tcp.h 55.7 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
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171

#define TCP_RESOURCE_PROBE_INTERVAL ((unsigned)(HZ/2U)) /* Maximal interval between probes
					                 * for local resources.
					                 */

#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
 */
172

L
Linus Torvalds 已提交
173 174 175 176 177 178 179
#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 */
180
#define TCPOPT_MD5SIG		19	/* MD5 Signature (RFC2385) */
181
#define TCPOPT_FASTOPEN		34	/* Fast open (RFC7413) */
Y
Yuchung Cheng 已提交
182 183 184 185 186
#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
187 188 189 190 191 192 193 194 195

/*
 *     TCP option lengths
 */

#define TCPOLEN_MSS            4
#define TCPOLEN_WINDOW         3
#define TCPOLEN_SACK_PERM      2
#define TCPOLEN_TIMESTAMP      10
196
#define TCPOLEN_MD5SIG         18
197
#define TCPOLEN_FASTOPEN_BASE  2
Y
Yuchung Cheng 已提交
198
#define TCPOLEN_EXP_FASTOPEN_BASE  4
L
Linus Torvalds 已提交
199 200 201 202 203 204 205 206

/* 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
207
#define TCPOLEN_MD5SIG_ALIGNED		20
A
Adam Langley 已提交
208
#define TCPOLEN_MSS_ALIGNED		4
L
Linus Torvalds 已提交
209 210 211 212

/* 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
213
#define TCP_NAGLE_PUSH		4	/* Cork is overridden for already queued data */
L
Linus Torvalds 已提交
214

A
Andreas Petlund 已提交
215 216 217
/* TCP thin-stream limits */
#define TCP_THIN_LINEAR_RETRIES 6       /* After 6 linear retries, do exp. backoff */

218
/* TCP initial congestion window as per rfc6928 */
219 220
#define TCP_INIT_CWND		10

221 222
/* Bit Flags for sysctl_tcp_fastopen */
#define	TFO_CLIENT_ENABLE	1
223
#define	TFO_SERVER_ENABLE	2
224
#define	TFO_CLIENT_NO_COOKIE	4	/* Data in SYN w/o cookie option */
225

226 227 228 229 230 231 232 233 234
/* 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
 * TCP_FASTOPEN socket option. SOCKOPT1/2 determine how to set max_qlen.
 */
#define	TFO_SERVER_WO_SOCKOPT1	0x400
#define	TFO_SERVER_WO_SOCKOPT2	0x800

235 236
extern struct inet_timewait_death_row tcp_death_row;

L
Linus Torvalds 已提交
237 238 239 240
/* 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
241
extern int sysctl_tcp_fastopen;
L
Linus Torvalds 已提交
242 243 244 245 246 247 248
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;
249
extern int sysctl_tcp_max_reordering;
L
Linus Torvalds 已提交
250
extern int sysctl_tcp_dsack;
251
extern long sysctl_tcp_mem[3];
L
Linus Torvalds 已提交
252 253 254 255 256 257 258 259 260 261
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_tw_reuse;
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;
262
extern int sysctl_tcp_workaround_signed_windows;
263
extern int sysctl_tcp_slow_start_after_idle;
A
Andreas Petlund 已提交
264
extern int sysctl_tcp_thin_linear_timeouts;
A
Andreas Petlund 已提交
265
extern int sysctl_tcp_thin_dupack;
Y
Yuchung Cheng 已提交
266
extern int sysctl_tcp_early_retrans;
E
Eric Dumazet 已提交
267
extern int sysctl_tcp_limit_output_bytes;
E
Eric Dumazet 已提交
268
extern int sysctl_tcp_challenge_ack_limit;
E
Eric Dumazet 已提交
269
extern int sysctl_tcp_min_tso_segs;
270
extern int sysctl_tcp_min_rtt_wlen;
E
Eric Dumazet 已提交
271
extern int sysctl_tcp_autocorking;
272
extern int sysctl_tcp_invalid_ratelimit;
273 274
extern int sysctl_tcp_pacing_ss_ratio;
extern int sysctl_tcp_pacing_ca_ratio;
L
Linus Torvalds 已提交
275

E
Eric Dumazet 已提交
276
extern atomic_long_t tcp_memory_allocated;
277
extern struct percpu_counter tcp_sockets_allocated;
L
Linus Torvalds 已提交
278 279
extern int tcp_memory_pressure;

280 281 282
/* optimized version of sk_under_memory_pressure() for TCP sockets */
static inline bool tcp_under_memory_pressure(const struct sock *sk)
{
283 284
	if (mem_cgroup_sockets_enabled && sk->sk_memcg &&
	    mem_cgroup_under_socket_pressure(sk->sk_memcg))
285
		return true;
286 287 288

	return tcp_memory_pressure;
}
L
Linus Torvalds 已提交
289 290 291 292 293
/*
 * The next routines deal with comparing 32 bit unsigned ints
 * and worry about wraparound (automatic with unsigned arithmetic).
 */

E
Eric Dumazet 已提交
294
static inline bool before(__u32 seq1, __u32 seq2)
L
Linus Torvalds 已提交
295
{
296
        return (__s32)(seq1-seq2) < 0;
L
Linus Torvalds 已提交
297
}
298
#define after(seq2, seq1) 	before(seq1, seq2)
L
Linus Torvalds 已提交
299 300

/* is s2<=s1<=s3 ? */
E
Eric Dumazet 已提交
301
static inline bool between(__u32 seq1, __u32 seq2, __u32 seq3)
L
Linus Torvalds 已提交
302 303 304 305
{
	return seq3 - seq2 >= seq1 - seq2;
}

A
Arun Sharma 已提交
306 307 308 309 310 311 312 313
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;
}

314 315
void sk_forced_mem_schedule(struct sock *sk, int size);

316
static inline bool tcp_too_many_orphans(struct sock *sk, int shift)
317
{
318 319 320 321 322 323 324 325 326
	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;
327
}
L
Linus Torvalds 已提交
328

329
bool tcp_check_oom(struct sock *sk, int shift);
A
Arun Sharma 已提交
330

331

L
Linus Torvalds 已提交
332 333
extern struct proto tcp_prot;

334
#define TCP_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.tcp_statistics, field)
E
Eric Dumazet 已提交
335
#define __TCP_INC_STATS(net, field)	__SNMP_INC_STATS((net)->mib.tcp_statistics, field)
336
#define TCP_DEC_STATS(net, field)	SNMP_DEC_STATS((net)->mib.tcp_statistics, field)
337
#define TCP_ADD_STATS(net, field, val)	SNMP_ADD_STATS((net)->mib.tcp_statistics, field, val)
L
Linus Torvalds 已提交
338

