2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1982, 1986, 1993
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.h 8.1 (Berkeley) 6/10/93
35 #ifndef _NETINET_TCP_H_
36 #define _NETINET_TCP_H_
38 #include <sys/cdefs.h>
39 #include <sys/types.h>
43 typedef u_int32_t tcp_seq;
45 #define tcp6_seq tcp_seq /* for KAME src sync over BSD*'s */
46 #define tcp6hdr tcphdr /* for KAME src sync over BSD*'s */
50 * Per RFC 793, September, 1981.
53 u_short th_sport; /* source port */
54 u_short th_dport; /* destination port */
55 tcp_seq th_seq; /* sequence number */
56 tcp_seq th_ack; /* acknowledgement number */
57 #if BYTE_ORDER == LITTLE_ENDIAN
58 u_char th_x2:4, /* (unused) */
59 th_off:4; /* data offset */
61 #if BYTE_ORDER == BIG_ENDIAN
62 u_char th_off:4, /* data offset */
63 th_x2:4; /* (unused) */
74 #define TH_AE 0x100 /* maps into th_x2 */
75 #define TH_FLAGS (TH_FIN|TH_SYN|TH_RST|TH_PUSH|TH_ACK|TH_URG|TH_ECE|TH_CWR)
76 #define PRINT_TH_FLAGS "\20\1FIN\2SYN\3RST\4PUSH\5ACK\6URG\7ECE\10CWR\11AE"
78 u_short th_win; /* window */
79 u_short th_sum; /* checksum */
80 u_short th_urp; /* urgent pointer */
83 #define PADTCPOLEN(len) ((((len) / 4) + !!((len) % 4)) * 4)
87 #define TCPOPT_PAD 0 /* padding after EOL */
91 #define TCPOPT_MAXSEG 2
92 #define TCPOLEN_MAXSEG 4
93 #define TCPOPT_WINDOW 3
94 #define TCPOLEN_WINDOW 3
95 #define TCPOPT_SACK_PERMITTED 4
96 #define TCPOLEN_SACK_PERMITTED 2
98 #define TCPOLEN_SACKHDR 2
99 #define TCPOLEN_SACK 8 /* 2*sizeof(tcp_seq) */
100 #define TCPOPT_TIMESTAMP 8
101 #define TCPOLEN_TIMESTAMP 10
102 #define TCPOLEN_TSTAMP_APPA (TCPOLEN_TIMESTAMP+2) /* appendix A */
103 #define TCPOPT_SIGNATURE 19 /* Keyed MD5: RFC 2385 */
104 #define TCPOLEN_SIGNATURE 18
105 #define TCPOPT_FAST_OPEN 34
106 #define TCPOLEN_FAST_OPEN_EMPTY 2
108 /* Miscellaneous constants */
109 #define MAX_SACK_BLKS 6 /* Max # SACK blocks stored at receiver side */
110 #define TCP_MAX_SACK 4 /* MAX # SACKs sent in any segment */
113 * The default maximum segment size (MSS) to be used for new TCP connections
114 * when path MTU discovery is not enabled.
116 * RFC879 derives the default MSS from the largest datagram size hosts are
117 * minimally required to handle directly or through IP reassembly minus the
118 * size of the IP and TCP header. With IPv6 the minimum MTU is specified
121 * For IPv4 the MSS is 576 - sizeof(struct tcpiphdr)
122 * For IPv6 the MSS is IPV6_MMTU - sizeof(struct ip6_hdr) - sizeof(struct tcphdr)
124 * We use explicit numerical definition here to avoid header pollution.
127 #define TCP6_MSS 1220
130 * Limit the lowest MSS we accept for path MTU discovery and the TCP SYN MSS
131 * option. Allowing low values of MSS can consume significant resources and
132 * be used to mount a resource exhaustion attack.
133 * Connections requesting lower MSS values will be rounded up to this value
134 * and the IP_DF flag will be cleared to allow fragmentation along the path.
136 * See tcp_subr.c tcp_minmss SYSCTL declaration for more comments. Setting
137 * it to "0" disables the minmss check.
139 * The default value is fine for TCP across the Internet's smallest official
140 * link MTU (256 bytes for AX.25 packet radio). However, a connection is very
141 * unlikely to come across such low MTU interfaces these days (anno domini 2003).
