tcp.h 58.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9
/*
 * 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 module.
 *
 * Version:	@(#)tcp.h	1.0.5	05/23/93
 *
10
 * Authors:	Ross Biro
L
Linus Torvalds 已提交
11 12 13 14 15 16 17 18 19 20 21 22 23 24
 *		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 _TCP_H
#define _TCP_H

#define FASTRETRANS_DEBUG 1

#include <linux/list.h>
#include <linux/tcp.h>
25
#include <linux/bug.h>
L
Linus Torvalds 已提交
26 27 28
#include <linux/slab.h>
#include <linux/cache.h>
#include <linux/percpu.h>
29
#include <linux/skbuff.h>
30
#include <linux/cryptohash.h>
31
#include <linux/kref.h>
32
#include <linux/ktime.h>
33 34

#include <net/inet_connection_sock.h>
35
#include <net/inet_timewait_sock.h>
36
#include <net/inet_hashtables.h>
L
Linus Torvalds 已提交
37
#include <net/checksum.h>
38
#include <net/request_sock.h>
L
Linus Torvalds 已提交
39 40 41
#include <net/sock.h>
#include <net/snmp.h>
#include <net/ip.h>
42
#include <net/tcp_states.h>
43
#include <net/inet_ecn.h>
44
#include <net/dst.h>
45

L
Linus Torvalds 已提交
46
#include <linux/seq_file.h>
47
#include <linux/memcontrol.h>
L
Linus Torvalds 已提交
48

49
extern struct inet_hashinfo tcp_hashinfo;
L
Linus Torvalds 已提交
50

51
extern struct percpu_counter tcp_orphan_count;
52
void tcp_time_wait(struct sock *sk, int state, int timeo);
L
Linus Torvalds 已提交
53 54

#define MAX_TCP_HEADER	(128 + MAX_HEADER)
A
Adam Langley 已提交
55
#define MAX_TCP_OPTION_SPACE 40
L
Linus Torvalds 已提交
56

57
/*
L
Linus Torvalds 已提交
58
 * Never offer a window over 32767 without using window scaling. Some
59
 * poor stacks do signed 16bit maths!
L
Linus Torvalds 已提交
60 61 62 63 64 65
 */
#define MAX_TCP_WINDOW		32767U

/* Minimal accepted MSS. It is (60+60+8) - (20+20). */
#define TCP_MIN_MSS		88U

J
John Heffner 已提交
66
/* The least MTU to use for probing */
67
#define TCP_BASE_MSS		1024
J
John Heffner 已提交
68

69 70 71
/* probing interval, default to 10 minutes as per RFC4821 */
#define TCP_PROBE_INTERVAL	600

72 73 74
/* Specify interval when tcp mtu probing will stop */
#define TCP_PROBE_THRESHOLD	8

L
Linus Torvalds 已提交
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
/* After receiving this amount of duplicate ACKs fast retransmit starts. */
#define TCP_FASTRETRANS_THRESH 3

/* Maximal number of ACKs sent quickly to accelerate slow-start. */
#define TCP_MAX_QUICKACKS	16U

/* urg_data states */
#define TCP_URG_VALID	0x0100
#define TCP_URG_NOTYET	0x0200
#define TCP_URG_READ	0x0400

#define TCP_RETR1	3	/*
				 * This is how many retries it does before it
				 * tries to figure out if the gateway is
				 * down. Minimal RFC value is 3; it corresponds
				 * to ~3sec-8min depending on RTO.
				 */

#define TCP_RETR2	15	/*
				 * This should take at least
				 * 90 minutes to time out.
				 * RFC1122 says that the limit is 100 sec.
				 * 15 is ~13-30min depending on RTO.
				 */

100 101 102 103 104 105 106 107
#define TCP_SYN_RETRIES	 6	/* This is how many retries are done
				 * when active opening a connection.
				 * RFC1122 says the minimum retry MUST
				 * be at least 180secs.  Nevertheless
				 * this value is corresponding to
				 * 63secs of retransmission with the
				 * current initial RTO.
				 */
L
Linus Torvalds 已提交
108

109 110 111 112 113 114
#define TCP_SYNACK_RETRIES 5	/* This is how may retries are done
				 * when passive opening a connection.
				 * This is corresponding to 31secs of
				 * retransmission with the current
				 * initial RTO.
				 */
L
Linus Torvalds 已提交
115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134

#define TCP_TIMEWAIT_LEN (60*HZ) /* how long to wait to destroy TIME-WAIT
				  * state, about 60 seconds	*/
#define TCP_FIN_TIMEOUT	TCP_TIMEWAIT_LEN
                                 /* BSD style FIN_WAIT2 deadlock breaker.
				  * It used to be 3min, new value is 60sec,
				  * to combine FIN-WAIT-2 timeout with
				  * TIME-WAIT timer.
				  */

#define TCP_DELACK_MAX	((unsigned)(HZ/5))	/* maximal time to delay before sending an ACK */
#if HZ >= 100
#define TCP_DELACK_MIN	((unsigned)(HZ/25))	/* minimal time to delay before sending an ACK */
#define TCP_ATO_MIN	((unsigned)(HZ/25))
#else
#define TCP_DELACK_MIN	4U
#define TCP_ATO_MIN	4U
#endif
#define TCP_RTO_MAX	((unsigned)(120*HZ))
#define TCP_RTO_MIN	((unsigned)(HZ/5))
135
#define TCP_TIMEOUT_INIT ((unsigned)(1*HZ))	/* RFC6298 2.1 initial RTO value	*/
136 137 138 139 140 141
#define TCP_TIMEOUT_FALLBACK ((unsigned)(3*HZ))	/* RFC 1122 initial RTO value, now
						 * used as a fallback RTO for the
						 * initial data transmission if no
						 * valid RTT sample has been acquired,
						 * most likely due to retrans in 3WHS.
						 */
L
Linus Torvalds 已提交
142 143 144 145

#define TCP_RESOURCE_PROBE_INTERVAL ((unsigned)(HZ/2U)) /* Maximal interval between probes
					                 * for local resources.
					                 */
146
#define TCP_REO_TIMEOUT_MIN	(2000) /* Min RACK reordering timeout in usec */
L
Linus Torvalds 已提交
147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172

#define TCP_KEEPALIVE_TIME	(120*60*HZ)	/* two hours */
#define TCP_KEEPALIVE_PROBES	9		/* Max of 9 keepalive probes	*/
#define TCP_KEEPALIVE_INTVL	(75*HZ)

#define MAX_TCP_KEEPIDLE	32767
#define MAX_TCP_KEEPINTVL	32767
#define MAX_TCP_KEEPCNT		127
#define MAX_TCP_SYNCNT		127

#define TCP_SYNQ_INTERVAL	(HZ/5)	/* Period of SYNACK timer */

#define TCP_PAWS_24DAYS	(60 * 60 * 24 * 24)
#define TCP_PAWS_MSL	60		/* Per-host timestamps are invalidated
					 * after this time. It should be equal
					 * (or greater than) TCP_TIMEWAIT_LEN
					 * to provide reliability equal to one
					 * provided by timewait state.
					 */
#define TCP_PAWS_WINDOW	1		/* Replay window for per-host
					 * timestamps. It must be less than
					 * minimal timewait lifetime.
					 */
/*
 *	TCP option
 */
173

L
Linus Torvalds 已提交
174 175 176 177 178 179 180
#define TCPOPT_NOP		1	/* Padding */
#define TCPOPT_EOL		0	/* End of options */
#define TCPOPT_MSS		2	/* Segment size negotiating */
#define TCPOPT_WINDOW		3	/* Window scaling */
#define TCPOPT_SACK_PERM        4       /* SACK Permitted */
#define TCPOPT_SACK             5       /* SACK Block */
#define TCPOPT_TIMESTAMP	8	/* Better RTT estimations/PAWS */
181
#define TCPOPT_MD5SIG		19	/* MD5 Signature (RFC2385) */
182
#define TCPOPT_FASTOPEN		34	/* Fast open (RFC7413) */
Y
Yuchung Cheng 已提交
183 184 185 186 187
#define TCPOPT_EXP		254	/* Experimental */
/* Magic number to be after the option value for sharing TCP
 * experimental options. See draft-ietf-tcpm-experimental-options-00.txt
 */
#define TCPOPT_FASTOPEN_MAGIC	0xF989
L
Linus Torvalds 已提交
188 189 190 191 192 193 194 195 196

/*
 *     TCP option lengths
 */

#define TCPOLEN_MSS            4
#define TCPOLEN_WINDOW         3
#define TCPOLEN_SACK_PERM      2
#define TCPOLEN_TIMESTAMP      10
197
#define TCPOLEN_MD5SIG         18
198
#define TCPOLEN_FASTOPEN_BASE  2
Y
Yuchung Cheng 已提交
199
#define TCPOLEN_EXP_FASTOPEN_BASE  4
L
Linus Torvalds 已提交
200 201 202 203 204 205 206 207

/* But this is what stacks really send out. */
#define TCPOLEN_TSTAMP_ALIGNED		12
#define TCPOLEN_WSCALE_ALIGNED		4
#define TCPOLEN_SACKPERM_ALIGNED	4
#define TCPOLEN_SACK_BASE		2
#define TCPOLEN_SACK_BASE_ALIGNED	4
#define TCPOLEN_SACK_PERBLOCK		8
208
#define TCPOLEN_MD5SIG_ALIGNED		20
A
Adam Langley 已提交
209
#define TCPOLEN_MSS_ALIGNED		4
L
Linus Torvalds 已提交
210 211 212 213

/* Flags in tp->nonagle */
#define TCP_NAGLE_OFF		1	/* Nagle's algo is disabled */
#define TCP_NAGLE_CORK		2	/* Socket is corked	    */
S
Stephen Hemminger 已提交
214
#define TCP_NAGLE_PUSH		4	/* Cork is overridden for already queued data */
L
Linus Torvalds 已提交
215

A
Andreas Petlund 已提交
216 217 218
/* TCP thin-stream limits */
#define TCP_THIN_LINEAR_RETRIES 6       /* After 6 linear retries, do exp. backoff */

219
/* TCP initial congestion window as per rfc6928 */
220 221
#define TCP_INIT_CWND		10

222 223
/* Bit Flags for sysctl_tcp_fastopen */
#define	TFO_CLIENT_ENABLE	1
224
#define	TFO_SERVER_ENABLE	2
225
#define	TFO_CLIENT_NO_COOKIE	4	/* Data in SYN w/o cookie option */
226

227 228 229 230
/* Accept SYN data w/o any cookie option */
#define	TFO_SERVER_COOKIE_NOT_REQD	0x200

/* Force enable TFO on all listeners, i.e., not requiring the
231
 * TCP_FASTOPEN socket option.
232 233 234
 */
#define	TFO_SERVER_WO_SOCKOPT1	0x400