339 340 341 342 343 344 345 346 347 348
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);
349
int tcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t size);
350 351 352 353 354 355 356
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);
357
int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb);
358 359 360 361 362 363 364 365
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
366

367 368
static inline void tcp_dec_quickack_mode(struct sock *sk,
					 const unsigned int pkts)
L
Linus Torvalds 已提交
369
{
370
	struct inet_connection_sock *icsk = inet_csk(sk);
371

372 373 374
	if (icsk->icsk_ack.quick) {
		if (pkts >= icsk->icsk_ack.quick) {
			icsk->icsk_ack.quick = 0;
375
			/* Leaving quickack mode we deflate ATO. */
376
			icsk->icsk_ack.ato   = TCP_ATO_MIN;
377
		} else
378
			icsk->icsk_ack.quick -= pkts;
L
Linus Torvalds 已提交
379 380 381
	}
}

382 383 384
#define	TCP_ECN_OK		1
#define	TCP_ECN_QUEUE_CWR	2
#define	TCP_ECN_DEMAND_CWR	4
385
#define	TCP_ECN_SEEN		8
386

E
Eric Dumazet 已提交
387
enum tcp_tw_status {
L
Linus Torvalds 已提交
388 389 390 391 392 393 394
	TCP_TW_SUCCESS = 0,
	TCP_TW_RST = 1,
	TCP_TW_ACK = 2,
	TCP_TW_SYN = 3
};


395 396 397 398
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,
399
			   struct request_sock *req, bool fastopen);
400 401
int tcp_child_process(struct sock *parent, struct sock *child,
		      struct sk_buff *skb);
402
void tcp_enter_loss(struct sock *sk);
403 404 405 406 407
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,
408
			bool paws_check, bool timestamps);
409 410 411 412 413 414 415 416 417 418 419 420 421
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
422
			  char __user *optval, int __user *optlen);
423
int compat_tcp_setsockopt(struct sock *sk, int level, int optname,
C
Changli Gao 已提交
424
			  char __user *optval, unsigned int optlen);
425
void tcp_set_keepalive(struct sock *sk, int val);
426
void tcp_syn_ack_timeout(const struct request_sock *req);
427 428
int tcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int nonblock,
		int flags, int *addr_len);
429 430 431 432
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);
433

L
Linus Torvalds 已提交
434 435 436 437
/*
 *	TCP v4 functions exported for the inet6 API
 */

438
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb);
439
void tcp_v4_mtu_reduced(struct sock *sk);
440
void tcp_req_err(struct sock *sk, u32 seq, bool abort);
441
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb);
442
struct sock *tcp_create_openreq_child(const struct sock *sk,
443 444
				      struct request_sock *req,
				      struct sk_buff *skb);
445
void tcp_ca_openreq_child(struct sock *sk, const struct dst_entry *dst);
446
struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
447
				  struct request_sock *req,
448 449 450
				  struct dst_entry *dst,
				  struct request_sock *req_unhash,
				  bool *own_req);
451 452 453
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);
454 455 456 457 458
enum tcp_synack_type {
	TCP_SYNACK_NORMAL,
	TCP_SYNACK_FASTOPEN,
	TCP_SYNACK_COOKIE,
};
459
struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
460
				struct request_sock *req,
461
				struct tcp_fastopen_cookie *foc,
462
				enum tcp_synack_type synack_type);
463
int tcp_disconnect(struct sock *sk, int flags);
L
Linus Torvalds 已提交
464

P
Pavel Emelyanov 已提交
465
void tcp_finish_connect(struct sock *sk, struct sk_buff *skb);
466
int tcp_send_rcvq(struct sock *sk, struct msghdr *msg, size_t size);
467
void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb);
L
Linus Torvalds 已提交
468 469

/* From syncookies.c */
470 471 472
struct sock *tcp_get_cookie_sock(struct sock *sk, struct sk_buff *skb,
				 struct request_sock *req,
				 struct dst_entry *dst);
473 474
int __cookie_v4_check(const struct iphdr *iph, const struct tcphdr *th,
		      u32 cookie);
C
Cong Wang 已提交
475
struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb);
476
#ifdef CONFIG_SYN_COOKIES
477

478
/* Syncookies use a monotonic timer which increments every 60 seconds.
479 480 481
 * 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,
482
 * i.e. a sent cookie is valid only at most for 2*60 seconds (or less if
483 484
 * the counter advances immediately after a cookie is generated).
 */
485 486 487 488 489 490
#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)
491
 * It is racy as we do not hold a lock, but race is very minor.
492
 */
493
static inline void tcp_synq_overflow(const struct sock *sk)
494 495 496 497 498 499 500 501 502 503 504 505 506 507 508
{
	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);
}
509 510 511

static inline u32 tcp_cookie_time(void)
{
512 513
	u64 val = get_jiffies_64();

514
	do_div(val, TCP_SYNCOOKIE_PERIOD);
515
	return val;
516 517
}

518 519
u32 __cookie_v4_init_sequence(const struct iphdr *iph, const struct tcphdr *th,
			      u16 *mssp);
520
__u32 cookie_v4_init_sequence(const struct sk_buff *skb, __u16 *mss);
521
__u32 cookie_init_timestamp(struct request_sock *req);
522 523
bool cookie_timestamp_decode(struct tcp_options_received *opt);
bool cookie_ecn_ok(const struct tcp_options_received *opt,
524
		   const struct net *net, const struct dst_entry *dst);
525

526
/* From net/ipv6/syncookies.c */
527 528 529
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);
530

531 532
u32 __cookie_v6_init_sequence(const struct ipv6hdr *iph,
			      const struct tcphdr *th, u16 *mssp);
533
__u32 cookie_v6_init_sequence(const struct sk_buff *skb, __u16 *mss);
534
#endif
L
Linus Torvalds 已提交
535 536
/* tcp_output.c */

537 538 539
void __tcp_push_pending_frames(struct sock *sk, unsigned int cur_mss,
			       int nonagle);
bool tcp_may_send_now(struct sock *sk);
540 541
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);
542 543 544 545
void tcp_retransmit_timer(struct sock *sk);
void tcp_xmit_retransmit_queue(struct sock *);
void tcp_simple_retransmit(struct sock *);
int tcp_trim_head(struct sock *, struct sk_buff *, u32);
546
int tcp_fragment(struct sock *, struct sk_buff *, u32, unsigned int, gfp_t);
547 548 549

void tcp_send_probe0(struct sock *);
void tcp_send_partial(struct sock *);
550
int tcp_write_wakeup(struct sock *, int mib);
551 552 553 554 555 556 557 558
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);
559 560
void tcp_skb_collapse_tstamp(struct sk_buff *skb,
			     const struct sk_buff *next_skb);
L
Linus Torvalds 已提交
561