143 #define TCP_MINMSS 216
145 #define TCP_MAXWIN 65535 /* largest value for (unscaled) window */
146 #define TTCP_CLIENT_SND_WND 4096 /* dflt send window for T/TCP client */
148 #define TCP_MAX_WINSHIFT 14 /* maximum window shift */
150 #define TCP_MAXBURST 4 /* maximum segments in a burst */
152 #define TCP_MAXHLEN (0xf<<2) /* max length of header in bytes */
153 #define TCP_MAXOLEN (TCP_MAXHLEN - sizeof(struct tcphdr))
154 /* max space left for options */
156 #define TCP_FASTOPEN_MIN_COOKIE_LEN 4 /* Per RFC7413 */
157 #define TCP_FASTOPEN_MAX_COOKIE_LEN 16 /* Per RFC7413 */
158 #define TCP_FASTOPEN_PSK_LEN 16 /* Same as TCP_FASTOPEN_KEY_LEN */
159 #endif /* __BSD_VISIBLE */
162 * User-settable options (used with setsockopt). These are discrete
163 * values and are not masked together. Some values appear to be
164 * bitmasks for historical reasons.
166 #define TCP_NODELAY 1 /* don't delay send to coalesce packets */
168 #define TCP_MAXSEG 2 /* set maximum segment size */
169 #define TCP_NOPUSH 4 /* don't push last block of write */
170 #define TCP_NOOPT 8 /* don't use TCP options */
171 #define TCP_MD5SIG 16 /* use MD5 digests (RFC2385) */
172 #define TCP_INFO 32 /* retrieve tcp_info structure */
173 #define TCP_STATS 33 /* retrieve stats blob structure */
174 #define TCP_LOG 34 /* configure event logging for connection */
175 #define TCP_LOGBUF 35 /* retrieve event log for connection */
176 #define TCP_LOGID 36 /* configure log ID to correlate connections */
177 #define TCP_LOGDUMP 37 /* dump connection log events to device */
178 #define TCP_LOGDUMPID 38 /* dump events from connections with same ID to
180 #define TCP_TXTLS_ENABLE 39 /* TLS framing and encryption for transmit */
181 #define TCP_TXTLS_MODE 40 /* Transmit TLS mode */
182 #define TCP_RXTLS_ENABLE 41 /* TLS framing and encryption for receive */
183 #define TCP_RXTLS_MODE 42 /* Receive TLS mode */
184 #define TCP_CONGESTION 64 /* get/set congestion control algorithm */
185 #define TCP_CCALGOOPT 65 /* get/set cc algorithm specific options */
186 #define TCP_DELACK 72 /* socket option for delayed ack */
187 #define TCP_FIN_IS_RST 73 /* A fin from the peer is treated has a RST */
188 #define TCP_LOG_LIMIT 74 /* Limit to number of records in tcp-log */
189 #define TCP_SHARED_CWND_ALLOWED 75 /* Use of a shared cwnd is allowed */
190 #define TCP_KEEPINIT 128 /* N, time to establish connection */
191 #define TCP_KEEPIDLE 256 /* L,N,X start keeplives after this period */
192 #define TCP_KEEPINTVL 512 /* L,N interval between keepalives */
193 #define TCP_KEEPCNT 1024 /* L,N number of keepalives before close */
194 #define TCP_FASTOPEN 1025 /* enable TFO / was created via TFO */
195 #define TCP_PCAP_OUT 2048 /* number of output packets to keep */
196 #define TCP_PCAP_IN 4096 /* number of input packets to keep */
197 #define TCP_FUNCTION_BLK 8192 /* Set the tcp function pointers to the specified stack */
198 /* Options for Rack and BBR */
199 #define TCP_RACK_MBUF_QUEUE 1050 /* Do we allow mbuf queuing if supported */
200 #define TCP_RACK_PROP 1051 /* RACK proportional rate reduction (bool) */
201 #define TCP_RACK_TLP_REDUCE 1052 /* RACK TLP cwnd reduction (bool) */
202 #define TCP_RACK_PACE_REDUCE 1053 /* RACK Pacing reduction factor (divisor) */
203 #define TCP_RACK_PACE_MAX_SEG 1054 /* Max TSO size we will send */
204 #define TCP_RACK_PACE_ALWAYS 1055 /* Use the always pace method */
