1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1982, 1986, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
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.
18 *
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
29 * SUCH DAMAGE.
30 */
31
32 #ifndef _NETINET_TCP_H_
33 #define _NETINET_TCP_H_
34
35 #include <sys/types.h>
36
37 #if __BSD_VISIBLE
38
39 typedef u_int32_t tcp_seq;
40
41 #define tcp6_seq tcp_seq /* for KAME src sync over BSD*'s */
42 #define tcp6hdr tcphdr /* for KAME src sync over BSD*'s */
43
44 /*
45 * TCP header.
46 * Per RFC 793, September, 1981.
47 */
48 struct tcphdr {
49 u_short th_sport; /* source port */
50 u_short th_dport; /* destination port */
51 tcp_seq th_seq; /* sequence number */
52 tcp_seq th_ack; /* acknowledgement number */
53 #if BYTE_ORDER == LITTLE_ENDIAN
54 u_char th_x2:4, /* upper 4 (reserved) flags */
55 th_off:4; /* data offset */
56 #endif
57 #if BYTE_ORDER == BIG_ENDIAN
58 u_char th_off:4, /* data offset */
59 th_x2:4; /* upper 4 (reserved) flags */
60 #endif
61 u_char th_flags;
62 #define TH_FIN 0x01
63 #define TH_SYN 0x02
64 #define TH_RST 0x04
65 #define TH_PUSH 0x08
66 #define TH_ACK 0x10
67 #define TH_URG 0x20
68 #define TH_ECE 0x40
69 #define TH_CWR 0x80
70 #define TH_AE 0x100 /* maps into th_x2 */
71 #define TH_RES3 0x200
72 #define TH_RES2 0x400
73 #define TH_RES1 0x800
74 #define TH_FLAGS (TH_FIN|TH_SYN|TH_RST|TH_PUSH|TH_ACK|TH_URG|TH_ECE|TH_CWR)
75 #define PRINT_TH_FLAGS "\20\1FIN\2SYN\3RST\4PUSH\5ACK\6URG\7ECE\10CWR\11AE"
76
77 u_short th_win; /* window */
78 u_short th_sum; /* checksum */
79 u_short th_urp; /* urgent pointer */
80 };
81
82 static __inline uint16_t
__tcp_get_flags(const struct tcphdr * th)83 __tcp_get_flags(const struct tcphdr *th)
84 {
85 return (((uint16_t)th->th_x2 << 8) | th->th_flags);
86 }
87
88 static __inline void
__tcp_set_flags(struct tcphdr * th,uint16_t flags)89 __tcp_set_flags(struct tcphdr *th, uint16_t flags)
90 {
91 th->th_x2 = (flags >> 8) & 0x0f;
92 th->th_flags = flags & 0xff;
93 }
94
95 #ifdef _KERNEL
96 #define tcp_get_flags(th) __tcp_get_flags(th)
97 #define tcp_set_flags(th, flags) __tcp_set_flags(th, flags)
98 #endif
99
100 #define PADTCPOLEN(len) ((((len) / 4) + !!((len) % 4)) * 4)
101
102 #define TCPOPT_EOL 0
103 #define TCPOLEN_EOL 1
104 #define TCPOPT_PAD 0 /* padding after EOL */
105 #define TCPOLEN_PAD 1
106 #define TCPOPT_NOP 1
107 #define TCPOLEN_NOP 1
108 #define TCPOPT_MAXSEG 2
109 #define TCPOLEN_MAXSEG 4
110 #define TCPOPT_WINDOW 3
111 #define TCPOLEN_WINDOW 3
112 #define TCPOPT_SACK_PERMITTED 4
113 #define TCPOLEN_SACK_PERMITTED 2
114 #define TCPOPT_SACK 5
115 #define TCPOLEN_SACKHDR 2
116 #define TCPOLEN_SACK 8 /* 2*sizeof(tcp_seq) */
117 #define TCPOPT_TIMESTAMP 8
118 #define TCPOLEN_TIMESTAMP 10
119 #define TCPOLEN_TSTAMP_APPA (TCPOLEN_TIMESTAMP+2) /* appendix A */
120 #define TCPOPT_SIGNATURE 19 /* Keyed MD5: RFC 2385 */
121 #define TCPOLEN_SIGNATURE 18
122 #define TCPOPT_FAST_OPEN 34
123 #define TCPOLEN_FAST_OPEN_EMPTY 2
124
125 #define MAX_TCPOPTLEN 40 /* Absolute maximum TCP options len */
126
127 /* Miscellaneous constants */
128 #define MAX_SACK_BLKS 6 /* Max # SACK blocks stored at receiver side */
129 #define TCP_MAX_SACK 4 /* MAX # SACKs sent in any segment */
130
131 /*
132 * The default maximum segment size (MSS) to be used for new TCP connections
133 * when path MTU discovery is not enabled.
