2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1982, 1986, 1993, 1994, 1995
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
35 #ifndef _NETINET_TCP_VAR_H_
36 #define _NETINET_TCP_VAR_H_
38 #include <netinet/tcp.h>
39 #include <netinet/tcp_fsm.h>
46 #define TCP_END_BYTE_INFO 8 /* Bytes that makeup the "end information array" */
47 /* Types of ending byte info */
48 #define TCP_EI_EMPTY_SLOT 0
49 #define TCP_EI_STATUS_CLIENT_FIN 0x1
50 #define TCP_EI_STATUS_CLIENT_RST 0x2
51 #define TCP_EI_STATUS_SERVER_FIN 0x3
52 #define TCP_EI_STATUS_SERVER_RST 0x4
53 #define TCP_EI_STATUS_RETRAN 0x5
54 #define TCP_EI_STATUS_PROGRESS 0x6
55 #define TCP_EI_STATUS_PERSIST_MAX 0x7
56 #define TCP_EI_STATUS_KEEP_MAX 0x8
57 #define TCP_EI_STATUS_DATA_A_CLOSE 0x9
58 #define TCP_EI_STATUS_RST_IN_FRONT 0xa
59 #define TCP_EI_STATUS_2MSL 0xb
60 #define TCP_EI_STATUS_MAX_VALUE 0xb
62 /************************************************/
63 /* Status bits we track to assure no duplicates,
64 * the bits here are not used by the code but
65 * for human representation. To check a bit we
66 * take and shift over by 1 minus the value (1-8).
68 /************************************************/
69 #define TCP_EI_BITS_CLIENT_FIN 0x001
70 #define TCP_EI_BITS_CLIENT_RST 0x002
71 #define TCP_EI_BITS_SERVER_FIN 0x004
72 #define TCP_EI_BITS_SERVER_RST 0x008
73 #define TCP_EI_BITS_RETRAN 0x010
74 #define TCP_EI_BITS_PROGRESS 0x020
75 #define TCP_EI_BITS_PRESIST_MAX 0x040
76 #define TCP_EI_BITS_KEEP_MAX 0x080
77 #define TCP_EI_BITS_DATA_A_CLO 0x100
78 #define TCP_EI_BITS_RST_IN_FR 0x200 /* a front state reset */
79 #define TCP_EI_BITS_2MS_TIMER 0x400 /* 2 MSL timer expired */
81 #if defined(_KERNEL) || defined(_WANT_TCPCB)
82 /* TCP segment queue entry */
84 TAILQ_ENTRY(tseg_qent) tqe_q;
85 struct mbuf *tqe_m; /* mbuf contains packet */
86 struct mbuf *tqe_last; /* last mbuf in chain */
87 tcp_seq tqe_start; /* TCP Sequence number start */
88 int tqe_len; /* TCP segment data length */
89 uint32_t tqe_flags; /* The flags from the th->th_flags */
90 uint32_t tqe_mbuf_cnt; /* Count of mbuf overhead */
92 TAILQ_HEAD(tsegqe_head, tseg_qent);
95 tcp_seq start; /* start seq no. of sack block */
96 tcp_seq end; /* end seq no. */
100 tcp_seq start; /* start seq no. of hole */
101 tcp_seq end; /* end seq no. */
102 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */
103 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */
107 struct sackhole *nexthole;
108 int32_t sack_bytes_rexmit;
109 tcp_seq last_sack_ack; /* Most recent/largest sacked ack */
111 int32_t delivered_data; /* Newly acked data from last SACK */
113 int32_t sacked_bytes; /* Total sacked bytes reported by the
114 * receiver via sack option
116 uint32_t recover_fs; /* Flight Size at the start of Loss recovery */
117 uint32_t prr_delivered; /* Total bytes delivered using PRR */
118 uint32_t prr_out; /* Bytes sent during IN_RECOVERY */
121 #define SEGQ_EMPTY(tp) TAILQ_EMPTY(&(tp)->t_segq)
123 STAILQ_HEAD(tcp_log_stailq, tcp_log_mem);
126 * Tcp control block, one per tcp; fields:
127 * Organized for 64 byte cacheline efficiency based
128 * on common tcp_input/tcp_output processing.
132 struct inpcb *t_inpcb; /* back pointer to internet pcb */
133 struct tcp_function_block *t_fb;/* TCP function call block */
134 void *t_fb_ptr; /* Pointer to t_fb specific data */
135 uint32_t t_maxseg:24, /* maximum segment size */
136 t_logstate:8; /* State of "black box" logging */
137 uint32_t t_port:16, /* Tunneling (over udp) port */
138 t_state:4, /* state of this connection */
140 t_delayed_ack: 7, /* Delayed ack variable */
141 t_fin_is_rst: 1, /* Are fin's treated as resets */
145 tcp_seq snd_una; /* sent but unacknowledged */
146 tcp_seq snd_max; /* highest sequence number sent;
147 * used to recognize retransmits
149 tcp_seq snd_nxt; /* send next */
150 tcp_seq snd_up; /* send urgent pointer */
151 uint32_t snd_wnd; /* send window */
152 uint32_t snd_cwnd; /* congestion-controlled window */
153 uint32_t t_peakrate_thr; /* pre-calculated peak rate threshold */
155 u_int32_t ts_offset; /* our timestamp offset */
156 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */
157 int rcv_numsacks; /* # distinct sack blks present */
158 u_int t_tsomax; /* TSO total burst length limit in bytes */
159 u_int t_tsomaxsegcount; /* TSO maximum segment count */
160 u_int t_tsomaxsegsize; /* TSO maximum segment size in bytes */
161 tcp_seq rcv_nxt; /* receive next */
162 tcp_seq rcv_adv; /* advertised window */
163 uint32_t rcv_wnd; /* receive window */
164 u_int t_flags2; /* More tcpcb flags storage */
165 int t_srtt; /* smoothed round-trip time */
166 int t_rttvar; /* variance in round-trip time */
167 u_int32_t ts_recent; /* timestamp echo data */
168 u_char snd_scale; /* window scaling for send window */
169 u_char rcv_scale; /* window scaling for recv window */
170 u_char snd_limited; /* segments limited transmitted */
171 u_char request_r_scale; /* pending window scaling */
172 tcp_seq last_ack_sent;
173 u_int t_rcvtime; /* inactivity time */
175 tcp_seq rcv_up; /* receive urgent pointer */
176 int t_segqlen; /* segment reassembly queue length */
177 uint32_t t_segqmbuflen; /* Count of bytes mbufs on all entries */
178 struct tsegqe_head t_segq; /* segment reassembly queue */
179 struct mbuf *t_in_pkt;
180 struct mbuf *t_tail_pkt;
