1/* $OpenBSD: tcp_var.h,v 1.196 2025/09/16 17:29:35 bluhm Exp $ */
2/* $NetBSD: tcp_var.h,v 1.17 1996/02/13 23:44:24 christos Exp $ */
3
4/*
5 * Copyright (c) 1982, 1986, 1993, 1994
6 * The Regents of the University of California. All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the University nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 *
32 * @(#)tcp_var.h 8.3 (Berkeley) 4/10/94
33 */
34
35#ifndef _NETINET_TCP_VAR_H_
36#define _NETINET_TCP_VAR_H_
37
38#include <sys/timeout.h>
39
40/*
41 * Kernel variables for tcp.
42 */
43
44struct sackblk {
45 tcp_seq start; /* start seq no. of sack block */
46 tcp_seq end; /* end seq no. */
47};
48
49struct sackhole {
50 tcp_seq start; /* start seq no. of hole */
51 tcp_seq end; /* end seq no. */
52 int dups; /* number of dup(s)acks for this hole */
53 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */
54 struct sackhole *next; /* next in list */
55};
56
57/*
58 * TCP sequence queue structures.
59 */
60TAILQ_HEAD(tcpqehead, tcpqent);
61struct tcpqent {
62 TAILQ_ENTRY(tcpqent) tcpqe_q;
63 struct tcphdr *tcpqe_tcp;
64 struct mbuf *tcpqe_m; /* mbuf contains packet */
65};
66
67/*
68 * Tcp control block, one per tcp; fields:
69 */
70struct tcpcb {
71 struct tcpqehead t_segq; /* sequencing queue */
72 struct timeout t_timer[TCPT_NTIMERS]; /* tcp timers */
73 short t_state; /* state of this connection */
74 short t_rxtshift; /* log(2) of rexmt exp. backoff */
75 int t_rxtcur; /* current retransmit value */
76 short t_dupacks; /* consecutive dup acks recd */
77 u_short t_maxseg; /* maximum segment size */
78 char t_force; /* 1 if forcing out a byte */
79 u_int t_flags;
80#define TF_ACKNOW 0x0001U /* ack peer immediately */
81#define TF_NODELAY 0x0004U /* don't delay packets to coalesce */
82#define TF_NOOPT 0x0008U /* don't use tcp options */
83#define TF_SENTFIN 0x0010U /* have sent FIN */
84#define TF_REQ_SCALE 0x0020U /* have/will request window scaling */
85#define TF_RCVD_SCALE 0x0040U /* other side has requested scaling */
86#define TF_REQ_TSTMP 0x0080U /* have/will request timestamps */
87#define TF_RCVD_TSTMP 0x0100U /* a timestamp was received in SYN */
88#define TF_SACK_PERMIT 0x0200U /* other side said I could SACK */
89#define TF_SIGNATURE 0x0400U /* require TCP MD5 signature */
90#ifdef TCP_ECN
91#define TF_ECN_PERMIT 0x00008000U /* other side said I could ECN */
92#define TF_RCVD_CE 0x00010000U /* send ECE in subsequent segs */
93#define TF_SEND_CWR 0x00020000U /* send CWR in next seg */
94#define TF_DISABLE_ECN 0x00040000U /* disable ECN for this connection */
95#endif
96#define TF_LASTIDLE 0x00100000U /* no outstanding ACK on last send */
97#define TF_PMTUD_PEND 0x00400000U /* Path MTU Discovery pending */
98#define TF_NEEDOUTPUT 0x00800000U /* call tcp_output after tcp_input */
99#define TF_NOPUSH 0x02000000U /* don't push */
100#define TF_TMR_REXMT 0x04000000U /* retransmit timer armed */
101#define TF_TMR_PERSIST 0x08000000U /* retransmit persistence timer armed */
102#define TF_TMR_KEEP 0x10000000U /* keep alive timer armed */
103#define TF_TMR_2MSL 0x20000000U /* 2*msl quiet time timer armed */
104#define TF_TMR_DELACK 0x40000000U /* delayed ack timer armed */
105#define TF_TIMER TF_TMR_REXMT /* used to shift with TCPT values */
106
107 struct mbuf *t_template; /* skeletal packet for transmit */
108 struct inpcb *t_inpcb; /* back pointer to internet pcb */
109/*
110 * The following fields are used as in the protocol specification.
111 * See RFC793, Dec. 1981, page 21.
