xref: /freebsd/sys/netinet/tcp.h (revision 732a02b4e77866604a120a275c082bb6221bd2ff)
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  *	@(#)tcp.h	8.1 (Berkeley) 6/10/93
32  * $FreeBSD$
33  */
34 
35 #ifndef _NETINET_TCP_H_
36 #define _NETINET_TCP_H_
37 
38 #include <sys/cdefs.h>
39 #include <sys/types.h>
40 
41 #if __BSD_VISIBLE
42 
43 typedef	u_int32_t tcp_seq;
44 
45 #define tcp6_seq	tcp_seq	/* for KAME src sync over BSD*'s */
46 #define tcp6hdr		tcphdr	/* for KAME src sync over BSD*'s */
47 
48 /*
49  * TCP header.
50  * Per RFC 793, September, 1981.
51  */
52 struct tcphdr {
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 */
60 #endif
61 #if BYTE_ORDER == BIG_ENDIAN
62 	u_char	th_off:4,		/* data offset */
63 		th_x2:4;		/* (unused) */
64 #endif
65 	u_char	th_flags;
66 #define	TH_FIN	0x01
67 #define	TH_SYN	0x02
68 #define	TH_RST	0x04
69 #define	TH_PUSH	0x08
70 #define	TH_ACK	0x10
71 #define	TH_URG	0x20
72 #define	TH_ECE	0x40
73 #define	TH_CWR	0x80
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"
77 
78 	u_short	th_win;			/* window */
79 	u_short	th_sum;			/* checksum */
80 	u_short	th_urp;			/* urgent pointer */
81 };
82 
83 #define	TCPOPT_EOL		0
84 #define	   TCPOLEN_EOL			1
85 #define	TCPOPT_PAD		0		/* padding after EOL */
86 #define	   TCPOLEN_PAD			1
87 #define	TCPOPT_NOP		1
88 #define	   TCPOLEN_NOP			1
89 #define	TCPOPT_MAXSEG		2
90 #define    TCPOLEN_MAXSEG		4
91 #define TCPOPT_WINDOW		3
92 #define    TCPOLEN_WINDOW		3
93 #define TCPOPT_SACK_PERMITTED	4
94 #define    TCPOLEN_SACK_PERMITTED	2
95 #define TCPOPT_SACK		5
96 #define	   TCPOLEN_SACKHDR		2
97 #define    TCPOLEN_SACK			8	/* 2*sizeof(tcp_seq) */
98 #define TCPOPT_TIMESTAMP	8
99 #define    TCPOLEN_TIMESTAMP		10
100 #define    TCPOLEN_TSTAMP_APPA		(TCPOLEN_TIMESTAMP+2) /* appendix A */
101 #define	TCPOPT_SIGNATURE	19		/* Keyed MD5: RFC 2385 */
102 #define	   TCPOLEN_SIGNATURE		18
103 #define	TCPOPT_FAST_OPEN	34
104 #define	   TCPOLEN_FAST_OPEN_EMPTY	2
105 
106 /* Miscellaneous constants */
107 #define	MAX_SACK_BLKS	6	/* Max # SACK blocks stored at receiver side */
108 #define	TCP_MAX_SACK	4	/* MAX # SACKs sent in any segment */
109 
110 
111 /*
112  * The default maximum segment size (MSS) to be used for new TCP connections
113  * when path MTU discovery is not enabled.
114  *
115  * RFC879 derives the default MSS from the largest datagram size hosts are
116  * minimally required to handle directly or through IP reassembly minus the
117  * size of the IP and TCP header.  With IPv6 the minimum MTU is specified
118  * in RFC2460.
119  *
120  * For IPv4 the MSS is 576 - sizeof(struct tcpiphdr)
121  * For IPv6 the MSS is IPV6_MMTU - sizeof(struct ip6_hdr) - sizeof(struct tcphdr)
122  *
123  * We use explicit numerical definition here to avoid header pollution.
124  */
125 #define	TCP_MSS		536
126 #define	TCP6_MSS	1220
127 
128 /*
129  * Limit the lowest MSS we accept for path MTU discovery and the TCP SYN MSS
130  * option.  Allowing low values of MSS can consume significant resources and
131  * be used to mount a resource exhaustion attack.
