tcp.h 62.7 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

L
Lawrence Brakmo 已提交
49 50 51 52
#include <linux/bpf.h>
#include <linux/filter.h>
#include <linux/bpf-cgroup.h>

53
extern struct inet_hashinfo tcp_hashinfo;
L
Linus Torvalds 已提交
54

55
extern struct percpu_counter tcp_orphan_count;
56
void tcp_time_wait(struct sock *sk, int state, int timeo);
L
Linus Torvalds 已提交
57 58

#define MAX_TCP_HEADER	(128 + MAX_HEADER)
A
Adam Langley 已提交
59
#define MAX_TCP_OPTION_SPACE 40
L
Linus Torvalds 已提交
60

61
/*
L
Linus Torvalds 已提交
62
 * Never offer a window over 32767 without using window scaling. Some
63
 * poor stacks do signed 16bit maths!
L
Linus Torvalds 已提交
64 65 66 67 68 69
 */
#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
70
/* The least MTU to use for probing */
71
#define TCP_BASE_MSS		1024
J
John Heffner 已提交
72

73 74 75
/* probing interval, default to 10 minutes as per RFC4821 */
#define TCP_PROBE_INTERVAL	600

76 77 78
/* Specify interval when tcp mtu probing will stop */
#define TCP_PROBE_THRESHOLD	8

L
Linus Torvalds 已提交
79 80 81 82 83 84
/* 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

85 86 87
/* Maximal number of window scale according to RFC1323 */
#define TCP_MAX_WSCALE		14U

L
Linus Torvalds 已提交
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
/* 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.
				 */

107 108 109 110 111 112 113 114
#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
115

116 117 118 119 120 121
#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
122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141

#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))
142
#define TCP_TIMEOUT_MIN	(2U) /* Min timeout for TCP timers in jiffies */
143
#define TCP_TIMEOUT_INIT ((unsigned)(1*HZ))	/* RFC6298 2.1 initial RTO value	*/
144 145 146 147 148 149
#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
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178

#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
 */
179

L
Linus Torvalds 已提交
180 181 182 183 184 185 186
#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 */
187
#define TCPOPT_MD5SIG		19	/* MD5 Signature (RFC2385) */
188
#define TCPOPT_FASTOPEN		34	/* Fast open (RFC7413) */
Y
Yuchung Cheng 已提交
189 190 191 192 193
#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
194 195 196 197 198 199 200 201 202

/*
 *     TCP option lengths
 */

#define TCPOLEN_MSS            4
#define TCPOLEN_WINDOW         3
#define TCPOLEN_SACK_PERM      2
#define TCPOLEN_TIMESTAMP      10
203
#define TCPOLEN_MD5SIG         18
204
#define TCPOLEN_FASTOPEN_BASE  2
Y
Yuchung Cheng 已提交
205
#define TCPOLEN_EXP_FASTOPEN_BASE  4
L
Linus Torvalds 已提交
206 207 208 209 210 211 212 213

/* 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
214
#define TCPOLEN_MD5SIG_ALIGNED		20
A
Adam Langley 已提交
215
#define TCPOLEN_MSS_ALIGNED		4
L
Linus Torvalds 已提交
216 217 218 219

/* 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
220
#define TCP_NAGLE_PUSH		4	/* Cork is overridden for already queued data */
L
Linus Torvalds 已提交
221

A
Andreas Petlund 已提交
222 223 224
/* TCP thin-stream limits */
#define TCP_THIN_LINEAR_RETRIES 6       /* After 6 linear retries, do exp. backoff */

225
/* TCP initial congestion window as per rfc6928 */
226 227
#define TCP_INIT_CWND		10

228 229
/* Bit Flags for sysctl_tcp_fastopen */
#define	TFO_CLIENT_ENABLE	1
230
#define	TFO_SERVER_ENABLE	2
231
#define	TFO_CLIENT_NO_COOKIE	4	/* Data in SYN w/o cookie option */
232

233 234 235 236
/* 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
237
 * TCP_FASTOPEN socket option.
238 239 240
 */
#define	TFO_SERVER_WO_SOCKOPT1	0x400

241

L
Linus Torvalds 已提交
242 243 244 245 246 247 248 249
/* sysctl variables for tcp */
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;
250
extern int sysctl_tcp_max_reordering;
L
Linus Torvalds 已提交
251
extern int sysctl_tcp_dsack;
252
extern long sysctl_tcp_mem[3];
L
Linus Torvalds 已提交
253 254 255 256 257 258 259 260
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_nometrics_save;
extern int sysctl_tcp_moderate_rcvbuf;
extern int sysctl_tcp_tso_win_divisor;
261
extern int sysctl_tcp_workaround_signed_windows;
262
extern int sysctl_tcp_slow_start_after_idle;
A
Andreas Petlund 已提交
263
extern int sysctl_tcp_thin_linear_timeouts;
A
Andreas Petlund 已提交
264
extern int sysctl_tcp_thin_dupack;
Y
Yuchung Cheng 已提交
265
extern int sysctl_tcp_early_retrans;
266 267 268
extern int sysctl_tcp_recovery;
#define TCP_RACK_LOSS_DETECTION  0x1 /* Use RACK to detect losses */

E
Eric Dumazet 已提交
269
extern int sysctl_tcp_limit_output_bytes;
E
Eric Dumazet 已提交
270
extern int sysctl_tcp_challenge_ack_limit;
E
Eric Dumazet 已提交
271
extern int sysctl_tcp_min_tso_segs;
272
extern int sysctl_tcp_min_rtt_wlen;
E
Eric Dumazet 已提交
273
extern int sysctl_tcp_autocorking;
274
extern int sysctl_tcp_invalid_ratelimit;
275 276
extern int sysctl_tcp_pacing_ss_ratio;
extern int sysctl_tcp_pacing_ca_ratio;
L
Linus Torvalds 已提交
277

E
Eric Dumazet 已提交
278
extern atomic_long_t tcp_memory_allocated;
279
extern struct percpu_counter tcp_sockets_allocated;
280
extern unsigned long tcp_memory_pressure;
L
Linus Torvalds 已提交
281

282 283 284
/* optimized version of sk_under_memory_pressure() for TCP sockets */
static inline bool tcp_under_memory_pressure(const struct sock *sk)
{
285 286
	if (mem_cgroup_sockets_enabled && sk->sk_memcg &&
	    mem_cgroup_under_socket_pressure(sk->sk_memcg))
287
		return true;
288 289 290

	return tcp_memory_pressure;
}
L
Linus Torvalds 已提交
291 292 293 294 295
/*
 * The next routines deal with comparing 32 bit unsigned ints
 * and worry about wraparound (automatic with unsigned arithmetic).
 */

E
Eric Dumazet 已提交
296
static inline bool before(__u32 seq1, __u32 seq2)
L
Linus Torvalds 已提交
297
{
298
        return (__s32)(seq1-seq2) < 0;
L
Linus Torvalds 已提交
299
}
300
#define after(seq2, seq1) 	before(seq1, seq2)
L
Linus Torvalds 已提交
301 302

/* is s2<=s1<=s3 ? */
E
Eric Dumazet 已提交
303
static inline bool between(__u32 seq1, __u32 seq2, __u32 seq3)
L
Linus Torvalds 已提交
304 305 306 307
{
	return seq3 - seq2 >= seq1 - seq2;
}

A
Arun Sharma 已提交
308 309 310 311 312 313 314 315
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;
}

316 317
void sk_forced_mem_schedule(struct sock *sk, int size);

318
static inline bool tcp_too_many_orphans(struct sock *sk, int shift)
319
{
320 321 322 323 324 325 326 327 328
	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;
329
}
L
Linus Torvalds 已提交
330

331
bool tcp_check_oom(struct sock *sk, int shift);
A
Arun Sharma 已提交
332

333

L
Linus Torvalds 已提交
334 335
extern struct proto tcp_prot;

336
#define TCP_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.tcp_statistics, field)
E
Eric Dumazet 已提交
337
#define __TCP_INC_STATS(net, field)	__SNMP_INC_STATS((net)->mib.tcp_statistics, field)
338
#define TCP_DEC_STATS(net, field)	SNMP_DEC_STATS((net)->mib.tcp_statistics, field)
339
#define TCP_ADD_STATS(net, field, val)	SNMP_ADD_STATS((net)->mib.tcp_statistics, field, val)
L
Linus Torvalds 已提交
340

341 342 343 344 345 346
void tcp_tasklet_init(void);

void tcp_v4_err(struct sk_buff *skb, u32);