134 *
135 * RFC879 derives the default MSS from the largest datagram size hosts are
136 * minimally required to handle directly or through IP reassembly minus the
137 * size of the IP and TCP header. With IPv6 the minimum MTU is specified
138 * in RFC2460.
139 *
140 * For IPv4 the MSS is 576 - sizeof(struct tcpiphdr)
141 * For IPv6 the MSS is IPV6_MMTU - sizeof(struct ip6_hdr) - sizeof(struct tcphdr)
142 *
143 * We use explicit numerical definition here to avoid header pollution.
144 */
145 #define TCP_MSS 536
146 #define TCP6_MSS 1220
147
148 /*
149 * Limit the lowest MSS we accept for path MTU discovery and the TCP SYN MSS
150 * option. Allowing low values of MSS can consume significant resources and
151 * be used to mount a resource exhaustion attack.
152 * Connections requesting lower MSS values will be rounded up to this value
153 * and the IP_DF flag will be cleared to allow fragmentation along the path.
154 *
155 * See tcp_subr.c tcp_minmss SYSCTL declaration for more comments. Setting
156 * it to "0" disables the minmss check.
157 *
158 * The default value is fine for TCP across the Internet's smallest official
159 * link MTU (256 bytes for AX.25 packet radio). However, a connection is very
160 * unlikely to come across such low MTU interfaces these days (anno domini 2003).
161 */
162 #define TCP_MINMSS 216
163
164 #define TCP_MAXWIN 65535 /* largest value for (unscaled) window */
165 #define TTCP_CLIENT_SND_WND 4096 /* dflt send window for T/TCP client */
166
167 #define TCP_MAX_WINSHIFT 14 /* maximum window shift */
168
169 #define TCP_MAXHLEN (0xf<<2) /* max length of header in bytes */
170 #define TCP_MAXOLEN (TCP_MAXHLEN - sizeof(struct tcphdr))
171 /* max space left for options */
172
173 #define TCP_FASTOPEN_MIN_COOKIE_LEN 4 /* Per RFC7413 */
174 #define TCP_FASTOPEN_MAX_COOKIE_LEN 16 /* Per RFC7413 */
175 #define TCP_FASTOPEN_PSK_LEN 16 /* Same as TCP_FASTOPEN_KEY_LEN */
176 #endif /* __BSD_VISIBLE */
177
178 /*
179 * User-settable options (used with setsockopt). These are discrete
180 * values and are not masked together. Some values appear to be
181 * bitmasks for historical reasons.