132  * Connections requesting lower MSS values will be rounded up to this value
133  * and the IP_DF flag will be cleared to allow fragmentation along the path.
134  *
135  * See tcp_subr.c tcp_minmss SYSCTL declaration for more comments.  Setting
136  * it to "0" disables the minmss check.
137  *
138  * The default value is fine for TCP across the Internet's smallest official
139  * link MTU (256 bytes for AX.25 packet radio).  However, a connection is very
140  * unlikely to come across such low MTU interfaces these days (anno domini 2003).
141  */
142 #define	TCP_MINMSS 216
143 
144 #define	TCP_MAXWIN	65535	/* largest value for (unscaled) window */
145 #define	TTCP_CLIENT_SND_WND	4096	/* dflt send window for T/TCP client */
146 
147 #define TCP_MAX_WINSHIFT	14	/* maximum window shift */
148 
149 #define TCP_MAXBURST		4	/* maximum segments in a burst */
150 
151 #define TCP_MAXHLEN	(0xf<<2)	/* max length of header in bytes */
152 #define TCP_MAXOLEN	(TCP_MAXHLEN - sizeof(struct tcphdr))
153 					/* max space left for options */
154 
155 #define TCP_FASTOPEN_MIN_COOKIE_LEN	4	/* Per RFC7413 */
156 #define TCP_FASTOPEN_MAX_COOKIE_LEN	16	/* Per RFC7413 */
157 #define TCP_FASTOPEN_PSK_LEN		16	/* Same as TCP_FASTOPEN_KEY_LEN */
158 #endif /* __BSD_VISIBLE */
159 
160 /*
161  * User-settable options (used with setsockopt).  These are discrete
162  * values and are not masked together.  Some values appear to be
163  * bitmasks for historical reasons.
164  */
165 #define	TCP_NODELAY	1	/* don't delay send to coalesce packets */
166 #if __BSD_VISIBLE
167 #define	TCP_MAXSEG	2	/* set maximum segment size */
168 #define TCP_NOPUSH	4	/* don't push last block of write */
169 #define TCP_NOOPT	8	/* don't use TCP options */
170 #define TCP_MD5SIG	16	/* use MD5 digests (RFC2385) */
171 #define	TCP_INFO	32	/* retrieve tcp_info structure */
172 #define	TCP_STATS	33	/* retrieve stats blob structure */
173 #define	TCP_LOG		34	/* configure event logging for connection */
174 #define	TCP_LOGBUF	35	/* retrieve event log for connection */
175 #define	TCP_LOGID	36	/* configure log ID to correlate connections */
176 #define	TCP_LOGDUMP	37	/* dump connection log events to device */
177 #define	TCP_LOGDUMPID	38	/* dump events from connections with same ID to
178 				   device */
179 #define	TCP_TXTLS_ENABLE 39	/* TLS framing and encryption for transmit */
180 #define	TCP_TXTLS_MODE	40	/* Transmit TLS mode */
181 #define	TCP_CONGESTION	64	/* get/set congestion control algorithm */
182 #define	TCP_CCALGOOPT	65	/* get/set cc algorithm specific options */
183 #define TCP_DELACK  	72	/* socket option for delayed ack */
184 #define	TCP_KEEPINIT	128	/* N, time to establish connection */
185 #define	TCP_KEEPIDLE	256	/* L,N,X start keeplives after this period */
186 #define	TCP_KEEPINTVL	512	/* L,N interval between keepalives */
187 #define	TCP_KEEPCNT	1024	/* L,N number of keepalives before close */
188 #define	TCP_FASTOPEN	1025	/* enable TFO / was created via TFO */
189 #define	TCP_PCAP_OUT	2048	/* number of output packets to keep */
