• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /* SPDX-License-Identifier: GPL-2.0+ WITH Linux-syscall-note */
2 /*
3  * INET		An implementation of the TCP/IP protocol suite for the LINUX
4  *		operating system.  INET is implemented using the  BSD Socket
5  *		interface as the means of communication with the user level.
6  *
7  *		Definitions for the TCP protocol.
8  *
9  * Version:	@(#)tcp.h	1.0.2	04/28/93
10  *
11  * Author:	Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12  *
13  *		This program is free software; you can redistribute it and/or
14  *		modify it under the terms of the GNU General Public License
15  *		as published by the Free Software Foundation; either version
16  *		2 of the License, or (at your option) any later version.
17  */
18 #ifndef _UAPI_LINUX_TCP_H
19 #define _UAPI_LINUX_TCP_H
20 
21 #include <linux/types.h>
22 #include <asm/byteorder.h>
23 #include <linux/socket.h>
24 
25 struct tcphdr {
26 	__be16	source;
27 	__be16	dest;
28 	__be32	seq;
29 	__be32	ack_seq;
30 #if defined(__LITTLE_ENDIAN_BITFIELD)
31 	__u16	res1:4,
32 		doff:4,
33 		fin:1,
34 		syn:1,
35 		rst:1,
36 		psh:1,
37 		ack:1,
38 		urg:1,
39 		ece:1,
40 		cwr:1;
41 #elif defined(__BIG_ENDIAN_BITFIELD)
42 	__u16	doff:4,
43 		res1:4,
44 		cwr:1,
45 		ece:1,
46 		urg:1,
47 		ack:1,
48 		psh:1,
49 		rst:1,
50 		syn:1,
51 		fin:1;
52 #else
53 #error	"Adjust your <asm/byteorder.h> defines"
54 #endif
55 	__be16	window;
56 	__sum16	check;
57 	__be16	urg_ptr;
58 };
59 
60 /*
61  *	The union cast uses a gcc extension to avoid aliasing problems
62  *  (union is compatible to any of its members)
63  *  This means this part of the code is -fstrict-aliasing safe now.
64  */
65 union tcp_word_hdr {
66 	struct tcphdr hdr;
67 	__be32 		  words[5];
68 };
69 
70 #define tcp_flag_word(tp) ( ((union tcp_word_hdr *)(tp))->words [3])
71 
72 enum {
73 	TCP_FLAG_CWR = __constant_cpu_to_be32(0x00800000),
74 	TCP_FLAG_ECE = __constant_cpu_to_be32(0x00400000),
75 	TCP_FLAG_URG = __constant_cpu_to_be32(0x00200000),
76 	TCP_FLAG_ACK = __constant_cpu_to_be32(0x00100000),
77 	TCP_FLAG_PSH = __constant_cpu_to_be32(0x00080000),
78 	TCP_FLAG_RST = __constant_cpu_to_be32(0x00040000),
79 	TCP_FLAG_SYN = __constant_cpu_to_be32(0x00020000),
80 	TCP_FLAG_FIN = __constant_cpu_to_be32(0x00010000),
81 	TCP_RESERVED_BITS = __constant_cpu_to_be32(0x0F000000),
82 	TCP_DATA_OFFSET = __constant_cpu_to_be32(0xF0000000)
83 };
84 
85 /*
86  * TCP general constants
87  */
88 #define TCP_MSS_DEFAULT		 536U	/* IPv4 (RFC1122, RFC2581) */
89 #define TCP_MSS_DESIRED		1220U	/* IPv6 (tunneled), EDNS0 (RFC3226) */
90 
91 /* TCP socket options */
92 #define TCP_NODELAY		1	/* Turn off Nagle's algorithm. */
93 #define TCP_MAXSEG		2	/* Limit MSS */
94 #define TCP_CORK		3	/* Never send partially complete segments */
95 #define TCP_KEEPIDLE		4	/* Start keeplives after this period */
96 #define TCP_KEEPINTVL		5	/* Interval between keepalives */
97 #define TCP_KEEPCNT		6	/* Number of keepalives before death */
98 #define TCP_SYNCNT		7	/* Number of SYN retransmits */
99 #define TCP_LINGER2		8	/* Life time of orphaned FIN-WAIT-2 state */
100 #define TCP_DEFER_ACCEPT	9	/* Wake up listener only when data arrive */
101 #define TCP_WINDOW_CLAMP	10	/* Bound advertised window */
102 #define TCP_INFO		11	/* Information about this connection. */
103 #define TCP_QUICKACK		12	/* Block/reenable quick acks */
104 #define TCP_CONGESTION		13	/* Congestion control algorithm */