182 */
183 #define TCP_NODELAY 1 /* don't delay send to coalesce packets */
184 #if __BSD_VISIBLE
185 #define TCP_MAXSEG 2 /* set maximum segment size */
186 #define TCP_NOPUSH 4 /* don't push last block of write */
187 #define TCP_NOOPT 8 /* don't use TCP options */
188 #define TCP_MD5SIG 16 /* use MD5 digests (RFC2385) */
189 #define TCP_INFO 32 /* retrieve tcp_info structure */
190 #define TCP_STATS 33 /* retrieve stats blob structure */
191 #define TCP_LOG 34 /* configure event logging for connection */
192 #define TCP_LOGBUF 35 /* retrieve event log for connection */
193 #define TCP_LOGID 36 /* configure log ID to correlate connections */
194 #define TCP_LOGDUMP 37 /* dump connection log events to device */
195 #define TCP_LOGDUMPID 38 /* dump events from connections with same ID to
196 device */
197 #define TCP_TXTLS_ENABLE 39 /* TLS framing and encryption for transmit */
198 #define TCP_TXTLS_MODE 40 /* Transmit TLS mode */
199 #define TCP_RXTLS_ENABLE 41 /* TLS framing and encryption for receive */
200 #define TCP_RXTLS_MODE 42 /* Receive TLS mode */
201 #define TCP_IWND_NB 43 /* Override initial window (units: bytes) */
202 #define TCP_IWND_NSEG 44 /* Override initial window (units: MSS segs) */
203 #ifdef _KERNEL
204 #define TCP_USE_DDP 45 /* Use direct data placement for so_rcvbuf */
205 #endif
206 #define TCP_LOGID_CNT 46 /* get number of connections with the same ID */
207 #define TCP_LOG_TAG 47 /* configure tag for grouping logs */
208 #define TCP_USER_LOG 48 /* userspace log event */
209 #define TCP_CONGESTION 64 /* get/set congestion control algorithm */
210 #define TCP_CCALGOOPT 65 /* get/set cc algorithm specific options */
211 #define TCP_MAXUNACKTIME 68 /* maximum time without making progress (sec) */
212 /* unused; was TCP_MAXPEAKRATE */
213 #define TCP_IDLE_REDUCE 70 /* Reduce cwnd on idle input */
214 #define TCP_REMOTE_UDP_ENCAPS_PORT 71 /* Enable TCP over UDP tunneling via the specified port */
215 #define TCP_DELACK 72 /* socket option for delayed ack */
216 #define TCP_FIN_IS_RST 73 /* A fin from the peer is treated has a RST */
217 #define TCP_LOG_LIMIT 74 /* Limit to number of records in tcp-log */
218 #define TCP_SHARED_CWND_ALLOWED 75 /* Use of a shared cwnd is allowed */
219 #define TCP_PROC_ACCOUNTING 76 /* Do accounting on tcp cpu usage and counts */
220 #define TCP_USE_CMP_ACKS 77 /* The transport can handle the Compressed mbuf acks */
221 #define TCP_PERF_INFO 78 /* retrieve accounting counters */
222 #define TCP_KEEPINIT 128 /* N, time to establish connection */
223 #define TCP_KEEPIDLE 256 /* L,N,X start keeplives after this period */
224 #define TCP_KEEPINTVL 512 /* L,N interval between keepalives */
225 #define TCP_KEEPCNT 1024 /* L,N number of keepalives before close */
226 #define TCP_FASTOPEN 1025 /* enable TFO / was created via TFO */
227 #define TCP_PCAP_OUT 2048 /* number of output packets to keep */
228 #define TCP_PCAP_IN 4096 /* number of input packets to keep */
229 #define TCP_FUNCTION_BLK 8192 /* Set the tcp function pointers to the specified stack */
230 #define TCP_FUNCTION_ALIAS 8193 /* Get the current tcp function pointer name alias */
231 /* Options for Rack and BBR */