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

347
int tcp_v4_early_demux(struct sk_buff *skb);
348 349 350
int tcp_v4_rcv(struct sk_buff *skb);

int tcp_v4_tw_remember_stamp(struct inet_timewait_sock *tw);
351
int tcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t size);
352
int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size);
353 354
int tcp_sendpage(struct sock *sk, struct page *page, int offset, size_t size,
		 int flags);
355 356
int tcp_sendpage_locked(struct sock *sk, struct page *page, int offset,
			size_t size, int flags);
357 358
ssize_t do_tcp_sendpages(struct sock *sk, struct page *page, int offset,
		 size_t size, int flags);
359 360 361 362 363
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);
364
int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb);
365
void tcp_rcv_established(struct sock *sk, struct sk_buff *skb,
366
			 const struct tcphdr *th);
367 368 369 370 371 372
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
373

374 375
static inline void tcp_dec_quickack_mode(struct sock *sk,
					 const unsigned int pkts)
L
Linus Torvalds 已提交
376
{
377
	struct inet_connection_sock *icsk = inet_csk(sk);
378

379 380 381
	if (icsk->icsk_ack.quick) {
		if (pkts >= icsk->icsk_ack.quick) {
			icsk->icsk_ack.quick = 0;
382
			/* Leaving quickack mode we deflate ATO. */
383
			icsk->icsk_ack.ato   = TCP_ATO_MIN;
384
		} else
385
			icsk->icsk_ack.quick -= pkts;
L
Linus Torvalds 已提交
386 387 388
	}
}

389 390 391
#define	TCP_ECN_OK		1
#define	TCP_ECN_QUEUE_CWR	2
#define	TCP_ECN_DEMAND_CWR	4
392
#define	TCP_ECN_SEEN		8
393

E
Eric Dumazet 已提交
394
enum tcp_tw_status {
L
Linus Torvalds 已提交
395 396 397 398 399 400 401
	TCP_TW_SUCCESS = 0,
	TCP_TW_RST = 1,
	TCP_TW_ACK = 2,
	TCP_TW_SYN = 3
};


402 403 404 405
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,
406
			   struct request_sock *req, bool fastopen);
407 408
int tcp_child_process(struct sock *parent, struct sock *child,
		      struct sk_buff *skb);
409
void tcp_enter_loss(struct sock *sk);
410
void tcp_cwnd_reduction(struct sock *sk, int newly_acked_sacked, int flag);
411 412 413 414
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);
415
bool tcp_peer_is_proven(struct request_sock *req, struct dst_entry *dst);
416 417 418
void tcp_disable_fack(struct tcp_sock *tp);
void tcp_close(struct sock *sk, long timeout);
void tcp_init_sock(struct sock *sk);
419
void tcp_init_transfer(struct sock *sk, int bpf_op);
420 421 422 423 424 425 426
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
427
			  char __user *optval, int __user *optlen);
428
int compat_tcp_setsockopt(struct sock *sk, int level, int optname,
C
Changli Gao 已提交
429
			  char __user *optval, unsigned int optlen);
430
void tcp_set_keepalive(struct sock *sk, int val);
431
void tcp_syn_ack_timeout(const struct request_sock *req);
432 433
int tcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int nonblock,
		int flags, int *addr_len);
434
void tcp_parse_options(const struct net *net, const struct sk_buff *skb,
435 436 437
		       struct tcp_options_received *opt_rx,
		       int estab, struct tcp_fastopen_cookie *foc);
const u8 *tcp_parse_md5sig_option(const struct tcphdr *th);
438

L
Linus Torvalds 已提交
439 440 441 442
/*
 *	TCP v4 functions exported for the inet6 API
 */

443
void tcp_v4_send_check(struct sock *sk, struct sk_buff *skb);
444
void tcp_v4_mtu_reduced(struct sock *sk);
445
void tcp_req_err(struct sock *sk, u32 seq, bool abort);
446
int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb);
447
struct sock *tcp_create_openreq_child(const struct sock *sk,
448 449
				      struct request_sock *req,
				      struct sk_buff *skb);
450
void tcp_ca_openreq_child(struct sock *sk, const struct dst_entry *dst);
451
struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
452
				  struct request_sock *req,
453 454 455
				  struct dst_entry *dst,
				  struct request_sock *req_unhash,
				  bool *own_req);
456 457 458
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);
459 460 461 462 463
enum tcp_synack_type {
	TCP_SYNACK_NORMAL,
	TCP_SYNACK_FASTOPEN,
	TCP_SYNACK_COOKIE,
};
464
struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
465
				struct request_sock *req,
466
				struct tcp_fastopen_cookie *foc,
467
				enum tcp_synack_type synack_type);
468
int tcp_disconnect(struct sock *sk, int flags);
L
Linus Torvalds 已提交
469

P
Pavel Emelyanov 已提交
470
void tcp_finish_connect(struct sock *sk, struct sk_buff *skb);
471
int tcp_send_rcvq(struct sock *sk, struct msghdr *msg, size_t size);
472
void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb);
L
Linus Torvalds 已提交
473 474

/* From syncookies.c */
475 476
struct sock *tcp_get_cookie_sock(struct sock *sk, struct sk_buff *skb,
				 struct request_sock *req,
477
				 struct dst_entry *dst, u32 tsoff);
478 479
int __cookie_v4_check(const struct iphdr *iph, const struct tcphdr *th,
		      u32 cookie);
C
Cong Wang 已提交
480
struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb);
481
#ifdef CONFIG_SYN_COOKIES
482

483
/* Syncookies use a monotonic timer which increments every 60 seconds.
484 485 486
 * 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,
487
 * i.e. a sent cookie is valid only at most for 2*60 seconds (or less if
488 489
 * the counter advances immediately after a cookie is generated).
 */
490 491 492 493 494 495
#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)
496
 * It is racy as we do not hold a lock, but race is very minor.
497
 */
498
static inline void tcp_synq_overflow(const struct sock *sk)
499 500 501 502 503 504 505 506 507 508 509 510 511 512 513
{
	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);
}
514 515 516

static inline u32 tcp_cookie_time(void)
{
517 518
	u64 val = get_jiffies_64();

519
	do_div(val, TCP_SYNCOOKIE_PERIOD);
520
	return val;
521 522
}

523 524
u32 __cookie_v4_init_sequence(const struct iphdr *iph, const struct tcphdr *th,
			      u16 *mssp);
525
__u32 cookie_v4_init_sequence(const struct sk_buff *skb, __u16 *mss);
526
u64 cookie_init_timestamp(struct request_sock *req);
E
Eric Dumazet 已提交
527 528
bool cookie_timestamp_decode(const struct net *net,
			     struct tcp_options_received *opt);
529
bool cookie_ecn_ok(const struct tcp_options_received *opt,
530
		   const struct net *net, const struct dst_entry *dst);
531

532
/* From net/ipv6/syncookies.c */
533 534 535
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);
536

537 538
u32 __cookie_v6_init_sequence(const struct ipv6hdr *iph,
			      const struct tcphdr *th, u16 *mssp);
539
__u32 cookie_v6_init_sequence(const struct sk_buff *skb, __u16 *mss);
540
#endif
L
Linus Torvalds 已提交
541 542
/* tcp_output.c */

543 544
u32 tcp_tso_autosize(const struct sock *sk, unsigned int mss_now,
		     int min_tso_segs);
545 546
void __tcp_push_pending_frames(struct sock *sk, unsigned int cur_mss,
			       int nonagle);
547 548
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);
549 550 551
void tcp_retransmit_timer(struct sock *sk);
void tcp_xmit_retransmit_queue(struct sock *);
void tcp_simple_retransmit(struct sock *);
552
void tcp_enter_recovery(struct sock *sk, bool ece_ack);
553
int tcp_trim_head(struct sock *, struct sk_buff *, u32);
554 555 556 557 558 559 560
enum tcp_queue {
	TCP_FRAG_IN_WRITE_QUEUE,
	TCP_FRAG_IN_RTX_QUEUE,
};
int tcp_fragment(struct sock *sk, enum tcp_queue tcp_queue,
		 struct sk_buff *skb, u32 len,
		 unsigned int mss_now, gfp_t gfp);
561 562 563

void tcp_send_probe0(struct sock *);
void tcp_send_partial(struct sock *);
564
int tcp_write_wakeup(struct sock *, int mib);
565 566 567 568 569 570 571 572
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);
573 574
void tcp_skb_collapse_tstamp(struct sk_buff *skb,
			     const struct sk_buff *next_skb);
L
Linus Torvalds 已提交
575

