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, <[email protected]> 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_AO_ADD_KEY 38 /* Add/Set MKT */ 133 #define TCP_AO_DEL_KEY 39 /* Delete MKT */ 134 #define TCP_AO_INFO 40 /* Set/list TCP-AO per-socket options */ 135 #define TCP_AO_GET_KEYS 41 /* List MKT(s) */ 136 137 #define TCP_REPAIR_ON 1 138 #define TCP_REPAIR_OFF 0 139 #define TCP_REPAIR_OFF_NO_WP -1 /* Turn off without window probes */ 140 141 struct tcp_repair_opt { 142 __u32 opt_code; 143 __u32 opt_val; 144 }; 145 146 struct tcp_repair_window { 147 __u32 snd_wl1; 148 __u32 snd_wnd; 149 __u32 max_window; 150 151 __u32 rcv_wnd; 152 __u32 rcv_wup; 153 }; 154 155 enum { 156 TCP_NO_QUEUE, 157 TCP_RECV_QUEUE, 158 TCP_SEND_QUEUE, 159 TCP_QUEUES_NR, 160 }; 161 162 /* why fastopen failed from client perspective */ 163 enum tcp_fastopen_client_fail { 164 TFO_STATUS_UNSPEC, /* catch-all */ 165 TFO_COOKIE_UNAVAILABLE, /* if not in TFO_CLIENT_NO_COOKIE mode */ 166 TFO_DATA_NOT_ACKED, /* SYN-ACK did not ack SYN data */ 167 TFO_SYN_RETRANSMITTED, /* SYN-ACK did not ack SYN data after timeout */ 168 }; 169 170 /* for TCP_INFO socket option */ 171 #define TCPI_OPT_TIMESTAMPS 1 172 #define TCPI_OPT_SACK 2 173 #define TCPI_OPT_WSCALE 4 174 #define TCPI_OPT_ECN 8 /* ECN was negociated at TCP session init */ 175 #define TCPI_OPT_ECN_SEEN 16 /* we received at least one packet with ECT */ 176 #define TCPI_OPT_SYN_DATA 32 /* SYN-ACK acked data in SYN sent or rcvd */ 177 #define TCPI_OPT_USEC_TS 64 /* usec timestamps */ 178 179 /* 180 * Sender's congestion state indicating normal or abnormal situations 181 * in the last round of packets sent. The state is driven by the ACK 182 * information and timer events. 183 */ 184 enum tcp_ca_state { 185 /* 186 * Nothing bad has been observed recently. 187 * No apparent reordering, packet loss, or ECN marks. 188 */ 189 TCP_CA_Open = 0, 190 #define TCPF_CA_Open (1<<TCP_CA_Open) 191 /* 192 * The sender enters disordered state when it has received DUPACKs or 193 * SACKs in the last round of packets sent. This could be due to packet 194 * loss or reordering but needs further information to confirm packets 195 * have been lost. 196 */ 197 TCP_CA_Disorder = 1, 198 #define TCPF_CA_Disorder (1<<TCP_CA_Disorder) 199 /* 200 * The sender enters Congestion Window Reduction (CWR) state when it 201 * has received ACKs with ECN-ECE marks, or has experienced congestion 202 * or packet discard on the sender host (e.g. qdisc). 203 */ 204 TCP_CA_CWR = 2, 205 #define TCPF_CA_CWR (1<<TCP_CA_CWR) 206 /* 207 * The sender is in fast recovery and retransmitting lost packets, 208 * typically triggered by ACK events. 209 */ 210 TCP_CA_Recovery = 3, 211 #define TCPF_CA_Recovery (1<<TCP_CA_Recovery) 212 /* 213 * The sender is in loss recovery triggered by retransmission timeout. 