562
/* tcp_input.c */
563 564
void tcp_resume_early_retransmit(struct sock *sk);
void tcp_rearm_rto(struct sock *sk);
Y
Yuchung Cheng 已提交
565
void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req);
566
void tcp_reset(struct sock *sk);
Y
Yuchung Cheng 已提交
567
void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, struct sk_buff *skb);
568
void tcp_fin(struct sock *sk);
569

L
Linus Torvalds 已提交
570
/* tcp_timer.c */
571
void tcp_init_xmit_timers(struct sock *);
572 573 574 575
static inline void tcp_clear_xmit_timers(struct sock *sk)
{
	inet_csk_clear_xmit_timers(sk);
}
L
Linus Torvalds 已提交
576

577 578
unsigned int tcp_sync_mss(struct sock *sk, u32 pmtu);
unsigned int tcp_current_mss(struct sock *sk);
I
Ilpo Järvinen 已提交
579 580 581 582

/* 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)
{
583 584 585 586 587 588 589 590 591
	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.
	 */
592
	if (tp->max_window > TCP_MSS_DEFAULT)
593 594 595 596 597 598
		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
599 600 601
	else
		return pktsize;
}
L
Linus Torvalds 已提交
602

603
/* tcp.c */
604
void tcp_get_info(struct sock *, struct tcp_info *);
L
Linus Torvalds 已提交
605 606 607 608

/* Read 'sendfile()'-style from a TCP socket */
typedef int (*sk_read_actor_t)(read_descriptor_t *, struct sk_buff *,
				unsigned int, size_t);
609 610
int tcp_read_sock(struct sock *sk, read_descriptor_t *desc,
		  sk_read_actor_t recv_actor);
L
Linus Torvalds 已提交
611

612
void tcp_initialize_rcv_mss(struct sock *sk);
L
Linus Torvalds 已提交
613

614 615 616 617
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
618

619 620 621 622 623 624 625 626
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)
{
627
	return usecs_to_jiffies((tp->srtt_us >> 3) + tp->rttvar_us);
628 629
}

S
Stephen Hemminger 已提交
630
static inline void __tcp_fast_path_on(struct tcp_sock *tp, u32 snd_wnd)
L
Linus Torvalds 已提交
631 632 633 634 635 636
{
	tp->pred_flags = htonl((tp->tcp_header_len << 26) |
			       ntohl(TCP_FLAG_ACK) |
			       snd_wnd);
}

S
Stephen Hemminger 已提交
637
static inline void tcp_fast_path_on(struct tcp_sock *tp)
L
Linus Torvalds 已提交
638 639 640 641
{
	__tcp_fast_path_on(tp, tp->snd_wnd >> tp->rx_opt.snd_wscale);
}

642
static inline void tcp_fast_path_check(struct sock *sk)
L
Linus Torvalds 已提交
643
{
644 645
	struct tcp_sock *tp = tcp_sk(sk);

646
	if (skb_queue_empty(&tp->out_of_order_queue) &&
L
Linus Torvalds 已提交
647 648 649 650 651 652
	    tp->rcv_wnd &&
	    atomic_read(&sk->sk_rmem_alloc) < sk->sk_rcvbuf &&
	    !tp->urg_data)
		tcp_fast_path_on(tp);
}

653 654 655
/* Compute the actual rto_min value */
static inline u32 tcp_rto_min(struct sock *sk)
{
656
	const struct dst_entry *dst = __sk_dst_get(sk);
657 658 659 660 661 662 663
	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;
}

664 665 666 667 668
static inline u32 tcp_rto_min_us(struct sock *sk)
{
	return jiffies_to_usecs(tcp_rto_min(sk));
}

669 670 671 672 673
static inline bool tcp_ca_dst_locked(const struct dst_entry *dst)
{
	return dst_metric_locked(dst, RTAX_CC_ALGO);
}

674 675 676 677 678 679
/* Minimum RTT in usec. ~0 means not available. */
static inline u32 tcp_min_rtt(const struct tcp_sock *tp)
{
	return tp->rtt_min[0].rtt;
}

L
Linus Torvalds 已提交
680 681 682 683
/* 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
684
static inline u32 tcp_receive_window(const struct tcp_sock *tp)
L
Linus Torvalds 已提交
685 686 687 688 689 690 691 692 693 694 695 696
{
	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.
 */
697
u32 __tcp_select_window(struct sock *sk);
L
Linus Torvalds 已提交
698

P
Pavel Emelyanov 已提交
699 700
void tcp_send_window_probe(struct sock *sk);

L
Linus Torvalds 已提交
701 702
/* 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
703 704
 * 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
705 706 707 708
 * casts with the following macro.
 */
#define tcp_time_stamp		((__u32)(jiffies))

E
Eric Dumazet 已提交
709 710 711 712 713 714
static inline u32 tcp_skb_timestamp(const struct sk_buff *skb)
{
	return skb->skb_mstamp.stamp_jiffies;
}


C
Changli Gao 已提交
715 716 717 718 719 720 721 722 723 724 725
#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

726 727
#define TCPHDR_SYN_ECN	(TCPHDR_SYN | TCPHDR_ECE | TCPHDR_CWR)

S
Stephen Hemminger 已提交
728
/* This is what the send packet queuing engine uses to pass
729 730 731 732
 * 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
733 734 735 736
 */
struct tcp_skb_cb {
	__u32		seq;		/* Starting sequence number	*/
	__u32		end_seq;	/* SEQ + FIN + SYN + datalen	*/
737 738 739 740
	union {
		/* Note : tcp_tw_isn is used in input path only
		 *	  (isn chosen by tcp_timewait_state_process())
		 *
741 742
		 * 	  tcp_gso_segs/size are used in write queue only,
		 *	  cf tcp_skb_pcount()/tcp_skb_mss()
743 744
		 */
		__u32		tcp_tw_isn;
745 746 747 748
		struct {
			u16	tcp_gso_segs;
			u16	tcp_gso_size;
		};
749
	};
E
Eric Dumazet 已提交
750
	__u8		tcp_flags;	/* TCP header flags. (tcp[13])	*/
751

L
Linus Torvalds 已提交
752 753 754 755 756
	__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			*/
757
#define TCPCB_REPAIRED		0x10	/* SKB repaired (no skb_mstamp)	*/
L
Linus Torvalds 已提交
758
#define TCPCB_EVER_RETRANS	0x80	/* Ever retransmitted frame	*/
759 760
#define TCPCB_RETRANS		(TCPCB_SACKED_RETRANS|TCPCB_EVER_RETRANS| \
				TCPCB_REPAIRED)
L
Linus Torvalds 已提交
761