112 */
113/* send sequence variables */
114 tcp_seq snd_una; /* send unacknowledged */
115 tcp_seq snd_nxt; /* send next */
116 tcp_seq snd_up; /* send urgent pointer */
117 tcp_seq snd_wl1; /* window update seg seq number */
118 tcp_seq snd_wl2; /* window update seg ack number */
119 tcp_seq iss; /* initial send sequence number */
120 u_long snd_wnd; /* send window */
121 int sack_enable; /* enable SACK for this connection */
122 int snd_numholes; /* number of holes seen by sender */
123 struct sackhole *snd_holes; /* linked list of holes (sorted) */
124 tcp_seq snd_last; /* for use in fast recovery */
125/* receive sequence variables */
126 u_long rcv_wnd; /* receive window */
127 tcp_seq rcv_nxt; /* receive next */
128 tcp_seq rcv_up; /* receive urgent pointer */
129 tcp_seq irs; /* initial receive sequence number */
130 tcp_seq rcv_lastsack; /* last seq number(+1) sack'd by rcv'r*/
131 int rcv_numsacks; /* # distinct sack blks present */
132 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
133
134/*
135 * Additional variables for this implementation.
136 */
137/* receive variables */
138 tcp_seq rcv_adv; /* advertised window */
139/* retransmit variables */
140 tcp_seq snd_max; /* highest sequence number sent;
141 * used to recognize retransmits
142 */
143/* congestion control (for slow start, source quench, retransmit after loss) */
144 u_long snd_cwnd; /* congestion-controlled window */
145 u_long snd_ssthresh; /* snd_cwnd size threshold for
146 * for slow start exponential to
147 * linear switch
148 */
149
150/* auto-sizing variables */
151 uint64_t rfbuf_ts; /* recv buffer autoscaling time stamp */
152 u_int rfbuf_cnt; /* recv buffer autoscaling byte count */
153
154 u_short t_maxopd; /* mss plus options */
155 u_short t_peermss; /* peer's maximum segment size */
156
157/*
158 * transmit timing stuff. See below for scale of srtt and rttvar.
159 * "Variance" is actually smoothed difference.
160 */
161 uint64_t t_rcvtime; /* time last segment received */
162 uint64_t t_rcvacktime; /* time last ack received */
163 uint64_t t_sndtime; /* time last segment sent */
164 uint64_t t_sndacktime; /* time last ack sent */
165 uint64_t t_rtttime; /* time we started measuring rtt */
166 tcp_seq t_rtseq; /* sequence number being timed */
167 int t_srtt; /* smoothed round-trip time */
168 int t_rttvar; /* variance in round-trip time */
169 u_int t_rttmin; /* minimum rtt allowed */
170 u_long max_sndwnd; /* largest window peer has offered */
171
172/* out-of-band data */
173 char t_oobflags; /* have some */
174 char t_iobc; /* input character */
175#define TCPOOB_HAVEDATA 0x01
176#define TCPOOB_HADDATA 0x02
177 short t_softerror; /* possible error not yet reported */
178
179/* RFC 1323 variables */
180 u_char snd_scale; /* window scaling for send window */
181 u_char rcv_scale; /* window scaling for recv window */
182 u_char request_r_scale; /* pending window scaling */
183 u_char requested_s_scale;
184 uint32_t ts_recent; /* timestamp echo data */
185 uint32_t ts_modulate; /* modulation on timestamp */
186 uint64_t ts_recent_age; /* when last updated */
187 tcp_seq last_ack_sent;
188
189/* pointer for syn cache entries*/
190 LIST_HEAD(, syn_cache) t_sc; /* list of entries by this tcb */
191
192/* Path-MTU Discovery Information */
193 u_int t_pmtud_mss_acked; /* MSS acked, lower bound for MTU */
194 u_int t_pmtud_mtu_sent; /* MTU used, upper bound for MTU */
195 tcp_seq t_pmtud_th_seq; /* TCP SEQ from ICMP payload */
196 u_int t_pmtud_nextmtu; /* Advertised Next-Hop MTU from ICMP */
197 u_short t_pmtud_ip_len; /* IP length from ICMP payload */
198 u_short t_pmtud_ip_hl; /* IP header length from ICMP payload */
199
200 int pf;
201
202/* maintain a few stats per connection: */
203 u_int t_rcvoopack; /* out-of-order packets received */
204 u_int t_sndrexmitpack; /* retransmit packets sent */
205 u_int t_sndzerowin; /* zero-window updates sent */
206};
207
208#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
209#define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
210
211#ifdef _KERNEL
212/*
213 * Handy way of passing around TCP option info.
214 */
215struct tcp_opt_info {
216 int ts_present;
217 u_int32_t ts_val;
218 u_int32_t ts_ecr;
219 u_int16_t maxseg;
220};
221
222/*
223 * Data for the TCP compressed state engine.