214 */ 215 TCP_CA_Loss = 4 216 #define TCPF_CA_Loss (1<<TCP_CA_Loss) 217 }; 218 219 struct tcp_info { 220 __u8 tcpi_state; 221 __u8 tcpi_ca_state; 222 __u8 tcpi_retransmits; 223 __u8 tcpi_probes; 224 __u8 tcpi_backoff; 225 __u8 tcpi_options; 226 __u8 tcpi_snd_wscale : 4, tcpi_rcv_wscale : 4; 227 __u8 tcpi_delivery_rate_app_limited:1, tcpi_fastopen_client_fail:2; 228 229 __u32 tcpi_rto; 230 __u32 tcpi_ato; 231 __u32 tcpi_snd_mss; 232 __u32 tcpi_rcv_mss; 233 234 __u32 tcpi_unacked; 235 __u32 tcpi_sacked; 236 __u32 tcpi_lost; 237 __u32 tcpi_retrans; 238 __u32 tcpi_fackets; 239 240 /* Times. */ 241 __u32 tcpi_last_data_sent; 242 __u32 tcpi_last_ack_sent; /* Not remembered, sorry. */ 243 __u32 tcpi_last_data_recv; 244 __u32 tcpi_last_ack_recv; 245 246 /* Metrics. */ 247 __u32 tcpi_pmtu; 248 __u32 tcpi_rcv_ssthresh; 249 __u32 tcpi_rtt; 250 __u32 tcpi_rttvar; 251 __u32 tcpi_snd_ssthresh; 252 __u32 tcpi_snd_cwnd; 253 __u32 tcpi_advmss; 254 __u32 tcpi_reordering; 255 256 __u32 tcpi_rcv_rtt; 257 __u32 tcpi_rcv_space; 258 259 __u32 tcpi_total_retrans; 260 261 __u64 tcpi_pacing_rate; 262 __u64 tcpi_max_pacing_rate; 263 __u64 tcpi_bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked */ 264 __u64 tcpi_bytes_received; /* RFC4898 tcpEStatsAppHCThruOctetsReceived */ 265 __u32 tcpi_segs_out; /* RFC4898 tcpEStatsPerfSegsOut */ 266 __u32 tcpi_segs_in; /* RFC4898 tcpEStatsPerfSegsIn */ 267 268 __u32 tcpi_notsent_bytes; 269 __u32 tcpi_min_rtt; 270 __u32 tcpi_data_segs_in; /* RFC4898 tcpEStatsDataSegsIn */ 271 __u32 tcpi_data_segs_out; /* RFC4898 tcpEStatsDataSegsOut */ 272 273 __u64 tcpi_delivery_rate; 274 275 __u64 tcpi_busy_time; /* Time (usec) busy sending data */ 276 __u64 tcpi_rwnd_limited; /* Time (usec) limited by receive window */ 277 __u64 tcpi_sndbuf_limited; /* Time (usec) limited by send buffer */ 278 279 __u32 tcpi_delivered; 280 __u32 tcpi_delivered_ce; 281 282 __u64 tcpi_bytes_sent; /* RFC4898 tcpEStatsPerfHCDataOctetsOut */ 283 __u64 tcpi_bytes_retrans; /* RFC4898 tcpEStatsPerfOctetsRetrans */ 284 __u32 tcpi_dsack_dups; /* RFC4898 tcpEStatsStackDSACKDups */ 285 __u32 tcpi_reord_seen; /* reordering events seen */ 286 287 __u32 tcpi_rcv_ooopack; /* Out-of-order packets received */ 288 289 __u32 tcpi_snd_wnd; /* peer's advertised receive window after 290 * scaling (bytes) 291 */ 292 __u32 tcpi_rcv_wnd; /* local advertised receive window after 293 * scaling (bytes) 294 */ 295 296 __u32 tcpi_rehash; /* PLB or timeout triggered rehash attempts */ 297 298 __u16 tcpi_total_rto; /* Total number of RTO timeouts, including 299 * SYN/SYN-ACK and recurring timeouts. 300 */ 301 __u16 tcpi_total_rto_recoveries; /* Total number of RTO 302 * recoveries, including any 303 * unfinished recovery. 304 */ 305 __u32 tcpi_total_rto_time; /* Total time spent in RTO recoveries 306 * in milliseconds, including any 307 * unfinished recovery. 308 */ 309 }; 310 311 /* netlink attributes types for SCM_TIMESTAMPING_OPT_STATS */ 312 enum { 313 TCP_NLA_PAD, 314 TCP_NLA_BUSY, /* Time (usec) busy sending data */ 315 TCP_NLA_RWND_LIMITED, /* Time (usec) limited by receive window */ 316 TCP_NLA_SNDBUF_LIMITED, /* Time (usec) limited by send buffer */ 317 TCP_NLA_DATA_SEGS_OUT, /* Data pkts sent including retransmission */ 318 TCP_NLA_TOTAL_RETRANS, /* Data pkts retransmitted */ 319 TCP_NLA_PACING_RATE, /* Pacing rate in bytes per second */ 320 TCP_NLA_DELIVERY_RATE, /* Delivery rate in bytes per second */ 321 TCP_NLA_SND_CWND, /* Sending congestion window */ 322 TCP_NLA_REORDERING, /* Reordering metric */ 323 TCP_NLA_MIN_RTT, /* minimum RTT */ 324 TCP_NLA_RECUR_RETRANS, /* Recurring retransmits for the current pkt */ 325 TCP_NLA_DELIVERY_RATE_APP_LMT, /* delivery rate application limited ? */ 