576
/* tcp_input.c */
577
void tcp_rearm_rto(struct sock *sk);
Y
Yuchung Cheng 已提交
578
void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req);
579
void tcp_reset(struct sock *sk);
Y
Yuchung Cheng 已提交
580
void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, struct sk_buff *skb);
581
void tcp_fin(struct sock *sk);
582

L
Linus Torvalds 已提交
583
/* tcp_timer.c */
584
void tcp_init_xmit_timers(struct sock *);
585 586
static inline void tcp_clear_xmit_timers(struct sock *sk)
{
587
	hrtimer_cancel(&tcp_sk(sk)->pacing_timer);
588 589
	inet_csk_clear_xmit_timers(sk);
}
L
Linus Torvalds 已提交
590

591 592
unsigned int tcp_sync_mss(struct sock *sk, u32 pmtu);
unsigned int tcp_current_mss(struct sock *sk);
I
Ilpo Järvinen 已提交
593 594 595 596

/* 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)
{
597 598 599 600 601 602 603 604 605
	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.
	 */
606
	if (tp->max_window > TCP_MSS_DEFAULT)
607 608 609 610 611 612
		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
613 614 615
	else
		return pktsize;
}
L
Linus Torvalds 已提交
616

617
/* tcp.c */
618
void tcp_get_info(struct sock *, struct tcp_info *);
L
Linus Torvalds 已提交
619 620

/* Read 'sendfile()'-style from a TCP socket */
621 622
int tcp_read_sock(struct sock *sk, read_descriptor_t *desc,
		  sk_read_actor_t recv_actor);
L
Linus Torvalds 已提交
623

624
void tcp_initialize_rcv_mss(struct sock *sk);
L
Linus Torvalds 已提交
625

626 627 628 629
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
630

631 632 633 634 635 636 637 638
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)
{
639
	return usecs_to_jiffies((tp->srtt_us >> 3) + tp->rttvar_us);
640 641
}

642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664
static inline void __tcp_fast_path_on(struct tcp_sock *tp, u32 snd_wnd)
{
	tp->pred_flags = htonl((tp->tcp_header_len << 26) |
			       ntohl(TCP_FLAG_ACK) |
			       snd_wnd);
}

static inline void tcp_fast_path_on(struct tcp_sock *tp)
{
	__tcp_fast_path_on(tp, tp->snd_wnd >> tp->rx_opt.snd_wscale);
}

static inline void tcp_fast_path_check(struct sock *sk)
{
	struct tcp_sock *tp = tcp_sk(sk);

	if (RB_EMPTY_ROOT(&tp->out_of_order_queue) &&
	    tp->rcv_wnd &&
	    atomic_read(&sk->sk_rmem_alloc) < sk->sk_rcvbuf &&
	    !tp->urg_data)
		tcp_fast_path_on(tp);
}

665 666 667
/* Compute the actual rto_min value */
static inline u32 tcp_rto_min(struct sock *sk)
{
668
	const struct dst_entry *dst = __sk_dst_get(sk);
669 670 671 672 673 674 675
	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;
}

676 677 678 679 680
static inline u32 tcp_rto_min_us(struct sock *sk)
{
	return jiffies_to_usecs(tcp_rto_min(sk));
}

681 682 683 684 685
static inline bool tcp_ca_dst_locked(const struct dst_entry *dst)
{
	return dst_metric_locked(dst, RTAX_CC_ALGO);
}

686 687 688
/* Minimum RTT in usec. ~0 means not available. */
static inline u32 tcp_min_rtt(const struct tcp_sock *tp)
{
689
	return minmax_get(&tp->rtt_min);
690 691
}

L
Linus Torvalds 已提交
692 693 694 695
/* 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
696
static inline u32 tcp_receive_window(const struct tcp_sock *tp)
L
Linus Torvalds 已提交
697 698 699 700 701 702 703 704 705 706 707 708
{
	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.
 */
709
u32 __tcp_select_window(struct sock *sk);
L
Linus Torvalds 已提交
710

P
Pavel Emelyanov 已提交
711 712
void tcp_send_window_probe(struct sock *sk);

E
Eric Dumazet 已提交
713 714 715 716 717 718
/* TCP uses 32bit jiffies to save some space.
 * Note that this is different from tcp_time_stamp, which
 * historically has been the same until linux-4.13.
 */
#define tcp_jiffies32 ((u32)jiffies)

719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
/*
 * Deliver a 32bit value for TCP timestamp option (RFC 7323)
 * It is no longer tied to jiffies, but to 1 ms clock.
 * Note: double check if you want to use tcp_jiffies32 instead of this.
 */
#define TCP_TS_HZ	1000

static inline u64 tcp_clock_ns(void)
{
	return local_clock();
}

static inline u64 tcp_clock_us(void)
{
	return div_u64(tcp_clock_ns(), NSEC_PER_USEC);
}

/* This should only be used in contexts where tp->tcp_mstamp is up to date */
static inline u32 tcp_time_stamp(const struct tcp_sock *tp)
{
	return div_u64(tp->tcp_mstamp, USEC_PER_SEC / TCP_TS_HZ);
}

/* Could use tcp_clock_us() / 1000, but this version uses a single divide */
static inline u32 tcp_time_stamp_raw(void)
{
	return div_u64(tcp_clock_ns(), NSEC_PER_SEC / TCP_TS_HZ);
}


/* Refresh 1us clock of a TCP socket,
 * ensuring monotically increasing values.
L
Linus Torvalds 已提交
751
 */
752 753 754 755 756 757 758 759 760 761 762 763
static inline void tcp_mstamp_refresh(struct tcp_sock *tp)
{
	u64 val = tcp_clock_us();

	if (val > tp->tcp_mstamp)
		tp->tcp_mstamp = val;
}

static inline u32 tcp_stamp_us_delta(u64 t1, u64 t0)
{
	return max_t(s64, t1 - t0, 0);
}
L
Linus Torvalds 已提交
764

E
Eric Dumazet 已提交
765 766
static inline u32 tcp_skb_timestamp(const struct sk_buff *skb)
{
767
	return div_u64(skb->skb_mstamp, USEC_PER_SEC / TCP_TS_HZ);
E
Eric Dumazet 已提交
768 769 770
}


C
Changli Gao 已提交
771 772 773 774 775 776 777 778 779 780 781
#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

782 783
#define TCPHDR_SYN_ECN	(TCPHDR_SYN | TCPHDR_ECE | TCPHDR_CWR)

S
Stephen Hemminger 已提交
784
/* This is what the send packet queuing engine uses to pass
785 786 787 788
 * 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
789 790 791 792
 */
struct tcp_skb_cb {
	__u32		seq;		/* Starting sequence number	*/
	__u32		end_seq;	/* SEQ + FIN + SYN + datalen	*/
793 794 795 796
	union {
		/* Note : tcp_tw_isn is used in input path only
		 *	  (isn chosen by tcp_timewait_state_process())
		 *
797 798
		 * 	  tcp_gso_segs/size are used in write queue only,
		 *	  cf tcp_skb_pcount()/tcp_skb_mss()
799 800
		 */
		__u32		tcp_tw_isn;
801 802 803 804
		struct {
			u16	tcp_gso_segs;
			u16	tcp_gso_size;
		};
805
	};
E
Eric Dumazet 已提交
806
	__u8		tcp_flags;	/* TCP header flags. (tcp[13])	*/
807

L
Linus Torvalds 已提交
808 809 810 811 812
	__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			*/
813
#define TCPCB_REPAIRED		0x10	/* SKB repaired (no skb_mstamp)	*/
L
Linus Torvalds 已提交
814
#define TCPCB_EVER_RETRANS	0x80	/* Ever retransmitted frame	*/
815 816
#define TCPCB_RETRANS		(TCPCB_SACKED_RETRANS|TCPCB_EVER_RETRANS| \
				TCPCB_REPAIRED)
L
Linus Torvalds 已提交
817