181 struct tcp_timer *t_timers; /* All the TCP timers in one struct */
182 struct vnet *t_vnet; /* back pointer to parent vnet */
183 uint32_t snd_ssthresh; /* snd_cwnd size threshold for
184 * for slow start exponential to
187 tcp_seq snd_wl1; /* window update seg seq number */
189 tcp_seq snd_wl2; /* window update seg ack number */
191 tcp_seq irs; /* initial receive sequence number */
192 tcp_seq iss; /* initial send sequence number */
193 u_int t_acktime; /* RACK and BBR incoming new data was acked */
194 u_int t_sndtime; /* time last data was sent */
195 u_int ts_recent_age; /* when last updated */
196 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */
197 uint16_t cl4_spare; /* Spare to adjust CL 4 */
198 char t_oobflags; /* have some */
199 char t_iobc; /* input character */
200 int t_rxtcur; /* current retransmit value (ticks) */
202 int t_rxtshift; /* log(2) of rexmt exp. backoff */
203 u_int t_rtttime; /* RTT measurement start time */
205 tcp_seq t_rtseq; /* sequence number being timed */
206 u_int t_starttime; /* time connection was established */
207 u_int t_fbyte_in; /* ticks time when first byte queued in */
208 u_int t_fbyte_out; /* ticks time when first byte queued out */
210 u_int t_pmtud_saved_maxseg; /* pre-blackhole MSS */
211 int t_blackhole_enter; /* when to enter blackhole detection */
212 int t_blackhole_exit; /* when to exit blackhole detection */
213 u_int t_rttmin; /* minimum rtt allowed */
215 u_int t_rttbest; /* best rtt we've seen */
217 int t_softerror; /* possible error not yet reported */
218 uint32_t max_sndwnd; /* largest window peer has offered */
220 uint32_t snd_cwnd_prev; /* cwnd prior to retransmit */
221 uint32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */
222 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */
223 int t_sndzerowin; /* zero-window updates sent */
224 u_long t_rttupdated; /* number of times rtt sampled */
225 int snd_numholes; /* number of holes seen by sender */
226 u_int t_badrxtwin; /* window for retransmit recovery */
227 TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
228 /* SACK scoreboard (sorted) */
229 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/
230 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
231 struct sackhint sackhint; /* SACK scoreboard hint */
232 int t_rttlow; /* smallest observerved RTT */
233 int rfbuf_cnt; /* recv buffer autoscaling byte count */
234 struct toedev *tod; /* toedev handling this connection */
235 int t_sndrexmitpack; /* retransmit packets sent */
236 int t_rcvoopack; /* out-of-order packets received */
237 void *t_toe; /* TOE pcb pointer */
238 struct cc_algo *cc_algo; /* congestion control algorithm */
239 struct cc_var *ccv; /* congestion control specific vars */
240 struct osd *osd; /* storage for Khelp module data */
241 int t_bytes_acked; /* # bytes acked during current RTT */
242 u_int t_maxunacktime;
243 u_int t_keepinit; /* time to establish connection */
244 u_int t_keepidle; /* time before keepalive probes begin */
245 u_int t_keepintvl; /* interval between keepalives */
246 u_int t_keepcnt; /* number of keepalives before close */
247 int t_dupacks; /* consecutive dup acks recd */
248 int t_lognum; /* Number of log entries */
249 int t_loglimit; /* Maximum number of log entries */
250 int64_t t_pacing_rate; /* bytes / sec, -1 => unlimited */
251 struct tcp_log_stailq t_logs; /* Log buffer */
252 struct tcp_log_id_node *t_lin;
253 struct tcp_log_id_bucket *t_lib;
254 const char *t_output_caller; /* Function that called tcp_output */
255 struct statsblob *t_stats; /* Per-connection stats */
256 uint32_t t_logsn; /* Log "serial number" */
257 uint32_t gput_ts; /* Time goodput measurement started */
258 tcp_seq gput_seq; /* Outbound measurement seq */
259 tcp_seq gput_ack; /* Inbound measurement ack */
260 int32_t t_stats_gput_prev; /* XXXLAS: Prev gput measurement */
261 uint32_t t_maxpeakrate; /* max peak rate set by user, in bytes/s */
262 uint32_t t_sndtlppack; /* tail loss probe packets sent */
263 uint64_t t_sndtlpbyte; /* total tail loss probe bytes sent */
264 uint64_t t_sndbytes; /* total bytes sent */
265 uint64_t t_snd_rxt_bytes; /* total bytes retransmitted */
267 uint8_t t_tfo_client_cookie_len; /* TCP Fast Open client cookie length */
268 uint32_t t_end_info_status; /* Status flag of end info */
269 unsigned int *t_tfo_pending; /* TCP Fast Open server pending counter */
271 uint8_t client[TCP_FASTOPEN_MAX_COOKIE_LEN];
273 } t_tfo_cookie; /* TCP Fast Open cookie to send */
275 uint8_t t_end_info_bytes[TCP_END_BYTE_INFO];
279 struct mbufq t_inpkts; /* List of saved input packets. */
280 struct mbufq t_outpkts; /* List of saved output packets. */
283 #endif /* _KERNEL || _WANT_TCPCB */
287 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
291 /* Enable TCP/UDP tunneling port */
292 #define TCP_TUNNELING_PORT_MIN 0
293 #define TCP_TUNNELING_PORT_MAX 65535
294 #define TCP_TUNNELING_PORT_DEFAULT 0
296 /* Enable TCP/UDP tunneling port */
297 #define TCP_TUNNELING_OVERHEAD_MIN sizeof(struct udphdr)
298 #define TCP_TUNNELING_OVERHEAD_MAX 1024
299 #define TCP_TUNNELING_OVERHEAD_DEFAULT TCP_TUNNELING_OVERHEAD_MIN
301 /* Minimum map entries limit value, if set */
302 #define TCP_MIN_MAP_ENTRIES_LIMIT 128
305 * TODO: We yet need to brave plowing in
306 * to tcp_input() and the pru_usrreq() block.