190 #define	TCP_PCAP_IN	4096	/* number of input packets to keep */
191 #define TCP_FUNCTION_BLK 8192	/* Set the tcp function pointers to the specified stack */
192 /* Options for Rack and BBR */
193 #define TCP_RACK_PROP	      1051 /* RACK proportional rate reduction (bool) */
194 #define TCP_RACK_TLP_REDUCE   1052 /* RACK TLP cwnd reduction (bool) */
195 #define TCP_RACK_PACE_REDUCE  1053 /* RACK Pacing reduction factor (divisor) */
196 #define TCP_RACK_PACE_MAX_SEG 1054 /* Max segments in a pace */
197 #define TCP_RACK_PACE_ALWAYS  1055 /* Use the always pace method */
198 #define TCP_RACK_PROP_RATE    1056 /* The proportional reduction rate */
199 #define TCP_RACK_PRR_SENDALOT 1057 /* Allow PRR to send more than one seg */
200 #define TCP_RACK_MIN_TO       1058 /* Minimum time between rack t-o's in ms */
201 #define TCP_RACK_EARLY_RECOV  1059 /* Should recovery happen early (bool) */
202 #define TCP_RACK_EARLY_SEG    1060 /* If early recovery max segments */
203 #define TCP_RACK_REORD_THRESH 1061 /* RACK reorder threshold (shift amount) */
204 #define TCP_RACK_REORD_FADE   1062 /* Does reordering fade after ms time */
205 #define TCP_RACK_TLP_THRESH   1063 /* RACK TLP theshold i.e. srtt+(srtt/N) */
206 #define TCP_RACK_PKT_DELAY    1064 /* RACK added ms i.e. rack-rtt + reord + N */
207 #define TCP_RACK_TLP_INC_VAR  1065 /* Does TLP include rtt variance in t-o */
208 #define TCP_BBR_IWINTSO	      1067 /* Initial TSO window for BBRs first sends */
209 #define TCP_BBR_RECFORCE      1068 /* Enter recovery force out a segment disregard pacer no longer valid */
210 #define TCP_BBR_STARTUP_PG    1069 /* Startup pacing gain */
211 #define TCP_BBR_DRAIN_PG      1070 /* Drain pacing gain */
212 #define TCP_BBR_RWND_IS_APP   1071 /* Rwnd limited is considered app limited */
213 #define TCP_BBR_PROBE_RTT_INT 1072 /* How long in useconds between probe-rtt */
214 #define TCP_BBR_ONE_RETRAN    1073 /* Is only one segment allowed out during retran */
215 #define TCP_BBR_STARTUP_LOSS_EXIT 1074	/* Do we exit a loss during startup if not 20% incr */
216 #define TCP_BBR_USE_LOWGAIN   1075 /* lower the gain in PROBE_BW enable */
217 #define TCP_BBR_LOWGAIN_THRESH 1076 /* Unused after 2.3 morphs to TSLIMITS >= 2.3 */
218 #define TCP_BBR_TSLIMITS 1076	   /* Do we use experimental Timestamp limiting for our algo */
219 #define TCP_BBR_LOWGAIN_HALF  1077 /* Unused after 2.3 */
220 #define TCP_BBR_PACE_OH        1077 /* Reused in 4.2 for pacing overhead setting */
221 #define TCP_BBR_LOWGAIN_FD    1078 /* Unused after 2.3 */
222 #define TCP_BBR_HOLD_TARGET 1078	/* For 4.3 on */
223 #define TCP_BBR_USEDEL_RATE   1079 /* Enable use of delivery rate for loss recovery */
224 #define TCP_BBR_MIN_RTO       1080 /* Min RTO in milliseconds */
225 #define TCP_BBR_MAX_RTO	      1081 /* Max RTO in milliseconds */
226 #define TCP_BBR_REC_OVER_HPTS 1082 /* Recovery override htps settings 0/1/3 */
227 #define TCP_BBR_UNLIMITED     1083 /* Not used before 2.3 and morphs to algorithm >= 2.3 */
228 #define TCP_BBR_ALGORITHM     1083 /* What measurement algo does BBR use netflix=0, google=1 */