232 #define TCP_REUSPORT_LB_NUMA 1026 /* set listen socket numa domain */
233 #define TCP_RACK_MBUF_QUEUE 1050 /* Do we allow mbuf queuing if supported */
234 /* unused; was TCP_RACK_PROP */
235 #define TCP_RACK_TLP_REDUCE 1052 /* RACK TLP cwnd reduction (bool) */
236 /* unused; was TCP_RACK_PACE_REDUCE */
237 #define TCP_RACK_PACE_MAX_SEG 1054 /* Max TSO size we will send */
238 #define TCP_RACK_PACE_ALWAYS 1055 /* Use the always pace method */
239 /* unused; was TCP_RACK_PROP_RATE */
240 #define TCP_RACK_PRR_SENDALOT 1057 /* Allow PRR to send more than one seg */
241 #define TCP_RACK_MIN_TO 1058 /* Minimum time between rack t-o's in ms */
242 /* unused; was TCP_RACK_EARLY_RECOV */
243 #define TCP_RACK_EARLY_SEG 1060 /* If early recovery max segments */
244 #define TCP_RACK_REORD_THRESH 1061 /* RACK reorder threshold (shift amount) */
245 #define TCP_RACK_REORD_FADE 1062 /* Does reordering fade after ms time */
246 #define TCP_RACK_TLP_THRESH 1063 /* RACK TLP theshold i.e. srtt+(srtt/N) */
247 #define TCP_RACK_PKT_DELAY 1064 /* RACK added ms i.e. rack-rtt + reord + N */
248 /* unused; was TCP_RACK_TLP_INC_VAR */
249 /* unused */
250 #define TCP_BBR_IWINTSO 1067 /* Initial TSO window for BBRs first sends */
251 /* unused; was TCP_BBR_RECFORCE */
252 #define TCP_BBR_STARTUP_PG 1069 /* Startup pacing gain */
253 #define TCP_BBR_DRAIN_PG 1070 /* Drain pacing gain */
254 /* unused; was TCP_BBR_RWND_IS_APP */
255 #define TCP_BBR_PROBE_RTT_INT 1072 /* How long in useconds between probe-rtt */
256 /* unused; was TCP_BBR_ONE_RETRAN */
257 #define TCP_BBR_STARTUP_LOSS_EXIT 1074 /* Do we exit a loss during startup if not 20% incr */
258 /* unused; was TCP_BBR_USE_LOWGAIN */
259 #define TCP_BBR_TSLIMITS 1076 /* Do we use experimental Timestamp limiting for our algo */
260 #define TCP_BBR_PACE_OH 1077 /* Reused in 4.2 for pacing overhead setting */
261 /* unused; was TCP_BBR_HOLD_TARGET */
262 #define TCP_BBR_USEDEL_RATE 1079 /* Enable use of delivery rate for loss recovery */
263 #define TCP_BBR_MIN_RTO 1080 /* Min RTO in milliseconds */
264 #define TCP_BBR_MAX_RTO 1081 /* Max RTO in milliseconds */
265 /* unused; was TCP_BBR_REC_OVER_HPTS */
266 #define TCP_BBR_ALGORITHM 1083 /* What measurement algo does BBR use netflix=0, google=1 */
267 /* unused; was TCP_BBR_DRAIN_INC_EXTRA */
268 /* unused; was TCP_BBR_STARTUP_EXIT_EPOCH */
269 #define TCP_BBR_PACE_PER_SEC 1086
270 #define TCP_BBR_PACE_DEL_TAR 1087
271 #define TCP_BBR_PACE_SEG_MAX 1088
272 #define TCP_BBR_PACE_SEG_MIN 1089
273 #define TCP_BBR_PACE_CROSS 1090
274 /* unused */
275 /* unused; was TCP_RACK_IDLE_REDUCE_HIGH */
276 /* unused; was TCP_RACK_MIN_PACE */
277 /* unused; was TCP_RACK_GP_INCREASE */
278 #define TCP_RACK_TLP_USE 1095
279 #define TCP_BBR_TMR_PACE_OH 1096 /* Recycled in 4.2 */
280 /* unused; was TCP_BBR_EXTRA_GAIN */
281 #define TCP_RACK_DO_DETECTION 1097 /* Recycle of extra gain for rack, attack detection */
282 #define TCP_BBR_RACK_RTT_USE 1098 /* what RTT should we use 0, 1, or 2? */
283 #define TCP_BBR_RETRAN_WTSO 1099
284 #define TCP_DATA_AFTER_CLOSE 1100
285 #define TCP_BBR_PROBE_RTT_GAIN 1101
286 #define TCP_BBR_PROBE_RTT_LEN 1102