235

L
Linus Torvalds 已提交
236 237 238 239
/* sysctl variables for tcp */
extern int sysctl_tcp_timestamps;
extern int sysctl_tcp_window_scaling;
extern int sysctl_tcp_sack;
Y
Yuchung Cheng 已提交
240
extern int sysctl_tcp_fastopen;
L
Linus Torvalds 已提交
241 242 243 244 245 246 247
extern int sysctl_tcp_retrans_collapse;
extern int sysctl_tcp_stdurg;
extern int sysctl_tcp_rfc1337;
extern int sysctl_tcp_abort_on_overflow;
extern int sysctl_tcp_max_orphans;
extern int sysctl_tcp_fack;
extern int sysctl_tcp_reordering;
248
extern int sysctl_tcp_max_reordering;
L
Linus Torvalds 已提交
249
extern int sysctl_tcp_dsack;
250
extern long sysctl_tcp_mem[3];
L
Linus Torvalds 已提交
251 252 253 254 255 256 257 258 259
extern int sysctl_tcp_wmem[3];
extern int sysctl_tcp_rmem[3];
extern int sysctl_tcp_app_win;
extern int sysctl_tcp_adv_win_scale;
extern int sysctl_tcp_frto;
extern int sysctl_tcp_low_latency;
extern int sysctl_tcp_nometrics_save;
extern int sysctl_tcp_moderate_rcvbuf;
extern int sysctl_tcp_tso_win_divisor;
260
extern int sysctl_tcp_workaround_signed_windows;
261
extern int sysctl_tcp_slow_start_after_idle;
A
Andreas Petlund 已提交
262
extern int sysctl_tcp_thin_linear_timeouts;
A
Andreas Petlund 已提交
263
extern int sysctl_tcp_thin_dupack;
Y
Yuchung Cheng 已提交
264
extern int sysctl_tcp_early_retrans;
265 266 267
extern int sysctl_tcp_recovery;
#define TCP_RACK_LOSS_DETECTION  0x1 /* Use RACK to detect losses */

E
Eric Dumazet 已提交
268
extern int sysctl_tcp_limit_output_bytes;
E
Eric Dumazet 已提交
269
extern int sysctl_tcp_challenge_ack_limit;
E
Eric Dumazet 已提交
270
extern int sysctl_tcp_min_tso_segs;
271
extern int sysctl_tcp_min_rtt_wlen;
E
Eric Dumazet 已提交
272
extern int sysctl_tcp_autocorking;
273
extern int sysctl_tcp_invalid_ratelimit;
274 275
extern int sysctl_tcp_pacing_ss_ratio;
extern int sysctl_tcp_pacing_ca_ratio;
L
Linus Torvalds 已提交
276

E
Eric Dumazet 已提交
277
extern atomic_long_t tcp_memory_allocated;
278
extern struct percpu_counter tcp_sockets_allocated;
L
Linus Torvalds 已提交
279 280
extern int tcp_memory_pressure;

281 282 283
/* optimized version of sk_under_memory_pressure() for TCP sockets */
static inline bool tcp_under_memory_pressure(const struct sock *sk)
{
284 285
	if (mem_cgroup_sockets_enabled && sk->sk_memcg &&
	    mem_cgroup_under_socket_pressure(sk->sk_memcg))
286
		return true;
287 288 289

	return tcp_memory_pressure;
}
L
Linus Torvalds 已提交
290 291 292 293 294
/*
 * The next routines deal with comparing 32 bit unsigned ints
 * and worry about wraparound (automatic with unsigned arithmetic).
 */

E
Eric Dumazet 已提交
295
static inline bool before(__u32 seq1, __u32 seq2)
L
Linus Torvalds 已提交
296
{
297
        return (__s32)(seq1-seq2) < 0;
L
Linus Torvalds 已提交
298
}
299
#define after(seq2, seq1) 	before(seq1, seq2)
L
Linus Torvalds 已提交
300 301

/* is s2<=s1<=s3 ? */
E
Eric Dumazet 已提交
302
static inline bool between(__u32 seq1, __u32 seq2, __u32 seq3)
L
Linus Torvalds 已提交
303 304 305 306
{
	return seq3 - seq2 >= seq1 - seq2;
}

A
Arun Sharma 已提交
307 308 309 310 311 312 313 314
static inline bool tcp_out_of_memory(struct sock *sk)
{
	if (sk->sk_wmem_queued > SOCK_MIN_SNDBUF &&
	    sk_memory_allocated(sk) > sk_prot_mem_limits(sk, 2))
		return true;
	return false;
}

315 316
void sk_forced_mem_schedule(struct sock *sk, int size);

317
static inline bool tcp_too_many_orphans(struct sock *sk, int shift)
318
{
319 320 321 322 323 324 325 326 327
	struct percpu_counter *ocp = sk->sk_prot->orphan_count;
	int orphans = percpu_counter_read_positive(ocp);

	if (orphans << shift > sysctl_tcp_max_orphans) {
		orphans = percpu_counter_sum_positive(ocp);
		if (orphans << shift > sysctl_tcp_max_orphans)
			return true;
	}
	return false;
328
}
L
Linus Torvalds 已提交
329

330
bool tcp_check_oom(struct sock *sk, int shift);
A
Arun Sharma 已提交
331

332

L
Linus Torvalds 已提交
333 334
extern struct proto tcp_prot;

335
#define TCP_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.tcp_statistics, field)
E
Eric Dumazet 已提交
336
#define __TCP_INC_STATS(net, field)	__SNMP_INC_STATS((net)->mib.tcp_statistics, field)
337
#define TCP_DEC_STATS(net, field)	SNMP_DEC_STATS((net)->mib.tcp_statistics, field)
338
#define TCP_ADD_STATS(net, field, val)	SNMP_ADD_STATS((net)->mib.tcp_statistics, field, val)
L
Linus Torvalds 已提交
339

340 341 342 343 344 345 346 347 348 349
void tcp_tasklet_init(void);

void tcp_v4_err(struct sk_buff *skb, u32);

void tcp_shutdown(struct sock *sk, int how);

void tcp_v4_early_demux(struct sk_buff *skb);
int tcp_v4_rcv(struct sk_buff *skb);

int tcp_v4_tw_remember_stamp(struct inet_timewait_sock *tw);
350
int tcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t size);
351 352 353 354 355 356 357
int tcp_sendpage(struct sock *sk, struct page *page, int offset, size_t size,
		 int flags);
void tcp_release_cb(struct sock *sk);
void tcp_wfree(struct sk_buff *skb);
void tcp_write_timer_handler(struct sock *sk);
void tcp_delack_timer_handler(struct sock *sk);
int tcp_ioctl(struct sock *sk, int cmd, unsigned long arg);
358
int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb);
359 360 361 362 363 364 365 366
void tcp_rcv_established(struct sock *sk, struct sk_buff *skb,
			 const struct tcphdr *th, unsigned int len);
void tcp_rcv_space_adjust(struct sock *sk);
int tcp_twsk_unique(struct sock *sk, struct sock *sktw, void *twp);
void tcp_twsk_destructor(struct sock *sk);
ssize_t tcp_splice_read(struct socket *sk, loff_t *ppos,
			struct pipe_inode_info *pipe, size_t len,
			unsigned int flags);
J
Jens Axboe 已提交
367

368 369
static inline void tcp_dec_quickack_mode(struct sock *sk,
					 const unsigned int pkts)
L
Linus Torvalds 已提交
370
{
371
	struct inet_connection_sock *icsk = inet_csk(sk);
372

373 374 375
	if (icsk->icsk_ack.quick) {
		if (pkts >= icsk->icsk_ack.quick) {
			icsk->icsk_ack.quick = 0;
376
			/* Leaving quickack mode we deflate ATO. */
377
			icsk->icsk_ack.ato   = TCP_ATO_MIN;
378
		} else
379
			icsk->icsk_ack.quick -= pkts;
L
Linus Torvalds 已提交
380 381 382
	}
}

383 384 385
#define	TCP_ECN_OK		1
#define	TCP_ECN_QUEUE_CWR	2
#define	TCP_ECN_DEMAND_CWR	4
386
#define	TCP_ECN_SEEN		8
387

E
Eric Dumazet 已提交
388
enum tcp_tw_status {
L
Linus Torvalds 已提交
389 390 391 392 393 394 395
	TCP_TW_SUCCESS = 0,
	TCP_TW_RST = 1,
	TCP_TW_ACK = 2,
	TCP_TW_SYN = 3
};


396 397 398 399
enum tcp_tw_status tcp_timewait_state_process(struct inet_timewait_sock *tw,
					      struct sk_buff *skb,
					      const struct tcphdr *th);
struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
400
			   struct request_sock *req, bool fastopen);
401 402
int tcp_child_process(struct sock *parent, struct sock *child,
		      struct sk_buff *skb);
403
void tcp_enter_loss(struct sock *sk);
404
void tcp_cwnd_reduction(struct sock *sk, int newly_acked_sacked, int flag);
405 406 407 408 409
void tcp_clear_retrans(struct tcp_sock *tp);
void tcp_update_metrics(struct sock *sk);
void tcp_init_metrics(struct sock *sk);
void tcp_metrics_init(void);
bool tcp_peer_is_proven(struct request_sock *req, struct dst_entry *dst,
410
			bool paws_check, bool timestamps);
411 412 413 414 415 416 417 418 419 420 421 422 423
bool tcp_remember_stamp(struct sock *sk);
bool tcp_tw_remember_stamp(struct inet_timewait_sock *tw);
void tcp_fetch_timewait_stamp(struct sock *sk, struct dst_entry *dst);
void tcp_disable_fack(struct tcp_sock *tp);
void tcp_close(struct sock *sk, long timeout);
void tcp_init_sock(struct sock *sk);
unsigned int tcp_poll(struct file *file, struct socket *sock,
		      struct poll_table_struct *wait);
int tcp_getsockopt(struct sock *sk, int level, int optname,
		   char __user *optval, int __user *optlen);
int tcp_setsockopt(struct sock *sk, int level, int optname,
		   char __user *optval, unsigned int optlen);
int compat_tcp_getsockopt(struct sock *sk, int level, int optname,
C
Changli Gao 已提交
424
			  char __user *optval, int __user *optlen);
425
int compat_tcp_setsockopt(struct sock *sk, int level, int optname,
C
Changli Gao 已提交
426
			  char __user *optval, unsigned int optlen);
427
void tcp_set_keepalive(struct sock *sk, int val);
428
void tcp_syn_ack_timeout(const struct request_sock *req);
429 430
int tcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int nonblock,
		int flags, int *addr_len);
431 432 433 434
void tcp_parse_options(const struct sk_buff *skb,
		       struct tcp_options_received *opt_rx,
		       int estab, struct tcp_fastopen_cookie *foc);
const u8 *tcp_parse_md5sig_option(const struct tcphdr *th);
435

L
Linus Torvalds 已提交
436 437 438 439
/*
 *	TCP v4 functions exported for the inet6 API
 */

440
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb);
441
void tcp_v4_mtu_reduced(struct sock *sk);
442
void tcp_req_err(struct sock *sk, u32 seq, bool abort);
443
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb);
444
struct sock *tcp_create_openreq_child(const struct sock *sk,
445 446
				      struct request_sock *req,
				      struct sk_buff *skb);
447
void tcp_ca_openreq_child(struct sock *sk, const struct dst_entry *dst);
448
struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
449
				  struct request_sock *req,
450 451 452
				  struct dst_entry *dst,
				  struct request_sock *req_unhash,
				  bool *own_req);
453 454 455
int tcp_v4_do_rcv(struct sock *sk, struct sk_buff *skb);
int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
int tcp_connect(struct sock *sk);
456 457 458 459 460
enum tcp_synack_type {
	TCP_SYNACK_NORMAL,
	TCP_SYNACK_FASTOPEN,
	TCP_SYNACK_COOKIE,
};
461
struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
462
				struct request_sock *req,
463
				struct tcp_fastopen_cookie *foc,
464
				enum tcp_synack_type synack_type);
465
int tcp_disconnect(struct sock *sk, int flags);
L
Linus Torvalds 已提交
466