818
	__u8		ip_dsfield;	/* IPv4 tos or IPv6 dsfield	*/
819
	__u8		txstamp_ack:1,	/* Record TX timestamp for ack? */
820
			eor:1,		/* Is skb MSG_EOR marked? */
821 822
			has_rxtstamp:1,	/* SKB has a RX timestamp	*/
			unused:5;
L
Linus Torvalds 已提交
823
	__u32		ack_seq;	/* Sequence number ACK'd	*/
824
	union {
825
		struct {
826
			/* There is space for up to 24 bytes */
827 828 829
			__u32 in_flight:30,/* Bytes in flight at transmit */
			      is_app_limited:1, /* cwnd not fully used? */
			      unused:1;
830 831 832
			/* pkts S/ACKed so far upon tx of skb, incl retrans: */
			__u32 delivered;
			/* start of send pipeline phase */
833
			u64 first_tx_mstamp;
834
			/* when we reached the "delivered" count */
835
			u64 delivered_mstamp;
836 837 838
		} tx;   /* only used for outgoing skbs */
		union {
			struct inet_skb_parm	h4;
839
#if IS_ENABLED(CONFIG_IPV6)
840
			struct inet6_skb_parm	h6;
841
#endif
842
		} header;	/* For incoming skbs */
843 844 845 846 847
		struct {
			__u32 key;
			__u32 flags;
			struct bpf_map *map;
		} bpf;
848
	};
L
Linus Torvalds 已提交
849 850 851 852
};

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

E
Eric Dumazet 已提交
853

854
#if IS_ENABLED(CONFIG_IPV6)
E
Eric Dumazet 已提交
855 856 857 858 859
/* 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)
{
860
	bool l3_slave = ipv6_l3mdev_skb(TCP_SKB_CB(skb)->header.h6.flags);
861 862

	return l3_slave ? skb->skb_iif : TCP_SKB_CB(skb)->header.h6.iif;
E
Eric Dumazet 已提交
863
}
864 865 866 867 868 869 870 871 872 873

/* TCP_SKB_CB reference means this can not be used from early demux */
static inline int tcp_v6_sdif(const struct sk_buff *skb)
{
#if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
	if (skb && ipv6_l3mdev_skb(TCP_SKB_CB(skb)->header.h6.flags))
		return TCP_SKB_CB(skb)->header.h6.iif;
#endif
	return 0;
}
874
#endif
E
Eric Dumazet 已提交
875

876 877 878 879 880
/* 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 &&
881
	    skb && ipv4_l3mdev_skb(TCP_SKB_CB(skb)->header.h4.flags))
882 883 884 885 886
		return true;
#endif
	return false;
}

887 888 889 890 891 892 893 894 895 896
/* TCP_SKB_CB reference means this can not be used from early demux */
static inline int tcp_v4_sdif(struct sk_buff *skb)
{
#if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
	if (skb && ipv4_l3mdev_skb(TCP_SKB_CB(skb)->header.h4.flags))
		return TCP_SKB_CB(skb)->header.h4.iif;
#endif
	return 0;
}

L
Linus Torvalds 已提交
897 898
/* Due to TSO, an SKB can be composed of multiple actual
 * packets.  To keep these tracked properly, we use this.
899
 */
L
Linus Torvalds 已提交
900
static inline int tcp_skb_pcount(const struct sk_buff *skb)
901
{
902 903
	return TCP_SKB_CB(skb)->tcp_gso_segs;
}
904

905 906 907
static inline void tcp_skb_pcount_set(struct sk_buff *skb, int segs)
{
	TCP_SKB_CB(skb)->tcp_gso_segs = segs;
908 909
}

910
static inline void tcp_skb_pcount_add(struct sk_buff *skb, int segs)
L
Linus Torvalds 已提交
911
{
912
	TCP_SKB_CB(skb)->tcp_gso_segs += segs;
L
Linus Torvalds 已提交
913 914
}

915
/* This is valid iff skb is in write queue and tcp_skb_pcount() > 1. */
L
Linus Torvalds 已提交
916 917
static inline int tcp_skb_mss(const struct sk_buff *skb)
{
918
	return TCP_SKB_CB(skb)->tcp_gso_size;
L
Linus Torvalds 已提交
919 920
}

921 922 923 924 925
static inline bool tcp_skb_can_collapse_to(const struct sk_buff *skb)
{
	return likely(!TCP_SKB_CB(skb)->eor);
}

926 927 928 929 930 931
/* 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 */
932 933 934 935
	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,
936 937
};

938
/* Information about inbound ACK, passed to cong_ops->in_ack_event() */
939
enum tcp_ca_ack_event_flags {
940 941 942
	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 */
943 944 945 946 947 948
};

/*
 * Interface for adding new TCP congestion control handlers
 */
#define TCP_CA_NAME_MAX	16
949 950 951
#define TCP_CA_MAX	128
#define TCP_CA_BUF_MAX	(TCP_CA_NAME_MAX*TCP_CA_MAX)

952 953
#define TCP_CA_UNSPEC	0

954
/* Algorithm can be set on socket without CAP_NET_ADMIN privileges */
955
#define TCP_CONG_NON_RESTRICTED 0x1
956 957
/* Requires ECN/ECT set on all packets */
#define TCP_CONG_NEEDS_ECN	0x2
958

959 960
union tcp_cc_info;

961 962 963
struct ack_sample {
	u32 pkts_acked;
	s32 rtt_us;
964
	u32 in_flight;
965 966
};

967 968 969 970 971 972 973 974 975
/* 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 {
976
	u64  prior_mstamp; /* starting timestamp for interval */
977 978 979 980 981 982 983
	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 */
984
	bool is_app_limited;	/* is sample from packet with bubble in pipe? */
985 986 987
	bool is_retrans;	/* is sample from retransmission? */
};

988 989
struct tcp_congestion_ops {
	struct list_head	list;
990 991
	u32 key;
	u32 flags;
992 993

	/* initialize private data (optional) */
994
	void (*init)(struct sock *sk);
995
	/* cleanup private data  (optional) */
996
	void (*release)(struct sock *sk);
997 998

	/* return slow start threshold (required) */
999
	u32 (*ssthresh)(struct sock *sk);
1000
	/* do new cwnd calculation (required) */
1001
	void (*cong_avoid)(struct sock *sk, u32 ack, u32 acked);
1002
	/* call before changing ca_state (optional) */
1003
	void (*set_state)(struct sock *sk, u8 new_state);
1004
	/* call when cwnd event occurs (optional) */
1005
	void (*cwnd_event)(struct sock *sk, enum tcp_ca_event ev);
1006 1007
	/* call when ack arrives (optional) */
	void (*in_ack_event)(struct sock *sk, u32 flags);
1008
	/* new value of cwnd after loss (required) */
1009
	u32  (*undo_cwnd)(struct sock *sk);
1010
	/* hook for packet ack accounting (optional) */
1011
	void (*pkts_acked)(struct sock *sk, const struct ack_sample *sample);
1012 1013
	/* suggest number of segments for each skb to transmit (optional) */
	u32 (*tso_segs_goal)(struct sock *sk);
1014 1015
	/* returns the multiplier used in tcp_sndbuf_expand (optional) */
	u32 (*sndbuf_expand)(struct sock *sk);
1016 1017 1018 1019
	/* 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);
1020
	/* get info for inet_diag (optional) */
1021 1022
	size_t (*get_info)(struct sock *sk, u32 ext, int *attr,
			   union tcp_cc_info *info);
1023 1024 1025 1026 1027

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

1028 1029
int tcp_register_congestion_control(struct tcp_congestion_ops *type);
void tcp_unregister_congestion_control(struct tcp_congestion_ops *type);
1030

1031
void tcp_assign_congestion_control(struct sock *sk);
1032 1033 1034 1035 1036 1037 1038
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);
1039
int tcp_set_congestion_control(struct sock *sk, const char *name, bool load, bool reinit);
N
Neal Cardwell 已提交
1040 1041
u32 tcp_slow_start(struct tcp_sock *tp, u32 acked);
void tcp_cong_avoid_ai(struct tcp_sock *tp, u32 w, u32 acked);
1042

1043
u32 tcp_reno_ssthresh(struct sock *sk);
1044
u32 tcp_reno_undo_cwnd(struct sock *sk);
1045
void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 acked);
1046
extern struct tcp_congestion_ops tcp_reno;
1047

1048
struct tcp_congestion_ops *tcp_ca_find_key(u32 key);
1049
u32 tcp_ca_get_key_by_name(const char *name, bool *ecn_ca);
1050
#ifdef CONFIG_INET
1051
char *tcp_ca_get_name_by_key(u32 key, char *buffer);
1052 1053 1054 1055 1056 1057
#else
static inline char *tcp_ca_get_name_by_key(u32 key, char *buffer)
{
	return NULL;
}
#endif
1058

1059 1060 1061 1062 1063 1064 1065
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;
}