287 #define TCP_BBR_SEND_IWND_IN_TSO 1103 /* Do we burst out whole iwin size chunks at start? */
288 #define TCP_BBR_USE_RACK_RR 1104 /* Do we use the rack rapid recovery for pacing rxt's */
289 #define TCP_BBR_USE_RACK_CHEAT TCP_BBR_USE_RACK_RR /* Compat. */
290 #define TCP_BBR_HDWR_PACE 1105 /* Enable/disable hardware pacing */
291 #define TCP_BBR_UTTER_MAX_TSO 1106 /* Do we enforce an utter max TSO size */
292 #define TCP_BBR_EXTRA_STATE 1107 /* Special exit-persist catch up */
293 #define TCP_BBR_FLOOR_MIN_TSO 1108 /* The min tso size */
294 #define TCP_BBR_MIN_TOPACEOUT 1109 /* Do we suspend pacing until */
295 #define TCP_BBR_TSTMP_RAISES 1110 /* Can a timestamp measurement raise the b/w */
296 #define TCP_BBR_POLICER_DETECT 1111 /* Turn on/off google mode policer detection */
297 #define TCP_BBR_RACK_INIT_RATE 1112 /* Set an initial pacing rate for when we have no b/w in kbits per sec */
298 #define TCP_RACK_RR_CONF 1113 /* Rack rapid recovery configuration control*/
299 #define TCP_RACK_GP_INCREASE_CA 1114 /* GP increase for Congestion Avoidance */
300 #define TCP_RACK_GP_INCREASE_SS 1115 /* GP increase for Slow Start */
301 #define TCP_RACK_GP_INCREASE_REC 1116 /* GP increase for Recovery */
302 #define TCP_RACK_FORCE_MSEG 1117 /* Override to use the user set max-seg value */
303 #define TCP_RACK_PACE_RATE_CA 1118 /* Pacing rate for Congestion Avoidance */
304 #define TCP_RACK_PACE_RATE_SS 1119 /* Pacing rate for Slow Start */
305 #define TCP_RACK_PACE_RATE_REC 1120 /* Pacing rate for Recovery */
306 #define TCP_NO_PRR 1122 /* If pacing, don't use prr */
307 #define TCP_RACK_NONRXT_CFG_RATE 1123 /* In recovery does a non-rxt use the cfg rate */
308 #define TCP_SHARED_CWND_ENABLE 1124 /* Use a shared cwnd if allowed */
309 #define TCP_TIMELY_DYN_ADJ 1125 /* Do we attempt dynamic multipler adjustment with timely. */
310 #define TCP_RACK_NO_PUSH_AT_MAX 1126 /* For timely do not push if we are over max rtt */
311 #define TCP_RACK_PACE_TO_FILL 1127 /* If we are not in recovery, always pace to fill the cwnd in 1 RTT */
312 #define TCP_SHARED_CWND_TIME_LIMIT 1128 /* we should limit to low time values the scwnd life */
313 #define TCP_RACK_PROFILE 1129 /* Select a profile that sets multiple options */
314 #define TCP_HDWR_RATE_CAP 1130 /* Allow hardware rates to cap pacing rate */
315 #define TCP_PACING_RATE_CAP 1131 /* Highest rate allowed in pacing in bytes per second (uint64_t) */
316 #define TCP_HDWR_UP_ONLY 1132 /* Allow the pacing rate to climb but not descend (with the exception of fill-cw */
317 #define TCP_RACK_ABC_VAL 1133 /* Set a local ABC value different then the system default */
318 #define TCP_REC_ABC_VAL 1134 /* Do we use the ABC value for recovery or the override one from sysctl */
319 #define TCP_RACK_MEASURE_CNT 1135 /* How many measurements are required in GP pacing */
320 #define TCP_DEFER_OPTIONS 1136 /* Defer options until the proper number of measurements occur, does not defer TCP_RACK_MEASURE_CNT */
321 /* unused; was TCP_FAST_RSM_HACK */
322 #define TCP_RACK_PACING_BETA 1138 /* Changing the beta for pacing */
323 #define TCP_RACK_PACING_BETA_ECN 1139 /* Changing the beta for ecn with pacing */