307 * Right now these go to the old standards which
308 * are somewhat ok, but in the long term may
309 * need to be changed. If we do tackle tcp_input()
310 * then we need to get rid of the tcp_do_segment()
313 /* Flags for tcp functions */
314 #define TCP_FUNC_BEING_REMOVED 0x01 /* Can no longer be referenced */
317 * If defining the optional tcp_timers, in the
318 * tfb_tcp_timer_stop call you must use the
319 * callout_async_drain() function with the
320 * tcp_timer_discard callback. You should check
321 * the return of callout_async_drain() and if 0
322 * increment tt_draincnt. Since the timer sub-system
323 * does not know your callbacks you must provide a
324 * stop_all function that loops through and calls
325 * tcp_timer_stop() with each of your defined timers.
326 * Adding a tfb_tcp_handoff_ok function allows the socket
327 * option to change stacks to query you even if the
328 * connection is in a later stage. You return 0 to
329 * say you can take over and run your stack, you return
330 * non-zero (an error number) to say no you can't.
331 * If the function is undefined you can only change
332 * in the early states (before connect or listen).
333 * tfb_tcp_fb_fini is changed to add a flag to tell
334 * the old stack if the tcb is being destroyed or
335 * not. A one in the flag means the TCB is being
336 * destroyed, a zero indicates its transitioning to
337 * another stack (via socket option).
339 struct tcp_function_block {
340 char tfb_tcp_block_name[TCP_FUNCTION_NAME_LEN_MAX];
341 int (*tfb_tcp_output)(struct tcpcb *);
342 int (*tfb_tcp_output_wtime)(struct tcpcb *, const struct timeval *);
343 void (*tfb_tcp_do_segment)(struct mbuf *, struct tcphdr *,
344 struct socket *, struct tcpcb *,
346 int (*tfb_do_queued_segments)(struct socket *, struct tcpcb *, int);
347 int (*tfb_do_segment_nounlock)(struct mbuf *, struct tcphdr *,
348 struct socket *, struct tcpcb *,
350 int, struct timeval *);
351 void (*tfb_tcp_hpts_do_segment)(struct mbuf *, struct tcphdr *,
352 struct socket *, struct tcpcb *,
354 int, struct timeval *);
355 int (*tfb_tcp_ctloutput)(struct socket *so, struct sockopt *sopt,
356 struct inpcb *inp, struct tcpcb *tp);
357 /* Optional memory allocation/free routine */
358 int (*tfb_tcp_fb_init)(struct tcpcb *);
359 void (*tfb_tcp_fb_fini)(struct tcpcb *, int);
360 /* Optional timers, must define all if you define one */
361 int (*tfb_tcp_timer_stop_all)(struct tcpcb *);
362 void (*tfb_tcp_timer_activate)(struct tcpcb *,
364 int (*tfb_tcp_timer_active)(struct tcpcb *, uint32_t);
365 void (*tfb_tcp_timer_stop)(struct tcpcb *, uint32_t);
366 void (*tfb_tcp_rexmit_tmr)(struct tcpcb *);
367 int (*tfb_tcp_handoff_ok)(struct tcpcb *);
368 void (*tfb_tcp_mtu_chg)(struct tcpcb *);
369 int (*tfb_pru_options)(struct tcpcb *, int);
370 void (*tfb_hwtls_change)(struct tcpcb *, int);
371 volatile uint32_t tfb_refcnt;
376 struct tcp_function {
377 TAILQ_ENTRY(tcp_function) tf_next;
378 char tf_name[TCP_FUNCTION_NAME_LEN_MAX];
379 struct tcp_function_block *tf_fb;
382 TAILQ_HEAD(tcp_funchead, tcp_function);
386 * Flags and utility macros for the t_flags field.
388 #define TF_ACKNOW 0x00000001 /* ack peer immediately */
389 #define TF_DELACK 0x00000002 /* ack, but try to delay it */
390 #define TF_NODELAY 0x00000004 /* don't delay packets to coalesce */
391 #define TF_NOOPT 0x00000008 /* don't use tcp options */
392 #define TF_SENTFIN 0x00000010 /* have sent FIN */
393 #define TF_REQ_SCALE 0x00000020 /* have/will request window scaling */
394 #define TF_RCVD_SCALE 0x00000040 /* other side has requested scaling */
395 #define TF_REQ_TSTMP 0x00000080 /* have/will request timestamps */
396 #define TF_RCVD_TSTMP 0x00000100 /* a timestamp was received in SYN */
397 #define TF_SACK_PERMIT 0x00000200 /* other side said I could SACK */
398 #define TF_NEEDSYN 0x00000400 /* send SYN (implicit state) */
399 #define TF_NEEDFIN 0x00000800 /* send FIN (implicit state) */
400 #define TF_NOPUSH 0x00001000 /* don't push */
401 #define TF_PREVVALID 0x00002000 /* saved values for bad rxmit valid */
402 #define TF_WAKESOR 0x00004000 /* wake up receive socket */
403 #define TF_GPUTINPROG 0x00008000 /* Goodput measurement in progress */
404 #define TF_MORETOCOME 0x00010000 /* More data to be appended to sock */
405 #define TF_LQ_OVERFLOW 0x00020000 /* listen queue overflow */
406 #define TF_LASTIDLE 0x00040000 /* connection was previously idle */
407 #define TF_RXWIN0SENT 0x00080000 /* sent a receiver win 0 in response */
408 #define TF_FASTRECOVERY 0x00100000 /* in NewReno Fast Recovery */
409 #define TF_WASFRECOVERY 0x00200000 /* was in NewReno Fast Recovery */
410 #define TF_SIGNATURE 0x00400000 /* require MD5 digests (RFC2385) */
411 #define TF_FORCEDATA 0x00800000 /* force out a byte */
412 #define TF_TSO 0x01000000 /* TSO enabled on this connection */
413 #define TF_TOE 0x02000000 /* this connection is offloaded */
414 #define TF_UNUSED0 0x04000000 /* unused */
415 #define TF_UNUSED1 0x08000000 /* unused */
416 #define TF_UNUSED2 0x10000000 /* unused */
417 #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */
418 #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */
419 #define TF_FASTOPEN 0x80000000 /* TCP Fast Open indication */
421 #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY)
422 #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY
423 #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY
425 #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY)
426 #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY
427 #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY
429 #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY))
430 #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY)
431 #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY)
433 #if defined(_KERNEL) && !defined(TCP_RFC7413)
434 #define IS_FASTOPEN(t_flags) (false)
436 #define IS_FASTOPEN(t_flags) (t_flags & TF_FASTOPEN)
439 #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una)
442 * Flags for the t_oobflags field.