P
Pavel Emelyanov 已提交
467
void tcp_finish_connect(struct sock *sk, struct sk_buff *skb);
468
int tcp_send_rcvq(struct sock *sk, struct msghdr *msg, size_t size);
469
void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb);
L
Linus Torvalds 已提交
470 471

/* From syncookies.c */
472 473 474
struct sock *tcp_get_cookie_sock(struct sock *sk, struct sk_buff *skb,
				 struct request_sock *req,
				 struct dst_entry *dst);
475 476
int __cookie_v4_check(const struct iphdr *iph, const struct tcphdr *th,
		      u32 cookie);
C
Cong Wang 已提交
477
struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb);
478
#ifdef CONFIG_SYN_COOKIES
479

480
/* Syncookies use a monotonic timer which increments every 60 seconds.
481 482 483
 * This counter is used both as a hash input and partially encoded into
 * the cookie value.  A cookie is only validated further if the delta
 * between the current counter value and the encoded one is less than this,
484
 * i.e. a sent cookie is valid only at most for 2*60 seconds (or less if
485 486
 * the counter advances immediately after a cookie is generated).
 */
487 488 489 490 491 492
#define MAX_SYNCOOKIE_AGE	2
#define TCP_SYNCOOKIE_PERIOD	(60 * HZ)
#define TCP_SYNCOOKIE_VALID	(MAX_SYNCOOKIE_AGE * TCP_SYNCOOKIE_PERIOD)

/* syncookies: remember time of last synqueue overflow
 * But do not dirty this field too often (once per second is enough)
493
 * It is racy as we do not hold a lock, but race is very minor.
494
 */
495
static inline void tcp_synq_overflow(const struct sock *sk)
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
{
	unsigned long last_overflow = tcp_sk(sk)->rx_opt.ts_recent_stamp;
	unsigned long now = jiffies;

	if (time_after(now, last_overflow + HZ))
		tcp_sk(sk)->rx_opt.ts_recent_stamp = now;
}

/* syncookies: no recent synqueue overflow on this listening socket? */
static inline bool tcp_synq_no_recent_overflow(const struct sock *sk)
{
	unsigned long last_overflow = tcp_sk(sk)->rx_opt.ts_recent_stamp;

	return time_after(jiffies, last_overflow + TCP_SYNCOOKIE_VALID);
}
511 512 513

static inline u32 tcp_cookie_time(void)
{
514 515
	u64 val = get_jiffies_64();

516
	do_div(val, TCP_SYNCOOKIE_PERIOD);
517
	return val;
518 519
}

520 521
u32 __cookie_v4_init_sequence(const struct iphdr *iph, const struct tcphdr *th,
			      u16 *mssp);
522
__u32 cookie_v4_init_sequence(const struct sk_buff *skb, __u16 *mss);
523
__u32 cookie_init_timestamp(struct request_sock *req);
524 525
bool cookie_timestamp_decode(struct tcp_options_received *opt);
bool cookie_ecn_ok(const struct tcp_options_received *opt,
526
		   const struct net *net, const struct dst_entry *dst);
527

528
/* From net/ipv6/syncookies.c */
529 530 531
int __cookie_v6_check(const struct ipv6hdr *iph, const struct tcphdr *th,
		      u32 cookie);
struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb);
532

533 534
u32 __cookie_v6_init_sequence(const struct ipv6hdr *iph,
			      const struct tcphdr *th, u16 *mssp);
535
__u32 cookie_v6_init_sequence(const struct sk_buff *skb, __u16 *mss);
536
#endif
L
Linus Torvalds 已提交
537 538
/* tcp_output.c */

539 540
u32 tcp_tso_autosize(const struct sock *sk, unsigned int mss_now,
		     int min_tso_segs);
541 542 543
void __tcp_push_pending_frames(struct sock *sk, unsigned int cur_mss,
			       int nonagle);
bool tcp_may_send_now(struct sock *sk);
544 545
int __tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb, int segs);
int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb, int segs);
546 547 548
void tcp_retransmit_timer(struct sock *sk);
void tcp_xmit_retransmit_queue(struct sock *);
void tcp_simple_retransmit(struct sock *);
549
void tcp_enter_recovery(struct sock *sk, bool ece_ack);
550
int tcp_trim_head(struct sock *, struct sk_buff *, u32);
551
int tcp_fragment(struct sock *, struct sk_buff *, u32, unsigned int, gfp_t);
552 553 554

void tcp_send_probe0(struct sock *);
void tcp_send_partial(struct sock *);
555
int tcp_write_wakeup(struct sock *, int mib);
556 557 558 559 560 561 562 563
void tcp_send_fin(struct sock *sk);
void tcp_send_active_reset(struct sock *sk, gfp_t priority);
int tcp_send_synack(struct sock *);
void tcp_push_one(struct sock *, unsigned int mss_now);
void tcp_send_ack(struct sock *sk);
void tcp_send_delayed_ack(struct sock *sk);
void tcp_send_loss_probe(struct sock *sk);
bool tcp_schedule_loss_probe(struct sock *sk);
564 565
void tcp_skb_collapse_tstamp(struct sk_buff *skb,
			     const struct sk_buff *next_skb);
L
Linus Torvalds 已提交
566

567
/* tcp_input.c */
568
void tcp_rearm_rto(struct sock *sk);
Y
Yuchung Cheng 已提交
569
void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req);
570
void tcp_reset(struct sock *sk);
Y
Yuchung Cheng 已提交
571
void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, struct sk_buff *skb);
572
void tcp_fin(struct sock *sk);
573

L
Linus Torvalds 已提交
574
/* tcp_timer.c */
575
void tcp_init_xmit_timers(struct sock *);
576 577 578 579
static inline void tcp_clear_xmit_timers(struct sock *sk)
{
	inet_csk_clear_xmit_timers(sk);
}
L
Linus Torvalds 已提交
580

581 582
unsigned int tcp_sync_mss(struct sock *sk, u32 pmtu);
unsigned int tcp_current_mss(struct sock *sk);
I
Ilpo Järvinen 已提交
583 584 585 586

/* Bound MSS / TSO packet size with the half of the window */
static inline int tcp_bound_to_half_wnd(struct tcp_sock *tp, int pktsize)
{
587 588 589 590 591 592 593 594 595
	int cutoff;

	/* When peer uses tiny windows, there is no use in packetizing
	 * to sub-MSS pieces for the sake of SWS or making sure there
	 * are enough packets in the pipe for fast recovery.
	 *
	 * On the other hand, for extremely large MSS devices, handling
	 * smaller than MSS windows in this way does make sense.
	 */
596
	if (tp->max_window > TCP_MSS_DEFAULT)
597 598 599 600 601 602
		cutoff = (tp->max_window >> 1);
	else
		cutoff = tp->max_window;

	if (cutoff && pktsize > cutoff)
		return max_t(int, cutoff, 68U - tp->tcp_header_len);
I
Ilpo Järvinen 已提交
603 604 605
	else
		return pktsize;
}
L
Linus Torvalds 已提交
606

607
/* tcp.c */
608
void tcp_get_info(struct sock *, struct tcp_info *);
L
Linus Torvalds 已提交
609 610

/* Read 'sendfile()'-style from a TCP socket */
611 612
int tcp_read_sock(struct sock *sk, read_descriptor_t *desc,
		  sk_read_actor_t recv_actor);
L
Linus Torvalds 已提交
613

614
void tcp_initialize_rcv_mss(struct sock *sk);
L
Linus Torvalds 已提交
615

616 617 618 619
int tcp_mtu_to_mss(struct sock *sk, int pmtu);
int tcp_mss_to_mtu(struct sock *sk, int mss);
void tcp_mtup_init(struct sock *sk);
void tcp_init_buffer_space(struct sock *sk);
J
John Heffner 已提交
620

621 622 623 624 625 626 627 628
static inline void tcp_bound_rto(const struct sock *sk)
{
	if (inet_csk(sk)->icsk_rto > TCP_RTO_MAX)
		inet_csk(sk)->icsk_rto = TCP_RTO_MAX;
}

static inline u32 __tcp_set_rto(const struct tcp_sock *tp)
{
629
	return usecs_to_jiffies((tp->srtt_us >> 3) + tp->rttvar_us);
630 631
}

S
Stephen Hemminger 已提交
632
static inline void __tcp_fast_path_on(struct tcp_sock *tp, u32 snd_wnd)
L
Linus Torvalds 已提交
633 634 635 636 637 638
{
	tp->pred_flags = htonl((tp->tcp_header_len << 26) |
			       ntohl(TCP_FLAG_ACK) |
			       snd_wnd);
}

S
Stephen Hemminger 已提交
639
static inline void tcp_fast_path_on(struct tcp_sock *tp)
L
Linus Torvalds 已提交
640 641 642 643
{
	__tcp_fast_path_on(tp, tp->snd_wnd >> tp->rx_opt.snd_wscale);
}

644
static inline void tcp_fast_path_check(struct sock *sk)
L
Linus Torvalds 已提交
645
{
646 647
	struct tcp_sock *tp = tcp_sk(sk);

648
	if (RB_EMPTY_ROOT(&tp->out_of_order_queue) &&
L
Linus Torvalds 已提交
649 650 651 652 653 654
	    tp->rcv_wnd &&
	    atomic_read(&sk->sk_rmem_alloc) < sk->sk_rcvbuf &&
	    !tp->urg_data)
		tcp_fast_path_on(tp);
}

655 656 657
/* Compute the actual rto_min value */
static inline u32 tcp_rto_min(struct sock *sk)
{
658
	const struct dst_entry *dst = __sk_dst_get(sk);
659 660 661 662 663 664 665
	u32 rto_min = TCP_RTO_MIN;

	if (dst && dst_metric_locked(dst, RTAX_RTO_MIN))
		rto_min = dst_metric_rtt(dst, RTAX_RTO_MIN);
	return rto_min;
}

666 667 668 669 670
static inline u32 tcp_rto_min_us(struct sock *sk)
{
	return jiffies_to_usecs(tcp_rto_min(sk));
}

671 672 673 674 675
static inline bool tcp_ca_dst_locked(const struct dst_entry *dst)
{
	return dst_metric_locked(dst, RTAX_CC_ALGO);
}

676 677 678
/* Minimum RTT in usec. ~0 means not available. */
static inline u32 tcp_min_rtt(const struct tcp_sock *tp)
{
679
	return minmax_get(&tp->rtt_min);
680 681
}

L
Linus Torvalds 已提交
682 683 684 685
/* Compute the actual receive window we are currently advertising.
 * Rcv_nxt can be after the window if our peer push more data
 * than the offered window.
 */
S
Stephen Hemminger 已提交
686
static inline u32 tcp_receive_window(const struct tcp_sock *tp)
L
Linus Torvalds 已提交
687 688 689 690 691 692 693 694 695 696 697 698
{
	s32 win = tp->rcv_wup + tp->rcv_wnd - tp->rcv_nxt;

	if (win < 0)
		win = 0;
	return (u32) win;
}

/* Choose a new window, without checks for shrinking, and without
 * scaling applied to the result.  The caller does these things
 * if necessary.  This is a "raw" window selection.
 */
699
u32 __tcp_select_window(struct sock *sk);
L
Linus Torvalds 已提交
700