224 */
225
226/*
227 * Locks used to protect global data and struct members:
228 * I immutable after creation
229 * N net lock
230 * S syn_cache_mtx tcp syn cache global mutex
231 * s so_lock socket lock of listen socket
232 */
233
234extern struct mutex syn_cache_mtx;
235
236#define TCP_SYN_HASH_SIZE 293
237#define TCP_SYN_BUCKET_SIZE 35
238
239union syn_cache_sa {
240 struct sockaddr sa;
241 struct sockaddr_in sin;
242 struct sockaddr_in6 sin6;
243};
244
245struct syn_cache {
246 TAILQ_ENTRY(syn_cache) sc_bucketq; /* [S] link on bucket list */
247 struct refcnt sc_refcnt; /* ref count list and timer */
248 struct timeout sc_timer; /* rexmt timer */
249 struct route sc_route; /* [s] cached route */
250 long sc_win; /* [I] advertised window */
251 struct syn_cache_head *sc_buckethead; /* [S] our bucket index */
252 struct syn_cache_set *sc_set; /* [S] our syn cache set */
253 u_int64_t sc_timestamp; /* [s] timestamp from SYN */
254 u_int32_t sc_hash; /* [S] */
255 u_int32_t sc_modulate; /* [I] our timestamp modulator */
256 union syn_cache_sa sc_src; /* [I] */
257 union syn_cache_sa sc_dst; /* [I] */
258 tcp_seq sc_irs; /* [I] */
259 tcp_seq sc_iss; /* [I] */
260 u_int sc_rtableid; /* [I] */
261 u_int sc_rxtcur; /* [S] current rxt timeout */
262 u_int sc_rxttot; /* [S] total time spend on queues */
263 u_int sc_rxtshift; /* [S] for computing backoff */
264 u_int sc_dynflags; /* [S] flags accessed with mutex */
265#define SCF_UNREACH 0x0001U /* we've had an unreach error */
266
267 u_short sc_fixflags; /* [I] set during initialization */
268#define SCF_TIMESTAMP 0x0010U /* peer will do timestamps */
269#define SCF_SACK_PERMIT 0x0020U /* permit sack */
270#define SCF_ECN_PERMIT 0x0040U /* permit ecn */
271#define SCF_SIGNATURE 0x0080U /* enforce tcp signatures */
272
273 struct mbuf *sc_ipopts; /* [s] IP options */
274 u_int16_t sc_peermaxseg; /* [I] */
275 u_int16_t sc_ourmaxseg; /* [I] */
276 u_int sc_request_r_scale : 4, /* [I] */
277 sc_requested_s_scale : 4; /* [I] */
278
279 struct inpcb *sc_inplisten; /* [S] inpcb for listening socket */
280 LIST_ENTRY(syn_cache) sc_tpq; /* [S] list of entries by same tp */
281};
282
283struct syn_cache_head {
284 TAILQ_HEAD(, syn_cache) sch_bucket; /* [S] bucket entries */
285 u_short sch_length; /* [S] # entries in bucket */
286};
287
288struct syn_cache_set {
289 struct syn_cache_head *scs_buckethead; /* [S] */
290 long scs_use; /* [S] */
291 int scs_size; /* [S] current size of hash table */
292 int scs_count; /* [S] */
293 u_int32_t scs_random[5]; /* [S] */
294};
295
296#endif /* _KERNEL */
297
298/*
299 * The smoothed round-trip time and estimated variance
300 * are stored as fixed point numbers scaled by the values below.
301 * For convenience, these scales are also used in smoothing the average
302 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
303 * With these scales, srtt has 5 bits to the right of the binary point,
304 * and thus an "ALPHA" of 0.875. rttvar has 4 bits to the right of the
305 * binary point, and is smoothed with an ALPHA of 0.75.
306 */
307#define TCP_RTT_SHIFT 3 /* shift for srtt; 5 bits frac. */
308#define TCP_RTTVAR_SHIFT 2 /* shift for rttvar; 4 bits */
309#define TCP_RTT_BASE_SHIFT 2 /* remaining 2 bit shift */
310#define TCP_RTT_MAX (1<<18) /* maximum rtt */
311
312/*
313 * The initial retransmission should happen at rtt + 4 * rttvar.
314 * Because of the way we do the smoothing, srtt and rttvar
315 * will each average +1/2 tick of bias. When we compute
316 * the retransmit timer, we want 1/2 tick of rounding and
317 * 1 extra tick because of +-1/2 tick uncertainty in the
318 * firing of the timer. The bias will give us exactly the
319 * 1.5 tick we need. But, because the bias is
320 * statistical, we have to test that we don't drop below
321 * the minimum feasible timer (which is 2 ticks).
322 * This macro assumes that the value of (1 << TCP_RTTVAR_SHIFT)
323 * is the same as the multiplier for rttvar.
324 */
325#define TCP_REXMTVAL(tp) \
326 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) \
327 >> TCP_RTT_BASE_SHIFT)
328
329/*
330 * TCP statistics.
331 * Many of these should be kept per connection,
332 * but that's inconvenient at the moment.