444 #define TCPOOB_HAVEDATA 0x01
445 #define TCPOOB_HADDATA 0x02
448 * Flags for the extended TCP flags field, t_flags2
450 #define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */
451 #define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */
452 #define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */
453 #define TF2_LOG_AUTO 0x00000008 /* Session is auto-logging. */
454 #define TF2_DROP_AF_DATA 0x00000010 /* Drop after all data ack'd */
455 #define TF2_ECN_PERMIT 0x00000020 /* connection ECN-ready */
456 #define TF2_ECN_SND_CWR 0x00000040 /* ECN CWR in queue */
457 #define TF2_ECN_SND_ECE 0x00000080 /* ECN ECE in queue */
458 #define TF2_ACE_PERMIT 0x00000100 /* Accurate ECN mode */
459 #define TF2_FBYTES_COMPLETE 0x00000400 /* We have first bytes in and out */
461 * Structure to hold TCP options that are only used during segment
462 * processing (in tcp_input), but not held in the tcpcb.
463 * It's basically used to reduce the number of parameters
464 * to tcp_dooptions and tcp_addoptions.
465 * The binary order of the to_flags is relevant for packing of the
466 * options in tcp_addoptions.
469 u_int32_t to_flags; /* which options are present */
470 #define TOF_MSS 0x0001 /* maximum segment size */
471 #define TOF_SCALE 0x0002 /* window scaling */
472 #define TOF_SACKPERM 0x0004 /* SACK permitted */
473 #define TOF_TS 0x0010 /* timestamp */
474 #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */
475 #define TOF_SACK 0x0080 /* Peer sent SACK option */
476 #define TOF_FASTOPEN 0x0100 /* TCP Fast Open (TFO) cookie */
477 #define TOF_MAXOPT 0x0200
478 u_int32_t to_tsval; /* new timestamp */
479 u_int32_t to_tsecr; /* reflected timestamp */
480 u_char *to_sacks; /* pointer to the first SACK blocks */
481 u_char *to_signature; /* pointer to the TCP-MD5 signature */
482 u_int8_t *to_tfo_cookie; /* pointer to the TFO cookie */
483 u_int16_t to_mss; /* maximum segment size */
484 u_int8_t to_wscale; /* window scaling */
485 u_int8_t to_nsacks; /* number of SACK blocks */
486 u_int8_t to_tfo_len; /* TFO cookie length */
487 u_int32_t to_spare; /* UTO */
491 * Flags for tcp_dooptions.
493 #define TO_SYN 0x01 /* parse SYN-only options */
495 struct hc_metrics_lite { /* must stay in sync with hc_metrics */
496 uint32_t rmx_mtu; /* MTU for this path */
497 uint32_t rmx_ssthresh; /* outbound gateway buffer limit */
498 uint32_t rmx_rtt; /* estimated round trip time */
499 uint32_t rmx_rttvar; /* estimated rtt variance */
500 uint32_t rmx_cwnd; /* congestion window */
501 uint32_t rmx_sendpipe; /* outbound delay-bandwidth product */
502 uint32_t rmx_recvpipe; /* inbound delay-bandwidth product */
506 * Used by tcp_maxmtu() to communicate interface specific features
507 * and limits at the time of connection setup.
512 u_int tsomaxsegcount;
516 #ifndef _NETINET_IN_PCB_H_
518 #endif /* _NETINET_IN_PCB_H_ */
521 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */
522 uint32_t t_port:16, /* UDP port number if TCPoUDP */
528 u_short last_win; /* cached window value */
529 short tw_so_options; /* copy of so_options */
530 struct ucred *tw_cred; /* user credentials */
532 u_int32_t ts_offset; /* our timestamp offset */
535 TAILQ_ENTRY(tcptw) tw_2msl;
536 void *tw_pspare; /* TCP_SIGNATURE */
537 u_int *tw_spare; /* TCP_SIGNATURE */
540 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
541 #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb)
542 #define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
545 * The smoothed round-trip time and estimated variance
546 * are stored as fixed point numbers scaled by the values below.
547 * For convenience, these scales are also used in smoothing the average
548 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
549 * With these scales, srtt has 3 bits to the right of the binary point,
550 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
551 * binary point, and is smoothed with an ALPHA of 0.75.
553 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
554 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */
555 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */
556 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */
557 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
560 * The initial retransmission should happen at rtt + 4 * rttvar.
561 * Because of the way we do the smoothing, srtt and rttvar
562 * will each average +1/2 tick of bias. When we compute
563 * the retransmit timer, we want 1/2 tick of rounding and
564 * 1 extra tick because of +-1/2 tick uncertainty in the
565 * firing of the timer. The bias will give us exactly the
566 * 1.5 tick we need. But, because the bias is
567 * statistical, we have to test that we don't drop below
568 * the minimum feasible timer (which is 2 ticks).
569 * This version of the macro adapted from a paper by Lawrence
570 * Brakmo and Larry Peterson which outlines a problem caused
571 * by insufficient precision in the original implementation,
572 * which results in inappropriately large RTO values for very
575 #define TCP_REXMTVAL(tp) \
576 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
577 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
581 * Many of these should be kept per connection,
582 * but that's inconvenient at the moment.