762
	__u8		ip_dsfield;	/* IPv4 tos or IPv6 dsfield	*/
763
	__u8		txstamp_ack:1,	/* Record TX timestamp for ack? */
764 765
			eor:1,		/* Is skb MSG_EOR marked? */
			unused:6;
L
Linus Torvalds 已提交
766
	__u32		ack_seq;	/* Sequence number ACK'd	*/
767
	union {
768 769
		struct {
			/* There is space for up to 20 bytes */
770
			__u32 in_flight;/* Bytes in flight when packet sent */
771 772 773
		} tx;   /* only used for outgoing skbs */
		union {
			struct inet_skb_parm	h4;
774
#if IS_ENABLED(CONFIG_IPV6)
775
			struct inet6_skb_parm	h6;
776
#endif
777 778
		} header;	/* For incoming skbs */
	};
L
Linus Torvalds 已提交
779 780 781 782
};

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

E
Eric Dumazet 已提交
783

784
#if IS_ENABLED(CONFIG_IPV6)
E
Eric Dumazet 已提交
785 786 787 788 789
/* 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)
{
790 791 792
	bool l3_slave = skb_l3mdev_slave(TCP_SKB_CB(skb)->header.h6.flags);

	return l3_slave ? skb->skb_iif : TCP_SKB_CB(skb)->header.h6.iif;
E
Eric Dumazet 已提交
793
}
794
#endif
E
Eric Dumazet 已提交
795

L
Linus Torvalds 已提交
796 797
/* Due to TSO, an SKB can be composed of multiple actual
 * packets.  To keep these tracked properly, we use this.
798
 */
L
Linus Torvalds 已提交
799
static inline int tcp_skb_pcount(const struct sk_buff *skb)
800
{
801 802
	return TCP_SKB_CB(skb)->tcp_gso_segs;
}
803

804 805 806
static inline void tcp_skb_pcount_set(struct sk_buff *skb, int segs)
{
	TCP_SKB_CB(skb)->tcp_gso_segs = segs;
807 808
}

809
static inline void tcp_skb_pcount_add(struct sk_buff *skb, int segs)
L
Linus Torvalds 已提交
810
{
811
	TCP_SKB_CB(skb)->tcp_gso_segs += segs;
L
Linus Torvalds 已提交
812 813
}

814
/* This is valid iff skb is in write queue and tcp_skb_pcount() > 1. */
L
Linus Torvalds 已提交
815 816
static inline int tcp_skb_mss(const struct sk_buff *skb)
{
817
	return TCP_SKB_CB(skb)->tcp_gso_size;
L
Linus Torvalds 已提交
818 819
}

820 821 822 823 824
static inline bool tcp_skb_can_collapse_to(const struct sk_buff *skb)
{
	return likely(!TCP_SKB_CB(skb)->eor);
}

825 826 827 828 829 830
/* 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 */
831 832 833 834
	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,
835 836
};

837
/* Information about inbound ACK, passed to cong_ops->in_ack_event() */
838
enum tcp_ca_ack_event_flags {
839 840 841
	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 */
842 843 844 845 846 847
};

/*
 * Interface for adding new TCP congestion control handlers
 */
#define TCP_CA_NAME_MAX	16
848 849 850
#define TCP_CA_MAX	128
#define TCP_CA_BUF_MAX	(TCP_CA_NAME_MAX*TCP_CA_MAX)

851 852
#define TCP_CA_UNSPEC	0

853
/* Algorithm can be set on socket without CAP_NET_ADMIN privileges */
854
#define TCP_CONG_NON_RESTRICTED 0x1
855 856
/* Requires ECN/ECT set on all packets */
#define TCP_CONG_NEEDS_ECN	0x2
857

858 859
union tcp_cc_info;

860 861 862
struct ack_sample {
	u32 pkts_acked;
	s32 rtt_us;
863
	u32 in_flight;
864 865
};

866 867
struct tcp_congestion_ops {
	struct list_head	list;
868 869
	u32 key;
	u32 flags;
870 871

	/* initialize private data (optional) */
872
	void (*init)(struct sock *sk);
873
	/* cleanup private data  (optional) */
874
	void (*release)(struct sock *sk);
875 876

	/* return slow start threshold (required) */
877
	u32 (*ssthresh)(struct sock *sk);
878
	/* do new cwnd calculation (required) */
879
	void (*cong_avoid)(struct sock *sk, u32 ack, u32 acked);
880
	/* call before changing ca_state (optional) */
881
	void (*set_state)(struct sock *sk, u8 new_state);
882
	/* call when cwnd event occurs (optional) */
883
	void (*cwnd_event)(struct sock *sk, enum tcp_ca_event ev);
884 885
	/* call when ack arrives (optional) */
	void (*in_ack_event)(struct sock *sk, u32 flags);
886
	/* new value of cwnd after loss (optional) */
887
	u32  (*undo_cwnd)(struct sock *sk);
888
	/* hook for packet ack accounting (optional) */
889
	void (*pkts_acked)(struct sock *sk, const struct ack_sample *sample);
890
	/* get info for inet_diag (optional) */
891 892
	size_t (*get_info)(struct sock *sk, u32 ext, int *attr,
			   union tcp_cc_info *info);
893 894 895 896 897

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

898 899
int tcp_register_congestion_control(struct tcp_congestion_ops *type);
void tcp_unregister_congestion_control(struct tcp_congestion_ops *type);
900

901
void tcp_assign_congestion_control(struct sock *sk);
902 903 904 905 906 907 908 909
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
910 911
u32 tcp_slow_start(struct tcp_sock *tp, u32 acked);
void tcp_cong_avoid_ai(struct tcp_sock *tp, u32 w, u32 acked);
912

913
u32 tcp_reno_ssthresh(struct sock *sk);
914
void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 acked);
915
extern struct tcp_congestion_ops tcp_reno;
916

917
struct tcp_congestion_ops *tcp_ca_find_key(u32 key);
918
u32 tcp_ca_get_key_by_name(const char *name, bool *ecn_ca);
919
#ifdef CONFIG_INET
920
char *tcp_ca_get_name_by_key(u32 key, char *buffer);
921 922 923 924 925 926
#else
static inline char *tcp_ca_get_name_by_key(u32 key, char *buffer)
{
	return NULL;
}
#endif
927

928 929 930 931 932 933 934
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;
}

935
static inline void tcp_set_ca_state(struct sock *sk, const u8 ca_state)
936
{
937 938 939 940 941
	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;
942 943
}

944
static inline void tcp_ca_event(struct sock *sk, const enum tcp_ca_event event)
945
{
946 947 948 949
	const struct inet_connection_sock *icsk = inet_csk(sk);

	if (icsk->icsk_ca_ops->cwnd_event)
		icsk->icsk_ca_ops->cwnd_event(sk, event);
950 951
}

