dumazet 已提交
404 405 406 407 408 409 410 411 412
enum tsq_flags {
	TSQF_THROTTLED			= (1UL << TSQ_THROTTLED),
	TSQF_QUEUED			= (1UL << TSQ_QUEUED),
	TCPF_TSQ_DEFERRED		= (1UL << TCP_TSQ_DEFERRED),
	TCPF_WRITE_TIMER_DEFERRED	= (1UL << TCP_WRITE_TIMER_DEFERRED),
	TCPF_DELACK_TIMER_DEFERRED	= (1UL << TCP_DELACK_TIMER_DEFERRED),
	TCPF_MTU_REDUCED_DEFERRED	= (1UL << TCP_MTU_REDUCED_DEFERRED),
};

L
Linus Torvalds 已提交
413 414 415 416 417
static inline struct tcp_sock *tcp_sk(const struct sock *sk)
{
	return (struct tcp_sock *)sk;
}

418 419
struct tcp_timewait_sock {
	struct inet_timewait_sock tw_sk;
420 421
#define tw_rcv_nxt tw_sk.__tw_common.skc_tw_rcv_nxt
#define tw_snd_nxt tw_sk.__tw_common.skc_tw_snd_nxt
422
	u32			  tw_rcv_wnd;
423
	u32			  tw_ts_offset;
424
	u32			  tw_ts_recent;
425 426 427 428

	/* The time we sent the last out-of-window ACK: */
	u32			  tw_last_oow_ack_time;

429
	long			  tw_ts_recent_stamp;
430
#ifdef CONFIG_TCP_MD5SIG
431
	struct tcp_md5sig_key	  *tw_md5_key;
432
#endif
433 434 435 436 437 438 439
};

static inline struct tcp_timewait_sock *tcp_twsk(const struct sock *sk)
{
	return (struct tcp_timewait_sock *)sk;
}

440 441 442 443 444 445
static inline bool tcp_passive_fastopen(const struct sock *sk)
{
	return (sk->sk_state == TCP_SYN_RECV &&
		tcp_sk(sk)->fastopen_rsk != NULL);
}

446
static inline void fastopen_queue_tune(struct sock *sk, int backlog)
447
{
448
	struct request_sock_queue *queue = &inet_csk(sk)->icsk_accept_queue;
E
Eric Dumazet 已提交
449
	int somaxconn = READ_ONCE(sock_net(sk)->core.sysctl_somaxconn);
450

E
Eric Dumazet 已提交
451
	queue->fastopenq.max_qlen = min_t(unsigned int, backlog, somaxconn);
452 453
}

E
Eric Dumazet 已提交
454 455 456 457 458 459 460
static inline void tcp_move_syn(struct tcp_sock *tp,
				struct request_sock *req)
{
	tp->saved_syn = req->saved_syn;
	req->saved_syn = NULL;
}

461 462 463 464 465 466
static inline void tcp_saved_syn_free(struct tcp_sock *tp)
{
	kfree(tp->saved_syn);
	tp->saved_syn = NULL;
}

467 468
struct sk_buff *tcp_get_timestamping_opt_stats(const struct sock *sk);

E
Eric Dumazet 已提交
469 470 471 472 473 474 475 476 477
static inline u16 tcp_mss_clamp(const struct tcp_sock *tp, u16 mss)
{
	/* We use READ_ONCE() here because socket might not be locked.
	 * This happens for listeners.
	 */
	u16 user_mss = READ_ONCE(tp->rx_opt.user_mss);

	return (user_mss && user_mss < mss) ? user_mss : mss;
}
L
Linus Torvalds 已提交
478
#endif	/* _LINUX_TCP_H */