1066
static inline void tcp_set_ca_state(struct sock *sk, const u8 ca_state)
1067
{
1068 1069 1070 1071 1072
	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;
1073 1074
}

1075
static inline void tcp_ca_event(struct sock *sk, const enum tcp_ca_event event)
1076
{
1077 1078 1079 1080
	const struct inet_connection_sock *icsk = inet_csk(sk);

	if (icsk->icsk_ca_ops->cwnd_event)
		icsk->icsk_ca_ops->cwnd_event(sk, event);
1081 1082
}

1083 1084 1085 1086 1087
/* 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,
1088
		  struct rate_sample *rs);
1089
void tcp_rate_check_app_limited(struct sock *sk);
1090

1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
/* 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
1104
static inline bool tcp_is_reno(const struct tcp_sock *tp)
1105 1106 1107 1108
{
	return !tcp_is_sack(tp);
}

E
Eric Dumazet 已提交
1109
static inline bool tcp_is_fack(const struct tcp_sock *tp)
1110
{
1111
	return tp->rx_opt.sack_ok & TCP_FACK_ENABLED;
1112 1113 1114 1115
}

static inline void tcp_enable_fack(struct tcp_sock *tp)
{
1116
	tp->rx_opt.sack_ok |= TCP_FACK_ENABLED;
1117 1118
}

1119 1120 1121 1122 1123
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
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137
/* 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
1138
static inline unsigned int tcp_packets_in_flight(const struct tcp_sock *tp)
L
Linus Torvalds 已提交
1139
{
1140
	return tp->packets_out - tcp_left_out(tp) + tp->retrans_out;
L
Linus Torvalds 已提交
1141 1142
}

I
Ilpo Järvinen 已提交
1143 1144
#define TCP_INFINITE_SSTHRESH	0x7fffffff

1145 1146
static inline bool tcp_in_slow_start(const struct tcp_sock *tp)
{
1147
	return tp->snd_cwnd < tp->snd_ssthresh;
1148 1149
}

I
Ilpo Järvinen 已提交
1150 1151 1152 1153 1154
static inline bool tcp_in_initial_slowstart(const struct tcp_sock *tp)
{
	return tp->snd_ssthresh >= TCP_INFINITE_SSTHRESH;
}

1155 1156 1157 1158 1159 1160
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
1161
/* If cwnd > ssthresh, we may raise ssthresh to be half-way to cwnd.
1162
 * The exception is cwnd reduction phase, when cwnd is decreasing towards
L
Linus Torvalds 已提交
1163 1164
 * ssthresh.
 */
1165
static inline __u32 tcp_current_ssthresh(const struct sock *sk)
L
Linus Torvalds 已提交
1166
{
1167
	const struct tcp_sock *tp = tcp_sk(sk);
1168

1169
	if (tcp_in_cwnd_reduction(sk))
L
Linus Torvalds 已提交
1170 1171 1172 1173 1174 1175 1176
		return tp->snd_ssthresh;
	else
		return max(tp->snd_ssthresh,
			   ((tp->snd_cwnd >> 1) +
			    (tp->snd_cwnd >> 2)));
}

1177 1178
/* 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
1179

1180
void tcp_enter_cwr(struct sock *sk);
1181
__u32 tcp_init_cwnd(const struct tcp_sock *tp, const struct dst_entry *dst);
L
Linus Torvalds 已提交
1182

1183 1184 1185 1186 1187 1188 1189 1190
/* 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;
}

1191 1192 1193 1194 1195
/* 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;
}
1196 1197 1198

/* We follow the spirit of RFC2861 to validate cwnd but implement a more
 * flexible approach. The RFC suggests cwnd should not be raised unless
1199 1200 1201
 * 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.
1202 1203 1204 1205 1206 1207 1208 1209
 * 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.
 */
1210
static inline bool tcp_is_cwnd_limited(const struct sock *sk)
1211 1212 1213
{
	const struct tcp_sock *tp = tcp_sk(sk);

1214
	/* If in slow start, ensure cwnd grows to twice what was ACKed. */
1215
	if (tcp_in_slow_start(tp))
1216 1217 1218
		return tp->snd_cwnd < 2 * tp->max_packets_out;

	return tp->is_cwnd_limited;
1219
}
1220

1221 1222 1223 1224 1225 1226 1227
/* 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
1228
{
1229 1230
	return max_t(unsigned long, inet_csk(sk)->icsk_rto, TCP_RTO_MIN);
}
1231

1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243
/* 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)
1244
		inet_csk_reset_xmit_timer(sk, ICSK_TIME_PROBE0,
1245
					  tcp_probe0_base(sk), TCP_RTO_MAX);
L
Linus Torvalds 已提交
1246 1247
}

1248
static inline void tcp_init_wl(struct tcp_sock *tp, u32 seq)
L
Linus Torvalds 已提交
1249 1250 1251 1252
{
	tp->snd_wl1 = seq;
}

1253
static inline void tcp_update_wl(struct tcp_sock *tp, u32 seq)
L
Linus Torvalds 已提交
1254 1255 1256 1257 1258 1259 1260
{
	tp->snd_wl1 = seq;
}

/*
 * Calculate(/check) TCP checksum
 */
1261 1262
static inline __sum16 tcp_v4_check(int len, __be32 saddr,
				   __be32 daddr, __wsum base)
L
Linus Torvalds 已提交
1263 1264 1265 1266
{
	return csum_tcpudp_magic(saddr,daddr,len,IPPROTO_TCP,base);
}

1267
static inline __sum16 __tcp_checksum_complete(struct sk_buff *skb)
L
Linus Torvalds 已提交
1268
{
1269
	return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
1270 1271
}

E
Eric Dumazet 已提交
1272
static inline bool tcp_checksum_complete(struct sk_buff *skb)
L
Linus Torvalds 已提交
1273
{
1274
	return !skb_csum_unnecessary(skb) &&
L
Linus Torvalds 已提交
1275 1276 1277
		__tcp_checksum_complete(skb);
}

E
Eric Dumazet 已提交
1278
bool tcp_add_backlog(struct sock *sk, struct sk_buff *skb);
1279
int tcp_filter(struct sock *sk, struct sk_buff *skb);
L
Linus Torvalds 已提交
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289

#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
1290
void tcp_set_state(struct sock *sk, int state);
L
Linus Torvalds 已提交
1291

1292
void tcp_done(struct sock *sk);
L
Linus Torvalds 已提交
1293

1294 1295
int tcp_abort(struct sock *sk, int err);

S
Stephen Hemminger 已提交
1296
static inline void tcp_sack_reset(struct tcp_options_received *rx_opt)
L
Linus Torvalds 已提交
1297 1298 1299 1300 1301
{
	rx_opt->dsack = 0;
	rx_opt->num_sacks = 0;
}

1302
u32 tcp_default_init_rwnd(u32 mss);
1303 1304 1305 1306
void tcp_cwnd_restart(struct sock *sk, s32 delta);

static inline void tcp_slow_start_after_idle_check(struct sock *sk)
{
1307
	const struct tcp_congestion_ops *ca_ops = inet_csk(sk)->icsk_ca_ops;
1308 1309 1310
	struct tcp_sock *tp = tcp_sk(sk);
	s32 delta;

1311 1312
	if (!sysctl_tcp_slow_start_after_idle || tp->packets_out ||
	    ca_ops->cong_control)
1313
		return;
1314
	delta = tcp_jiffies32 - tp->lsndtime;
1315 1316 1317
	if (delta > inet_csk(sk)->icsk_rto)
		tcp_cwnd_restart(sk, delta);
}
1318

L
Linus Torvalds 已提交
1319
/* Determine a window scaling and initial window to offer. */
1320 1321 1322
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
1323 1324 1325

static inline int tcp_win_from_space(int space)
{
1326 1327 1328 1329 1330
	int tcp_adv_win_scale = sysctl_tcp_adv_win_scale;

	return tcp_adv_win_scale <= 0 ?
		(space>>(-tcp_adv_win_scale)) :
		space - (space>>tcp_adv_win_scale);
L
Linus Torvalds 已提交
1331 1332
}

1333
/* Note: caller must be prepared to deal with negative returns */
L
Linus Torvalds 已提交
1334 1335 1336 1337
static inline int tcp_space(const struct sock *sk)
{
	return tcp_win_from_space(sk->sk_rcvbuf -
				  atomic_read(&sk->sk_rmem_alloc));
1338
}
L
Linus Torvalds 已提交
1339 1340 1341

static inline int tcp_full_space(const struct sock *sk)
{
1342
	return tcp_win_from_space(sk->sk_rcvbuf);
L
Linus Torvalds 已提交
1343 1344
}