326 TCP_NLA_SNDQ_SIZE, /* Data (bytes) pending in send queue */ 327 TCP_NLA_CA_STATE, /* ca_state of socket */ 328 TCP_NLA_SND_SSTHRESH, /* Slow start size threshold */ 329 TCP_NLA_DELIVERED, /* Data pkts delivered incl. out-of-order */ 330 TCP_NLA_DELIVERED_CE, /* Like above but only ones w/ CE marks */ 331 TCP_NLA_BYTES_SENT, /* Data bytes sent including retransmission */ 332 TCP_NLA_BYTES_RETRANS, /* Data bytes retransmitted */ 333 TCP_NLA_DSACK_DUPS, /* DSACK blocks received */ 334 TCP_NLA_REORD_SEEN, /* reordering events seen */ 335 TCP_NLA_SRTT, /* smoothed RTT in usecs */ 336 TCP_NLA_TIMEOUT_REHASH, /* Timeout-triggered rehash attempts */ 337 TCP_NLA_BYTES_NOTSENT, /* Bytes in write queue not yet sent */ 338 TCP_NLA_EDT, /* Earliest departure time (CLOCK_MONOTONIC) */ 339 TCP_NLA_TTL, /* TTL or hop limit of a packet received */ 340 TCP_NLA_REHASH, /* PLB and timeout triggered rehash attempts */ 341 }; 342 343 /* for TCP_MD5SIG socket option */ 344 #define TCP_MD5SIG_MAXKEYLEN 80 345 346 /* tcp_md5sig extension flags for TCP_MD5SIG_EXT */ 347 #define TCP_MD5SIG_FLAG_PREFIX 0x1 /* address prefix length */ 348 #define TCP_MD5SIG_FLAG_IFINDEX 0x2 /* ifindex set */ 349 350 struct tcp_md5sig { 351 struct __kernel_sockaddr_storage tcpm_addr; /* address associated */ 352 __u8 tcpm_flags; /* extension flags */ 353 __u8 tcpm_prefixlen; /* address prefix */ 354 __u16 tcpm_keylen; /* key length */ 355 int tcpm_ifindex; /* device index for scope */ 356 __u8 tcpm_key[TCP_MD5SIG_MAXKEYLEN]; /* key (binary) */ 357 }; 358 359 /* INET_DIAG_MD5SIG */ 360 struct tcp_diag_md5sig { 361 __u8 tcpm_family; 362 __u8 tcpm_prefixlen; 363 __u16 tcpm_keylen; 364 __be32 tcpm_addr[4]; 365 __u8 tcpm_key[TCP_MD5SIG_MAXKEYLEN]; 366 }; 367 368 #define TCP_AO_MAXKEYLEN 80 369 370 #define TCP_AO_KEYF_IFINDEX (1 << 0) /* L3 ifindex for VRF */ 371 #define TCP_AO_KEYF_EXCLUDE_OPT (1 << 1) /* "Indicates whether TCP 372 * options other than TCP-AO 373 * are included in the MAC 374 * calculation" 375 */ 376 377 struct tcp_ao_add { /* setsockopt(TCP_AO_ADD_KEY) */ 378 struct __kernel_sockaddr_storage addr; /* peer's address for the key */ 379 char alg_name[64]; /* crypto hash algorithm to use */ 380 __s32 ifindex; /* L3 dev index for VRF */ 381 __u32 set_current :1, /* set key as Current_key at once */ 382 set_rnext :1, /* request it from peer with RNext_key */ 383 reserved :30; /* must be 0 */ 384 __u16 reserved2; /* padding, must be 0 */ 385 __u8 prefix; /* peer's address prefix */ 386 __u8 sndid; /* SendID for outgoing segments */ 387 __u8 rcvid; /* RecvID to match for incoming seg */ 388 __u8 maclen; /* length of authentication code (hash) */ 389 __u8 keyflags; /* see TCP_AO_KEYF_ */ 390 __u8 keylen; /* length of ::key */ 391 __u8 key[TCP_AO_MAXKEYLEN]; 392 } __attribute__((aligned(8))); 393 394 struct tcp_ao_del { /* setsockopt(TCP_AO_DEL_KEY) */ 395 struct __kernel_sockaddr_storage addr; /* peer's address for the key */ 396 __s32 ifindex; /* L3 dev index for VRF */ 397 __u32 set_current :1, /* corresponding ::current_key */ 398 set_rnext :1, /* corresponding ::rnext */ 399 reserved :30; /* must be 0 */ 400 __u16 reserved2; /* padding, must be 0 */ 401 __u8 prefix; /* peer's address prefix */ 402 __u8 sndid; /* SendID for outgoing segments */ 403 __u8 