585 uint64_t tcps_connattempt; /* connections initiated */
586 uint64_t tcps_accepts; /* connections accepted */
587 uint64_t tcps_connects; /* connections established */
588 uint64_t tcps_drops; /* connections dropped */
589 uint64_t tcps_conndrops; /* embryonic connections dropped */
590 uint64_t tcps_minmssdrops; /* average minmss too low drops */
591 uint64_t tcps_closed; /* conn. closed (includes drops) */
592 uint64_t tcps_segstimed; /* segs where we tried to get rtt */
593 uint64_t tcps_rttupdated; /* times we succeeded */
594 uint64_t tcps_delack; /* delayed acks sent */
595 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
596 uint64_t tcps_rexmttimeo; /* retransmit timeouts */
597 uint64_t tcps_persisttimeo; /* persist timeouts */
598 uint64_t tcps_keeptimeo; /* keepalive timeouts */
599 uint64_t tcps_keepprobe; /* keepalive probes sent */
600 uint64_t tcps_keepdrops; /* connections dropped in keepalive */
602 uint64_t tcps_sndtotal; /* total packets sent */
603 uint64_t tcps_sndpack; /* data packets sent */
604 uint64_t tcps_sndbyte; /* data bytes sent */
605 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */
606 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */
607 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */
608 uint64_t tcps_sndacks; /* ack-only packets sent */
609 uint64_t tcps_sndprobe; /* window probes sent */
610 uint64_t tcps_sndurg; /* packets sent with URG only */
611 uint64_t tcps_sndwinup; /* window update-only packets sent */
612 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
614 uint64_t tcps_rcvtotal; /* total packets received */
615 uint64_t tcps_rcvpack; /* packets received in sequence */
616 uint64_t tcps_rcvbyte; /* bytes received in sequence */
617 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */
618 uint64_t tcps_rcvbadoff; /* packets received with bad offset */
619 uint64_t tcps_rcvreassfull; /* packets dropped for no reass space */
620 uint64_t tcps_rcvshort; /* packets received too short */
621 uint64_t tcps_rcvduppack; /* duplicate-only packets received */
622 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */
623 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */
624 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
625 uint64_t tcps_rcvoopack; /* out-of-order packets received */
626 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */
627 uint64_t tcps_rcvpackafterwin; /* packets with data after window */
628 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */
629 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */
630 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */
631 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */
632 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */
633 uint64_t tcps_rcvackpack; /* rcvd ack packets */
634 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */
635 uint64_t tcps_rcvwinupd; /* rcvd window update packets */
636 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */
637 uint64_t tcps_predack; /* times hdr predict ok for acks */
638 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */
639 uint64_t tcps_pcbcachemiss;
640 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */
641 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */
642 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */
643 uint64_t tcps_usedrtt; /* times RTT initialized from route */
644 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */
645 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/
646 uint64_t tcps_persistdrop; /* timeout in persist state */
647 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */
648 uint64_t tcps_mturesent; /* resends due to MTU discovery */
649 uint64_t tcps_listendrop; /* listen queue overflows */
650 uint64_t tcps_badrst; /* ignored RSTs in the window */
652 uint64_t tcps_sc_added; /* entry added to syncache */
653 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */
654 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */
655 uint64_t tcps_sc_dropped; /* could not reply to packet */
656 uint64_t tcps_sc_completed; /* successful extraction of entry */
657 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */
658 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */
659 uint64_t tcps_sc_reset; /* RST removed entry from syncache */
660 uint64_t tcps_sc_stale; /* timed out or listen socket gone */
661 uint64_t tcps_sc_aborted; /* syncache entry aborted */
662 uint64_t tcps_sc_badack; /* removed due to bad ACK */
663 uint64_t tcps_sc_unreach; /* ICMP unreachable received */
664 uint64_t tcps_sc_zonefail; /* zalloc() failed */
665 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */
666 uint64_t tcps_sc_recvcookie; /* SYN cookie received */
668 uint64_t tcps_hc_added; /* entry added to hostcache */
669 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */
671 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */
673 /* SACK related stats */
674 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */
675 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */
676 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */
677 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */
678 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */
679 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */
681 /* ECN related stats */
682 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */
683 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */
684 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */
685 uint64_t tcps_ecn_shs; /* ECN successful handshakes */
686 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */
688 /* TCP_SIGNATURE related stats */
689 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */
690 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */
691 uint64_t tcps_sig_err_buildsig; /* Failed to make signature */
692 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */
693 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */
695 /* Path MTU Discovery Black Hole Detection related stats */
696 uint64_t tcps_pmtud_blackhole_activated; /* Black Hole Count */
697 uint64_t tcps_pmtud_blackhole_activated_min_mss; /* BH at min MSS Count */
698 uint64_t tcps_pmtud_blackhole_failed; /* Black Hole Failure Count */
700 uint64_t tcps_tunneled_pkts; /* Packets encap's in UDP received */
701 uint64_t tcps_tunneled_errs; /* Packets that had errors that were UDP encaped */
703 uint64_t _pad[10]; /* 6 UTO, 6 TBD */
706 #define tcps_rcvmemdrop tcps_rcvreassfull /* compat */
709 #define TI_UNLOCKED 1
711 #include <sys/counter.h>
713 VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */
715 * In-kernel consumers can use these accessor macros directly to update
718 #define TCPSTAT_ADD(name, val) \
719 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val))
720 #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1)
723 * Kernel module consumers must use this accessor macro.
725 void kmod_tcpstat_add(int statnum, int val);
726 #define KMOD_TCPSTAT_ADD(name, val) \
727 kmod_tcpstat_add(offsetof(struct tcpstat, name) / sizeof(uint64_t), val)
728 #define KMOD_TCPSTAT_INC(name) KMOD_TCPSTAT_ADD(name, 1)
731 * Running TCP connection count by state.
733 VNET_DECLARE(counter_u64_t, tcps_states[TCP_NSTATES]);
734 #define V_tcps_states VNET(tcps_states)
735 #define TCPSTATES_INC(state) counter_u64_add(V_tcps_states[state], 1)
736 #define TCPSTATES_DEC(state) counter_u64_add(V_tcps_states[state], -1)
739 * TCP specific helper hook point identifiers.
741 #define HHOOK_TCP_EST_IN 0
742 #define HHOOK_TCP_EST_OUT 1
743 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT
745 struct tcp_hhook_data {
754 void hhook_run_tcp_est_out(struct tcpcb *tp,
755 struct tcphdr *th, struct tcpopt *to,
756 uint32_t len, int tso);
761 * TCB structure exported to user-land via sysctl(3).