1345
extern void tcp_openreq_init_rwin(struct request_sock *req,
1346 1347
				  const struct sock *sk_listener,
				  const struct dst_entry *dst);
1348

1349
void tcp_enter_memory_pressure(struct sock *sk);
1350
void tcp_leave_memory_pressure(struct sock *sk);
L
Linus Torvalds 已提交
1351 1352 1353

static inline int keepalive_intvl_when(const struct tcp_sock *tp)
{
1354 1355 1356
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_intvl ? : net->ipv4.sysctl_tcp_keepalive_intvl;
L
Linus Torvalds 已提交
1357 1358 1359 1360
}

static inline int keepalive_time_when(const struct tcp_sock *tp)
{
1361 1362 1363
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_time ? : net->ipv4.sysctl_tcp_keepalive_time;
L
Linus Torvalds 已提交
1364 1365
}

E
Eric Dumazet 已提交
1366 1367
static inline int keepalive_probes(const struct tcp_sock *tp)
{
1368 1369 1370
	struct net *net = sock_net((struct sock *)tp);

	return tp->keepalive_probes ? : net->ipv4.sysctl_tcp_keepalive_probes;
E
Eric Dumazet 已提交
1371 1372
}

1373 1374 1375 1376
static inline u32 keepalive_time_elapsed(const struct tcp_sock *tp)
{
	const struct inet_connection_sock *icsk = &tp->inet_conn;

1377 1378
	return min_t(u32, tcp_jiffies32 - icsk->icsk_ack.lrcvtime,
			  tcp_jiffies32 - tp->rcv_tstamp);
1379 1380
}

1381
static inline int tcp_fin_time(const struct sock *sk)
L
Linus Torvalds 已提交
1382
{
1383
	int fin_timeout = tcp_sk(sk)->linger2 ? : sock_net(sk)->ipv4.sysctl_tcp_fin_timeout;
1384
	const int rto = inet_csk(sk)->icsk_rto;
L
Linus Torvalds 已提交
1385

1386 1387
	if (fin_timeout < (rto << 2) - (rto >> 1))
		fin_timeout = (rto << 2) - (rto >> 1);
L
Linus Torvalds 已提交
1388 1389 1390 1391

	return fin_timeout;
}

E
Eric Dumazet 已提交
1392 1393
static inline bool tcp_paws_check(const struct tcp_options_received *rx_opt,
				  int paws_win)
L
Linus Torvalds 已提交
1394
{
I
Ilpo Järvinen 已提交
1395
	if ((s32)(rx_opt->ts_recent - rx_opt->rcv_tsval) <= paws_win)
E
Eric Dumazet 已提交
1396
		return true;
I
Ilpo Järvinen 已提交
1397
	if (unlikely(get_seconds() >= rx_opt->ts_recent_stamp + TCP_PAWS_24DAYS))
E
Eric Dumazet 已提交
1398
		return true;
E
Eric Dumazet 已提交
1399 1400 1401 1402 1403 1404
	/*
	 * 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
1405 1406
		return true;
	return false;
I
Ilpo Järvinen 已提交
1407 1408
}

E
Eric Dumazet 已提交
1409 1410
static inline bool tcp_paws_reject(const struct tcp_options_received *rx_opt,
				   int rst)
I
Ilpo Järvinen 已提交
1411 1412
{
	if (tcp_paws_check(rx_opt, 0))
E
Eric Dumazet 已提交
1413
		return false;
L
Linus Torvalds 已提交
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426

	/* 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.
	 */
1427
	if (rst && get_seconds() >= rx_opt->ts_recent_stamp + TCP_PAWS_MSL)
E
Eric Dumazet 已提交
1428 1429
		return false;
	return true;
L
Linus Torvalds 已提交
1430 1431
}

1432 1433
bool tcp_oow_rate_limited(struct net *net, const struct sk_buff *skb,
			  int mib_idx, u32 *last_oow_ack_time);
1434

P
Pavel Emelyanov 已提交
1435
static inline void tcp_mib_init(struct net *net)
L
Linus Torvalds 已提交
1436 1437
{
	/* See RFC 2012 */
1438 1439 1440 1441
	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
1442 1443
}

1444
/* from STCP */
1445
static inline void tcp_clear_retrans_hints_partial(struct tcp_sock *tp)
D
David S. Miller 已提交
1446
{
1447
	tp->lost_skb_hint = NULL;
1448 1449 1450 1451 1452
}

static inline void tcp_clear_all_retrans_hints(struct tcp_sock *tp)
{
	tcp_clear_retrans_hints_partial(tp);
1453
	tp->retransmit_skb_hint = NULL;
1454 1455
}

E
Eric Dumazet 已提交
1456 1457 1458 1459 1460 1461 1462
union tcp_md5_addr {
	struct in_addr  a4;
#if IS_ENABLED(CONFIG_IPV6)
	struct in6_addr	a6;
#endif
};

1463 1464
/* - key database */
struct tcp_md5sig_key {
E
Eric Dumazet 已提交
1465
	struct hlist_node	node;
1466
	u8			keylen;
E
Eric Dumazet 已提交
1467 1468
	u8			family; /* AF_INET or AF_INET6 */
	union tcp_md5_addr	addr;
1469
	u8			prefixlen;
E
Eric Dumazet 已提交
1470 1471
	u8			key[TCP_MD5SIG_MAXKEYLEN];
	struct rcu_head		rcu;
1472 1473 1474 1475
};

/* - sock block */
struct tcp_md5sig_info {
E
Eric Dumazet 已提交
1476
	struct hlist_head	head;
1477
	struct rcu_head		rcu;
1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497
};

/* - 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
1498
#if IS_ENABLED(CONFIG_IPV6)
1499 1500 1501 1502 1503 1504
	struct tcp6_pseudohdr ip6;
#endif
};

/* - pool: digest algorithm, hash description and scratch buffer */
struct tcp_md5sig_pool {
H
Herbert Xu 已提交
1505
	struct ahash_request	*md5_req;
1506
	void			*scratch;
1507 1508 1509
};

/* - functions */
1510 1511
int tcp_v4_md5_hash_skb(char *md5_hash, const struct tcp_md5sig_key *key,
			const struct sock *sk, const struct sk_buff *skb);
1512
int tcp_md5_do_add(struct sock *sk, const union tcp_md5_addr *addr,
1513 1514
		   int family, u8 prefixlen, const u8 *newkey, u8 newkeylen,
		   gfp_t gfp);
1515
int tcp_md5_do_del(struct sock *sk, const union tcp_md5_addr *addr,
1516
		   int family, u8 prefixlen);
1517
struct tcp_md5sig_key *tcp_v4_md5_lookup(const struct sock *sk,
1518
					 const struct sock *addr_sk);
1519

1520
#ifdef CONFIG_TCP_MD5SIG
1521
struct tcp_md5sig_key *tcp_md5_do_lookup(const struct sock *sk,
1522 1523
					 const union tcp_md5_addr *addr,
					 int family);
E
Eric Dumazet 已提交
1524
#define tcp_twsk_md5_key(twsk)	((twsk)->tw_md5_key)
1525
#else
1526
static inline struct tcp_md5sig_key *tcp_md5_do_lookup(const struct sock *sk,
E
Eric Dumazet 已提交
1527 1528 1529 1530 1531
					 const union tcp_md5_addr *addr,
					 int family)
{
	return NULL;
}
1532 1533 1534
#define tcp_twsk_md5_key(twsk)	NULL
#endif

1535
bool tcp_alloc_md5sig_pool(void);
1536

1537
struct tcp_md5sig_pool *tcp_get_md5sig_pool(void);
1538 1539 1540 1541
static inline void tcp_put_md5sig_pool(void)
{
	local_bh_enable();
}
E
Eric Dumazet 已提交
1542