P
Pavel Emelyanov 已提交
701 702
void tcp_send_window_probe(struct sock *sk);

L
Linus Torvalds 已提交
703 704
/* TCP timestamps are only 32-bits, this causes a slight
 * complication on 64-bit systems since we store a snapshot
S
Stephen Hemminger 已提交
705 706
 * of jiffies in the buffer control blocks below.  We decided
 * to use only the low 32-bits of jiffies and hide the ugly
L
Linus Torvalds 已提交
707 708 709 710
 * casts with the following macro.
 */
#define tcp_time_stamp		((__u32)(jiffies))

E
Eric Dumazet 已提交
711 712 713 714 715 716
static inline u32 tcp_skb_timestamp(const struct sk_buff *skb)
{
	return skb->skb_mstamp.stamp_jiffies;
}


C
Changli Gao 已提交
717 718 719 720 721 722 723 724 725 726 727
#define tcp_flag_byte(th) (((u_int8_t *)th)[13])

#define TCPHDR_FIN 0x01
#define TCPHDR_SYN 0x02
#define TCPHDR_RST 0x04
#define TCPHDR_PSH 0x08
#define TCPHDR_ACK 0x10
#define TCPHDR_URG 0x20
#define TCPHDR_ECE 0x40
#define TCPHDR_CWR 0x80

728 729
#define TCPHDR_SYN_ECN	(TCPHDR_SYN | TCPHDR_ECE | TCPHDR_CWR)

S
Stephen Hemminger 已提交
730
/* This is what the send packet queuing engine uses to pass
731 732 733 734
 * TCP per-packet control information to the transmission code.
 * We also store the host-order sequence numbers in here too.
 * This is 44 bytes if IPV6 is enabled.
 * If this grows please adjust skbuff.h:skbuff->cb[xxx] size appropriately.
L
Linus Torvalds 已提交
735 736 737 738
 */
struct tcp_skb_cb {
	__u32		seq;		/* Starting sequence number	*/
	__u32		end_seq;	/* SEQ + FIN + SYN + datalen	*/
739 740 741 742
	union {
		/* Note : tcp_tw_isn is used in input path only
		 *	  (isn chosen by tcp_timewait_state_process())
		 *
743 744
		 * 	  tcp_gso_segs/size are used in write queue only,
		 *	  cf tcp_skb_pcount()/tcp_skb_mss()
745 746
		 */
		__u32		tcp_tw_isn;
747 748 749 750
		struct {
			u16	tcp_gso_segs;
			u16	tcp_gso_size;
		};
751
	};
E
Eric Dumazet 已提交
752
	__u8		tcp_flags;	/* TCP header flags. (tcp[13])	*/
753

L
Linus Torvalds 已提交
754 755 756 757 758
	__u8		sacked;		/* State flags for SACK/FACK.	*/
#define TCPCB_SACKED_ACKED	0x01	/* SKB ACK'd by a SACK block	*/
#define TCPCB_SACKED_RETRANS	0x02	/* SKB retransmitted		*/
#define TCPCB_LOST		0x04	/* SKB is lost			*/
#define TCPCB_TAGBITS		0x07	/* All tag bits			*/
759
#define TCPCB_REPAIRED		0x10	/* SKB repaired (no skb_mstamp)	*/
L
Linus Torvalds 已提交
760
#define TCPCB_EVER_RETRANS	0x80	/* Ever retransmitted frame	*/
761 762
#define TCPCB_RETRANS		(TCPCB_SACKED_RETRANS|TCPCB_EVER_RETRANS| \
				TCPCB_REPAIRED)
L
Linus Torvalds 已提交
763

764
	__u8		ip_dsfield;	/* IPv4 tos or IPv6 dsfield	*/
765
	__u8		txstamp_ack:1,	/* Record TX timestamp for ack? */
766 767
			eor:1,		/* Is skb MSG_EOR marked? */
			unused:6;
L
Linus Torvalds 已提交
768
	__u32		ack_seq;	/* Sequence number ACK'd	*/
769
	union {
770
		struct {
771
			/* There is space for up to 24 bytes */
772 773 774
			__u32 in_flight:30,/* Bytes in flight at transmit */
			      is_app_limited:1, /* cwnd not fully used? */
			      unused:1;
775 776 777 778 779 780
			/* pkts S/ACKed so far upon tx of skb, incl retrans: */
			__u32 delivered;
			/* start of send pipeline phase */
			struct skb_mstamp first_tx_mstamp;
			/* when we reached the "delivered" count */
			struct skb_mstamp delivered_mstamp;
781 782 783
		} tx;   /* only used for outgoing skbs */
		union {
			struct inet_skb_parm	h4;
784
#if IS_ENABLED(CONFIG_IPV6)
785
			struct inet6_skb_parm	h6;
786
#endif
787 788
		} header;	/* For incoming skbs */
	};
L
Linus Torvalds 已提交
789 790 791 792
};

#define TCP_SKB_CB(__skb)	((struct tcp_skb_cb *)&((__skb)->cb[0]))

E
Eric Dumazet 已提交
793

794
#if IS_ENABLED(CONFIG_IPV6)
E
Eric Dumazet 已提交
795 796 797 798 799
/* This is the variant of inet6_iif() that must be used by TCP,
 * as TCP moves IP6CB into a different location in skb->cb[]
 */
static inline int tcp_v6_iif(const struct sk_buff *skb)
{
800
	bool l3_slave = ipv6_l3mdev_skb(TCP_SKB_CB(skb)->header.h6.flags);
801 802

	return l3_slave ? skb->skb_iif : TCP_SKB_CB(skb)->header.h6.iif;
E
Eric Dumazet 已提交
803
}
804
#endif
E
Eric Dumazet 已提交
805

806 807 808 809 810
/* TCP_SKB_CB reference means this can not be used from early demux */
static inline bool inet_exact_dif_match(struct net *net, struct sk_buff *skb)
{
#if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
	if (!net->ipv4.sysctl_tcp_l3mdev_accept &&
811
	    skb && ipv4_l3mdev_skb(TCP_SKB_CB(skb)->header.h4.flags))
812 813 814 815 816
		return true;
#endif
	return false;
}

L
Linus Torvalds 已提交
817 818
/* Due to TSO, an SKB can be composed of multiple actual
 * packets.  To keep these tracked properly, we use this.
819
 */
L
Linus Torvalds 已提交
820
static inline int tcp_skb_pcount(const struct sk_buff *skb)
821
{
822 823
	return TCP_SKB_CB(skb)->tcp_gso_segs;
}
824

825 826 827
static inline void tcp_skb_pcount_set(struct sk_buff *skb, int segs)
{
	TCP_SKB_CB(skb)->tcp_gso_segs = segs;
828 829
}

830
static inline void tcp_skb_pcount_add(struct sk_buff *skb, int segs)
L
Linus Torvalds 已提交
831
{
832
	TCP_SKB_CB(skb)->tcp_gso_segs += segs;
L
Linus Torvalds 已提交
833 834
}

835
/* This is valid iff skb is in write queue and tcp_skb_pcount() > 1. */
L
Linus Torvalds 已提交
836 837
static inline int tcp_skb_mss(const struct sk_buff *skb)
{
838
	return TCP_SKB_CB(skb)->tcp_gso_size;
L
Linus Torvalds 已提交
839 840
}

841 842 843 844 845
static inline bool tcp_skb_can_collapse_to(const struct sk_buff *skb)
{
	return likely(!TCP_SKB_CB(skb)->eor);
}

846 847 848 849 850 851
/* Events passed to congestion control interface */
enum tcp_ca_event {
	CA_EVENT_TX_START,	/* first transmit when no packets in flight */
	CA_EVENT_CWND_RESTART,	/* congestion window restart */
	CA_EVENT_COMPLETE_CWR,	/* end of congestion recovery */
	CA_EVENT_LOSS,		/* loss timeout */
852 853 854 855
	CA_EVENT_ECN_NO_CE,	/* ECT set, but not CE marked */
	CA_EVENT_ECN_IS_CE,	/* received CE marked IP packet */
	CA_EVENT_DELAYED_ACK,	/* Delayed ack is sent */
	CA_EVENT_NON_DELAYED_ACK,
856 857
};

858
/* Information about inbound ACK, passed to cong_ops->in_ack_event() */
859
enum tcp_ca_ack_event_flags {
860 861 862
	CA_ACK_SLOWPATH		= (1 << 0),	/* In slow path processing */
	CA_ACK_WIN_UPDATE	= (1 << 1),	/* ACK updated window */
	CA_ACK_ECE		= (1 << 2),	/* ECE bit is set on ack */
863 864 865 866 867 868
};

/*
 * Interface for adding new TCP congestion control handlers
 */
#define TCP_CA_NAME_MAX	16
869 870 871
#define TCP_CA_MAX	128
#define TCP_CA_BUF_MAX	(TCP_CA_NAME_MAX*TCP_CA_MAX)

872 873
#define TCP_CA_UNSPEC	0

874
/* Algorithm can be set on socket without CAP_NET_ADMIN privileges */
875
#define TCP_CONG_NON_RESTRICTED 0x1
876 877
/* Requires ECN/ECT set on all packets */
#define TCP_CONG_NEEDS_ECN	0x2
878

879 880
union tcp_cc_info;

881 882 883
struct ack_sample {
	u32 pkts_acked;
	s32 rtt_us;
884
	u32 in_flight;
885 886
};

887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903
/* A rate sample measures the number of (original/retransmitted) data
 * packets delivered "delivered" over an interval of time "interval_us".
 * The tcp_rate.c code fills in the rate sample, and congestion
 * control modules that define a cong_control function to run at the end
 * of ACK processing can optionally chose to consult this sample when
 * setting cwnd and pacing rate.
 * A sample is invalid if "delivered" or "interval_us" is negative.
 */
struct rate_sample {
	struct	skb_mstamp prior_mstamp; /* starting timestamp for interval */
	u32  prior_delivered;	/* tp->delivered at "prior_mstamp" */
	s32  delivered;		/* number of packets delivered over interval */
	long interval_us;	/* time for tp->delivered to incr "delivered" */
	long rtt_us;		/* RTT of last (S)ACKed packet (or -1) */
	int  losses;		/* number of packets marked lost upon ACK */
	u32  acked_sacked;	/* number of packets newly (S)ACKed upon ACK */
	u32  prior_in_flight;	/* in flight before this ACK */
904
	bool is_app_limited;	/* is sample from packet with bubble in pipe? */
905 906 907
	bool is_retrans;	/* is sample from retransmission? */
};

908 909
struct tcp_congestion_ops {
	struct list_head	list;
910 911
	u32 key;
	u32 flags;
912 913

	/* initialize private data (optional) */
914
	void (*init)(struct sock *sk);
915
	/* cleanup private data  (optional) */
916
	void (*release)(struct sock *sk);
917 918