952 953 954 955 956 957 958 959 960 961 962 963 964
/* 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
965
static inline bool tcp_is_reno(const struct tcp_sock *tp)
966 967 968 969
{
	return !tcp_is_sack(tp);
}

E
Eric Dumazet 已提交
970
static inline bool tcp_is_fack(const struct tcp_sock *tp)
971
{
972
	return tp->rx_opt.sack_ok & TCP_FACK_ENABLED;
973 974 975 976
}

static inline void tcp_enable_fack(struct tcp_sock *tp)
{
977
	tp->rx_opt.sack_ok |= TCP_FACK_ENABLED;
978 979
}

Y
Yuchung Cheng 已提交
980 981 982 983 984
/* TCP early-retransmit (ER) is similar to but more conservative than
 * the thin-dupack feature.  Enable ER only if thin-dupack is disabled.
 */
static inline void tcp_enable_early_retrans(struct tcp_sock *tp)
{
985 986
	struct net *net = sock_net((struct sock *)tp);

Y
Yuchung Cheng 已提交
987
	tp->do_early_retrans = sysctl_tcp_early_retrans &&
N
Nandita Dukkipati 已提交
988
		sysctl_tcp_early_retrans < 4 && !sysctl_tcp_thin_dupack &&
989
		net->ipv4.sysctl_tcp_reordering == 3;
Y
Yuchung Cheng 已提交
990 991 992 993 994 995 996
}

static inline void tcp_disable_early_retrans(struct tcp_sock *tp)
{
	tp->do_early_retrans = 0;
}

997 998 999 1000 1001
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
1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
/* 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
1016
static inline unsigned int tcp_packets_in_flight(const struct tcp_sock *tp)
L
Linus Torvalds 已提交
1017
{
1018
	return tp->packets_out - tcp_left_out(tp) + tp->retrans_out;
L
Linus Torvalds 已提交
1019 1020
}

I
Ilpo Järvinen 已提交
1021 1022
#define TCP_INFINITE_SSTHRESH	0x7fffffff

1023 1024
static inline bool tcp_in_slow_start(const struct tcp_sock *tp)
{
1025
	return tp->snd_cwnd < tp->snd_ssthresh;
1026 1027
}

I
Ilpo Järvinen 已提交
1028 1029 1030 1031 1032
static inline bool tcp_in_initial_slowstart(const struct tcp_sock *tp)
{
	return tp->snd_ssthresh >= TCP_INFINITE_SSTHRESH;
}

1033 1034 1035 1036 1037 1038
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
1039
/* If cwnd > ssthresh, we may raise ssthresh to be half-way to cwnd.
1040
 * The exception is cwnd reduction phase, when cwnd is decreasing towards
L
Linus Torvalds 已提交
1041 1042
 * ssthresh.
 */
1043
static inline __u32 tcp_current_ssthresh(const struct sock *sk)
L
Linus Torvalds 已提交
1044
{
1045
	const struct tcp_sock *tp = tcp_sk(sk);
1046

1047
	if (tcp_in_cwnd_reduction(sk))
L
Linus Torvalds 已提交
1048 1049 1050 1051 1052 1053 1054
		return tp->snd_ssthresh;
	else
		return max(tp->snd_ssthresh,
			   ((tp->snd_cwnd >> 1) +
			    (tp->snd_cwnd >> 2)));
}

1055 1056
/* 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
1057

1058
void tcp_enter_cwr(struct sock *sk);
1059
__u32 tcp_init_cwnd(const struct tcp_sock *tp, const struct dst_entry *dst);
L
Linus Torvalds 已提交
1060

1061 1062 1063 1064 1065 1066 1067 1068
/* 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;
}

1069 1070 1071 1072 1073
/* 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;
}
1074 1075 1076

/* We follow the spirit of RFC2861 to validate cwnd but implement a more
 * flexible approach. The RFC suggests cwnd should not be raised unless
1077 1078 1079
 * 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.
1080 1081 1082 1083 1084 1085 1086 1087
 * 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.
 */
1088
static inline bool tcp_is_cwnd_limited(const struct sock *sk)
1089 1090 1091
{
	const struct tcp_sock *tp = tcp_sk(sk);

1092
	/* If in slow start, ensure cwnd grows to twice what was ACKed. */
1093
	if (tcp_in_slow_start(tp))
1094 1095 1096
		return tp->snd_cwnd < 2 * tp->max_packets_out;

	return tp->is_cwnd_limited;
1097
}
1098

1099 1100 1101 1102 1103 1104 1105
/* 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
1106
{
1107 1108
	return max_t(unsigned long, inet_csk(sk)->icsk_rto, TCP_RTO_MIN);
}
1109

1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121
/* 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)
1122
		inet_csk_reset_xmit_timer(sk, ICSK_TIME_PROBE0,
1123
					  tcp_probe0_base(sk), TCP_RTO_MAX);
L
Linus Torvalds 已提交
1124 1125
}

1126
static inline void tcp_init_wl(struct tcp_sock *tp, u32 seq)
L
Linus Torvalds 已提交
1127 1128 1129 1130
{
	tp->snd_wl1 = seq;
}

1131
static inline void tcp_update_wl(struct tcp_sock *tp, u32 seq)
L
Linus Torvalds 已提交
1132 1133 1134 1135 1136 1137 1138
{
	tp->snd_wl1 = seq;
}

/*
 * Calculate(/check) TCP checksum
 */
1139 1140
static inline __sum16 tcp_v4_check(int len, __be32 saddr,
				   __be32 daddr, __wsum base)
L
Linus Torvalds 已提交
1141 1142 1143 1144
{
	return csum_tcpudp_magic(saddr,daddr,len,IPPROTO_TCP,base);
}

1145
static inline __sum16 __tcp_checksum_complete(struct sk_buff *skb)
L
Linus Torvalds 已提交
1146
{
1147
	return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1148 1149
}

E
Eric Dumazet 已提交
1150
static inline bool tcp_checksum_complete(struct sk_buff *skb)
L
Linus Torvalds 已提交
1151
{
1152
	return !skb_csum_unnecessary(skb) &&
L
Linus Torvalds 已提交
1153 1154 1155 1156 1157
		__tcp_checksum_complete(skb);
}

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

S
Stephen Hemminger 已提交
1158
static inline void tcp_prequeue_init(struct tcp_sock *tp)
L
Linus Torvalds 已提交
1159 1160 1161 1162 1163 1164 1165
{
	tp->ucopy.task = NULL;
	tp->ucopy.len = 0;
	tp->ucopy.memory = 0;
	skb_queue_head_init(&tp->ucopy.prequeue);
}