229 #define TCP_BBR_DRAIN_INC_EXTRA 1084 /* Does the 3/4 drain target include the extra gain */
230 #define TCP_BBR_STARTUP_EXIT_EPOCH 1085 /* what epoch gets us out of startup */
231 #define TCP_BBR_PACE_PER_SEC   1086
232 #define TCP_BBR_PACE_DEL_TAR   1087
233 #define TCP_BBR_PACE_SEG_MAX   1088
234 #define TCP_BBR_PACE_SEG_MIN   1089
235 #define TCP_BBR_PACE_CROSS     1090
236 #define TCP_RACK_IDLE_REDUCE_HIGH 1092  /* Reduce the highest cwnd seen to IW on idle */
237 #define TCP_RACK_MIN_PACE      1093 	/* Do we enforce rack min pace time */
238 #define TCP_RACK_MIN_PACE_SEG  1094	/* If so what is the seg threshould */
239 #define TCP_RACK_GP_INCREASE   1094	/* After 4.1 its the GP increase */
240 #define TCP_RACK_TLP_USE       1095
241 #define TCP_BBR_ACK_COMP_ALG   1096 	/* Not used */
242 #define TCP_BBR_TMR_PACE_OH    1096	/* Recycled in 4.2 */
243 #define TCP_BBR_EXTRA_GAIN     1097
244 #define TCP_RACK_DO_DETECTION  1097	/* Recycle of extra gain for rack, attack detection */
245 #define TCP_BBR_RACK_RTT_USE   1098	/* what RTT should we use 0, 1, or 2? */
246 #define TCP_BBR_RETRAN_WTSO    1099
247 #define TCP_DATA_AFTER_CLOSE   1100
248 #define TCP_BBR_PROBE_RTT_GAIN 1101
249 #define TCP_BBR_PROBE_RTT_LEN  1102
250 #define TCP_BBR_SEND_IWND_IN_TSO 1103	/* Do we burst out whole iwin size chunks at start? */
251 #define TCP_BBR_USE_RACK_CHEAT 1104	/* Do we use the rack cheat for pacing rxt's */
252 #define TCP_BBR_HDWR_PACE      1105	/* Enable/disable hardware pacing */
253 #define TCP_BBR_UTTER_MAX_TSO  1106	/* Do we enforce an utter max TSO size */
254 #define TCP_BBR_EXTRA_STATE    1107	/* Special exit-persist catch up */
255 #define TCP_BBR_FLOOR_MIN_TSO  1108     /* The min tso size */
256 #define TCP_BBR_MIN_TOPACEOUT  1109	/* Do we suspend pacing until */
257 #define TCP_BBR_TSTMP_RAISES   1110	/* Can a timestamp measurement raise the b/w */
258 #define TCP_BBR_POLICER_DETECT 1111	/* Turn on/off google mode policer detection */
259 
260 
261 /* Start of reserved space for third-party user-settable options. */
262 #define	TCP_VENDOR	SO_VENDOR
263 
264 #define	TCP_CA_NAME_MAX	16	/* max congestion control name length */
265 
266 #define	TCPI_OPT_TIMESTAMPS	0x01
267 #define	TCPI_OPT_SACK		0x02
268 #define	TCPI_OPT_WSCALE		0x04
269 #define	TCPI_OPT_ECN		0x08
270 #define	TCPI_OPT_TOE		0x10
271 
272 /* Maximum length of log ID. */
273 #define TCP_LOG_ID_LEN	64
274 
275 /*
276  * The TCP_INFO socket option comes from the Linux 2.6 TCP API, and permits
277  * the caller to query certain information about the state of a TCP
278  * connection.  We provide an overlapping set of fields with the Linux
279  * implementation, but since this is a fixed size structure, room has been
280  * left for growth.  In order to maximize potential future compatibility with
281  * the Linux API, the same variable names and order have been adopted, and
282  * padding left to make room for omitted fields in case they are added later.
283  *
284  * XXX: This is currently an unstable ABI/API, in that it is expected to
285  * change.