333 */
334struct tcpstat {
335 u_int32_t tcps_connattempt; /* connections initiated */
336 u_int32_t tcps_accepts; /* connections accepted */
337 u_int32_t tcps_connects; /* connections established */
338 u_int32_t tcps_drops; /* connections dropped */
339 u_int32_t tcps_conndrops; /* embryonic connections dropped */
340 u_int32_t tcps_closed; /* conn. closed (includes drops) */
341 u_int32_t tcps_segstimed; /* segs where we tried to get rtt */
342 u_int32_t tcps_rttupdated; /* times we succeeded */
343 u_int32_t tcps_delack; /* delayed acks sent */
344 u_int32_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
345 u_int32_t tcps_rexmttimeo; /* retransmit timeouts */
346 u_int32_t tcps_persisttimeo; /* persist timeouts */
347 u_int32_t tcps_persistdrop; /* connections dropped in persist */
348 u_int32_t tcps_keeptimeo; /* keepalive timeouts */
349 u_int32_t tcps_keepprobe; /* keepalive probes sent */
350 u_int32_t tcps_keepdrops; /* connections dropped in keepalive */
351
352 u_int32_t tcps_sndtotal; /* total packets sent */
353 u_int32_t tcps_sndpack; /* data packets sent */
354 u_int64_t tcps_sndbyte; /* data bytes sent */
355 u_int32_t tcps_sndrexmitpack; /* data packets retransmitted */
356 u_int64_t tcps_sndrexmitbyte; /* data bytes retransmitted */
357 u_int64_t tcps_sndrexmitfast; /* Fast retransmits */
358 u_int32_t tcps_sndacks; /* ack-only packets sent */
359 u_int32_t tcps_sndprobe; /* window probes sent */
360 u_int32_t tcps_sndurg; /* packets sent with URG only */
361 u_int32_t tcps_sndwinup; /* window update-only packets sent */
362 u_int32_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
363
364 u_int32_t tcps_rcvtotal; /* total packets received */
365 u_int32_t tcps_rcvpack; /* packets received in sequence */
366 u_int64_t tcps_rcvbyte; /* bytes received in sequence */
367 u_int32_t tcps_rcvbadsum; /* packets received with ccksum errs */
368 u_int32_t tcps_rcvbadoff; /* packets received with bad offset */
369 u_int32_t tcps_rcvmemdrop; /* packets dropped for lack of memory */
370 u_int32_t tcps_rcvnosec; /* packets dropped for lack of ipsec */
371 u_int32_t tcps_rcvshort; /* packets received too short */
372 u_int32_t tcps_rcvduppack; /* duplicate-only packets received */
373 u_int64_t tcps_rcvdupbyte; /* duplicate-only bytes received */
374 u_int32_t tcps_rcvpartduppack; /* packets with some duplicate data */
375 u_int64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
376 u_int32_t tcps_rcvoopack; /* out-of-order packets received */
377 u_int64_t tcps_rcvoobyte; /* out-of-order bytes received */
378 u_int32_t tcps_rcvpackafterwin; /* packets with data after window */
379 u_int64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */
380 u_int32_t tcps_rcvafterclose; /* packets rcvd after "close" */
381 u_int32_t tcps_rcvwinprobe; /* rcvd window probe packets */
382 u_int32_t tcps_rcvdupack; /* rcvd duplicate acks */
383 u_int32_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */
384 u_int32_t tcps_rcvacktooold; /* rcvd acks for old data */
385 u_int32_t tcps_rcvackpack; /* rcvd ack packets */
386 u_int64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */
387 u_int32_t tcps_rcvwinupd; /* rcvd window update packets */
388 u_int32_t tcps_pawsdrop; /* segments dropped due to PAWS */
389 u_int32_t tcps_predack; /* times hdr predict ok for acks */
390 u_int32_t tcps_preddat; /* times hdr predict ok for data pkts */
391
392 u_int32_t tcps_pcbhashmiss; /* input packets missing pcb hash */
393 u_int32_t tcps_noport; /* no socket on port */
394 u_int32_t tcps_closing; /* inpcb exists, socket is closing */
395 u_int32_t tcps_badsyn; /* SYN packet with src==dst rcv'ed */
396 u_int32_t tcps_dropsyn; /* SYN packet dropped */
397