1543 1544 1545 1546
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);
1547

1548
/* From tcp_fastopen.c */
1549 1550 1551 1552
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,
1553 1554
			    struct tcp_fastopen_cookie *cookie, bool syn_lost,
			    u16 try_exp);
1555 1556 1557 1558
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 */
1559 1560
	size_t				size;
	int				copied;	/* queued in tcp_connect() */
1561 1562
};
void tcp_free_fastopen_req(struct tcp_sock *tp);
1563
void tcp_fastopen_destroy_cipher(struct sock *sk);
1564
void tcp_fastopen_ctx_destroy(struct net *net);
1565 1566
int tcp_fastopen_reset_cipher(struct net *net, struct sock *sk,
			      void *key, unsigned int len);
1567
void tcp_fastopen_add_skb(struct sock *sk, struct sk_buff *skb);
1568 1569
struct sock *tcp_try_fastopen(struct sock *sk, struct sk_buff *skb,
			      struct request_sock *req,
1570
			      struct tcp_fastopen_cookie *foc);
1571
void tcp_fastopen_init_key_once(struct net *net);
1572 1573
bool tcp_fastopen_cookie_check(struct sock *sk, u16 *mss,
			     struct tcp_fastopen_cookie *cookie);
W
Wei Wang 已提交
1574
bool tcp_fastopen_defer_connect(struct sock *sk, int *err);
1575 1576 1577 1578
#define TCP_FASTOPEN_KEY_LENGTH 16

/* Fastopen key context */
struct tcp_fastopen_context {
1579 1580 1581
	struct crypto_cipher	*tfm;
	__u8			key[TCP_FASTOPEN_KEY_LENGTH];
	struct rcu_head		rcu;
1582 1583
};

1584
extern unsigned int sysctl_tcp_fastopen_blackhole_timeout;
1585
void tcp_fastopen_active_disable(struct sock *sk);
1586 1587 1588 1589
bool tcp_fastopen_active_should_disable(struct sock *sk);
void tcp_fastopen_active_disable_ofo_check(struct sock *sk);
void tcp_fastopen_active_timeout_reset(void);

1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603
/* 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);

1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620
/* This helper is needed, because skb->tcp_tsorted_anchor uses
 * the same memory storage than skb->destructor/_skb_refdst
 */
static inline void tcp_skb_tsorted_anchor_cleanup(struct sk_buff *skb)
{
	skb->destructor = NULL;
	skb->_skb_refdst = 0UL;
}

#define tcp_skb_tsorted_save(skb) {		\
	unsigned long _save = skb->_skb_refdst;	\
	skb->_skb_refdst = 0UL;

#define tcp_skb_tsorted_restore(skb)		\
	skb->_skb_refdst = _save;		\
}

1621
void tcp_write_queue_purge(struct sock *sk);
1622

1623 1624 1625 1626 1627
static inline struct sk_buff *tcp_rtx_queue_head(const struct sock *sk)
{
	return skb_rb_first(&sk->tcp_rtx_queue);
}

1628
static inline struct sk_buff *tcp_write_queue_head(const struct sock *sk)
1629
{
1630
	return skb_peek(&sk->sk_write_queue);
1631 1632
}

1633
static inline struct sk_buff *tcp_write_queue_tail(const struct sock *sk)
1634
{
1635
	return skb_peek_tail(&sk->sk_write_queue);
1636 1637
}

1638
#define tcp_for_write_queue_from_safe(skb, tmp, sk)			\
1639
	skb_queue_walk_from_safe(&(sk)->sk_write_queue, skb, tmp)
1640

1641
static inline struct sk_buff *tcp_send_head(const struct sock *sk)
1642
{
1643
	return skb_peek(&sk->sk_write_queue);
1644 1645
}

1646 1647 1648 1649 1650 1651
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);
}

1652
static inline bool tcp_write_queue_empty(const struct sock *sk)
1653
{
1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664
	return skb_queue_empty(&sk->sk_write_queue);
}

static inline bool tcp_rtx_queue_empty(const struct sock *sk)
{
	return RB_EMPTY_ROOT(&sk->tcp_rtx_queue);
}

static inline bool tcp_rtx_and_write_queues_empty(const struct sock *sk)
{
	return tcp_rtx_queue_empty(sk) && tcp_write_queue_empty(sk);
1665 1666 1667 1668
}

static inline void tcp_check_send_head(struct sock *sk, struct sk_buff *skb_unlinked)
{
1669
	if (tcp_write_queue_empty(sk))
1670
		tcp_chrono_stop(sk, TCP_CHRONO_BUSY);
1671

1672 1673
	if (tcp_sk(sk)->highest_sack == skb_unlinked)
		tcp_sk(sk)->highest_sack = NULL;
1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
}

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. */
1686
	if (sk->sk_write_queue.next == skb) {
1687
		tcp_chrono_start(sk, TCP_CHRONO_BUSY);
1688 1689 1690 1691

		if (tcp_sk(sk)->highest_sack == NULL)
			tcp_sk(sk)->highest_sack = skb;
	}
1692 1693
}

1694
/* Insert new before skb on the write queue of sk.  */
1695 1696 1697 1698
static inline void tcp_insert_write_queue_before(struct sk_buff *new,
						  struct sk_buff *skb,
						  struct sock *sk)
{
1699
	__skb_queue_before(&sk->sk_write_queue, skb, new);
1700 1701 1702 1703
}

static inline void tcp_unlink_write_queue(struct sk_buff *skb, struct sock *sk)
{
E
Eric Dumazet 已提交
1704
	tcp_skb_tsorted_anchor_cleanup(skb);
1705 1706 1707
	__skb_unlink(skb, &sk->sk_write_queue);
}

1708 1709 1710
void tcp_rbtree_insert(struct rb_root *root, struct sk_buff *skb);

static inline void tcp_rtx_queue_unlink(struct sk_buff *skb, struct sock *sk)
1711
{
1712 1713 1714 1715 1716 1717 1718 1719 1720
	tcp_skb_tsorted_anchor_cleanup(skb);
	rb_erase(&skb->rbnode, &sk->tcp_rtx_queue);
}

static inline void tcp_rtx_queue_unlink_and_free(struct sk_buff *skb, struct sock *sk)
{
	list_del(&skb->tcp_tsorted_anchor);
	tcp_rtx_queue_unlink(skb, sk);
	sk_wmem_free_skb(sk, skb);
1721 1722
}

1723 1724 1725 1726 1727 1728 1729 1730 1731
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);
	}
}

1732 1733 1734
/* 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.
1735 1736 1737 1738 1739
 */
static inline u32 tcp_highest_sack_seq(struct tcp_sock *tp)
{
	if (!tp->sacked_out)
		return tp->snd_una;
1740 1741 1742 1743

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

1744 1745 1746
	return TCP_SKB_CB(tp->highest_sack)->seq;
}

1747 1748
static inline void tcp_advance_highest_sack(struct sock *sk, struct sk_buff *skb)
{
1749 1750 1751
	struct sk_buff *next = skb_rb_next(skb);

	tcp_sk(sk)->highest_sack = next ?: tcp_send_head(sk);
1752 1753 1754 1755 1756 1757 1758 1759 1760
}

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)
{
1761 1762 1763
	struct sk_buff *skb = tcp_rtx_queue_head(sk);

	tcp_sk(sk)->highest_sack = skb ?: tcp_send_head(sk);
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774
}

/* 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;
}

1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786
/* 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
1787 1788 1789
/* 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
1790
static inline bool tcp_stream_is_thin(struct tcp_sock *tp)
A
Andreas Petlund 已提交
1791 1792 1793 1794
{
	return tp->packets_out < 4 && !tcp_in_initial_slowstart(tp);
}

L
Linus Torvalds 已提交
1795 1796 1797 1798 1799 1800
/* /proc */
enum tcp_seq_states {
	TCP_SEQ_STATE_LISTENING,
	TCP_SEQ_STATE_ESTABLISHED,
};

1801 1802
int tcp_seq_open(struct inode *inode, struct file *file);

L
Linus Torvalds 已提交
1803
struct tcp_seq_afinfo {
1804 1805 1806 1807
	char				*name;
	sa_family_t			family;
	const struct file_operations	*seq_fops;
	struct seq_operations		seq_ops;
L
Linus Torvalds 已提交
1808 1809 1810
};

struct tcp_iter_state {
1811
	struct seq_net_private	p;
L
Linus Torvalds 已提交
1812 1813 1814
	sa_family_t		family;
	enum tcp_seq_states	state;
	struct sock		*syn_wait_sk;
1815
	int			bucket, offset, sbucket, num;
1816
	loff_t			last_pos;
L
Linus Torvalds 已提交
1817 1818
};

1819 1820
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
1821

1822
extern struct request_sock_ops tcp_request_sock_ops;
1823
extern struct request_sock_ops tcp6_request_sock_ops;
1824

1825
void tcp_v4_destroy_sock(struct sock *sk);
1826

E
Eric Dumazet 已提交
1827
struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
1828 1829 1830
				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);
1831

1832
void __tcp_v4_send_check(struct sk_buff *skb, __be32 saddr, __be32 daddr);
H
Herbert Xu 已提交
1833