	/* return slow start threshold (required) */
919
	u32 (*ssthresh)(struct sock *sk);
920
	/* do new cwnd calculation (required) */
921
	void (*cong_avoid)(struct sock *sk, u32 ack, u32 acked);
922
	/* call before changing ca_state (optional) */
923
	void (*set_state)(struct sock *sk, u8 new_state);
924
	/* call when cwnd event occurs (optional) */
925
	void (*cwnd_event)(struct sock *sk, enum tcp_ca_event ev);
926 927
	/* call when ack arrives (optional) */
	void (*in_ack_event)(struct sock *sk, u32 flags);
928
	/* new value of cwnd after loss (optional) */
929
	u32  (*undo_cwnd)(struct sock *sk);
930
	/* hook for packet ack accounting (optional) */
931
	void (*pkts_acked)(struct sock *sk, const struct ack_sample *sample);
932 933
	/* suggest number of segments for each skb to transmit (optional) */
	u32 (*tso_segs_goal)(struct sock *sk);
934 935
	/* returns the multiplier used in tcp_sndbuf_expand (optional) */
	u32 (*sndbuf_expand)(struct sock *sk);
936 937 938 939
	/* call when packets are delivered to update cwnd and pacing rate,
	 * after all the ca_state processing. (optional)
	 */
	void (*cong_control)(struct sock *sk, const struct rate_sample *rs);
940
	/* get info for inet_diag (optional) */
941 942
	size_t (*get_info)(struct sock *sk, u32 ext, int *attr,
			   union tcp_cc_info *info);
943 944 945 946 947

	char 		name[TCP_CA_NAME_MAX];
	struct module 	*owner;
};

948 949
int tcp_register_congestion_control(struct tcp_congestion_ops *type);
void tcp_unregister_congestion_control(struct tcp_congestion_ops *type);
950

951
void tcp_assign_congestion_control(struct sock *sk);
952 953 954 955 956 957 958 959
void tcp_init_congestion_control(struct sock *sk);
void tcp_cleanup_congestion_control(struct sock *sk);
int tcp_set_default_congestion_control(const char *name);
void tcp_get_default_congestion_control(char *name);
void tcp_get_available_congestion_control(char *buf, size_t len);
void tcp_get_allowed_congestion_control(char *buf, size_t len);
int tcp_set_allowed_congestion_control(char *allowed);
int tcp_set_congestion_control(struct sock *sk, const char *name);
N
Neal Cardwell 已提交
960 961
u32 tcp_slow_start(struct tcp_sock *tp, u32 acked);
void tcp_cong_avoid_ai(struct tcp_sock *tp, u32 w, u32 acked);
962

963
u32 tcp_reno_ssthresh(struct sock *sk);
964
u32 tcp_reno_undo_cwnd(struct sock *sk);
965
void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 acked);
966
extern struct tcp_congestion_ops tcp_reno;
967

968
struct tcp_congestion_ops *tcp_ca_find_key(u32 key);
969
u32 tcp_ca_get_key_by_name(const char *name, bool *ecn_ca);
970
#ifdef CONFIG_INET
971
char *tcp_ca_get_name_by_key(u32 key, char *buffer);
972 973 974 975 976 977
#else
static inline char *tcp_ca_get_name_by_key(u32 key, char *buffer)
{
	return NULL;
}
#endif
978

979 980 981 982 983 984 985
static inline bool tcp_ca_needs_ecn(const struct sock *sk)
{
	const struct inet_connection_sock *icsk = inet_csk(sk);

	return icsk->icsk_ca_ops->flags & TCP_CONG_NEEDS_ECN;
}

986
static inline void tcp_set_ca_state(struct sock *sk, const u8 ca_state)
987
{
988 989 990 991 992
	struct inet_connection_sock *icsk = inet_csk(sk);

	if (icsk->icsk_ca_ops->set_state)
		icsk->icsk_ca_ops->set_state(sk, ca_state);
	icsk->icsk_ca_state = ca_state;
993 994
}

995
static inline void tcp_ca_event(struct sock *sk, const enum tcp_ca_event event)
996
{
997 998 999 1000
	const struct inet_connection_sock *icsk = inet_csk(sk);

	if (icsk->icsk_ca_ops->cwnd_event)
		icsk->icsk_ca_ops->cwnd_event(sk, event);
1001 1002
}

1003 1004 1005 1006 1007 1008
/* From tcp_rate.c */
void tcp_rate_skb_sent(struct sock *sk, struct sk_buff *skb);
void tcp_rate_skb_delivered(struct sock *sk, struct sk_buff *skb,
			    struct rate_sample *rs);
void tcp_rate_gen(struct sock *sk, u32 delivered, u32 lost,
		  struct skb_mstamp *now, struct rate_sample *rs);
1009
void tcp_rate_check_app_limited(struct sock *sk);
1010

1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023
/* These functions determine how the current flow behaves in respect of SACK
 * handling. SACK is negotiated with the peer, and therefore it can vary
 * between different flows.
 *
 * tcp_is_sack - SACK enabled
 * tcp_is_reno - No SACK
 * tcp_is_fack - FACK enabled, implies SACK enabled
 */
static inline int tcp_is_sack(const struct tcp_sock *tp)
{
	return tp->rx_opt.sack_ok;
}

E
Eric Dumazet 已提交
1024
static inline bool tcp_is_reno(const struct tcp_sock *tp)
1025 1026 1027 1028
{
	return !tcp_is_sack(tp);
}

E
Eric Dumazet 已提交
1029
static inline bool tcp_is_fack(const struct tcp_sock *tp)
1030
{
1031
	return tp->rx_opt.sack_ok & TCP_FACK_ENABLED;
1032 1033 1034 1035
}

static inline void tcp_enable_fack(struct tcp_sock *tp)
{
1036
	tp->rx_opt.sack_ok |= TCP_FACK_ENABLED;
1037 1038
}

1039 1040 1041 1042 1043
static inline unsigned int tcp_left_out(const struct tcp_sock *tp)
{
	return tp->sacked_out + tp->lost_out;
}

L
Linus Torvalds 已提交
1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
/* This determines how many packets are "in the network" to the best
 * of our knowledge.  In many cases it is conservative, but where
 * detailed information is available from the receiver (via SACK
 * blocks etc.) we can make more aggressive calculations.
 *
 * Use this for decisions involving congestion control, use just
 * tp->packets_out to determine if the send queue is empty or not.
 *
 * Read this equation as:
 *
 *	"Packets sent once on transmission queue" MINUS
 *	"Packets left network, but not honestly ACKed yet" PLUS
 *	"Packets fast retransmitted"
 */
S
Stephen Hemminger 已提交
1058
static inline unsigned int tcp_packets_in_flight(const struct tcp_sock *tp)
L
Linus Torvalds 已提交
1059
{
1060
	return tp->packets_out - tcp_left_out(tp) + tp->retrans_out;
L
Linus Torvalds 已提交
1061 1062
}

I
Ilpo Järvinen 已提交
1063 1064
#define TCP_INFINITE_SSTHRESH	0x7fffffff

1065 1066
static inline bool tcp_in_slow_start(const struct tcp_sock *tp)
{
1067
	return tp->snd_cwnd < tp->snd_ssthresh;
1068 1069
}

I
Ilpo Järvinen 已提交
1070 1071 1072 1073 1074
static inline bool tcp_in_initial_slowstart(const struct tcp_sock *tp)
{
	return tp->snd_ssthresh >= TCP_INFINITE_SSTHRESH;
}

1075 1076 1077 1078 1079 1080
static inline bool tcp_in_cwnd_reduction(const struct sock *sk)
{
	return (TCPF_CA_CWR | TCPF_CA_Recovery) &
	       (1 << inet_csk(sk)->icsk_ca_state);
}

L
Linus Torvalds 已提交
1081
/* If cwnd > ssthresh, we may raise ssthresh to be half-way to cwnd.
1082
 * The exception is cwnd reduction phase, when cwnd is decreasing towards
L
Linus Torvalds 已提交
1083 1084
 * ssthresh.
 */
1085
static inline __u32 tcp_current_ssthresh(const struct sock *sk)
L
Linus Torvalds 已提交
1086
{
1087
	const struct tcp_sock *tp = tcp_sk(sk);
1088

1089
	if (tcp_in_cwnd_reduction(sk))
L
Linus Torvalds 已提交
1090 1091 1092 1093 1094 1095 1096
		return tp->snd_ssthresh;
	else
		return max(tp->snd_ssthresh,
			   ((tp->snd_cwnd >> 1) +
			    (tp->snd_cwnd >> 2)));
}

1097 1098
/* Use define here intentionally to get WARN_ON location shown at the caller */
#define tcp_verify_left_out(tp)	WARN_ON(tcp_left_out(tp) > tp->packets_out)
L
Linus Torvalds 已提交
1099

1100
void tcp_enter_cwr(struct sock *sk);
1101
__u32 tcp_init_cwnd(const struct tcp_sock *tp, const struct dst_entry *dst);
L
Linus Torvalds 已提交
1102

1103 1104 1105 1106 1107 1108 1109 1110
/* The maximum number of MSS of available cwnd for which TSO defers
 * sending if not using sysctl_tcp_tso_win_divisor.
 */
static inline __u32 tcp_max_tso_deferred_mss(const struct tcp_sock *tp)
{
	return 3;
}

1111 1112 1113 1114 1115
/* Returns end sequence number of the receiver's advertised window */
static inline u32 tcp_wnd_end(const struct tcp_sock *tp)
{
	return tp->snd_una + tp->snd_wnd;
}
1116 1117 1118

/* We follow the spirit of RFC2861 to validate cwnd but implement a more
 * flexible approach. The RFC suggests cwnd should not be raised unless
1119 1120 1121
 * it was fully used previously. And that's exactly what we do in
 * congestion avoidance mode. But in slow start we allow cwnd to grow
 * as long as the application has used half the cwnd.
1122 1123 1124 1125 1126 1127 1128 1129
 * Example :
 *    cwnd is 10 (IW10), but application sends 9 frames.
 *    We allow cwnd to reach 18 when all frames are ACKed.
 * This check is safe because it's as aggressive as slow start which already
 * risks 100% overshoot. The advantage is that we discourage application to
 * either send more filler packets or data to artificially blow up the cwnd
 * usage, and allow application-limited process to probe bw more aggressively.
 */
1130
static inline bool tcp_is_cwnd_limited(const struct sock *sk)
1131 1132 1133
{
	const struct tcp_sock *tp = tcp_sk(sk);

1134
	/* If in slow start, ensure cwnd grows to twice what was ACKed. */
1135
	if (tcp_in_slow_start(tp))
1136 1137 1138
		return tp->snd_cwnd < 2 * tp->max_packets_out;

	return tp->is_cwnd_limited;
1139
}
1140

1141 1142 1143 1144 1145 1146 1147
/* Something is really bad, we could not queue an additional packet,
 * because qdisc is full or receiver sent a 0 window.
 * We do not want to add fuel to the fire, or abort too early,
 * so make sure the timer we arm now is at least 200ms in the future,
 * regardless of current icsk_rto value (as it could be ~2ms)
 */
static inline unsigned long tcp_probe0_base(const struct sock *sk)
L
Linus Torvalds 已提交
1148
{
1149 1150
	return max_t(unsigned long, inet_csk(sk)->icsk_rto, TCP_RTO_MIN);
}
1151

1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163
/* Variant of inet_csk_rto_backoff() used for zero window probes */
static inline unsigned long tcp_probe0_when(const struct sock *sk,
					    unsigned long max_when)
{
	u64 when = (u64)tcp_probe0_base(sk) << inet_csk(sk)->icsk_backoff;

	return (unsigned long)min_t(u64, when, max_when);
}

static inline void tcp_check_probe_timer(struct sock *sk)
{
	if (!tcp_sk(sk)->packets_out && !inet_csk(sk)->icsk_pending)
1164
		inet_csk_reset_xmit_timer(sk, ICSK_TIME_PROBE0,
1165
					  tcp_probe0_base(sk), TCP_RTO_MAX);
L
Linus Torvalds 已提交
1166 1167
}