763 * Fields prefixed with "xt_" are unique to the export structure, and fields
764 * with "t_" or other prefixes match corresponding fields of 'struct tcpcb'.
767 * (s) - used by userland utilities in src
768 * (p) - used by utilities in ports
769 * (3) - is known to be used by third party software not in ports
770 * (n) - no known usage
772 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
773 * included. Not all of our clients do.
775 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
777 ksize_t xt_len; /* length of this structure */
778 struct xinpcb xt_inp;
779 char xt_stack[TCP_FUNCTION_NAME_LEN_MAX]; /* (s) */
780 char xt_logid[TCP_LOG_ID_LEN]; /* (s) */
781 char xt_cc[TCP_CA_NAME_MAX]; /* (s) */
783 int32_t t_state; /* (s,p) */
784 uint32_t t_flags; /* (s,p) */
785 int32_t t_sndzerowin; /* (s) */
786 int32_t t_sndrexmitpack; /* (s) */
787 int32_t t_rcvoopack; /* (s) */
788 int32_t t_rcvtime; /* (s) */
789 int32_t tt_rexmt; /* (s) */
790 int32_t tt_persist; /* (s) */
791 int32_t tt_keep; /* (s) */
792 int32_t tt_2msl; /* (s) */
793 int32_t tt_delack; /* (s) */
794 int32_t t_logstate; /* (3) */
795 uint32_t t_snd_cwnd; /* (s) */
796 uint32_t t_snd_ssthresh; /* (s) */
797 uint32_t t_maxseg; /* (s) */
798 uint32_t t_rcv_wnd; /* (s) */
799 uint32_t t_snd_wnd; /* (s) */
800 uint32_t xt_ecn; /* (s) */
801 uint16_t xt_encaps_port; /* (s) */
807 void tcp_inptoxtp(const struct inpcb *, struct xtcpcb *);
812 * TCP function information (name-to-id mapping, aliases, and refcnt)
813 * exported to user-land via sysctl(3).
815 struct tcp_function_info {
818 char tfi_name[TCP_FUNCTION_NAME_LEN_MAX];
819 char tfi_alias[TCP_FUNCTION_NAME_LEN_MAX];
823 * Identifiers for TCP sysctl nodes
825 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
826 #define TCPCTL_MSSDFLT 3 /* MSS default */
827 #define TCPCTL_STATS 4 /* statistics */
828 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */
829 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
830 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
831 #define TCPCTL_SENDSPACE 8 /* send buffer space */
832 #define TCPCTL_RECVSPACE 9 /* receive buffer space */
833 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
834 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
835 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
836 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
837 #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */
838 #define TCPCTL_DROP 15 /* drop tcp connection */
839 #define TCPCTL_STATES 16 /* connection counts by TCP state */
843 SYSCTL_DECL(_net_inet_tcp);
844 SYSCTL_DECL(_net_inet_tcp_sack);
845 MALLOC_DECLARE(M_TCPLOG);
848 VNET_DECLARE(int, tcp_log_in_vain);
849 #define V_tcp_log_in_vain VNET(tcp_log_in_vain)
852 * Global TCP tunables shared between different stacks.
853 * Please keep the list sorted.
855 VNET_DECLARE(int, drop_synfin);
856 VNET_DECLARE(int, path_mtu_discovery);
857 VNET_DECLARE(int, tcp_abc_l_var);
858 VNET_DECLARE(int, tcp_autorcvbuf_max);
859 VNET_DECLARE(int, tcp_autosndbuf_inc);
860 VNET_DECLARE(int, tcp_autosndbuf_max);
861 VNET_DECLARE(int, tcp_delack_enabled);
862 VNET_DECLARE(int, tcp_do_autorcvbuf);
863 VNET_DECLARE(int, tcp_do_autosndbuf);
864 VNET_DECLARE(int, tcp_do_ecn);
865 VNET_DECLARE(int, tcp_do_prr);
866 VNET_DECLARE(int, tcp_do_prr_conservative);
867 VNET_DECLARE(int, tcp_do_newcwv);
868 VNET_DECLARE(int, tcp_do_rfc1323);
869 VNET_DECLARE(int, tcp_tolerate_missing_ts);
870 VNET_DECLARE(int, tcp_do_rfc3042);
871 VNET_DECLARE(int, tcp_do_rfc3390);
872 VNET_DECLARE(int, tcp_do_rfc3465);
873 VNET_DECLARE(int, tcp_do_rfc6675_pipe);
874 VNET_DECLARE(int, tcp_do_sack);
875 VNET_DECLARE(int, tcp_do_tso);
876 VNET_DECLARE(int, tcp_ecn_maxretries);
877 VNET_DECLARE(int, tcp_initcwnd_segments);
878 VNET_DECLARE(int, tcp_insecure_rst);
879 VNET_DECLARE(int, tcp_insecure_syn);
880 VNET_DECLARE(uint32_t, tcp_map_entries_limit);
881 VNET_DECLARE(uint32_t, tcp_map_split_limit);
882 VNET_DECLARE(int, tcp_minmss);
883 VNET_DECLARE(int, tcp_mssdflt);
885 VNET_DECLARE(int, tcp_perconn_stats_dflt_tpl);
886 VNET_DECLARE(int, tcp_perconn_stats_enable);
888 VNET_DECLARE(int, tcp_recvspace);
889 VNET_DECLARE(int, tcp_sack_globalholes);
890 VNET_DECLARE(int, tcp_sack_globalmaxholes);
891 VNET_DECLARE(int, tcp_sack_maxholes);
892 VNET_DECLARE(int, tcp_sc_rst_sock_fail);
893 VNET_DECLARE(int, tcp_sendspace);
894 VNET_DECLARE(int, tcp_udp_tunneling_overhead);
895 VNET_DECLARE(int, tcp_udp_tunneling_port);
896 VNET_DECLARE(struct inpcbhead, tcb);
897 VNET_DECLARE(struct inpcbinfo, tcbinfo);
899 #define V_tcp_do_prr VNET(tcp_do_prr)
900 #define V_tcp_do_prr_conservative VNET(tcp_do_prr_conservative)
901 #define V_tcp_do_newcwv VNET(tcp_do_newcwv)
902 #define V_drop_synfin VNET(drop_synfin)
903 #define V_path_mtu_discovery VNET(path_mtu_discovery)
904 #define V_tcb VNET(tcb)
905 #define V_tcbinfo VNET(tcbinfo)