324 #define TCP_RACK_TIMER_SLOP 1140 /* Set or get the timer slop used */
325 #define TCP_RACK_DSACK_OPT 1141 /* How do we setup rack timer DSACK options bit 1/2 */
326 #define TCP_RACK_ENABLE_HYSTART 1142 /* Do we allow hystart in the CC modules */
327 #define TCP_RACK_SET_RXT_OPTIONS 1143 /* Set the bits in the retransmit options */
328 #define TCP_RACK_HI_BETA 1144 /* Turn on/off high beta */
329 #define TCP_RACK_SPLIT_LIMIT 1145 /* Set a split limit for split allocations */
330 #define TCP_RACK_PACING_DIVISOR 1146 /* Pacing divisor given to rate-limit code for burst sizing */
331 #define TCP_RACK_PACE_MIN_SEG 1147 /* Pacing min seg size rack will use */
332 #define TCP_RACK_DGP_IN_REC 1148 /* Do we use full DGP in recovery? */
333 /* unused; was TCP_POLICER_DETECT */
334 #define TCP_HYBRID_PACING 1150 /* Hybrid pacing enablement */
335 #define TCP_PACING_DND 1151 /* When pacing with rr_config=3 can sacks disturb us */
336 #define TCP_SS_EEXIT 1152 /* Do we do early exit from slowtart if no b/w growth */
337 #define TCP_DGP_UPPER_BOUNDS 1153 /* SS and CA upper bound in percentage */
338 #define TCP_NO_TIMELY 1154 /* Disable/enable Timely */
339 #define TCP_HONOR_HPTS_MIN 1155 /* Do we honor hpts min to */
340 #define TCP_REC_IS_DYN 1156 /* Do we allow timely to change recovery multiplier? */
341 #define TCP_SIDECHAN_DIS 1157 /* Disable/enable the side-channel */
342 #define TCP_FILLCW_RATE_CAP 1158 /* Set a cap for DGP's fillcw */
343 /* unused; was TCP_POLICER_MSS */
344 #define TCP_STACK_SPEC_INFO 1160 /* Get stack specific information (if present) */
345 #define RACK_CSPR_IS_FCC 1161
346 #define TCP_GP_USE_LTBW 1162 /* how we use lt_bw 0=not, 1=min, 2=max */
347
348
349 /* Start of reserved space for third-party user-settable options. */
350 #define TCP_VENDOR SO_VENDOR
351
352 #define TCP_CA_NAME_MAX 16 /* max congestion control name length */
353
354 #define TCPI_OPT_TIMESTAMPS 0x01
355 #define TCPI_OPT_SACK 0x02
356 #define TCPI_OPT_WSCALE 0x04
357 #define TCPI_OPT_ECN 0x08
358 #define TCPI_OPT_TOE 0x10
359 #define TCPI_OPT_TFO 0x20
360 #define TCPI_OPT_ACE 0x40
361
362 /* Maximum length of log ID. */
363 #define TCP_LOG_ID_LEN 64
364
365 /* TCP accounting counters */
366 #define TCP_NUM_PROC_COUNTERS 11
367 #define TCP_NUM_CNT_COUNTERS 13
368
369 /* Must match counter array sizes in tcpcb */
370 struct tcp_perf_info {
371 uint64_t tcp_cnt_counters[TCP_NUM_CNT_COUNTERS];
372 uint64_t tcp_proc_time[TCP_NUM_CNT_COUNTERS];
373 uint64_t timebase; /* timebase for tcp_proc_time */
374 uint8_t tb_is_stable; /* timebase is stable/invariant */
375 };
376
377 /*
378 * The TCP_INFO socket option comes from the Linux 2.6 TCP API, and permits
379 * the caller to query certain information about the state of a TCP
380 * connection. We provide an overlapping set of fields with the Linux
381 * implementation, but since this is a fixed size structure, room has been
382 * left for growth. In order to maximize potential future compatibility with
383 * the Linux API, the same variable names and order have been adopted, and
384 * padding left to make room for omitted fields in case they are added later.
385 *
386 * XXX: This is currently an unstable ABI/API, in that it is expected to