286  */
287 struct tcp_info {
288 	u_int8_t	tcpi_state;		/* TCP FSM state. */
289 	u_int8_t	__tcpi_ca_state;
290 	u_int8_t	__tcpi_retransmits;
291 	u_int8_t	__tcpi_probes;
292 	u_int8_t	__tcpi_backoff;
293 	u_int8_t	tcpi_options;		/* Options enabled on conn. */
294 	u_int8_t	tcpi_snd_wscale:4,	/* RFC1323 send shift value. */
295 			tcpi_rcv_wscale:4;	/* RFC1323 recv shift value. */
296 
297 	u_int32_t	tcpi_rto;		/* Retransmission timeout (usec). */
298 	u_int32_t	__tcpi_ato;
299 	u_int32_t	tcpi_snd_mss;		/* Max segment size for send. */
300 	u_int32_t	tcpi_rcv_mss;		/* Max segment size for receive. */
301 
302 	u_int32_t	__tcpi_unacked;
303 	u_int32_t	__tcpi_sacked;
304 	u_int32_t	__tcpi_lost;
305 	u_int32_t	__tcpi_retrans;
306 	u_int32_t	__tcpi_fackets;
307 
308 	/* Times; measurements in usecs. */
309 	u_int32_t	__tcpi_last_data_sent;
310 	u_int32_t	__tcpi_last_ack_sent;	/* Also unimpl. on Linux? */
311 	u_int32_t	tcpi_last_data_recv;	/* Time since last recv data. */
312 	u_int32_t	__tcpi_last_ack_recv;
313 
314 	/* Metrics; variable units. */
315 	u_int32_t	__tcpi_pmtu;
316 	u_int32_t	__tcpi_rcv_ssthresh;
317 	u_int32_t	tcpi_rtt;		/* Smoothed RTT in usecs. */
318 	u_int32_t	tcpi_rttvar;		/* RTT variance in usecs. */
319 	u_int32_t	tcpi_snd_ssthresh;	/* Slow start threshold. */
320 	u_int32_t	tcpi_snd_cwnd;		/* Send congestion window. */
321 	u_int32_t	__tcpi_advmss;
322 	u_int32_t	__tcpi_reordering;
323 
324 	u_int32_t	__tcpi_rcv_rtt;
325 	u_int32_t	tcpi_rcv_space;		/* Advertised recv window. */
326 
327 	/* FreeBSD extensions to tcp_info. */
328 	u_int32_t	tcpi_snd_wnd;		/* Advertised send window. */
329 	u_int32_t	tcpi_snd_bwnd;		/* No longer used. */
330 	u_int32_t	tcpi_snd_nxt;		/* Next egress seqno */
331 	u_int32_t	tcpi_rcv_nxt;		/* Next ingress seqno */
332 	u_int32_t	tcpi_toe_tid;		/* HWTID for TOE endpoints */
333 	u_int32_t	tcpi_snd_rexmitpack;	/* Retransmitted packets */
334 	u_int32_t	tcpi_rcv_ooopack;	/* Out-of-order packets */
335 	u_int32_t	tcpi_snd_zerowin;	/* Zero-sized windows sent */
336 
337 	/* Padding to grow without breaking ABI. */
338 	u_int32_t	__tcpi_pad[26];		/* Padding. */
339 };
340 
341 /*
342  * If this structure is provided when setting the TCP_FASTOPEN socket
343  * option, and the enable member is non-zero, a subsequent connect will use
344  * pre-shared key (PSK) mode using the provided key.
345  */
346 struct tcp_fastopen {
347 	int enable;
348 	uint8_t psk[TCP_FASTOPEN_PSK_LEN];
349 };
350 #endif
351 #define TCP_FUNCTION_NAME_LEN_MAX 32
352 
353 struct tcp_function_set {
354 	char function_set_name[TCP_FUNCTION_NAME_LEN_MAX];
355 	uint32_t pcbcnt;
356 };
357 
358 /* TLS modes for TCP_TXTLS_MODE */
359 #define	TCP_TLS_MODE_NONE	0
360 #define	TCP_TLS_MODE_SW		1
361 #define	TCP_TLS_MODE_IFNET	2
362 #define	TCP_TLS_MODE_TOE	3
363 
364 /*
365  * TCP Control message types
366  */
367 #define	TLS_SET_RECORD_TYPE	1
368 
369 /*
370  * TCP specific variables of interest for tp->t_stats stats(9) accounting.
371  */
372 #define	VOI_TCP_TXPB		0 /* Transmit payload bytes */
373 #define	VOI_TCP_RETXPB		1 /* Retransmit payload bytes */
374 #define	VOI_TCP_FRWIN		2 /* Foreign receive window */
375 #define	VOI_TCP_LCWIN		3 /* Local congesiton window */
376 #define	VOI_TCP_RTT		4 /* Round trip time */
377 #define	VOI_TCP_CSIG		5 /* Congestion signal */
378 #define	VOI_TCP_GPUT		6 /* Goodput */
379 #define	VOI_TCP_CALCFRWINDIFF	7 /* Congestion avoidance LCWIN - FRWIN */
380 #define	VOI_TCP_GPUT_ND		8 /* Goodput normalised delta */
381 #define	VOI_TCP_ACKLEN		9 /* Average ACKed bytes per ACK */
382 
383 #endif /* !_NETINET_TCP_H_ */
384