906 #define V_tcp_abc_l_var VNET(tcp_abc_l_var)
907 #define V_tcp_autorcvbuf_max VNET(tcp_autorcvbuf_max)
908 #define V_tcp_autosndbuf_inc VNET(tcp_autosndbuf_inc)
909 #define V_tcp_autosndbuf_max VNET(tcp_autosndbuf_max)
910 #define V_tcp_delack_enabled VNET(tcp_delack_enabled)
911 #define V_tcp_do_autorcvbuf VNET(tcp_do_autorcvbuf)
912 #define V_tcp_do_autosndbuf VNET(tcp_do_autosndbuf)
913 #define V_tcp_do_ecn VNET(tcp_do_ecn)
914 #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323)
915 #define V_tcp_tolerate_missing_ts VNET(tcp_tolerate_missing_ts)
916 #define V_tcp_ts_offset_per_conn VNET(tcp_ts_offset_per_conn)
917 #define V_tcp_do_rfc3042 VNET(tcp_do_rfc3042)
918 #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390)
919 #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465)
920 #define V_tcp_do_rfc6675_pipe VNET(tcp_do_rfc6675_pipe)
921 #define V_tcp_do_sack VNET(tcp_do_sack)
922 #define V_tcp_do_tso VNET(tcp_do_tso)
923 #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries)
924 #define V_tcp_initcwnd_segments VNET(tcp_initcwnd_segments)
925 #define V_tcp_insecure_rst VNET(tcp_insecure_rst)
926 #define V_tcp_insecure_syn VNET(tcp_insecure_syn)
927 #define V_tcp_map_entries_limit VNET(tcp_map_entries_limit)
928 #define V_tcp_map_split_limit VNET(tcp_map_split_limit)
929 #define V_tcp_minmss VNET(tcp_minmss)
930 #define V_tcp_mssdflt VNET(tcp_mssdflt)
932 #define V_tcp_perconn_stats_dflt_tpl VNET(tcp_perconn_stats_dflt_tpl)
933 #define V_tcp_perconn_stats_enable VNET(tcp_perconn_stats_enable)
935 #define V_tcp_recvspace VNET(tcp_recvspace)
936 #define V_tcp_sack_globalholes VNET(tcp_sack_globalholes)
937 #define V_tcp_sack_globalmaxholes VNET(tcp_sack_globalmaxholes)
938 #define V_tcp_sack_maxholes VNET(tcp_sack_maxholes)
939 #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail)
940 #define V_tcp_sendspace VNET(tcp_sendspace)
941 #define V_tcp_udp_tunneling_overhead VNET(tcp_udp_tunneling_overhead)
942 #define V_tcp_udp_tunneling_port VNET(tcp_udp_tunneling_port)
945 VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]);
946 #define V_tcp_hhh VNET(tcp_hhh)
949 int tcp_addoptions(struct tcpopt *, u_char *);
950 int tcp_ccalgounload(struct cc_algo *unload_algo);
952 tcp_close(struct tcpcb *);
953 void tcp_discardcb(struct tcpcb *);
954 void tcp_twstart(struct tcpcb *);
955 void tcp_twclose(struct tcptw *, int);
956 void tcp_ctlinput(int, struct sockaddr *, void *);
957 int tcp_ctloutput(struct socket *, struct sockopt *);
958 void tcp_ctlinput_viaudp(int, struct sockaddr *, void *, void *);
960 tcp_drop(struct tcpcb *, int);
961 void tcp_drain(void);
963 void tcp_fini(void *);
964 char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *,
966 char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *,
968 int tcp_reass(struct tcpcb *, struct tcphdr *, tcp_seq *, int *,
970 void tcp_reass_global_init(void);
971 void tcp_reass_flush(struct tcpcb *);
972 void tcp_dooptions(struct tcpopt *, u_char *, int, int);
973 void tcp_dropwithreset(struct mbuf *, struct tcphdr *,
974 struct tcpcb *, int, int);
975 void tcp_pulloutofband(struct socket *,
976 struct tcphdr *, struct mbuf *, int);
977 void tcp_xmit_timer(struct tcpcb *, int);
978 void tcp_newreno_partial_ack(struct tcpcb *, struct tcphdr *);
979 void cc_ack_received(struct tcpcb *tp, struct tcphdr *th,
980 uint16_t nsegs, uint16_t type);
981 void cc_conn_init(struct tcpcb *tp);
982 void cc_post_recovery(struct tcpcb *tp, struct tcphdr *th);
983 void cc_ecnpkt_handler(struct tcpcb *tp, struct tcphdr *th, uint8_t iptos);
984 void cc_ecnpkt_handler_flags(struct tcpcb *tp, uint16_t flags, uint8_t iptos);
985 void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type);
987 void hhook_run_tcp_est_in(struct tcpcb *tp,
988 struct tcphdr *th, struct tcpopt *to);
991 int tcp_input(struct mbuf **, int *, int);
992 int tcp_autorcvbuf(struct mbuf *, struct tcphdr *, struct socket *,
993 struct tcpcb *, int);
994 int tcp_input_with_port(struct mbuf **, int *, int, uint16_t);
995 void tcp_handle_wakeup(struct tcpcb *, struct socket *);
996 void tcp_do_segment(struct mbuf *, struct tcphdr *,
997 struct socket *, struct tcpcb *, int, int, uint8_t);
999 int register_tcp_functions(struct tcp_function_block *blk, int wait);
1000 int register_tcp_functions_as_names(struct tcp_function_block *blk,
1001 int wait, const char *names[], int *num_names);
1002 int register_tcp_functions_as_name(struct tcp_function_block *blk,
1003 const char *name, int wait);
1004 int deregister_tcp_functions(struct tcp_function_block *blk, bool quiesce,
1006 struct tcp_function_block *find_and_ref_tcp_functions(struct tcp_function_set *fs);
1007 void tcp_switch_back_to_default(struct tcpcb *tp);
1008 struct tcp_function_block *
1009 find_and_ref_tcp_fb(struct tcp_function_block *fs);
1010 int tcp_default_ctloutput(struct socket *so, struct sockopt *sopt, struct inpcb *inp, struct tcpcb *tp);