1168
static inline void tcp_init_wl(struct tcp_sock *tp, u32 seq)
L
Linus Torvalds 已提交
1169 1170 1171 1172
{
	tp->snd_wl1 = seq;
}

1173
static inline void tcp_update_wl(struct tcp_sock *tp, u32 seq)
L
Linus Torvalds 已提交
1174 1175 1176 1177 1178 1179 1180
{
	tp->snd_wl1 = seq;
}

/*
 * Calculate(/check) TCP checksum
 */
1181 1182
static inline __sum16 tcp_v4_check(int len, __be32 saddr,
				   __be32 daddr, __wsum base)
L
Linus Torvalds 已提交
1183 1184 1185 1186
{
	return csum_tcpudp_magic(saddr,daddr,len,IPPROTO_TCP,base);
}

1187
static inline __sum16 __tcp_checksum_complete(struct sk_buff *skb)
L
Linus Torvalds 已提交
1188
{
1189
	return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1190 1191
}

E
Eric Dumazet 已提交
1192
static inline bool tcp_checksum_complete(struct sk_buff *skb)
L
Linus Torvalds 已提交
1193
{
1194
	return !skb_csum_unnecessary(skb) &&
L
Linus Torvalds 已提交
1195 1196 1197 1198 1199
		__tcp_checksum_complete(skb);
}

/* Prequeue for VJ style copy to user, combined with checksumming. */

S
Stephen Hemminger 已提交
1200
static inline void tcp_prequeue_init(struct tcp_sock *tp)
L
Linus Torvalds 已提交
1201 1202 1203 1204 1205 1206 1207
{
	tp->ucopy.task = NULL;
	tp->ucopy.len = 0;
	tp->ucopy.memory = 0;
	skb_queue_head_init(&tp->ucopy.prequeue);
}

1208
bool tcp_prequeue(struct sock *sk, struct sk_buff *skb);
E
Eric Dumazet 已提交
1209
bool tcp_add_backlog(struct sock *sk, struct sk_buff *skb);
1210
int tcp_filter(struct sock *sk, struct sk_buff *skb);
L
Linus Torvalds 已提交
1211 1212 1213 1214 1215 1216 1217 1218 1219 1220

#undef STATE_TRACE

#ifdef STATE_TRACE
static const char *statename[]={
	"Unused","Established","Syn Sent","Syn Recv",
	"Fin Wait 1","Fin Wait 2","Time Wait", "Close",
	"Close Wait","Last ACK","Listen","Closing"
};
#endif
1221
void tcp_set_state(struct sock *sk, int state);
L
Linus Torvalds 已提交
1222

1223
void tcp_done(struct sock *sk);
L
Linus Torvalds 已提交
1224

1225 1226
int tcp_abort(struct sock *sk, int err);

S
Stephen Hemminger 已提交
1227
static inline void tcp_sack_reset(struct tcp_options_received *rx_opt)
L
Linus Torvalds 已提交
1228 1229 1230 1231 1232
{
	rx_opt->dsack = 0;
	rx_opt->num_sacks = 0;
}

1233
u32 tcp_default_init_rwnd(u32 mss);
1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246
void tcp_cwnd_restart(struct sock *sk, s32 delta);

static inline void tcp_slow_start_after_idle_check(struct sock *sk)
{
	struct tcp_sock *tp = tcp_sk(sk);
	s32 delta;

	if (!sysctl_tcp_slow_start_after_idle || tp->packets_out)
		return;
	delta = tcp_time_stamp - tp->lsndtime;
	if (delta > inet_csk(sk)->icsk_rto)
		tcp_cwnd_restart(sk, delta);
}
1247

L
Linus Torvalds 已提交
1248
/* Determine a window scaling and initial window to offer. */
1249 1250 1251
void tcp_select_initial_window(int __space, __u32 mss, __u32 *rcv_wnd,
			       __u32 *window_clamp, int wscale_ok,
			       __u8 *rcv_wscale, __u32 init_rcv_wnd);
L
Linus Torvalds 已提交
1252 1253 1254 1255 1256 1257 1258 1259

static inline int tcp_win_from_space(int space)
{
	return sysctl_tcp_adv_win_scale<=0 ?
		(space>>(-sysctl_tcp_adv_win_scale)) :
		space - (space>>sysctl_tcp_adv_win_scale);
}

1260
/* Note: caller must be prepared to deal with negative returns */
L
Linus Torvalds 已提交
1261 1262 1263 1264
static inline int tcp_space(const struct sock *sk)
{
	return tcp_win_from_space(sk->sk_rcvbuf -
				  atomic_read(&sk->sk_rmem_alloc));
1265
}
L
Linus Torvalds 已提交
1266 1267 1268

static inline int tcp_full_space(const struct sock *sk)
{
1269
	return tcp_win_from_space(sk->sk_rcvbuf);
L
Linus Torvalds 已提交
1270 1271
}

1272
extern void tcp_openreq_init_rwin(struct request_sock *req,
1273 1274
				  const struct sock *sk_listener,
				  const struct dst_entry *dst);
1275

1276
void tcp_enter_memory_pressure(struct sock *sk);
L
Linus Torvalds 已提交
1277 1278 1279

static inline int keepalive_intvl_when(const struct tcp_sock *tp)
{
1280 1281 1282
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_intvl ? : net->ipv4.sysctl_tcp_keepalive_intvl;
L
Linus Torvalds 已提交
1283 1284 1285 1286
}

static inline int keepalive_time_when(const struct tcp_sock *tp)
{
1287 1288 1289
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_time ? : net->ipv4.sysctl_tcp_keepalive_time;
L
Linus Torvalds 已提交
1290 1291
}

E
Eric Dumazet 已提交
1292 1293
static inline int keepalive_probes(const struct tcp_sock *tp)
{
1294 1295 1296
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_probes ? : net->ipv4.sysctl_tcp_keepalive_probes;
E
Eric Dumazet 已提交
1297 1298
}

1299 1300 1301 1302 1303 1304 1305 1306
static inline u32 keepalive_time_elapsed(const struct tcp_sock *tp)
{
	const struct inet_connection_sock *icsk = &tp->inet_conn;

	return min_t(u32, tcp_time_stamp - icsk->icsk_ack.lrcvtime,
			  tcp_time_stamp - tp->rcv_tstamp);
}

1307
static inline int tcp_fin_time(const struct sock *sk)
L
Linus Torvalds 已提交
1308
{
1309
	int fin_timeout = tcp_sk(sk)->linger2 ? : sock_net(sk)->ipv4.sysctl_tcp_fin_timeout;
1310
	const int rto = inet_csk(sk)->icsk_rto;
L
Linus Torvalds 已提交
1311

1312 1313
	if (fin_timeout < (rto << 2) - (rto >> 1))
		fin_timeout = (rto << 2) - (rto >> 1);
L
Linus Torvalds 已提交
1314 1315 1316 1317

	return fin_timeout;
}

E
Eric Dumazet 已提交
1318 1319
static inline bool tcp_paws_check(const struct tcp_options_received *rx_opt,
				  int paws_win)
L
Linus Torvalds 已提交
1320
{
I
Ilpo Järvinen 已提交
1321
	if ((s32)(rx_opt->ts_recent - rx_opt->rcv_tsval) <= paws_win)
E
Eric Dumazet 已提交
1322
		return true;
I
Ilpo Järvinen 已提交
1323
	if (unlikely(get_seconds() >= rx_opt->ts_recent_stamp + TCP_PAWS_24DAYS))
E
Eric Dumazet 已提交
1324
		return true;
E
Eric Dumazet 已提交
1325 1326 1327 1328 1329 1330
	/*
	 * Some OSes send SYN and SYNACK messages with tsval=0 tsecr=0,
	 * then following tcp messages have valid values. Ignore 0 value,
	 * or else 'negative' tsval might forbid us to accept their packets.
	 */
	if (!rx_opt->ts_recent)
E
Eric Dumazet 已提交
1331 1332
		return true;
	return false;
I
Ilpo Järvinen 已提交
1333 1334
}

E
Eric Dumazet 已提交
1335 1336
static inline bool tcp_paws_reject(const struct tcp_options_received *rx_opt,
				   int rst)
I
Ilpo Järvinen 已提交
1337 1338
{
	if (tcp_paws_check(rx_opt, 0))
E
Eric Dumazet 已提交
1339
		return false;
L
Linus Torvalds 已提交
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352

	/* RST segments are not recommended to carry timestamp,
	   and, if they do, it is recommended to ignore PAWS because
	   "their cleanup function should take precedence over timestamps."
	   Certainly, it is mistake. It is necessary to understand the reasons
	   of this constraint to relax it: if peer reboots, clock may go
	   out-of-sync and half-open connections will not be reset.
	   Actually, the problem would be not existing if all
	   the implementations followed draft about maintaining clock
	   via reboots. Linux-2.2 DOES NOT!

	   However, we can relax time bounds for RST segments to MSL.
	 */
1353
	if (rst && get_seconds() >= rx_opt->ts_recent_stamp + TCP_PAWS_MSL)
E
Eric Dumazet 已提交
1354 1355
		return false;
	return true;
L
Linus Torvalds 已提交
1356 1357
}

1358 1359
bool tcp_oow_rate_limited(struct net *net, const struct sk_buff *skb,
			  int mib_idx, u32 *last_oow_ack_time);
1360

P
Pavel Emelyanov 已提交
1361
static inline void tcp_mib_init(struct net *net)
L
Linus Torvalds 已提交
1362 1363
{
	/* See RFC 2012 */
1364 1365 1366 1367
	TCP_ADD_STATS(net, TCP_MIB_RTOALGORITHM, 1);
	TCP_ADD_STATS(net, TCP_MIB_RTOMIN, TCP_RTO_MIN*1000/HZ);
	TCP_ADD_STATS(net, TCP_MIB_RTOMAX, TCP_RTO_MAX*1000/HZ);
	TCP_ADD_STATS(net, TCP_MIB_MAXCONN, -1);
L
Linus Torvalds 已提交
1368 1369
}

1370
/* from STCP */
1371
static inline void tcp_clear_retrans_hints_partial(struct tcp_sock *tp)
D
David S. Miller 已提交
1372
{
1373
	tp->lost_skb_hint = NULL;
1374 1375 1376 1377 1378
}

static inline void tcp_clear_all_retrans_hints(struct tcp_sock *tp)
{
	tcp_clear_retrans_hints_partial(tp);
1379
	tp->retransmit_skb_hint = NULL;
1380 1381
}

E
Eric Dumazet 已提交
1382 1383 1384 1385 1386 1387 1388
union tcp_md5_addr {
	struct in_addr  a4;
#if IS_ENABLED(CONFIG_IPV6)
	struct in6_addr	a6;
#endif
};

1389 1390
/* - key database */
struct tcp_md5sig_key {
E
Eric Dumazet 已提交
1391
	struct hlist_node	node;
1392
	u8			keylen;
E
Eric Dumazet 已提交
1393 1394 1395 1396
	u8			family; /* AF_INET or AF_INET6 */
	union tcp_md5_addr	addr;
	u8			key[TCP_MD5SIG_MAXKEYLEN];
	struct rcu_head		rcu;
1397 1398 1399 1400
};