1834 1835
static inline u32 tcp_notsent_lowat(const struct tcp_sock *tp)
{
1836 1837
	struct net *net = sock_net((struct sock *)tp);
	return tp->notsent_lowat ?: net->ipv4.sysctl_tcp_notsent_lowat;
1838 1839 1840 1841 1842 1843 1844 1845 1846 1847
}

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);
}

1848
#ifdef CONFIG_PROC_FS
1849 1850
int tcp4_proc_init(void);
void tcp4_proc_exit(void);
1851 1852
#endif

1853
int tcp_rtx_synack(const struct sock *sk, struct request_sock *req);
O
Octavian Purdila 已提交
1854 1855 1856
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);
1857

1858 1859 1860
/* TCP af-specific functions */
struct tcp_sock_af_ops {
#ifdef CONFIG_TCP_MD5SIG
1861
	struct tcp_md5sig_key	*(*md5_lookup) (const struct sock *sk,
1862
						const struct sock *addr_sk);
1863 1864 1865 1866 1867
	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,
1868
				     int optname,
1869 1870
				     char __user *optval,
				     int optlen);
1871 1872 1873 1874
#endif
};

struct tcp_request_sock_ops {
1875
	u16 mss_clamp;
1876
#ifdef CONFIG_TCP_MD5SIG
1877
	struct tcp_md5sig_key *(*req_md5_lookup)(const struct sock *sk,
1878
						 const struct sock *addr_sk);
1879 1880 1881 1882
	int		(*calc_md5_hash) (char *location,
					  const struct tcp_md5sig_key *md5,
					  const struct sock *sk,
					  const struct sk_buff *skb);
1883
#endif
1884 1885
	void (*init_req)(struct request_sock *req,
			 const struct sock *sk_listener,
1886
			 struct sk_buff *skb);
1887
#ifdef CONFIG_SYN_COOKIES
1888
	__u32 (*cookie_init_seq)(const struct sk_buff *skb,
1889 1890
				 __u16 *mss);
#endif
1891
	struct dst_entry *(*route_req)(const struct sock *sk, struct flowi *fl,
1892
				       const struct request_sock *req);
1893
	u32 (*init_seq)(const struct sk_buff *skb);
1894
	u32 (*init_ts_off)(const struct net *net, const struct sk_buff *skb);
1895
	int (*send_synack)(const struct sock *sk, struct dst_entry *dst,
1896
			   struct flowi *fl, struct request_sock *req,
1897
			   struct tcp_fastopen_cookie *foc,
1898
			   enum tcp_synack_type synack_type);
1899 1900
};

1901 1902
#ifdef CONFIG_SYN_COOKIES
static inline __u32 cookie_init_sequence(const struct tcp_request_sock_ops *ops,
1903
					 const struct sock *sk, struct sk_buff *skb,
1904 1905
					 __u16 *mss)
{
1906
	tcp_synq_overflow(sk);
1907
	__NET_INC_STATS(sock_net(sk), LINUX_MIB_SYNCOOKIESSENT);
1908
	return ops->cookie_init_seq(skb, mss);
1909 1910 1911
}
#else
static inline __u32 cookie_init_sequence(const struct tcp_request_sock_ops *ops,
1912
					 const struct sock *sk, struct sk_buff *skb,
1913 1914 1915 1916 1917 1918
					 __u16 *mss)
{
	return 0;
}
#endif

1919
int tcpv4_offload_init(void);
1920

1921 1922
void tcp_v4_init(void);
void tcp_init(void);
1923

1924
/* tcp_recovery.c */
1925
extern void tcp_rack_mark_lost(struct sock *sk);
1926
extern void tcp_rack_advance(struct tcp_sock *tp, u8 sacked, u32 end_seq,
1927
			     u64 xmit_time);
1928
extern void tcp_rack_reo_timeout(struct sock *sk);
1929

1930 1931 1932
/* At how many usecs into the future should the RTO fire? */
static inline s64 tcp_rto_delta_us(const struct sock *sk)
{
1933
	const struct sk_buff *skb = tcp_rtx_queue_head(sk);
1934 1935 1936 1937 1938 1939
	u32 rto = inet_csk(sk)->icsk_rto;
	u64 rto_time_stamp_us = skb->skb_mstamp + jiffies_to_usecs(rto);

	return rto_time_stamp_us - tcp_sk(sk)->tcp_mstamp;
}

1940 1941 1942
/*
 * Save and compile IPv4 options, return a pointer to it
 */
1943 1944
static inline struct ip_options_rcu *tcp_v4_save_options(struct net *net,
							 struct sk_buff *skb)
1945 1946 1947 1948
{
	const struct ip_options *opt = &TCP_SKB_CB(skb)->header.h4.opt;
	struct ip_options_rcu *dopt = NULL;

C
Cong Wang 已提交
1949
	if (opt->optlen) {
1950 1951 1952
		int opt_size = sizeof(*dopt) + opt->optlen;

		dopt = kmalloc(opt_size, GFP_ATOMIC);
1953
		if (dopt && __ip_options_echo(net, &dopt->opt, skb, opt)) {
1954 1955 1956 1957 1958 1959 1960
			kfree(dopt);
			dopt = NULL;
		}
	}
	return dopt;
}

E
Eric Dumazet 已提交
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975
/* 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;
}

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
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;
}

2000 2001
int tcp_peek_len(struct socket *sock);

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
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;
}

2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
/*
 * 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);
2022
	__NET_INC_STATS(sock_net(sk), LINUX_MIB_LISTENDROPS);
2023 2024
}

2025 2026
enum hrtimer_restart tcp_pace_kick(struct hrtimer *timer);

D
Dave Watson 已提交
2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051
/*
 * Interface for adding Upper Level Protocols over TCP
 */

#define TCP_ULP_NAME_MAX	16
#define TCP_ULP_MAX		128
#define TCP_ULP_BUF_MAX		(TCP_ULP_NAME_MAX*TCP_ULP_MAX)

struct tcp_ulp_ops {
	struct list_head	list;

	/* initialize ulp */
	int (*init)(struct sock *sk);
	/* cleanup ulp */
	void (*release)(struct sock *sk);

	char		name[TCP_ULP_NAME_MAX];
	struct module	*owner;
};
int tcp_register_ulp(struct tcp_ulp_ops *type);
void tcp_unregister_ulp(struct tcp_ulp_ops *type);
int tcp_set_ulp(struct sock *sk, const char *name);
void tcp_get_available_ulp(char *buf, size_t len);
void tcp_cleanup_ulp(struct sock *sk);

L
Lawrence Brakmo 已提交
2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083
/* Call BPF_SOCK_OPS program that returns an int. If the return value
 * is < 0, then the BPF op failed (for example if the loaded BPF
 * program does not support the chosen operation or there is no BPF
 * program loaded).
 */
#ifdef CONFIG_BPF
static inline int tcp_call_bpf(struct sock *sk, int op)
{
	struct bpf_sock_ops_kern sock_ops;
	int ret;

	if (sk_fullsock(sk))
		sock_owned_by_me(sk);

	memset(&sock_ops, 0, sizeof(sock_ops));
	sock_ops.sk = sk;
	sock_ops.op = op;

	ret = BPF_CGROUP_RUN_PROG_SOCK_OPS(&sock_ops);
	if (ret == 0)
		ret = sock_ops.reply;
	else
		ret = -1;
	return ret;
}
#else
static inline int tcp_call_bpf(struct sock *sk, int op)
{
	return -EPERM;
}
#endif

2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094
static inline u32 tcp_timeout_init(struct sock *sk)
{
	int timeout;

	timeout = tcp_call_bpf(sk, BPF_SOCK_OPS_TIMEOUT_INIT);

	if (timeout <= 0)
		timeout = TCP_TIMEOUT_INIT;
	return timeout;
}

2095 2096 2097 2098 2099 2100 2101 2102 2103 2104
static inline u32 tcp_rwnd_init_bpf(struct sock *sk)
{
	int rwnd;

	rwnd = tcp_call_bpf(sk, BPF_SOCK_OPS_RWND_INIT);

	if (rwnd < 0)
		rwnd = 0;
	return rwnd;
}
2105 2106 2107 2108 2109

static inline bool tcp_bpf_ca_needs_ecn(struct sock *sk)
{
	return (tcp_call_bpf(sk, BPF_SOCK_OPS_NEEDS_ECN) == 1);
}
L
Linus Torvalds 已提交
2110
#endif	/* _TCP_H */