1166
bool tcp_prequeue(struct sock *sk, struct sk_buff *skb);
L
Linus Torvalds 已提交
1167 1168 1169 1170 1171 1172 1173 1174 1175 1176

#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
1177
void tcp_set_state(struct sock *sk, int state);
L
Linus Torvalds 已提交
1178

1179
void tcp_done(struct sock *sk);
L
Linus Torvalds 已提交
1180

1181 1182
int tcp_abort(struct sock *sk, int err);

S
Stephen Hemminger 已提交
1183
static inline void tcp_sack_reset(struct tcp_options_received *rx_opt)
L
Linus Torvalds 已提交
1184 1185 1186 1187 1188
{
	rx_opt->dsack = 0;
	rx_opt->num_sacks = 0;
}

1189
u32 tcp_default_init_rwnd(u32 mss);
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202
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);
}
1203

L
Linus Torvalds 已提交
1204
/* Determine a window scaling and initial window to offer. */
1205 1206 1207
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
1208 1209 1210 1211 1212 1213 1214 1215

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);
}

1216
/* Note: caller must be prepared to deal with negative returns */
L
Linus Torvalds 已提交
1217 1218 1219 1220
static inline int tcp_space(const struct sock *sk)
{
	return tcp_win_from_space(sk->sk_rcvbuf -
				  atomic_read(&sk->sk_rmem_alloc));
1221
}
L
Linus Torvalds 已提交
1222 1223 1224

static inline int tcp_full_space(const struct sock *sk)
{
1225
	return tcp_win_from_space(sk->sk_rcvbuf);
L
Linus Torvalds 已提交
1226 1227
}

1228
extern void tcp_openreq_init_rwin(struct request_sock *req,
1229 1230
				  const struct sock *sk_listener,
				  const struct dst_entry *dst);
1231

1232
void tcp_enter_memory_pressure(struct sock *sk);
L
Linus Torvalds 已提交
1233 1234 1235

static inline int keepalive_intvl_when(const struct tcp_sock *tp)
{
1236 1237 1238
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_intvl ? : net->ipv4.sysctl_tcp_keepalive_intvl;
L
Linus Torvalds 已提交
1239 1240 1241 1242
}

static inline int keepalive_time_when(const struct tcp_sock *tp)
{
1243 1244 1245
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_time ? : net->ipv4.sysctl_tcp_keepalive_time;
L
Linus Torvalds 已提交
1246 1247
}

E
Eric Dumazet 已提交
1248 1249
static inline int keepalive_probes(const struct tcp_sock *tp)
{
1250 1251 1252
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_probes ? : net->ipv4.sysctl_tcp_keepalive_probes;
E
Eric Dumazet 已提交
1253 1254
}

1255 1256 1257 1258 1259 1260 1261 1262
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);
}

1263
static inline int tcp_fin_time(const struct sock *sk)
L
Linus Torvalds 已提交
1264
{
1265
	int fin_timeout = tcp_sk(sk)->linger2 ? : sock_net(sk)->ipv4.sysctl_tcp_fin_timeout;
1266
	const int rto = inet_csk(sk)->icsk_rto;
L
Linus Torvalds 已提交
1267

1268 1269
	if (fin_timeout < (rto << 2) - (rto >> 1))
		fin_timeout = (rto << 2) - (rto >> 1);
L
Linus Torvalds 已提交
1270 1271 1272 1273

	return fin_timeout;
}

E
Eric Dumazet 已提交
1274 1275
static inline bool tcp_paws_check(const struct tcp_options_received *rx_opt,
				  int paws_win)
L
Linus Torvalds 已提交
1276
{
I
Ilpo Järvinen 已提交
1277
	if ((s32)(rx_opt->ts_recent - rx_opt->rcv_tsval) <= paws_win)
E
Eric Dumazet 已提交
1278
		return true;
I
Ilpo Järvinen 已提交
1279
	if (unlikely(get_seconds() >= rx_opt->ts_recent_stamp + TCP_PAWS_24DAYS))
E
Eric Dumazet 已提交
1280
		return true;
E
Eric Dumazet 已提交
1281 1282 1283 1284 1285 1286
	/*
	 * 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
1287 1288
		return true;
	return false;
I
Ilpo Järvinen 已提交
1289 1290
}

E
Eric Dumazet 已提交
1291 1292
static inline bool tcp_paws_reject(const struct tcp_options_received *rx_opt,
				   int rst)
I
Ilpo Järvinen 已提交
1293 1294
{
	if (tcp_paws_check(rx_opt, 0))
E
Eric Dumazet 已提交
1295
		return false;
L
Linus Torvalds 已提交
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308

	/* 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.
	 */
1309
	if (rst && get_seconds() >= rx_opt->ts_recent_stamp + TCP_PAWS_MSL)
E
Eric Dumazet 已提交
1310 1311
		return false;
	return true;
L
Linus Torvalds 已提交
1312 1313
}

1314 1315
bool tcp_oow_rate_limited(struct net *net, const struct sk_buff *skb,
			  int mib_idx, u32 *last_oow_ack_time);
1316

P
Pavel Emelyanov 已提交
1317
static inline void tcp_mib_init(struct net *net)
L
Linus Torvalds 已提交
1318 1319
{
	/* See RFC 2012 */
1320 1321 1322 1323
	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
1324 1325
}

1326
/* from STCP */
1327
static inline void tcp_clear_retrans_hints_partial(struct tcp_sock *tp)
D
David S. Miller 已提交
1328
{
1329
	tp->lost_skb_hint = NULL;
1330 1331 1332 1333 1334
}

static inline void tcp_clear_all_retrans_hints(struct tcp_sock *tp)
{
	tcp_clear_retrans_hints_partial(tp);
1335
	tp->retransmit_skb_hint = NULL;
1336 1337
}

E
Eric Dumazet 已提交
1338 1339 1340 1341 1342 1343 1344
union tcp_md5_addr {
	struct in_addr  a4;
#if IS_ENABLED(CONFIG_IPV6)
	struct in6_addr	a6;
#endif
};

1345 1346
/* - key database */
struct tcp_md5sig_key {
E
Eric Dumazet 已提交
1347
	struct hlist_node	node;
1348
	u8			keylen;
E
Eric Dumazet 已提交
1349 1350 1351 1352
	u8			family; /* AF_INET or AF_INET6 */
	union tcp_md5_addr	addr;
	u8			key[TCP_MD5SIG_MAXKEYLEN];
	struct rcu_head		rcu;
1353 1354 1355 1356
};

/* - sock block */
struct tcp_md5sig_info {
E
Eric Dumazet 已提交
1357
	struct hlist_head	head;
1358
	struct rcu_head		rcu;
1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378
};