rcvid; /* RecvID to match for incoming seg */ 404 __u8 current_key; /* KeyID to set as Current_key */ 405 __u8 rnext; /* KeyID to set as Rnext_key */ 406 __u8 keyflags; /* see TCP_AO_KEYF_ */ 407 } __attribute__((aligned(8))); 408 409 struct tcp_ao_info_opt { /* setsockopt(TCP_AO_INFO), getsockopt(TCP_AO_INFO) */ 410 /* Here 'in' is for setsockopt(), 'out' is for getsockopt() */ 411 __u32 set_current :1, /* in/out: corresponding ::current_key */ 412 set_rnext :1, /* in/out: corresponding ::rnext */ 413 ao_required :1, /* in/out: don't accept non-AO connects */ 414 set_counters :1, /* in: set/clear ::pkt_* counters */ 415 accept_icmps :1, /* in/out: accept incoming ICMPs */ 416 reserved :27; /* must be 0 */ 417 __u16 reserved2; /* padding, must be 0 */ 418 __u8 current_key; /* in/out: KeyID of Current_key */ 419 __u8 rnext; /* in/out: keyid of RNext_key */ 420 __u64 pkt_good; /* in/out: verified segments */ 421 __u64 pkt_bad; /* in/out: failed verification */ 422 __u64 pkt_key_not_found; /* in/out: could not find a key to verify */ 423 __u64 pkt_ao_required; /* in/out: segments missing TCP-AO sign */ 424 __u64 pkt_dropped_icmp; /* in/out: ICMPs that were ignored */ 425 } __attribute__((aligned(8))); 426 427 struct tcp_ao_getsockopt { /* getsockopt(TCP_AO_GET_KEYS) */ 428 struct __kernel_sockaddr_storage addr; /* in/out: dump keys for peer 429 * with this address/prefix 430 */ 431 char alg_name[64]; /* out: crypto hash algorithm */ 432 __u8 key[TCP_AO_MAXKEYLEN]; 433 __u32 nkeys; /* in: size of the userspace buffer 434 * @optval, measured in @optlen - the 435 * sizeof(struct tcp_ao_getsockopt) 436 * out: number of keys that matched 437 */ 438 __u16 is_current :1, /* in: match and dump Current_key, 439 * out: the dumped key is Current_key 440 */ 441 442 is_rnext :1, /* in: match and dump RNext_key, 443 * out: the dumped key is RNext_key 444 */ 445 get_all :1, /* in: dump all keys */ 446 reserved :13; /* padding, must be 0 */ 447 __u8 sndid; /* in/out: dump keys with SendID */ 448 __u8 rcvid; /* in/out: dump keys with RecvID */ 449 __u8 prefix; /* in/out: dump keys with address/prefix */ 450 __u8 maclen; /* out: key's length of authentication 451 * code (hash) 452 */ 453 __u8 keyflags; /* in/out: see TCP_AO_KEYF_ */ 454 __u8 keylen; /* out: length of ::key */ 455 __s32 ifindex; /* in/out: L3 dev index for VRF */ 456 __u64 pkt_good; /* out: verified segments */ 457 __u64 pkt_bad; /* out: segments that failed verification */ 458 } __attribute__((aligned(8))); 459 460 /* setsockopt(fd, IPPROTO_TCP, TCP_ZEROCOPY_RECEIVE, ...) */ 461 462 #define TCP_RECEIVE_ZEROCOPY_FLAG_TLB_CLEAN_HINT 0x1 463 struct tcp_zerocopy_receive { 464 __u64 address; /* in: address of mapping */ 465 __u32 length; /* in/out: number of bytes to map/mapped */ 466 __u32 recv_skip_hint; /* out: amount of bytes to skip */ 467 __u32 inq; /* out: amount of bytes in read queue */ 468 __s32 err; /* out: socket error */ 469 __u64 copybuf_address; /* in: copybuf address (small reads) */ 470 __s32 copybuf_len; /* in/out: copybuf bytes avail/used or error */ 471 __u32 flags; /* in: flags */ 472 __u64 msg_control; /* ancillary data */ 473 __u64 msg_controllen; 474 __u32 msg_flags; 475 __u32 reserved; /* set to 0 for now */ 476 }; 477 #endif /* _UAPI_LINUX_TCP_H */ 478