398 u_int32_t tcps_rcvbadsig; /* rcvd bad/missing TCP signatures */
399 u_int64_t tcps_rcvgoodsig; /* rcvd good TCP signatures */
400 u_int32_t tcps_inswcsum; /* input software-checksummed pkts */
401 u_int32_t tcps_outswcsum; /* output software-checksummed pkts */
402
403 /* ECN stats */
404 u_int32_t tcps_ecn_accepts; /* ecn connections accepted */
405 u_int32_t tcps_ecn_rcvece; /* # of rcvd ece */
406 u_int32_t tcps_ecn_rcvcwr; /* # of rcvd cwr */
407 u_int32_t tcps_ecn_rcvce; /* # of rcvd ce in ip header */
408 u_int32_t tcps_ecn_sndect; /* # of cwr sent */
409 u_int32_t tcps_ecn_sndece; /* # of ece sent */
410 u_int32_t tcps_ecn_sndcwr; /* # of cwr sent */
411 u_int32_t tcps_cwr_ecn; /* # of cwnd reduced by ecn */
412 u_int32_t tcps_cwr_frecovery; /* # of cwnd reduced by fastrecovery */
413 u_int32_t tcps_cwr_timeout; /* # of cwnd reduced by timeout */
414
415 /* These statistics deal with the SYN cache. */
416 u_int64_t tcps_sc_added; /* # of entries added */
417 u_int64_t tcps_sc_completed; /* # of connections completed */
418 u_int64_t tcps_sc_timed_out; /* # of entries timed out */
419 u_int64_t tcps_sc_overflowed; /* # dropped due to overflow */
420 u_int64_t tcps_sc_reset; /* # dropped due to RST */
421 u_int64_t tcps_sc_unreach; /* # dropped due to ICMP unreach */
422 u_int64_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
423 u_int64_t tcps_sc_aborted; /* # of entries aborted (no mem) */
424 u_int64_t tcps_sc_dupesyn; /* # of duplicate SYNs received */
425 u_int64_t tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */
426 u_int64_t tcps_sc_collisions; /* # of hash collisions */
427 u_int64_t tcps_sc_retransmitted;/* # of retransmissions */
428 u_int64_t tcps_sc_seedrandom; /* # of syn cache seeds with random */
429 u_int64_t tcps_sc_hash_size; /* hash buckets in current syn cache */
430 u_int64_t tcps_sc_entry_count; /* # of entries in current syn cache */
431 u_int64_t tcps_sc_entry_limit; /* limit of syn cache entries */
432 u_int64_t tcps_sc_bucket_maxlen;/* maximum # of entries in any bucket */
433 u_int64_t tcps_sc_bucket_limit; /* limit of syn cache bucket list */
434 int64_t tcps_sc_uses_left; /* use counter of current syn cache */
435
436 u_int64_t tcps_conndrained; /* # of connections drained */
437
438 u_int64_t tcps_sack_recovery_episode; /* SACK recovery episodes */
439 u_int64_t tcps_sack_rexmits; /* SACK rexmit segments */
440 u_int64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */
441 u_int64_t tcps_sack_rcv_opts; /* SACK options received */
442 u_int64_t tcps_sack_snd_opts; /* SACK options sent */
443 u_int64_t tcps_sack_drop_opts; /* SACK options dropped */
444
445 u_int32_t tcps_outswtso; /* output tso chopped in software */
446 u_int32_t tcps_outhwtso; /* output tso processed by hardware */
447 u_int32_t tcps_outpkttso; /* packets generated by tso */
448 u_int32_t tcps_outbadtso; /* output tso failed, packet dropped */
449 u_int32_t tcps_inswlro; /* input lro on pseudo device */
450 u_int32_t tcps_inhwlro; /* input lro from hardware */
451 u_int32_t tcps_inpktlro; /* packets coalesced by hardware lro */
452 u_int32_t tcps_inbadlro; /* input bad lro packets */
453};
454
455/*
456 * Names for TCP sysctl objects.
457 */
458
459#define TCPCTL_RFC1323 1 /* enable RFC1323 timestamps/scaling */
460#define TCPCTL_KEEPINITTIME 2 /* TCPT_KEEP value */
461#define TCPCTL_KEEPIDLE 3 /* allow tcp_keepidle to be changed */
462#define TCPCTL_KEEPINTVL 4 /* allow tcp_keepintvl to be changed */
463#define TCPCTL_SLOWHZ 5 /* return kernel idea of PR_SLOWHZ */
464#define TCPCTL_BADDYNAMIC 6 /* return bad dynamic port bitmap */
465#define TCPCTL_RECVSPACE 7 /* receive buffer space */
466#define TCPCTL_SENDSPACE 8 /* send buffer space */