105 #define TCP_MD5SIG		14	/* TCP MD5 Signature (RFC2385) */
106 #define TCP_THIN_LINEAR_TIMEOUTS 16      /* Use linear timeouts for thin streams*/
107 #define TCP_THIN_DUPACK         17      /* Fast retrans. after 1 dupack */
108 #define TCP_USER_TIMEOUT	18	/* How long for loss retry before timeout */
109 #define TCP_REPAIR		19	/* TCP sock is under repair right now */
110 #define TCP_REPAIR_QUEUE	20
111 #define TCP_QUEUE_SEQ		21
112 #define TCP_REPAIR_OPTIONS	22
113 #define TCP_FASTOPEN		23	/* Enable FastOpen on listeners */
114 #define TCP_TIMESTAMP		24
115 #define TCP_NOTSENT_LOWAT	25	/* limit number of unsent bytes in write queue */
116 #define TCP_CC_INFO		26	/* Get Congestion Control (optional) info */
117 #define TCP_SAVE_SYN		27	/* Record SYN headers for new connections */
118 #define TCP_SAVED_SYN		28	/* Get SYN headers recorded for connection */
119 #define TCP_REPAIR_WINDOW	29	/* Get/set window parameters */
120 #define TCP_FASTOPEN_CONNECT	30	/* Attempt FastOpen with connect */
121 #define TCP_ULP			31	/* Attach a ULP to a TCP connection */
122 #define TCP_MD5SIG_EXT		32	/* TCP MD5 Signature with extensions */
123 #define TCP_FASTOPEN_KEY	33	/* Set the key for Fast Open (cookie) */
124 #define TCP_FASTOPEN_NO_COOKIE	34	/* Enable TFO without a TFO cookie */
125 #define TCP_ZEROCOPY_RECEIVE	35
126 #define TCP_INQ			36	/* Notify bytes available to read as a cmsg on read */
127 
128 #define TCP_CM_INQ		TCP_INQ
129 
130 #define TCP_TX_DELAY		37	/* delay outgoing packets by XX usec */
131 
132 #define TCP_REPAIR_ON		1
133 #define TCP_REPAIR_OFF		0
134 #define TCP_REPAIR_OFF_NO_WP	-1	/* Turn off without window probes */
135 
136 #define TCP_NATA_URC		101	/* Nata ultar-reliable connection */
137 struct tcp_nata_urc {
138 	__u8	nata_urc_enabled;
139 	__u8	nata_syn_retries;
140 	__u16	nata_data_retries;
141 	__u32	nata_rto_ms;
142 };
143 
144 #define TCP_NATA_STL		102	/* Nata satellite connection */
145 struct tcp_nata_stl {
146 	__u8	nata_stl_enabled;
147 	__u8	nata_syn_retries;
148 	__u16	nata_data_retries;
149 	__u32	nata_syn_rto_ms;
150 	__u32	nata_data_rto_ms;
151 };
152 
153 struct tcp_repair_opt {
154 	__u32	opt_code;
155 	__u32	opt_val;
156 };
157 
158 struct tcp_repair_window {
159 	__u32	snd_wl1;
160 	__u32	snd_wnd;
161 	__u32	max_window;
162 
163 	__u32	rcv_wnd;
164 	__u32	rcv_wup;
165 };
166 
167 enum {
168 	TCP_NO_QUEUE,
169 	TCP_RECV_QUEUE,
170 	TCP_SEND_QUEUE,
171 	TCP_QUEUES_NR,
172 };
173 
174 /* why fastopen failed from client perspective */
175 enum tcp_fastopen_client_fail {
176 	TFO_STATUS_UNSPEC, /* catch-all */
177 	TFO_COOKIE_UNAVAILABLE, /* if not in TFO_CLIENT_NO_COOKIE mode */
178 	TFO_DATA_NOT_ACKED, /* SYN-ACK did not ack SYN data */
179 	TFO_SYN_RETRANSMITTED, /* SYN-ACK did not ack SYN data after timeout */
180 };
181 
182 /* for TCP_INFO socket option */
183 #define TCPI_OPT_TIMESTAMPS	1
184 #define TCPI_OPT_SACK		2
185 #define TCPI_OPT_WSCALE		4
186 #define TCPI_OPT_ECN		8 /* ECN was negociated at TCP session init */
187 #define TCPI_OPT_ECN_SEEN	16 /* we received at least one packet with ECT */
188 #define TCPI_OPT_SYN_DATA	32 /* SYN-ACK acked data in SYN sent or rcvd */
189 
190 /*
191  * Sender's congestion state indicating normal or abnormal situations
192  * in the last round of packets sent. The state is driven by the ACK
193  * information and timer events.
194  */
195 enum tcp_ca_state {
196 	/*
197 	 * Nothing bad has been observed recently.
198 	 * No apparent reordering, packet loss, or ECN marks.