387 * change.
388 */
389 struct tcp_info {
390 u_int8_t tcpi_state; /* TCP FSM state. */
391 u_int8_t __tcpi_ca_state;
392 u_int8_t __tcpi_retransmits;
393 u_int8_t __tcpi_probes;
394 u_int8_t __tcpi_backoff;
395 u_int8_t tcpi_options; /* Options enabled on conn. */
396 u_int8_t tcpi_snd_wscale:4, /* RFC1323 send shift value. */
397 tcpi_rcv_wscale:4; /* RFC1323 recv shift value. */
398
399 u_int32_t tcpi_rto; /* Retransmission timeout (usec). */
400 u_int32_t __tcpi_ato;
401 u_int32_t tcpi_snd_mss; /* Max segment size for send. */
402 u_int32_t tcpi_rcv_mss; /* Max segment size for receive. */
403
404 u_int32_t __tcpi_unacked;
405 u_int32_t __tcpi_sacked;
406 u_int32_t __tcpi_lost;
407 u_int32_t __tcpi_retrans;
408 u_int32_t __tcpi_fackets;
409
410 /* Times; measurements in usecs. */
411 u_int32_t __tcpi_last_data_sent;
412 u_int32_t __tcpi_last_ack_sent; /* Also unimpl. on Linux? */
413 u_int32_t tcpi_last_data_recv; /* Time since last recv data. */
414 u_int32_t __tcpi_last_ack_recv;
415
416 /* Metrics; variable units. */
417 u_int32_t __tcpi_pmtu;
418 u_int32_t __tcpi_rcv_ssthresh;
419 u_int32_t tcpi_rtt; /* Smoothed RTT in usecs. */
420 u_int32_t tcpi_rttvar; /* RTT variance in usecs. */
421 u_int32_t tcpi_snd_ssthresh; /* Slow start threshold. */
422 u_int32_t tcpi_snd_cwnd; /* Send congestion window. */
423 u_int32_t __tcpi_advmss;
424 u_int32_t __tcpi_reordering;
425
426 u_int32_t __tcpi_rcv_rtt;
427 u_int32_t tcpi_rcv_space; /* Advertised recv window. */
428
429 /* FreeBSD extensions to tcp_info. */
430 u_int32_t tcpi_snd_wnd; /* Advertised send window. */
431 u_int32_t tcpi_snd_bwnd; /* No longer used. */
432 u_int32_t tcpi_snd_nxt; /* Next egress seqno */
433 u_int32_t tcpi_rcv_nxt; /* Next ingress seqno */
434 u_int32_t tcpi_toe_tid; /* HWTID for TOE endpoints */
435 u_int32_t tcpi_snd_rexmitpack; /* Retransmitted packets */
436 u_int32_t tcpi_rcv_ooopack; /* Out-of-order packets */
437 u_int32_t tcpi_snd_zerowin; /* Zero-sized windows sent */
438
439 /* Accurate ECN counters. */
440 u_int32_t tcpi_delivered_ce;
441 u_int32_t tcpi_received_ce; /* # of CE marks received */
442 u_int32_t __tcpi_delivered_e1_bytes;
443 u_int32_t __tcpi_delivered_e0_bytes;
444 u_int32_t __tcpi_delivered_ce_bytes;
445 u_int32_t __tcpi_received_e1_bytes;
446 u_int32_t __tcpi_received_e0_bytes;
447 u_int32_t __tcpi_received_ce_bytes;
448
449 u_int32_t tcpi_total_tlp; /* tail loss probes sent */
450 u_int64_t tcpi_total_tlp_bytes; /* tail loss probe bytes sent */
451
452 u_int32_t tcpi_snd_una; /* Unacked seqno sent */
453 u_int32_t tcpi_snd_max; /* Highest seqno sent */
454 u_int32_t tcpi_rcv_numsacks; /* Distinct SACK blks present */
455 u_int32_t tcpi_rcv_adv; /* Peer advertised window */
456 u_int32_t tcpi_dupacks; /* Consecutive dup ACKs recvd */
457
458 u_int32_t tcpi_rttmin; /* Min observed RTT */
459 /* Padding to grow without breaking ABI. */
460 u_int32_t __tcpi_pad[14]; /* Padding. */
461 };
462
463 /*
464 * If this structure is provided when setting the TCP_FASTOPEN socket
465 * option, and the enable member is non-zero, a subsequent connect will use
466 * pre-shared key (PSK) mode using the provided key.