205 #define TCP_RACK_PROP_RATE 1056 /* The proportional reduction rate */
206 #define TCP_RACK_PRR_SENDALOT 1057 /* Allow PRR to send more than one seg */
207 #define TCP_RACK_MIN_TO 1058 /* Minimum time between rack t-o's in ms */
208 #define TCP_RACK_EARLY_RECOV 1059 /* Should recovery happen early (bool) */
209 #define TCP_RACK_EARLY_SEG 1060 /* If early recovery max segments */
210 #define TCP_RACK_REORD_THRESH 1061 /* RACK reorder threshold (shift amount) */
211 #define TCP_RACK_REORD_FADE 1062 /* Does reordering fade after ms time */
212 #define TCP_RACK_TLP_THRESH 1063 /* RACK TLP theshold i.e. srtt+(srtt/N) */
213 #define TCP_RACK_PKT_DELAY 1064 /* RACK added ms i.e. rack-rtt + reord + N */
214 #define TCP_RACK_TLP_INC_VAR 1065 /* Does TLP include rtt variance in t-o */
215 #define TCP_BBR_IWINTSO 1067 /* Initial TSO window for BBRs first sends */
216 #define TCP_BBR_RECFORCE 1068 /* Enter recovery force out a segment disregard pacer no longer valid */
217 #define TCP_BBR_STARTUP_PG 1069 /* Startup pacing gain */
218 #define TCP_BBR_DRAIN_PG 1070 /* Drain pacing gain */
219 #define TCP_BBR_RWND_IS_APP 1071 /* Rwnd limited is considered app limited */
220 #define TCP_BBR_PROBE_RTT_INT 1072 /* How long in useconds between probe-rtt */
221 #define TCP_BBR_ONE_RETRAN 1073 /* Is only one segment allowed out during retran */
222 #define TCP_BBR_STARTUP_LOSS_EXIT 1074 /* Do we exit a loss during startup if not 20% incr */
223 #define TCP_BBR_USE_LOWGAIN 1075 /* lower the gain in PROBE_BW enable */
224 #define TCP_BBR_LOWGAIN_THRESH 1076 /* Unused after 2.3 morphs to TSLIMITS >= 2.3 */
225 #define TCP_BBR_TSLIMITS 1076 /* Do we use experimental Timestamp limiting for our algo */
226 #define TCP_BBR_LOWGAIN_HALF 1077 /* Unused after 2.3 */
227 #define TCP_BBR_PACE_OH 1077 /* Reused in 4.2 for pacing overhead setting */
228 #define TCP_BBR_LOWGAIN_FD 1078 /* Unused after 2.3 */
229 #define TCP_BBR_HOLD_TARGET 1078 /* For 4.3 on */
230 #define TCP_BBR_USEDEL_RATE 1079 /* Enable use of delivery rate for loss recovery */
231 #define TCP_BBR_MIN_RTO 1080 /* Min RTO in milliseconds */
232 #define TCP_BBR_MAX_RTO 1081 /* Max RTO in milliseconds */
233 #define TCP_BBR_REC_OVER_HPTS 1082 /* Recovery override htps settings 0/1/3 */
234 #define TCP_BBR_UNLIMITED 1083 /* Not used before 2.3 and morphs to algorithm >= 2.3 */
235 #define TCP_BBR_ALGORITHM 1083 /* What measurement algo does BBR use netflix=0, google=1 */
236 #define TCP_BBR_DRAIN_INC_EXTRA 1084 /* Does the 3/4 drain target include the extra gain */
237 #define TCP_BBR_STARTUP_EXIT_EPOCH 1085 /* what epoch gets us out of startup */
238 #define TCP_BBR_PACE_PER_SEC 1086
239 #define TCP_BBR_PACE_DEL_TAR 1087
240 #define TCP_BBR_PACE_SEG_MAX 1088
241 #define TCP_BBR_PACE_SEG_MIN 1089
242 #define TCP_BBR_PACE_CROSS 1090
243 #define TCP_RACK_IDLE_REDUCE_HIGH 1092 /* Reduce the highest cwnd seen to IW on idle */
244 #define TCP_RACK_MIN_PACE 1093 /* Do we enforce rack min pace time */
245 #define TCP_RACK_MIN_PACE_SEG 1094 /* If so what is the seg threshould */
246 #define TCP_RACK_GP_INCREASE 1094 /* After 4.1 its the GP increase in older rack */
247 #define TCP_RACK_TLP_USE 1095
248 #define TCP_BBR_ACK_COMP_ALG 1096 /* Not used */
249 #define TCP_BBR_TMR_PACE_OH 1096 /* Recycled in 4.2 */
250 #define TCP_BBR_EXTRA_GAIN 1097
251 #define TCP_RACK_DO_DETECTION 1097 /* Recycle of extra gain for rack, attack detection */
252 #define TCP_BBR_RACK_RTT_USE 1098 /* what RTT should we use 0, 1, or 2? */