467#define TCPCTL_IDENT 9 /* get connection owner */
468#define TCPCTL_SACK 10 /* selective acknowledgement, rfc 2018 */
469#define TCPCTL_MSSDFLT 11 /* Default maximum segment size */
470#define TCPCTL_RSTPPSLIMIT 12 /* RST pps limit */
471#define TCPCTL_ACK_ON_PUSH 13 /* ACK immediately on PUSH */
472#define TCPCTL_ECN 14 /* RFC3168 ECN */
473#define TCPCTL_SYN_CACHE_LIMIT 15 /* max size of comp. state engine */
474#define TCPCTL_SYN_BUCKET_LIMIT 16 /* max size of hash bucket */
475#define TCPCTL_RFC3390 17 /* enable/disable RFC3390 increased cwnd */
476#define TCPCTL_REASS_LIMIT 18 /* max entries for tcp reass queues */
477#define TCPCTL_DROP 19 /* drop tcp connection */
478#define TCPCTL_SACKHOLE_LIMIT 20 /* max entries for tcp sack queues */
479#define TCPCTL_STATS 21 /* TCP statistics */
480#define TCPCTL_ALWAYS_KEEPALIVE 22 /* assume SO_KEEPALIVE is always set */
481#define TCPCTL_SYN_USE_LIMIT 23 /* number of uses before reseeding hash */
482#define TCPCTL_ROOTONLY 24 /* return root only port bitmap */
483#define TCPCTL_SYN_HASH_SIZE 25 /* number of buckets in the hash */
484#define TCPCTL_TSO 26 /* enable TCP segmentation offload */
485#define TCPCTL_MAXID 27
486
487#define TCPCTL_NAMES { \
488 { 0, 0 }, \
489 { "rfc1323", CTLTYPE_INT }, \
490 { "keepinittime", CTLTYPE_INT }, \
491 { "keepidle", CTLTYPE_INT }, \
492 { "keepintvl", CTLTYPE_INT }, \
493 { NULL, 0 }, \
494 { "baddynamic", CTLTYPE_STRUCT }, \
495 { NULL, 0 }, \
496 { NULL, 0 }, \
497 { "ident", CTLTYPE_STRUCT }, \
498 { "sack", CTLTYPE_INT }, \
499 { "mssdflt", CTLTYPE_INT }, \
500 { "rstppslimit", CTLTYPE_INT }, \
501 { "ackonpush", CTLTYPE_INT }, \
502 { "ecn", CTLTYPE_INT }, \
503 { "syncachelimit", CTLTYPE_INT }, \
504 { "synbucketlimit", CTLTYPE_INT }, \
505 { "rfc3390", CTLTYPE_INT }, \
506 { "reasslimit", CTLTYPE_INT }, \
507 { "drop", CTLTYPE_STRUCT }, \
508 { "sackholelimit", CTLTYPE_INT }, \
509 { "stats", CTLTYPE_STRUCT }, \
510 { "always_keepalive", CTLTYPE_INT }, \
511 { "synuselimit", CTLTYPE_INT }, \
512 { "rootonly", CTLTYPE_STRUCT }, \
513 { "synhashsize", CTLTYPE_INT }, \
514 { "tso", CTLTYPE_INT }, \
515}
516
517struct tcp_ident_mapping {
518 struct sockaddr_storage faddr, laddr;
519 int euid, ruid;
520 u_int rdomain;
521};
522
523#ifdef _KERNEL
524
525#include <sys/percpu.h>
526#include <sys/stat.h>
527
528enum tcpstat_counters {
529 tcps_connattempt,
530 tcps_accepts,
531 tcps_connects,
532 tcps_drops,
533 tcps_conndrops,
534 tcps_closed,
535 tcps_segstimed,
536 tcps_rttupdated,
537 tcps_delack,
538 tcps_timeoutdrop,
539 tcps_rexmttimeo,
540 tcps_persisttimeo,
541 tcps_persistdrop,
542 tcps_keeptimeo,
543 tcps_keepprobe,
544 tcps_keepdrops,
545 tcps_sndtotal,
546 tcps_sndpack,
547 tcps_sndbyte,
548 tcps_sndrexmitpack,
549 tcps_sndrexmitbyte,
550 tcps_sndrexmitfast,
551 tcps_sndacks,
552 tcps_sndprobe,
553 tcps_sndurg,
554 tcps_sndwinup,
555 tcps_sndctrl,
556 tcps_rcvtotal,
557 tcps_rcvpack,
558 tcps_rcvbyte,
559 tcps_rcvbadsum,
560 tcps_rcvbadoff,
561 tcps_rcvmemdrop,
562 tcps_rcvnosec,
563 tcps_rcvshort,
564 tcps_rcvduppack,
565 tcps_rcvdupbyte,
566 tcps_rcvpartduppack,
567 tcps_rcvpartdupbyte,
568 tcps_rcvoopack,
569 tcps_rcvoobyte,
570 tcps_rcvpackafterwin,
571 tcps_rcvbyteafterwin,
572 tcps_rcvafterclose,
573 tcps_rcvwinprobe,
574 tcps_rcvdupack,
575 tcps_rcvacktoomuch,
576 tcps_rcvacktooold,
577 tcps_rcvackpack,
578 tcps_rcvackbyte,
579 tcps_rcvwinupd,
580 tcps_pawsdrop,
581 tcps_predack,
582 tcps_preddat,
583 tcps_pcbhashmiss,
584 tcps_noport,
585 tcps_closing,
586 tcps_badsyn,
587 tcps_dropsyn,
588 tcps_rcvbadsig,
589 tcps_rcvgoodsig,
590 tcps_inswcsum,
591 tcps_outswcsum,
592 tcps_ecn_accepts,
593 tcps_ecn_rcvece,
594 tcps_ecn_rcvcwr,
595 tcps_ecn_rcvce,
596 tcps_ecn_sndect,
597 tcps_ecn_sndece,