467 */
468 struct tcp_fastopen {
469 int enable;
470 uint8_t psk[TCP_FASTOPEN_PSK_LEN];
471 };
472
473 #define TCP_FUNCTION_NAME_LEN_MAX 32
474
475 struct stack_specific_info {
476 char stack_name[TCP_FUNCTION_NAME_LEN_MAX];
477 uint64_t policer_last_bw; /* Only valid if detection enabled and policer detected */
478 uint64_t bytes_transmitted;
479 uint64_t bytes_retransmitted;
480 uint32_t policer_detection_enabled: 1,
481 policer_detected : 1, /* transport thinks a policer is on path */
482 highly_buffered : 1, /* transport considers the path highly buffered */
483 spare : 29;
484 uint32_t policer_bucket_size; /* Only valid if detection enabled and policer detected */
485 uint32_t current_round;
486 uint32_t _rack_i_pad[18];
487 };
488
489 struct tcp_function_set {
490 char function_set_name[TCP_FUNCTION_NAME_LEN_MAX];
491 uint32_t pcbcnt;
492 };
493
494 /* TLS modes for TCP_TXTLS_MODE */
495 #define TCP_TLS_MODE_NONE 0
496 #define TCP_TLS_MODE_SW 1
497 #define TCP_TLS_MODE_IFNET 2
498 #define TCP_TLS_MODE_TOE 3
499
500 /*
501 * TCP Control message types
502 */
503 #define TLS_SET_RECORD_TYPE 1
504 #define TLS_GET_RECORD 2
505
506 /*
507 * TCP log user opaque
508 */
509 struct tcp_snd_req {
510 uint64_t timestamp;
511 uint64_t start;
512 uint64_t end;
513 uint32_t flags;
514 uint32_t playout_ms;
515 };
516
517 union tcp_log_userdata {
518 struct tcp_snd_req tcp_req;
519 };
520
521 struct tcp_log_user {
522 uint32_t type;
523 uint32_t subtype;
524 union tcp_log_userdata data;
525 };
526
527 /* user types, i.e. apps */
528 #define TCP_LOG_USER_HTTPD 1
529
530 /* user subtypes */
531 #define TCP_LOG_HTTPD_TS 1 /* client timestamp */
532 #define TCP_LOG_HTTPD_TS_REQ 2 /* client timestamp and request info */
533
534 /* HTTPD REQ flags */
535 #define TCP_LOG_HTTPD_RANGE_START 0x0001
536 #define TCP_LOG_HTTPD_RANGE_END 0x0002
537
538 /* Flags for hybrid pacing */
539 #define TCP_HYBRID_PACING_CU 0x0001 /* Enable catch-up mode */
540 #define TCP_HYBRID_PACING_DTL 0x0002 /* Enable Detailed logging */
541 #define TCP_HYBRID_PACING_CSPR 0x0004 /* A client suggested rate is present */
542 #define TCP_HYBRID_PACING_H_MS 0x0008 /* A client hint for maxseg is present */
543 #define TCP_HYBRID_PACING_ENABLE 0x0010 /* We are enabling hybrid pacing else disable */
544 #define TCP_HYBRID_PACING_S_MSS 0x0020 /* Clent wants us to set the mss overriding gp est in CU */
545 #define TCP_HAS_PLAYOUT_MS 0x0040 /* The client included the chunk playout milliseconds: deprecate */
546 /* the below are internal only flags */
547 #define TCP_HYBRID_PACING_USER_MASK 0x0FFF /* Non-internal flags mask */
548 #define TCP_HYBRID_PACING_SETMSS 0x1000 /* Internal flag that tells us we set the mss on this entry */
549 #define TCP_HYBRID_PACING_WASSET 0x2000 /* We init to this to know if a hybrid command was issued */
550 #define TCP_HYBRID_PACING_SENDTIME 0x4000 /* Duplicate tm to last, use sendtime for catch up mode */
551
552 struct tcp_hybrid_req {
553 struct tcp_snd_req req;
554 uint64_t cspr;
555 uint32_t hint_maxseg;
556 uint32_t hybrid_flags;
557 };
558
559 /*
560 * TCP specific variables of interest for tp->t_stats stats(9) accounting.
561 */
562 #define VOI_TCP_TXPB 0 /* Transmit payload bytes */
563 #define VOI_TCP_RETXPB 1 /* Retransmit payload bytes */
564 #define VOI_TCP_FRWIN 2 /* Foreign receive window */
565 #define VOI_TCP_LCWIN 3 /* Local congesiton window */
566 #define VOI_TCP_RTT 4 /* Round trip time */
567 #define VOI_TCP_CSIG 5 /* Congestion signal */
568 #define VOI_TCP_GPUT 6 /* Goodput */
569 #define VOI_TCP_CALCFRWINDIFF 7 /* Congestion avoidance LCWIN - FRWIN */
570 #define VOI_TCP_GPUT_ND 8 /* Goodput normalised delta */
571 #define VOI_TCP_ACKLEN 9 /* Average ACKed bytes per ACK */
572 #define VOI_TCP_PATHRTT 10 /* The path RTT based on ACK arrival */
573
574 #define TCP_REUSPORT_LB_NUMA_NODOM (-2) /* remove numa binding */
575 #define TCP_REUSPORT_LB_NUMA_CURDOM (-1) /* bind to current domain */
576
577 #endif /* __BSD_VISIBLE */
578 #endif /* !_NETINET_TCP_H_ */
579