199 	 */
200 	TCP_CA_Open = 0,
201 #define TCPF_CA_Open	(1<<TCP_CA_Open)
202 	/*
203 	 * The sender enters disordered state when it has received DUPACKs or
204 	 * SACKs in the last round of packets sent. This could be due to packet
205 	 * loss or reordering but needs further information to confirm packets
206 	 * have been lost.
207 	 */
208 	TCP_CA_Disorder = 1,
209 #define TCPF_CA_Disorder (1<<TCP_CA_Disorder)
210 	/*
211 	 * The sender enters Congestion Window Reduction (CWR) state when it
212 	 * has received ACKs with ECN-ECE marks, or has experienced congestion
213 	 * or packet discard on the sender host (e.g. qdisc).
214 	 */
215 	TCP_CA_CWR = 2,
216 #define TCPF_CA_CWR	(1<<TCP_CA_CWR)
217 	/*
218 	 * The sender is in fast recovery and retransmitting lost packets,
219 	 * typically triggered by ACK events.
220 	 */
221 	TCP_CA_Recovery = 3,
222 #define TCPF_CA_Recovery (1<<TCP_CA_Recovery)
223 	/*
224 	 * The sender is in loss recovery triggered by retransmission timeout.
225 	 */
226 	TCP_CA_Loss = 4
227 #define TCPF_CA_Loss	(1<<TCP_CA_Loss)
228 };
229 
230 struct tcp_info {
231 	__u8	tcpi_state;
232 	__u8	tcpi_ca_state;
233 	__u8	tcpi_retransmits;
234 	__u8	tcpi_probes;
235 	__u8	tcpi_backoff;
236 	__u8	tcpi_options;
237 	__u8	tcpi_snd_wscale : 4, tcpi_rcv_wscale : 4;
238 	__u8	tcpi_delivery_rate_app_limited:1, tcpi_fastopen_client_fail:2;
239 
240 	__u32	tcpi_rto;
241 	__u32	tcpi_ato;
242 	__u32	tcpi_snd_mss;
243 	__u32	tcpi_rcv_mss;
244 
245 	__u32	tcpi_unacked;
246 	__u32	tcpi_sacked;
247 	__u32	tcpi_lost;
248 	__u32	tcpi_retrans;
249 	__u32	tcpi_fackets;
250 
251 	/* Times. */
252 	__u32	tcpi_last_data_sent;
253 	__u32	tcpi_last_ack_sent;     /* Not remembered, sorry. */
254 	__u32	tcpi_last_data_recv;
255 	__u32	tcpi_last_ack_recv;
256 
257 	/* Metrics. */
258 	__u32	tcpi_pmtu;
259 	__u32	tcpi_rcv_ssthresh;
260 	__u32	tcpi_rtt;
261 	__u32	tcpi_rttvar;
262 	__u32	tcpi_snd_ssthresh;
263 	__u32	tcpi_snd_cwnd;
264 	__u32	tcpi_advmss;
265 	__u32	tcpi_reordering;
266 
267 	__u32	tcpi_rcv_rtt;
268 	__u32	tcpi_rcv_space;
269 
270 	__u32	tcpi_total_retrans;
271 
272 	__u64	tcpi_pacing_rate;
273 	__u64	tcpi_max_pacing_rate;
274 	__u64	tcpi_bytes_acked;    /* RFC4898 tcpEStatsAppHCThruOctetsAcked */
275 	__u64	tcpi_bytes_received; /* RFC4898 tcpEStatsAppHCThruOctetsReceived */
276 	__u32	tcpi_segs_out;	     /* RFC4898 tcpEStatsPerfSegsOut */
277 	__u32	tcpi_segs_in;	     /* RFC4898 tcpEStatsPerfSegsIn */
278 
279 	__u32	tcpi_notsent_bytes;
280 	__u32	tcpi_min_rtt;
281 	__u32	tcpi_data_segs_in;	/* RFC4898 tcpEStatsDataSegsIn */
282 	__u32	tcpi_data_segs_out;	/* RFC4898 tcpEStatsDataSegsOut */
283 
284 	__u64   tcpi_delivery_rate;
285 
286 	__u64	tcpi_busy_time;      /* Time (usec) busy sending data */
287 	__u64	tcpi_rwnd_limited;   /* Time (usec) limited by receive window */
288 	__u64	tcpi_sndbuf_limited; /* Time (usec) limited by send buffer */
289 
290 	__u32	tcpi_delivered;
291 	__u32	tcpi_delivered_ce;
292 
293 	__u64	tcpi_bytes_sent;     /* RFC4898 tcpEStatsPerfHCDataOctetsOut */
294 	__u64	tcpi_bytes_retrans;  /* RFC4898 tcpEStatsPerfOctetsRetrans */
295 	__u32	tcpi_dsack_dups;     /* RFC4898 tcpEStatsStackDSACKDups */
296 	__u32	tcpi_reord_seen;     /* reordering events seen */
297 