/* - 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
1379
#if IS_ENABLED(CONFIG_IPV6)
1380 1381 1382 1383 1384 1385
	struct tcp6_pseudohdr ip6;
#endif
};

/* - pool: digest algorithm, hash description and scratch buffer */
struct tcp_md5sig_pool {
H
Herbert Xu 已提交
1386
	struct ahash_request	*md5_req;
1387 1388 1389 1390
	union tcp_md5sum_block	md5_blk;
};

/* - functions */
1391 1392
int tcp_v4_md5_hash_skb(char *md5_hash, const struct tcp_md5sig_key *key,
			const struct sock *sk, const struct sk_buff *skb);
1393 1394 1395 1396
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);
1397
struct tcp_md5sig_key *tcp_v4_md5_lookup(const struct sock *sk,
1398
					 const struct sock *addr_sk);
1399

1400
#ifdef CONFIG_TCP_MD5SIG
1401
struct tcp_md5sig_key *tcp_md5_do_lookup(const struct sock *sk,
1402 1403
					 const union tcp_md5_addr *addr,
					 int family);
E
Eric Dumazet 已提交
1404
#define tcp_twsk_md5_key(twsk)	((twsk)->tw_md5_key)
1405
#else
1406
static inline struct tcp_md5sig_key *tcp_md5_do_lookup(const struct sock *sk,
E
Eric Dumazet 已提交
1407 1408 1409 1410 1411
					 const union tcp_md5_addr *addr,
					 int family)
{
	return NULL;
}
1412 1413 1414
#define tcp_twsk_md5_key(twsk)	NULL
#endif

1415
bool tcp_alloc_md5sig_pool(void);
1416

1417
struct tcp_md5sig_pool *tcp_get_md5sig_pool(void);
1418 1419 1420 1421
static inline void tcp_put_md5sig_pool(void)
{
	local_bh_enable();
}
E
Eric Dumazet 已提交
1422

1423 1424 1425 1426 1427
int tcp_md5_hash_header(struct tcp_md5sig_pool *, const struct tcphdr *);
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);
1428

1429
/* From tcp_fastopen.c */
1430 1431 1432 1433
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,
1434 1435
			    struct tcp_fastopen_cookie *cookie, bool syn_lost,
			    u16 try_exp);
1436 1437 1438 1439
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 */
1440 1441
	size_t				size;
	int				copied;	/* queued in tcp_connect() */
1442 1443 1444
};
void tcp_free_fastopen_req(struct tcp_sock *tp);

1445 1446
extern struct tcp_fastopen_context __rcu *tcp_fastopen_ctx;
int tcp_fastopen_reset_cipher(void *key, unsigned int len);
1447
void tcp_fastopen_add_skb(struct sock *sk, struct sk_buff *skb);
1448 1449 1450 1451
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);
1452
void tcp_fastopen_init_key_once(bool publish);
1453 1454 1455 1456
#define TCP_FASTOPEN_KEY_LENGTH 16

/* Fastopen key context */
struct tcp_fastopen_context {
1457 1458 1459
	struct crypto_cipher	*tfm;
	__u8			key[TCP_FASTOPEN_KEY_LENGTH];
	struct rcu_head		rcu;
1460 1461
};

1462 1463 1464 1465 1466 1467
/* write queue abstraction */
static inline void tcp_write_queue_purge(struct sock *sk)
{
	struct sk_buff *skb;

	while ((skb = __skb_dequeue(&sk->sk_write_queue)) != NULL)
1468 1469
		sk_wmem_free_skb(sk, skb);
	sk_mem_reclaim(sk);
1470
	tcp_clear_all_retrans_hints(tcp_sk(sk));
1471 1472
}

1473
static inline struct sk_buff *tcp_write_queue_head(const struct sock *sk)
1474
{
1475
	return skb_peek(&sk->sk_write_queue);
1476 1477
}

1478
static inline struct sk_buff *tcp_write_queue_tail(const struct sock *sk)
1479
{
1480
	return skb_peek_tail(&sk->sk_write_queue);
1481 1482
}

1483 1484
static inline struct sk_buff *tcp_write_queue_next(const struct sock *sk,
						   const struct sk_buff *skb)
1485
{
1486
	return skb_queue_next(&sk->sk_write_queue, skb);
1487 1488
}

1489 1490
static inline struct sk_buff *tcp_write_queue_prev(const struct sock *sk,
						   const struct sk_buff *skb)
1491 1492 1493 1494
{
	return skb_queue_prev(&sk->sk_write_queue, skb);
}

1495
#define tcp_for_write_queue(skb, sk)					\
1496
	skb_queue_walk(&(sk)->sk_write_queue, skb)
1497 1498

#define tcp_for_write_queue_from(skb, sk)				\
1499
	skb_queue_walk_from(&(sk)->sk_write_queue, skb)
1500

1501
#define tcp_for_write_queue_from_safe(skb, tmp, sk)			\
1502
	skb_queue_walk_from_safe(&(sk)->sk_write_queue, skb, tmp)
1503

1504
static inline struct sk_buff *tcp_send_head(const struct sock *sk)
1505 1506 1507 1508
{
	return sk->sk_send_head;
}

1509 1510 1511 1512 1513 1514
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);
}

1515
static inline void tcp_advance_send_head(struct sock *sk, const struct sk_buff *skb)
1516
{
1517
	if (tcp_skb_is_last(sk, skb))
1518
		sk->sk_send_head = NULL;
1519 1520
	else
		sk->sk_send_head = tcp_write_queue_next(sk, skb);
1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543
}

static inline void tcp_check_send_head(struct sock *sk, struct sk_buff *skb_unlinked)
{
	if (sk->sk_send_head == skb_unlinked)
		sk->sk_send_head = NULL;
}

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. */
1544
	if (sk->sk_send_head == NULL) {
1545
		sk->sk_send_head = skb;
1546 1547 1548 1549

		if (tcp_sk(sk)->highest_sack == NULL)
			tcp_sk(sk)->highest_sack = skb;
	}
1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561
}

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)
{
1562
	__skb_queue_after(&sk->sk_write_queue, skb, buff);
1563 1564
}

1565
/* Insert new before skb on the write queue of sk.  */
1566 1567 1568 1569
static inline void tcp_insert_write_queue_before(struct sk_buff *new,
						  struct sk_buff *skb,
						  struct sock *sk)
{
1570
	__skb_queue_before(&sk->sk_write_queue, skb, new);
1571 1572 1573

	if (sk->sk_send_head == skb)
		sk->sk_send_head = new;
1574 1575 1576 1577 1578 1579 1580
}

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
1581
static inline bool tcp_write_queue_empty(struct sock *sk)
1582 1583 1584 1585
{
	return skb_queue_empty(&sk->sk_write_queue);
}