598 tcps_ecn_sndcwr,
599 tcps_cwr_ecn,
600 tcps_cwr_frecovery,
601 tcps_cwr_timeout,
602 tcps_sc_added,
603 tcps_sc_completed,
604 tcps_sc_timed_out,
605 tcps_sc_overflowed,
606 tcps_sc_reset,
607 tcps_sc_unreach,
608 tcps_sc_bucketoverflow,
609 tcps_sc_aborted,
610 tcps_sc_dupesyn,
611 tcps_sc_dropped,
612 tcps_sc_collisions,
613 tcps_sc_retransmitted,
614 tcps_sc_seedrandom,
615 tcps_sc_hash_size,
616 tcps_sc_entry_count,
617 tcps_sc_entry_limit,
618 tcps_sc_bucket_maxlen,
619 tcps_sc_bucket_limit,
620 tcps_sc_uses_left,
621 tcps_conndrained,
622 tcps_sack_recovery_episode,
623 tcps_sack_rexmits,
624 tcps_sack_rexmit_bytes,
625 tcps_sack_rcv_opts,
626 tcps_sack_snd_opts,
627 tcps_sack_drop_opts,
628 tcps_outswtso,
629 tcps_outhwtso,
630 tcps_outpkttso,
631 tcps_outbadtso,
632 tcps_inswlro,
633 tcps_inhwlro,
634 tcps_inpktlro,
635 tcps_inbadlro,
636 tcps_ncounters,
637};
638
639extern struct cpumem *tcpcounters;
640
641static inline void
642tcpstat_inc(enum tcpstat_counters c)
643{
644 int s = splnet();
645 counters_inc(tcpcounters, c);
646 splx(s);
647}
648
649static inline void
650tcpstat_add(enum tcpstat_counters c, uint64_t v)
651{
652 int s = splnet();
653 counters_add(tcpcounters, c, v);
654 splx(s);
655}
656
657static inline void
658tcpstat_pkt(enum tcpstat_counters pcounter, enum tcpstat_counters bcounter,
659 uint64_t v)
660{
661 int s = splnet();
662 counters_pkt(tcpcounters, pcounter, bcounter, v);
663 splx(s);
664}
665
666extern uint64_t tcp_starttime;
667
668static inline uint64_t
669tcp_now(void)
670{
671 /* TCP time ticks in 63 bit milliseconds with 63 bit random offset. */
672 return tcp_starttime + (getnsecruntime() / 1000000ULL);
673}
674
675#define TCP_TIME(_sec) ((_sec) * 1000) /* tcp_now() is in milliseconds */
676
677extern struct mutex tcp_timer_mtx;
678extern const struct pr_usrreqs tcp_usrreqs;
679
680#ifdef INET6
681extern const struct pr_usrreqs tcp6_usrreqs;
682#endif
683
684extern struct pool tcpcb_pool;
685extern struct inpcbtable tcbtable, tcb6table; /* queue of active tcpcb's */
686extern int tcp_do_rfc1323; /* [a] enabled/disabled? */
687extern const int tcprexmtthresh;
688extern int tcp_mssdflt; /* [a] default maximum segment size */
689extern int tcp_rst_ppslim; /* [a] maximum outgoing RST packet per second */
690extern int tcp_ack_on_push; /* [a] ACK immediately on PUSH */
691extern int tcp_do_sack; /* [a] SACK enabled/disabled */
692extern struct pool sackhl_pool;
693extern int tcp_sackhole_limit; /* max entries for tcp sack queues */
694extern int tcp_do_ecn; /* [a] RFC3168 ECN enabled/disabled? */
695extern int tcp_do_rfc3390; /* [a] RFC3390 Increasing TCP Initial Window */
696extern int tcp_do_tso; /* [a] enable TSO for TCP output packets */
697
698extern struct pool tcpqe_pool;
699extern int tcp_reass_limit; /* max entries for tcp reass queues */
700
701extern int tcp_syn_hash_size; /* adjustable size of the hash array */
702extern int tcp_syn_cache_limit; /* max entries for compressed state engine */
703extern int tcp_syn_bucket_limit;/* max entries per hash bucket */
704extern int tcp_syn_use_limit; /* number of uses before reseeding hash */
705extern struct syn_cache_set tcp_syn_cache[];
706extern int tcp_syn_cache_active; /* active syn cache, may be 0 or 1 */
707
708struct tdb;
709
710void tcp_canceltimers(struct tcpcb *);
711struct tcpcb *
712 tcp_close(struct tcpcb *);
713int tcp_freeq(struct tcpcb *);
714#ifdef INET6
715void tcp6_ctlinput(int, struct sockaddr *, u_int, void *);
716#endif
717void tcp_ctlinput(int, struct sockaddr *, u_int, void *);
718int tcp_ctloutput(int, struct socket *, int, int, struct mbuf *);
719struct tcpcb *
720 tcp_dodisconnect(struct tcpcb *);
721struct tcpcb *
722 tcp_drop(struct tcpcb *, int);