298 	__u32	tcpi_rcv_ooopack;    /* Out-of-order packets received */
299 
300 	__u32	tcpi_snd_wnd;	     /* peer's advertised receive window after
301 				      * scaling (bytes)
302 				      */
303 };
304 
305 /* netlink attributes types for SCM_TIMESTAMPING_OPT_STATS */
306 enum {
307 	TCP_NLA_PAD,
308 	TCP_NLA_BUSY,		/* Time (usec) busy sending data */
309 	TCP_NLA_RWND_LIMITED,	/* Time (usec) limited by receive window */
310 	TCP_NLA_SNDBUF_LIMITED,	/* Time (usec) limited by send buffer */
311 	TCP_NLA_DATA_SEGS_OUT,	/* Data pkts sent including retransmission */
312 	TCP_NLA_TOTAL_RETRANS,	/* Data pkts retransmitted */
313 	TCP_NLA_PACING_RATE,    /* Pacing rate in bytes per second */
314 	TCP_NLA_DELIVERY_RATE,  /* Delivery rate in bytes per second */
315 	TCP_NLA_SND_CWND,       /* Sending congestion window */
316 	TCP_NLA_REORDERING,     /* Reordering metric */
317 	TCP_NLA_MIN_RTT,        /* minimum RTT */
318 	TCP_NLA_RECUR_RETRANS,  /* Recurring retransmits for the current pkt */
319 	TCP_NLA_DELIVERY_RATE_APP_LMT, /* delivery rate application limited ? */
320 	TCP_NLA_SNDQ_SIZE,	/* Data (bytes) pending in send queue */
321 	TCP_NLA_CA_STATE,	/* ca_state of socket */
322 	TCP_NLA_SND_SSTHRESH,	/* Slow start size threshold */
323 	TCP_NLA_DELIVERED,	/* Data pkts delivered incl. out-of-order */
324 	TCP_NLA_DELIVERED_CE,	/* Like above but only ones w/ CE marks */
325 	TCP_NLA_BYTES_SENT,	/* Data bytes sent including retransmission */
326 	TCP_NLA_BYTES_RETRANS,	/* Data bytes retransmitted */
327 	TCP_NLA_DSACK_DUPS,	/* DSACK blocks received */
328 	TCP_NLA_REORD_SEEN,	/* reordering events seen */
329 	TCP_NLA_SRTT,		/* smoothed RTT in usecs */
330 	TCP_NLA_TIMEOUT_REHASH, /* Timeout-triggered rehash attempts */
331 	TCP_NLA_BYTES_NOTSENT,	/* Bytes in write queue not yet sent */
332 	TCP_NLA_EDT,		/* Earliest departure time (CLOCK_MONOTONIC) */
333 };
334 
335 /* for TCP_MD5SIG socket option */
336 #define TCP_MD5SIG_MAXKEYLEN	80
337 
338 /* tcp_md5sig extension flags for TCP_MD5SIG_EXT */
339 #define TCP_MD5SIG_FLAG_PREFIX		0x1	/* address prefix length */
340 #define TCP_MD5SIG_FLAG_IFINDEX		0x2	/* ifindex set */
341 
342 struct tcp_md5sig {
343 	struct __kernel_sockaddr_storage tcpm_addr;	/* address associated */
344 	__u8	tcpm_flags;				/* extension flags */
345 	__u8	tcpm_prefixlen;				/* address prefix */
346 	__u16	tcpm_keylen;				/* key length */
347 	int	tcpm_ifindex;				/* device index for scope */
348 	__u8	tcpm_key[TCP_MD5SIG_MAXKEYLEN];		/* key (binary) */
349 };
350 
351 /* INET_DIAG_MD5SIG */
352 struct tcp_diag_md5sig {
353 	__u8	tcpm_family;
354 	__u8	tcpm_prefixlen;
355 	__u16	tcpm_keylen;
356 	__be32	tcpm_addr[4];
357 	__u8	tcpm_key[TCP_MD5SIG_MAXKEYLEN];
358 };
359 
360 /* setsockopt(fd, IPPROTO_TCP, TCP_ZEROCOPY_RECEIVE, ...) */
361 
362 struct tcp_zerocopy_receive {
363 	__u64 address;		/* in: address of mapping */
364 	__u32 length;		/* in/out: number of bytes to map/mapped */
365 	__u32 recv_skip_hint;	/* out: amount of bytes to skip */
366 	__u32 inq; /* out: amount of bytes in read queue */
367 	__s32 err; /* out: socket error */
368 	__u64 copybuf_address;	/* in: copybuf address (small reads) */
369 	__s32 copybuf_len; /* in/out: copybuf bytes avail/used or error */
370 };
371 #endif /* _UAPI_LINUX_TCP_H */
372