/* - sock block */
struct tcp_md5sig_info {
E
Eric Dumazet 已提交
1401
	struct hlist_head	head;
1402
	struct rcu_head		rcu;
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422
};

/* - pseudo header */
struct tcp4_pseudohdr {
	__be32		saddr;
	__be32		daddr;
	__u8		pad;
	__u8		protocol;
	__be16		len;
};

struct tcp6_pseudohdr {
	struct in6_addr	saddr;
	struct in6_addr daddr;
	__be32		len;
	__be32		protocol;	/* including padding */
};

union tcp_md5sum_block {
	struct tcp4_pseudohdr ip4;
E
Eric Dumazet 已提交
1423
#if IS_ENABLED(CONFIG_IPV6)
1424 1425 1426 1427 1428 1429
	struct tcp6_pseudohdr ip6;
#endif
};

/* - pool: digest algorithm, hash description and scratch buffer */
struct tcp_md5sig_pool {
H
Herbert Xu 已提交
1430
	struct ahash_request	*md5_req;
1431
	void			*scratch;
1432 1433 1434
};

/* - functions */
1435 1436
int tcp_v4_md5_hash_skb(char *md5_hash, const struct tcp_md5sig_key *key,
			const struct sock *sk, const struct sk_buff *skb);
1437 1438 1439 1440
int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
		   int family, const u8 *newkey, u8 newkeylen, gfp_t gfp);
int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr,
		   int family);
1441
struct tcp_md5sig_key *tcp_v4_md5_lookup(const struct sock *sk,
1442
					 const struct sock *addr_sk);
1443

1444
#ifdef CONFIG_TCP_MD5SIG
1445
struct tcp_md5sig_key *tcp_md5_do_lookup(const struct sock *sk,
1446 1447
					 const union tcp_md5_addr *addr,
					 int family);
E
Eric Dumazet 已提交
1448
#define tcp_twsk_md5_key(twsk)	((twsk)->tw_md5_key)
1449
#else
1450
static inline struct tcp_md5sig_key *tcp_md5_do_lookup(const struct sock *sk,
E
Eric Dumazet 已提交
1451 1452 1453 1454 1455
					 const union tcp_md5_addr *addr,
					 int family)
{
	return NULL;
}
1456 1457 1458
#define tcp_twsk_md5_key(twsk)	NULL
#endif

1459
bool tcp_alloc_md5sig_pool(void);
1460

1461
struct tcp_md5sig_pool *tcp_get_md5sig_pool(void);
1462 1463 1464 1465
static inline void tcp_put_md5sig_pool(void)
{
	local_bh_enable();
}
E
Eric Dumazet 已提交
1466

1467 1468 1469 1470
int tcp_md5_hash_skb_data(struct tcp_md5sig_pool *, const struct sk_buff *,
			  unsigned int header_len);
int tcp_md5_hash_key(struct tcp_md5sig_pool *hp,
		     const struct tcp_md5sig_key *key);
1471

1472
/* From tcp_fastopen.c */
1473 1474 1475 1476
void tcp_fastopen_cache_get(struct sock *sk, u16 *mss,
			    struct tcp_fastopen_cookie *cookie, int *syn_loss,
			    unsigned long *last_syn_loss);
void tcp_fastopen_cache_set(struct sock *sk, u16 mss,
1477 1478
			    struct tcp_fastopen_cookie *cookie, bool syn_lost,
			    u16 try_exp);
1479 1480 1481 1482
struct tcp_fastopen_request {
	/* Fast Open cookie. Size 0 means a cookie request */
	struct tcp_fastopen_cookie	cookie;
	struct msghdr			*data;  /* data in MSG_FASTOPEN */
1483 1484
	size_t				size;
	int				copied;	/* queued in tcp_connect() */
1485 1486 1487
};
void tcp_free_fastopen_req(struct tcp_sock *tp);

1488 1489
extern struct tcp_fastopen_context __rcu *tcp_fastopen_ctx;
int tcp_fastopen_reset_cipher(void *key, unsigned int len);
1490
void tcp_fastopen_add_skb(struct sock *sk, struct sk_buff *skb);
1491 1492 1493 1494
struct sock *tcp_try_fastopen(struct sock *sk, struct sk_buff *skb,
			      struct request_sock *req,
			      struct tcp_fastopen_cookie *foc,
			      struct dst_entry *dst);
1495
void tcp_fastopen_init_key_once(bool publish);
1496 1497
bool tcp_fastopen_cookie_check(struct sock *sk, u16 *mss,
			     struct tcp_fastopen_cookie *cookie);
W
Wei Wang 已提交
1498
bool tcp_fastopen_defer_connect(struct sock *sk, int *err);
1499 1500 1501 1502
#define TCP_FASTOPEN_KEY_LENGTH 16

/* Fastopen key context */
struct tcp_fastopen_context {
1503 1504 1505
	struct crypto_cipher	*tfm;
	__u8			key[TCP_FASTOPEN_KEY_LENGTH];
	struct rcu_head		rcu;
1506 1507
};

1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521
/* Latencies incurred by various limits for a sender. They are
 * chronograph-like stats that are mutually exclusive.
 */
enum tcp_chrono {
	TCP_CHRONO_UNSPEC,
	TCP_CHRONO_BUSY, /* Actively sending data (non-empty write queue) */
	TCP_CHRONO_RWND_LIMITED, /* Stalled by insufficient receive window */
	TCP_CHRONO_SNDBUF_LIMITED, /* Stalled by insufficient send buffer */
	__TCP_CHRONO_MAX,
};

void tcp_chrono_start(struct sock *sk, const enum tcp_chrono type);
void tcp_chrono_stop(struct sock *sk, const enum tcp_chrono type);

1522 1523 1524 1525 1526
/* write queue abstraction */
static inline void tcp_write_queue_purge(struct sock *sk)
{
	struct sk_buff *skb;

1527
	tcp_chrono_stop(sk, TCP_CHRONO_BUSY);
1528
	while ((skb = __skb_dequeue(&sk->sk_write_queue)) != NULL)
1529 1530
		sk_wmem_free_skb(sk, skb);
	sk_mem_reclaim(sk);
1531
	tcp_clear_all_retrans_hints(tcp_sk(sk));
1532 1533
}

1534
static inline struct sk_buff *tcp_write_queue_head(const struct sock *sk)
1535
{
1536
	return skb_peek(&sk->sk_write_queue);
1537 1538
}

1539
static inline struct sk_buff *tcp_write_queue_tail(const struct sock *sk)
1540
{
1541
	return skb_peek_tail(&sk->sk_write_queue);
1542 1543
}

1544 1545
static inline struct sk_buff *tcp_write_queue_next(const struct sock *sk,
						   const struct sk_buff *skb)
1546
{
1547
	return skb_queue_next(&sk->sk_write_queue, skb);
1548 1549
}

1550 1551
static inline struct sk_buff *tcp_write_queue_prev(const struct sock *sk,
						   const struct sk_buff *skb)
1552 1553 1554 1555
{
	return skb_queue_prev(&sk->sk_write_queue, skb);
}

1556
#define tcp_for_write_queue(skb, sk)					\
1557
	skb_queue_walk(&(sk)->sk_write_queue, skb)
1558 1559

#define tcp_for_write_queue_from(skb, sk)				\
1560
	skb_queue_walk_from(&(sk)->sk_write_queue, skb)
1561

1562
#define tcp_for_write_queue_from_safe(skb, tmp, sk)			\
1563
	skb_queue_walk_from_safe(&(sk)->sk_write_queue, skb, tmp)
1564

1565
static inline struct sk_buff *tcp_send_head(const struct sock *sk)
1566 1567 1568 1569
{
	return sk->sk_send_head;
}

1570 1571 1572 1573 1574 1575
static inline bool tcp_skb_is_last(const struct sock *sk,
				   const struct sk_buff *skb)
{
	return skb_queue_is_last(&sk->sk_write_queue, skb);
}

1576
static inline void tcp_advance_send_head(struct sock *sk, const struct sk_buff *skb)
1577
{
1578
	if (tcp_skb_is_last(sk, skb))
1579
		sk->sk_send_head = NULL;
1580 1581
	else
		sk->sk_send_head = tcp_write_queue_next(sk, skb);
1582 1583 1584 1585
}

static inline void tcp_check_send_head(struct sock *sk, struct sk_buff *skb_unlinked)
{
1586
	if (sk->sk_send_head == skb_unlinked) {
1587
		sk->sk_send_head = NULL;
1588 1589
		tcp_chrono_stop(sk, TCP_CHRONO_BUSY);
	}
1590 1591
	if (tcp_sk(sk)->highest_sack == skb_unlinked)
		tcp_sk(sk)->highest_sack = NULL;
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
}

static inline void tcp_init_send_head(struct sock *sk)
{
	sk->sk_send_head = NULL;
}

static inline void __tcp_add_write_queue_tail(struct sock *sk, struct sk_buff *skb)
{
	__skb_queue_tail(&sk->sk_write_queue, skb);
}

static inline void tcp_add_write_queue_tail(struct sock *sk, struct sk_buff *skb)
{
	__tcp_add_write_queue_tail(sk, skb);

	/* Queue it, remembering where we must start sending. */
1609
	if (sk->sk_send_head == NULL) {
1610
		sk->sk_send_head = skb;
1611
		tcp_chrono_start(sk, TCP_CHRONO_BUSY);
1612 1613 1614 1615

		if (tcp_sk(sk)->highest_sack == NULL)
			tcp_sk(sk)->highest_sack = skb;
	}
1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627
}

static inline void __tcp_add_write_queue_head(struct sock *sk, struct sk_buff *skb)
{
	__skb_queue_head(&sk->sk_write_queue, skb);
}

/* Insert buff after skb on the write queue of sk.  */
static inline void tcp_insert_write_queue_after(struct sk_buff *skb,
						struct sk_buff *buff,
						struct sock *sk)
{
1628
	__skb_queue_after(&sk->sk_write_queue, skb, buff);
1629 1630
}

1631
/* Insert new before skb on the write queue of sk.  */
1632 1633 1634 1635
static inline void tcp_insert_write_queue_before(struct sk_buff *new,
						  struct sk_buff *skb,
						  struct sock *sk)
{
1636
	__skb_queue_before(&sk->sk_write_queue, skb, new);
1637 1638 1639

	if (sk->sk_send_head == skb)
		sk->sk_send_head = new;
1640 1641 1642 1643 1644 1645 1646
}

static inline void tcp_unlink_write_queue(struct sk_buff *skb, struct sock *sk)
{
	__skb_unlink(skb, &sk->sk_write_queue);
}

E
Eric Dumazet 已提交
1647
static inline bool tcp_write_queue_empty(struct sock *sk)
1648 1649 1650 1651
{
	return skb_queue_empty(&sk->sk_write_queue);
}

1652 1653 1654 1655 1656 1657 1658 1659 1660
static inline void tcp_push_pending_frames(struct sock *sk)
{
	if (tcp_send_head(sk)) {
		struct tcp_sock *tp = tcp_sk(sk);

		__tcp_push_pending_frames(sk, tcp_current_mss(sk), tp->nonagle);
	}
}

1661 1662 1663
/* Start sequence of the skb just after the highest skb with SACKed
 * bit, valid only if sacked_out > 0 or when the caller has ensured
 * validity by itself.
1664 1665 1666 1667 1668
 */
static inline u32 tcp_highest_sack_seq(struct tcp_sock *tp)
{
	if (!tp->sacked_out)
		return tp->snd_una;
1669 1670 1671 1672

	if (tp->highest_sack == NULL)
		return tp->snd_nxt;