1012 extern counter_u64_t tcp_inp_lro_direct_queue;
1013 extern counter_u64_t tcp_inp_lro_wokeup_queue;
1014 extern counter_u64_t tcp_inp_lro_compressed;
1015 extern counter_u64_t tcp_inp_lro_locks_taken;
1016 extern counter_u64_t tcp_extra_mbuf;
1017 extern counter_u64_t tcp_would_have_but;
1018 extern counter_u64_t tcp_comp_total;
1019 extern counter_u64_t tcp_uncomp_total;
1021 #ifdef NETFLIX_EXP_DETECTION
1022 /* Various SACK attack thresholds */
1023 extern int32_t tcp_force_detection;
1024 extern int32_t tcp_sack_to_ack_thresh;
1025 extern int32_t tcp_sack_to_move_thresh;
1026 extern int32_t tcp_restoral_thresh;
1027 extern int32_t tcp_sad_decay_val;
1028 extern int32_t tcp_sad_pacing_interval;
1029 extern int32_t tcp_sad_low_pps;
1030 extern int32_t tcp_map_minimum;
1031 extern int32_t tcp_attack_on_turns_on_logging;
1033 extern uint32_t tcp_ack_war_time_window;
1034 extern uint32_t tcp_ack_war_cnt;
1036 uint32_t tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *);
1037 uint32_t tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *);
1038 u_int tcp_maxseg(const struct tcpcb *);
1039 u_int tcp_fixed_maxseg(const struct tcpcb *);
1040 void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *,
1041 struct tcp_ifcap *);
1042 void tcp_mss(struct tcpcb *, int);
1043 int tcp_mssopt(struct in_conninfo *);
1045 tcp_drop_syn_sent(struct inpcb *, int);
1047 tcp_newtcpcb(struct inpcb *);
1048 int tcp_output(struct tcpcb *);
1049 void tcp_state_change(struct tcpcb *, int);
1050 void tcp_respond(struct tcpcb *, void *,
1051 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
1052 void tcp_tw_init(void);
1054 void tcp_tw_destroy(void);
1056 void tcp_tw_zone_change(void);
1057 int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *,
1058 struct mbuf *, int);
1059 void tcp_setpersist(struct tcpcb *);
1060 void tcp_slowtimo(void);
1062 tcpip_maketemplate(struct inpcb *);
1063 void tcpip_fillheaders(struct inpcb *, uint16_t, void *, void *);
1064 void tcp_timer_activate(struct tcpcb *, uint32_t, u_int);
1065 int tcp_timer_suspend(struct tcpcb *, uint32_t);
1066 void tcp_timers_unsuspend(struct tcpcb *, uint32_t);
1067 int tcp_timer_active(struct tcpcb *, uint32_t);
1068 void tcp_timer_stop(struct tcpcb *, uint32_t);
1069 void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int);
1070 int inp_to_cpuid(struct inpcb *inp);
1072 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
1074 void tcp_hc_init(void);
1076 void tcp_hc_destroy(void);
1078 void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
1079 uint32_t tcp_hc_getmtu(struct in_conninfo *);
1080 void tcp_hc_updatemtu(struct in_conninfo *, uint32_t);
1081 void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
1083 extern struct pr_usrreqs tcp_usrreqs;
1085 uint32_t tcp_new_ts_offset(struct in_conninfo *);
1086 tcp_seq tcp_new_isn(struct in_conninfo *);
1088 int tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
1089 int tcp_dsack_block_exists(struct tcpcb *);
1090 void tcp_update_dsack_list(struct tcpcb *, tcp_seq, tcp_seq);
1091 void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
1092 void tcp_clean_dsack_blocks(struct tcpcb *tp);
1093 void tcp_clean_sackreport(struct tcpcb *tp);
1094 void tcp_sack_adjust(struct tcpcb *tp);
1095 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
1096 void tcp_do_prr_ack(struct tcpcb *, struct tcphdr *, struct tcpopt *);
1097 void tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
1098 void tcp_free_sackholes(struct tcpcb *tp);
1099 int tcp_newreno(struct tcpcb *, struct tcphdr *);
1100 int tcp_compute_pipe(struct tcpcb *);
1101 uint32_t tcp_compute_initwnd(uint32_t);
1102 void tcp_sndbuf_autoscale(struct tcpcb *, struct socket *, uint32_t);
1103 int tcp_stats_sample_rollthedice(struct tcpcb *tp, void *seed_bytes,
1105 int tcp_can_enable_pacing(void);
1106 void tcp_decrement_paced_conn(void);
1109 tcp_m_copym(struct mbuf *m, int32_t off0, int32_t *plen,
1110 int32_t seglimit, int32_t segsize, struct sockbuf *sb, bool hw_tls);
1112 int tcp_stats_init(void);
1113 void tcp_log_end_status(struct tcpcb *tp, uint8_t status);
1116 tcp_fields_to_host(struct tcphdr *th)
1119 th->th_seq = ntohl(th->th_seq);
1120 th->th_ack = ntohl(th->th_ack);
1121 th->th_win = ntohs(th->th_win);
1122 th->th_urp = ntohs(th->th_urp);
1126 tcp_fields_to_net(struct tcphdr *th)
1129 th->th_seq = htonl(th->th_seq);
1130 th->th_ack = htonl(th->th_ack);
1131 th->th_win = htons(th->th_win);
1132 th->th_urp = htons(th->th_urp);
1136 tcp_account_for_send(struct tcpcb *tp, uint32_t len, uint8_t is_rxt,
1137 uint8_t is_tlp, int hw_tls __unused)
1141 tp->t_sndtlpbyte += len;
1143 /* To get total bytes sent you must add t_snd_rxt_bytes to t_sndbytes */
1145 tp->t_snd_rxt_bytes += len;
1147 tp->t_sndbytes += len;
1149 #endif /* _KERNEL */
1151 #endif /* _NETINET_TCP_VAR_H_ */