253 #define TCP_BBR_RETRAN_WTSO 1099
254 #define TCP_DATA_AFTER_CLOSE 1100
255 #define TCP_BBR_PROBE_RTT_GAIN 1101
256 #define TCP_BBR_PROBE_RTT_LEN 1102
257 #define TCP_BBR_SEND_IWND_IN_TSO 1103 /* Do we burst out whole iwin size chunks at start? */
258 #define TCP_BBR_USE_RACK_RR 1104 /* Do we use the rack rapid recovery for pacing rxt's */
259 #define TCP_BBR_USE_RACK_CHEAT TCP_BBR_USE_RACK_RR /* Compat. */
260 #define TCP_BBR_HDWR_PACE 1105 /* Enable/disable hardware pacing */
261 #define TCP_BBR_UTTER_MAX_TSO 1106 /* Do we enforce an utter max TSO size */
262 #define TCP_BBR_EXTRA_STATE 1107 /* Special exit-persist catch up */
263 #define TCP_BBR_FLOOR_MIN_TSO 1108 /* The min tso size */
264 #define TCP_BBR_MIN_TOPACEOUT 1109 /* Do we suspend pacing until */
265 #define TCP_BBR_TSTMP_RAISES 1110 /* Can a timestamp measurement raise the b/w */
266 #define TCP_BBR_POLICER_DETECT 1111 /* Turn on/off google mode policer detection */
267 #define TCP_BBR_RACK_INIT_RATE 1112 /* Set an initial pacing rate for when we have no b/w in kbits per sec */
268 #define TCP_RACK_RR_CONF 1113 /* Rack rapid recovery configuration control*/
269 #define TCP_RACK_CHEAT_NOT_CONF_RATE TCP_RACK_RR_CONF
270 #define TCP_RACK_GP_INCREASE_CA 1114 /* GP increase for Congestion Avoidance */
271 #define TCP_RACK_GP_INCREASE_SS 1115 /* GP increase for Slow Start */
272 #define TCP_RACK_GP_INCREASE_REC 1116 /* GP increase for Recovery */
273 #define TCP_RACK_FORCE_MSEG 1117 /* Override to use the user set max-seg value */
274 #define TCP_RACK_PACE_RATE_CA 1118 /* Pacing rate for Congestion Avoidance */
275 #define TCP_RACK_PACE_RATE_SS 1119 /* Pacing rate for Slow Start */
276 #define TCP_RACK_PACE_RATE_REC 1120 /* Pacing rate for Recovery */
277 #define TCP_NO_PRR 1122 /* If pacing, don't use prr */
278 #define TCP_RACK_NONRXT_CFG_RATE 1123 /* In recovery does a non-rxt use the cfg rate */
279 #define TCP_SHARED_CWND_ENABLE 1124 /* Use a shared cwnd if allowed */
280 #define TCP_TIMELY_DYN_ADJ 1125 /* Do we attempt dynamic multipler adjustment with timely. */
281 #define TCP_RACK_NO_PUSH_AT_MAX 1126 /* For timely do not push if we are over max rtt */
282 #define TCP_RACK_PACE_TO_FILL 1127 /* If we are not in recovery, always pace to fill the cwnd in 1 RTT */
283 #define TCP_SHARED_CWND_TIME_LIMIT 1128 /* we should limit to low time values the scwnd life */
284 #define TCP_RACK_PROFILE 1129 /* Select a profile that sets multiple options */
286 /* Start of reserved space for third-party user-settable options. */
287 #define TCP_VENDOR SO_VENDOR
289 #define TCP_CA_NAME_MAX 16 /* max congestion control name length */
291 #define TCPI_OPT_TIMESTAMPS 0x01
292 #define TCPI_OPT_SACK 0x02
293 #define TCPI_OPT_WSCALE 0x04
294 #define TCPI_OPT_ECN 0x08
295 #define TCPI_OPT_TOE 0x10
297 /* Maximum length of log ID. */
298 #define TCP_LOG_ID_LEN 64
301 * The TCP_INFO socket option comes from the Linux 2.6 TCP API, and permits
302 * the caller to query certain information about the state of a TCP
303 * connection. We provide an overlapping set of fields with the Linux
304 * implementation, but since this is a fixed size structure, room has been
305 * left for growth. In order to maximize potential future compatibility with
306 * the Linux API, the same variable names and order have been adopted, and
307 * padding left to make room for omitted fields in case they are added later.