1586 1587 1588 1589 1590 1591 1592 1593 1594
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);
	}
}

1595 1596 1597
/* 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.
1598 1599 1600 1601 1602
 */
static inline u32 tcp_highest_sack_seq(struct tcp_sock *tp)
{
	if (!tp->sacked_out)
		return tp->snd_una;
1603 1604 1605 1606

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

1607 1608 1609
	return TCP_SKB_CB(tp->highest_sack)->seq;
}

1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634
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;
}

1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646
/* 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
1647 1648 1649
/* 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
1650
static inline bool tcp_stream_is_thin(struct tcp_sock *tp)
A
Andreas Petlund 已提交
1651 1652 1653 1654
{
	return tp->packets_out < 4 && !tcp_in_initial_slowstart(tp);
}

L
Linus Torvalds 已提交
1655 1656 1657 1658 1659 1660
/* /proc */
enum tcp_seq_states {
	TCP_SEQ_STATE_LISTENING,
	TCP_SEQ_STATE_ESTABLISHED,
};

1661 1662
int tcp_seq_open(struct inode *inode, struct file *file);

L
Linus Torvalds 已提交
1663
struct tcp_seq_afinfo {
1664 1665 1666 1667
	char				*name;
	sa_family_t			family;
	const struct file_operations	*seq_fops;
	struct seq_operations		seq_ops;
L
Linus Torvalds 已提交
1668 1669 1670
};

struct tcp_iter_state {
1671
	struct seq_net_private	p;
L
Linus Torvalds 已提交
1672 1673 1674
	sa_family_t		family;
	enum tcp_seq_states	state;
	struct sock		*syn_wait_sk;
1675
	int			bucket, offset, sbucket, num;
1676
	loff_t			last_pos;
L
Linus Torvalds 已提交
1677 1678
};

1679 1680
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
1681

1682
extern struct request_sock_ops tcp_request_sock_ops;
1683
extern struct request_sock_ops tcp6_request_sock_ops;
1684

1685
void tcp_v4_destroy_sock(struct sock *sk);
1686

E
Eric Dumazet 已提交
1687
struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
1688 1689 1690
				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);
1691

1692
void __tcp_v4_send_check(struct sk_buff *skb, __be32 saddr, __be32 daddr);
H
Herbert Xu 已提交
1693

1694 1695
static inline u32 tcp_notsent_lowat(const struct tcp_sock *tp)
{
1696 1697
	struct net *net = sock_net((struct sock *)tp);
	return tp->notsent_lowat ?: net->ipv4.sysctl_tcp_notsent_lowat;
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707
}

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);
}

1708
#ifdef CONFIG_PROC_FS
1709 1710
int tcp4_proc_init(void);
void tcp4_proc_exit(void);
1711 1712
#endif

1713
int tcp_rtx_synack(const struct sock *sk, struct request_sock *req);
O
Octavian Purdila 已提交
1714 1715 1716
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);
1717

1718 1719 1720
/* TCP af-specific functions */
struct tcp_sock_af_ops {
#ifdef CONFIG_TCP_MD5SIG
1721
	struct tcp_md5sig_key	*(*md5_lookup) (const struct sock *sk,
1722
						const struct sock *addr_sk);
1723 1724 1725 1726 1727 1728 1729
	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);
1730 1731 1732 1733
#endif
};

struct tcp_request_sock_ops {
1734
	u16 mss_clamp;
1735
#ifdef CONFIG_TCP_MD5SIG
1736
	struct tcp_md5sig_key *(*req_md5_lookup)(const struct sock *sk,
1737
						 const struct sock *addr_sk);
1738 1739 1740 1741
	int		(*calc_md5_hash) (char *location,
					  const struct tcp_md5sig_key *md5,
					  const struct sock *sk,
					  const struct sk_buff *skb);
1742
#endif
1743 1744
	void (*init_req)(struct request_sock *req,
			 const struct sock *sk_listener,
1745
			 struct sk_buff *skb);
1746
#ifdef CONFIG_SYN_COOKIES
1747
	__u32 (*cookie_init_seq)(const struct sk_buff *skb,
1748 1749
				 __u16 *mss);
#endif
1750
	struct dst_entry *(*route_req)(const struct sock *sk, struct flowi *fl,
1751 1752
				       const struct request_sock *req,
				       bool *strict);
1753
	__u32 (*init_seq)(const struct sk_buff *skb);
1754
	int (*send_synack)(const struct sock *sk, struct dst_entry *dst,
1755
			   struct flowi *fl, struct request_sock *req,
1756
			   struct tcp_fastopen_cookie *foc,
1757
			   enum tcp_synack_type synack_type);
1758 1759
};

1760 1761
#ifdef CONFIG_SYN_COOKIES
static inline __u32 cookie_init_sequence(const struct tcp_request_sock_ops *ops,
1762
					 const struct sock *sk, struct sk_buff *skb,
1763 1764
					 __u16 *mss)
{
1765
	tcp_synq_overflow(sk);
1766
	__NET_INC_STATS(sock_net(sk), LINUX_MIB_SYNCOOKIESSENT);
1767
	return ops->cookie_init_seq(skb, mss);
1768 1769 1770
}
#else
static inline __u32 cookie_init_sequence(const struct tcp_request_sock_ops *ops,
1771
					 const struct sock *sk, struct sk_buff *skb,
1772 1773 1774 1775 1776 1777
					 __u16 *mss)
{
	return 0;
}
#endif

1778
int tcpv4_offload_init(void);
1779

1780 1781
void tcp_v4_init(void);
void tcp_init(void);
1782

1783 1784
/* tcp_recovery.c */

Y
Yuchung Cheng 已提交
1785 1786 1787 1788 1789 1790 1791 1792
/* Flags to enable various loss recovery features. See below */
extern int sysctl_tcp_recovery;

/* Use TCP RACK to detect (some) tail and retransmit losses */
#define TCP_RACK_LOST_RETRANS  0x1

extern int tcp_rack_mark_lost(struct sock *sk);

1793 1794 1795
extern void tcp_rack_advance(struct tcp_sock *tp,
			     const struct skb_mstamp *xmit_time, u8 sacked);

1796 1797 1798 1799 1800 1801 1802 1803
/*
 * 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
1804
	if (opt->optlen) {
1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815
		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
1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830
/* 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;
}

1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854
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;
}

1855 1856 1857 1858 1859 1860 1861 1862 1863 1864
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;
}

1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
/*
 * 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);
1875
	__NET_INC_STATS(sock_net(sk), LINUX_MIB_LISTENDROPS);
1876 1877
}

L
Linus Torvalds 已提交
1878
#endif	/* _TCP_H */