1673 1674 1675
	return TCP_SKB_CB(tp->highest_sack)->seq;
}

1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700
static inline void tcp_advance_highest_sack(struct sock *sk, struct sk_buff *skb)
{
	tcp_sk(sk)->highest_sack = tcp_skb_is_last(sk, skb) ? NULL :
						tcp_write_queue_next(sk, skb);
}

static inline struct sk_buff *tcp_highest_sack(struct sock *sk)
{
	return tcp_sk(sk)->highest_sack;
}

static inline void tcp_highest_sack_reset(struct sock *sk)
{
	tcp_sk(sk)->highest_sack = tcp_write_queue_head(sk);
}

/* Called when old skb is about to be deleted (to be combined with new skb) */
static inline void tcp_highest_sack_combine(struct sock *sk,
					    struct sk_buff *old,
					    struct sk_buff *new)
{
	if (tcp_sk(sk)->sacked_out && (old == tcp_sk(sk)->highest_sack))
		tcp_sk(sk)->highest_sack = new;
}

1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712
/* This helper checks if socket has IP_TRANSPARENT set */
static inline bool inet_sk_transparent(const struct sock *sk)
{
	switch (sk->sk_state) {
	case TCP_TIME_WAIT:
		return inet_twsk(sk)->tw_transparent;
	case TCP_NEW_SYN_RECV:
		return inet_rsk(inet_reqsk(sk))->no_srccheck;
	}
	return inet_sk(sk)->transparent;
}

A
Andreas Petlund 已提交
1713 1714 1715
/* Determines whether this is a thin stream (which may suffer from
 * increased latency). Used to trigger latency-reducing mechanisms.
 */
E
Eric Dumazet 已提交
1716
static inline bool tcp_stream_is_thin(struct tcp_sock *tp)
A
Andreas Petlund 已提交
1717 1718 1719 1720
{
	return tp->packets_out < 4 && !tcp_in_initial_slowstart(tp);
}

L
Linus Torvalds 已提交
1721 1722 1723 1724 1725 1726
/* /proc */
enum tcp_seq_states {
	TCP_SEQ_STATE_LISTENING,
	TCP_SEQ_STATE_ESTABLISHED,
};

1727 1728
int tcp_seq_open(struct inode *inode, struct file *file);

L
Linus Torvalds 已提交
1729
struct tcp_seq_afinfo {
1730 1731 1732 1733
	char				*name;
	sa_family_t			family;
	const struct file_operations	*seq_fops;
	struct seq_operations		seq_ops;
L
Linus Torvalds 已提交
1734 1735 1736
};

struct tcp_iter_state {
1737
	struct seq_net_private	p;
L
Linus Torvalds 已提交
1738 1739 1740
	sa_family_t		family;
	enum tcp_seq_states	state;
	struct sock		*syn_wait_sk;
1741
	int			bucket, offset, sbucket, num;
1742
	loff_t			last_pos;
L
Linus Torvalds 已提交
1743 1744
};

1745 1746
int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo);
void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo);
L
Linus Torvalds 已提交
1747

1748
extern struct request_sock_ops tcp_request_sock_ops;
1749
extern struct request_sock_ops tcp6_request_sock_ops;
1750

1751
void tcp_v4_destroy_sock(struct sock *sk);
1752

E
Eric Dumazet 已提交
1753
struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
1754 1755 1756
				netdev_features_t features);
struct sk_buff **tcp_gro_receive(struct sk_buff **head, struct sk_buff *skb);
int tcp_gro_complete(struct sk_buff *skb);
1757

1758
void __tcp_v4_send_check(struct sk_buff *skb, __be32 saddr, __be32 daddr);
H
Herbert Xu 已提交
1759

1760 1761
static inline u32 tcp_notsent_lowat(const struct tcp_sock *tp)
{
1762 1763
	struct net *net = sock_net((struct sock *)tp);
	return tp->notsent_lowat ?: net->ipv4.sysctl_tcp_notsent_lowat;
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773
}

static inline bool tcp_stream_memory_free(const struct sock *sk)
{
	const struct tcp_sock *tp = tcp_sk(sk);
	u32 notsent_bytes = tp->write_seq - tp->snd_nxt;

	return notsent_bytes < tcp_notsent_lowat(tp);
}

1774
#ifdef CONFIG_PROC_FS
1775 1776
int tcp4_proc_init(void);
void tcp4_proc_exit(void);
1777 1778
#endif

1779
int tcp_rtx_synack(const struct sock *sk, struct request_sock *req);
O
Octavian Purdila 已提交
1780 1781 1782
int tcp_conn_request(struct request_sock_ops *rsk_ops,
		     const struct tcp_request_sock_ops *af_ops,
		     struct sock *sk, struct sk_buff *skb);
1783

1784 1785 1786
/* TCP af-specific functions */
struct tcp_sock_af_ops {
#ifdef CONFIG_TCP_MD5SIG
1787
	struct tcp_md5sig_key	*(*md5_lookup) (const struct sock *sk,
1788
						const struct sock *addr_sk);
1789 1790 1791 1792 1793 1794 1795
	int		(*calc_md5_hash)(char *location,
					 const struct tcp_md5sig_key *md5,
					 const struct sock *sk,
					 const struct sk_buff *skb);
	int		(*md5_parse)(struct sock *sk,
				     char __user *optval,
				     int optlen);
1796 1797 1798 1799
#endif
};

struct tcp_request_sock_ops {
1800
	u16 mss_clamp;
1801
#ifdef CONFIG_TCP_MD5SIG
1802
	struct tcp_md5sig_key *(*req_md5_lookup)(const struct sock *sk,
1803
						 const struct sock *addr_sk);
1804 1805 1806 1807
	int		(*calc_md5_hash) (char *location,
					  const struct tcp_md5sig_key *md5,
					  const struct sock *sk,
					  const struct sk_buff *skb);
1808
#endif
1809 1810
	void (*init_req)(struct request_sock *req,
			 const struct sock *sk_listener,
1811
			 struct sk_buff *skb);
1812
#ifdef CONFIG_SYN_COOKIES
1813
	__u32 (*cookie_init_seq)(const struct sk_buff *skb,
1814 1815
				 __u16 *mss);
#endif
1816
	struct dst_entry *(*route_req)(const struct sock *sk, struct flowi *fl,
1817 1818
				       const struct request_sock *req,
				       bool *strict);
1819
	__u32 (*init_seq)(const struct sk_buff *skb, u32 *tsoff);
1820
	int (*send_synack)(const struct sock *sk, struct dst_entry *dst,
1821
			   struct flowi *fl, struct request_sock *req,
1822
			   struct tcp_fastopen_cookie *foc,
1823
			   enum tcp_synack_type synack_type);
1824 1825
};

1826 1827
#ifdef CONFIG_SYN_COOKIES
static inline __u32 cookie_init_sequence(const struct tcp_request_sock_ops *ops,
1828
					 const struct sock *sk, struct sk_buff *skb,
1829 1830
					 __u16 *mss)
{
1831
	tcp_synq_overflow(sk);
1832
	__NET_INC_STATS(sock_net(sk), LINUX_MIB_SYNCOOKIESSENT);
1833
	return ops->cookie_init_seq(skb, mss);
1834 1835 1836
}
#else
static inline __u32 cookie_init_sequence(const struct tcp_request_sock_ops *ops,
1837
					 const struct sock *sk, struct sk_buff *skb,
1838 1839 1840 1841 1842 1843
					 __u16 *mss)
{
	return 0;
}
#endif

1844
int tcpv4_offload_init(void);
1845

1846 1847
void tcp_v4_init(void);
void tcp_init(void);
1848

1849
/* tcp_recovery.c */
1850
extern void tcp_rack_mark_lost(struct sock *sk, const struct skb_mstamp *now);
1851
extern void tcp_rack_advance(struct tcp_sock *tp, u8 sacked, u32 end_seq,
1852 1853
			     const struct skb_mstamp *xmit_time,
			     const struct skb_mstamp *ack_time);
1854
extern void tcp_rack_reo_timeout(struct sock *sk);
1855

1856 1857 1858 1859 1860 1861 1862 1863
/*
 * Save and compile IPv4 options, return a pointer to it
 */
static inline struct ip_options_rcu *tcp_v4_save_options(struct sk_buff *skb)
{
	const struct ip_options *opt = &TCP_SKB_CB(skb)->header.h4.opt;
	struct ip_options_rcu *dopt = NULL;

C
Cong Wang 已提交
1864
	if (opt->optlen) {
1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875
		int opt_size = sizeof(*dopt) + opt->optlen;

		dopt = kmalloc(opt_size, GFP_ATOMIC);
		if (dopt && __ip_options_echo(&dopt->opt, skb, opt)) {
			kfree(dopt);
			dopt = NULL;
		}
	}
	return dopt;
}

E
Eric Dumazet 已提交
1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890
/* locally generated TCP pure ACKs have skb->truesize == 2
 * (check tcp_send_ack() in net/ipv4/tcp_output.c )
 * This is much faster than dissecting the packet to find out.
 * (Think of GRE encapsulations, IPv4, IPv6, ...)
 */
static inline bool skb_is_tcp_pure_ack(const struct sk_buff *skb)
{
	return skb->truesize == 2;
}

static inline void skb_set_tcp_pure_ack(struct sk_buff *skb)
{
	skb->truesize = 2;
}

1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914
static inline int tcp_inq(struct sock *sk)
{
	struct tcp_sock *tp = tcp_sk(sk);
	int answ;

	if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) {
		answ = 0;
	} else if (sock_flag(sk, SOCK_URGINLINE) ||
		   !tp->urg_data ||
		   before(tp->urg_seq, tp->copied_seq) ||
		   !before(tp->urg_seq, tp->rcv_nxt)) {

		answ = tp->rcv_nxt - tp->copied_seq;

		/* Subtract 1, if FIN was received */
		if (answ && sock_flag(sk, SOCK_DONE))
			answ--;
	} else {
		answ = tp->urg_seq - tp->copied_seq;
	}

	return answ;
}

1915 1916
int tcp_peek_len(struct socket *sock);

1917 1918 1919 1920 1921 1922 1923 1924 1925 1926
static inline void tcp_segs_in(struct tcp_sock *tp, const struct sk_buff *skb)
{
	u16 segs_in;

	segs_in = max_t(u16, 1, skb_shinfo(skb)->gso_segs);
	tp->segs_in += segs_in;
	if (skb->len > tcp_hdrlen(skb))
		tp->data_segs_in += segs_in;
}

1927 1928 1929 1930 1931 1932 1933 1934 1935 1936
/*
 * TCP listen path runs lockless.
 * We forced "struct sock" to be const qualified to make sure
 * we don't modify one of its field by mistake.
 * Here, we increment sk_drops which is an atomic_t, so we can safely
 * make sock writable again.
 */
static inline void tcp_listendrop(const struct sock *sk)
{
	atomic_inc(&((struct sock *)sk)->sk_drops);
1937
	__NET_INC_STATS(sock_net(sk), LINUX_MIB_LISTENDROPS);
1938 1939
}

L
Linus Torvalds 已提交
1940
#endif	/* _TCP_H */