309 * XXX: This is currently an unstable ABI/API, in that it is expected to
313 u_int8_t tcpi_state; /* TCP FSM state. */
314 u_int8_t __tcpi_ca_state;
315 u_int8_t __tcpi_retransmits;
316 u_int8_t __tcpi_probes;
317 u_int8_t __tcpi_backoff;
318 u_int8_t tcpi_options; /* Options enabled on conn. */
319 u_int8_t tcpi_snd_wscale:4, /* RFC1323 send shift value. */
320 tcpi_rcv_wscale:4; /* RFC1323 recv shift value. */
322 u_int32_t tcpi_rto; /* Retransmission timeout (usec). */
323 u_int32_t __tcpi_ato;
324 u_int32_t tcpi_snd_mss; /* Max segment size for send. */
325 u_int32_t tcpi_rcv_mss; /* Max segment size for receive. */
327 u_int32_t __tcpi_unacked;
328 u_int32_t __tcpi_sacked;
329 u_int32_t __tcpi_lost;
330 u_int32_t __tcpi_retrans;
331 u_int32_t __tcpi_fackets;
333 /* Times; measurements in usecs. */
334 u_int32_t __tcpi_last_data_sent;
335 u_int32_t __tcpi_last_ack_sent; /* Also unimpl. on Linux? */
336 u_int32_t tcpi_last_data_recv; /* Time since last recv data. */
337 u_int32_t __tcpi_last_ack_recv;
339 /* Metrics; variable units. */
340 u_int32_t __tcpi_pmtu;
341 u_int32_t __tcpi_rcv_ssthresh;
342 u_int32_t tcpi_rtt; /* Smoothed RTT in usecs. */
343 u_int32_t tcpi_rttvar; /* RTT variance in usecs. */
344 u_int32_t tcpi_snd_ssthresh; /* Slow start threshold. */
345 u_int32_t tcpi_snd_cwnd; /* Send congestion window. */
346 u_int32_t __tcpi_advmss;
347 u_int32_t __tcpi_reordering;
349 u_int32_t __tcpi_rcv_rtt;
350 u_int32_t tcpi_rcv_space; /* Advertised recv window. */
352 /* FreeBSD extensions to tcp_info. */
353 u_int32_t tcpi_snd_wnd; /* Advertised send window. */
354 u_int32_t tcpi_snd_bwnd; /* No longer used. */
355 u_int32_t tcpi_snd_nxt; /* Next egress seqno */
356 u_int32_t tcpi_rcv_nxt; /* Next ingress seqno */
357 u_int32_t tcpi_toe_tid; /* HWTID for TOE endpoints */
358 u_int32_t tcpi_snd_rexmitpack; /* Retransmitted packets */
359 u_int32_t tcpi_rcv_ooopack; /* Out-of-order packets */
360 u_int32_t tcpi_snd_zerowin; /* Zero-sized windows sent */
362 /* Padding to grow without breaking ABI. */
363 u_int32_t __tcpi_pad[26]; /* Padding. */
367 * If this structure is provided when setting the TCP_FASTOPEN socket
368 * option, and the enable member is non-zero, a subsequent connect will use
369 * pre-shared key (PSK) mode using the provided key.
371 struct tcp_fastopen {
373 uint8_t psk[TCP_FASTOPEN_PSK_LEN];
376 #define TCP_FUNCTION_NAME_LEN_MAX 32
378 struct tcp_function_set {
379 char function_set_name[TCP_FUNCTION_NAME_LEN_MAX];
383 /* TLS modes for TCP_TXTLS_MODE */
384 #define TCP_TLS_MODE_NONE 0
385 #define TCP_TLS_MODE_SW 1
386 #define TCP_TLS_MODE_IFNET 2
387 #define TCP_TLS_MODE_TOE 3
390 * TCP Control message types
392 #define TLS_SET_RECORD_TYPE 1
393 #define TLS_GET_RECORD 2
396 * TCP specific variables of interest for tp->t_stats stats(9) accounting.
398 #define VOI_TCP_TXPB 0 /* Transmit payload bytes */
399 #define VOI_TCP_RETXPB 1 /* Retransmit payload bytes */
400 #define VOI_TCP_FRWIN 2 /* Foreign receive window */
401 #define VOI_TCP_LCWIN 3 /* Local congesiton window */
402 #define VOI_TCP_RTT 4 /* Round trip time */
403 #define VOI_TCP_CSIG 5 /* Congestion signal */
404 #define VOI_TCP_GPUT 6 /* Goodput */
405 #define VOI_TCP_CALCFRWINDIFF 7 /* Congestion avoidance LCWIN - FRWIN */
406 #define VOI_TCP_GPUT_ND 8 /* Goodput normalised delta */
407 #define VOI_TCP_ACKLEN 9 /* Average ACKed bytes per ACK */
409 #endif /* !_NETINET_TCP_H_ */