723int tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *,
724 struct mbuf *, int, struct tcp_opt_info *, u_int, uint64_t);
725void tcp_init(void);
726int tcp_input(struct mbuf **, int *, int, int, struct netstack *);
727void tcp_input_mlist(struct mbuf_list *, int);
728int tcp_mss(struct tcpcb *, int);
729void tcp_mss_update(struct tcpcb *);
730void tcp_softlro_glue(struct mbuf_list *, struct mbuf *, struct ifnet *);
731u_int tcp_hdrsz(struct tcpcb *);
732void tcp_mtudisc(struct inpcb *, int);
733void tcp_mtudisc_increase(struct inpcb *, int);
734#ifdef INET6
735void tcp6_mtudisc_callback(struct sockaddr_in6 *, u_int);
736#endif
737struct tcpcb *
738 tcp_newtcpcb(struct inpcb *, int);
739void tcp_notify(struct inpcb *, int);
740int tcp_output(struct tcpcb *);
741int tcp_softtso_chop(struct mbuf_list *, struct mbuf *, struct ifnet *,
742 u_int);
743int tcp_if_output_tso(struct ifnet *, struct mbuf **, struct sockaddr *,
744 struct rtentry *, uint32_t, u_int);
745void tcp_pulloutofband(struct socket *, u_int, struct mbuf *, int);
746int tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *);
747void tcp_rscale(struct tcpcb *, u_long);
748void tcp_respond(struct tcpcb *, caddr_t, struct tcphdr *, tcp_seq,
749 tcp_seq, int, u_int, uint64_t);
750void tcp_setpersist(struct tcpcb *);
751void tcp_update_sndspace(struct tcpcb *);
752void tcp_update_rcvspace(struct tcpcb *);
753void tcp_slowtimo(void);
754struct mbuf *
755 tcp_template(struct tcpcb *);
756#ifndef SMALL_KERNEL
757void tcp_trace(short, short, struct tcpcb *, struct tcpcb *, caddr_t,
758 int, int);
759#endif
760struct tcpcb *
761 tcp_usrclosed(struct tcpcb *);
762int tcp_sysctl(int *, u_int, void *, size_t *, void *, size_t);
763int tcp_attach(struct socket *, int, int);
764int tcp_detach(struct socket *);
765int tcp_bind(struct socket *, struct mbuf *, struct proc *);
766int tcp_listen(struct socket *);
767int tcp_connect(struct socket *, struct mbuf *);
768int tcp_accept(struct socket *, struct mbuf *);
769int tcp_disconnect(struct socket *);
770int tcp_shutdown(struct socket *);
771void tcp_rcvd(struct socket *);
772int tcp_send(struct socket *, struct mbuf *, struct mbuf *,
773 struct mbuf *);
774void tcp_abort(struct socket *);
775int tcp_sockaddr(struct socket *, struct mbuf *);
776int tcp_peeraddr(struct socket *, struct mbuf *);
777int tcp_sense(struct socket *, struct stat *);
778int tcp_rcvoob(struct socket *, struct mbuf *, int);
779int tcp_sendoob(struct socket *, struct mbuf *, struct mbuf *,
780 struct mbuf *);
781void tcp_xmit_timer(struct tcpcb *, int32_t);
782void tcp_sack_option(struct tcpcb *,struct tcphdr *,u_char *,int);
783void tcp_update_sack_list(struct tcpcb *tp, tcp_seq, tcp_seq);
784void tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
785void tcp_clean_sackreport(struct tcpcb *tp);
786void tcp_sack_adjust(struct tcpcb *tp);
787struct sackhole *
788 tcp_sack_output(struct tcpcb *tp);
789#ifdef DEBUG
790void tcp_print_holes(struct tcpcb *tp);
791#endif
792u_long tcp_seq_subtract(u_long, u_long );
793#ifdef TCP_SIGNATURE
794int tcp_signature_apply(caddr_t, caddr_t, unsigned int);
795int tcp_signature(struct tdb *, int, struct mbuf *, struct tcphdr *,
796 int, int, char *);
797#endif /* TCP_SIGNATURE */
798void tcp_set_iss_tsm(struct tcpcb *);
799
800void syn_cache_unreach(const struct sockaddr *, const struct sockaddr *,
801 struct tcphdr *, u_int);
802void syn_cache_init(void);
803void syn_cache_cleanup(struct tcpcb *);
804
805#ifdef SMALL_KERNEL
806static inline void
807tcp_trace(short act, short ostate, struct tcpcb *tp, struct tcpcb *otp,
808 caddr_t headers, int req, int len)
809{
810}
811#endif
812
813#endif /* _KERNEL */
814#endif /* _NETINET_TCP_VAR_H_ */