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 /* Modify TCP-AO per-socket options */ 135 136 #define TCP_REPAIR_ON 1 137 #define TCP_REPAIR_OFF 0 138 #define TCP_REPAIR_OFF_NO_WP -1 /* Turn off without window probes */ 139 140 struct tcp_repair_opt { 141 __u32 opt_code; 142 __u32 opt_val; 143 }; 144 145 struct tcp_repair_window { 146 __u32 snd_wl1; 147 __u32 snd_wnd; 148 __u32 max_window; 149 150 __u32 rcv_wnd; 151 __u32 rcv_wup; 152 }; 153 154 enum { 155 TCP_NO_QUEUE, 156 TCP_RECV_QUEUE, 157 TCP_SEND_QUEUE, 158 TCP_QUEUES_NR, 159 }; 160 161 /* why fastopen failed from client perspective */ 162 enum tcp_fastopen_client_fail { 163 TFO_STATUS_UNSPEC, /* catch-all */ 164 TFO_COOKIE_UNAVAILABLE, /* if not in TFO_CLIENT_NO_COOKIE mode */ 165 TFO_DATA_NOT_ACKED, /* SYN-ACK did not ack SYN data */ 166 TFO_SYN_RETRANSMITTED, /* SYN-ACK did not ack SYN data after timeout */ 167 }; 168 169 /* for TCP_INFO socket option */ 170 #define TCPI_OPT_TIMESTAMPS 1 171 #define TCPI_OPT_SACK 2 172 #define TCPI_OPT_WSCALE 4 173 #define TCPI_OPT_ECN 8 /* ECN was negociated at TCP session init */ 174 #define TCPI_OPT_ECN_SEEN 16 /* we received at least one packet with ECT */ 175 #define TCPI_OPT_SYN_DATA 32 /* SYN-ACK acked data in SYN sent or rcvd */ 176 #define TCPI_OPT_USEC_TS 64 /* usec timestamps */ 177 178 /* 179 * Sender's congestion state indicating normal or abnormal situations 180 * in the last round of packets sent. The state is driven by the ACK 181 * information and timer events. 182 */ 183 enum tcp_ca_state { 184 /* 185 * Nothing bad has been observed recently. 186 * No apparent reordering, packet loss, or ECN marks. 187 */ 188 TCP_CA_Open = 0, 189 #define TCPF_CA_Open (1<<TCP_CA_Open) 190 /* 191 * The sender enters disordered state when it has received DUPACKs or 192 * SACKs in the last round of packets sent. This could be due to packet 193 * loss or reordering but needs further information to confirm packets 194 * have been lost. 195 */ 196 TCP_CA_Disorder = 1, 197 #define TCPF_CA_Disorder (1<<TCP_CA_Disorder) 198 /* 199 * The sender enters Congestion Window Reduction (CWR) state when it 200 * has received ACKs with ECN-ECE marks, or has experienced congestion 201 * or packet discard on the sender host (e.g. qdisc). 202 */ 203 TCP_CA_CWR = 2, 204 #define TCPF_CA_CWR (1<<TCP_CA_CWR) 205 /* 206 * The sender is in fast recovery and retransmitting lost packets, 207 * typically triggered by ACK events. 208 */ 209 TCP_CA_Recovery = 3, 210 #define TCPF_CA_Recovery (1<<TCP_CA_Recovery) 211 /* 212 * The sender is in loss recovery triggered by retransmission timeout. 213 */ 214 TCP_CA_Loss = 4 215 #define TCPF_CA_Loss (1<<TCP_CA_Loss) 216 }; 217 218 struct tcp_info { 219 __u8 tcpi_state; 220 __u8 tcpi_ca_state; 221 __u8 tcpi_retransmits; 222 __u8 tcpi_probes; 223 __u8 tcpi_backoff; 224 __u8 tcpi_options; 225 __u8 tcpi_snd_wscale : 4, tcpi_rcv_wscale : 4; 226 __u8 tcpi_delivery_rate_app_limited:1, tcpi_fastopen_client_fail:2; 227 228 __u32 tcpi_rto; 229 __u32 tcpi_ato; 230 __u32 tcpi_snd_mss; 231 __u32 tcpi_rcv_mss; 232 233 __u32 tcpi_unacked; 234 __u32 tcpi_sacked; 235 __u32 tcpi_lost; 236 __u32 tcpi_retrans; 237 __u32 tcpi_fackets; 238 239 /* Times. */ 240 __u32 tcpi_last_data_sent; 241 __u32 tcpi_last_ack_sent; /* Not remembered, sorry. */ 242 __u32 tcpi_last_data_recv; 243 __u32 tcpi_last_ack_recv; 244 245 /* Metrics. */ 246 __u32 tcpi_pmtu; 247 __u32 tcpi_rcv_ssthresh; 248 __u32 tcpi_rtt; 249 __u32 tcpi_rttvar; 250 __u32 tcpi_snd_ssthresh; 251 __u32 tcpi_snd_cwnd; 252 __u32 tcpi_advmss; 253 __u32 tcpi_reordering; 254 255 __u32 tcpi_rcv_rtt; 256 __u32 tcpi_rcv_space; 257 258 __u32 tcpi_total_retrans; 259 260 __u64 tcpi_pacing_rate; 261 __u64 tcpi_max_pacing_rate; 262 __u64 tcpi_bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked */ 263 __u64 tcpi_bytes_received; /* RFC4898 tcpEStatsAppHCThruOctetsReceived */ 264 __u32 tcpi_segs_out; /* RFC4898 tcpEStatsPerfSegsOut */ 265 __u32 tcpi_segs_in; /* RFC4898 tcpEStatsPerfSegsIn */ 266 267 __u32 tcpi_notsent_bytes; 268 __u32 tcpi_min_rtt; 269 __u32 tcpi_data_segs_in; /* RFC4898 tcpEStatsDataSegsIn */ 270 __u32 tcpi_data_segs_out; /* RFC4898 tcpEStatsDataSegsOut */ 271 272 __u64 tcpi_delivery_rate; 273 274 __u64 tcpi_busy_time; /* Time (usec) busy sending data */ 275 __u64 tcpi_rwnd_limited; /* Time (usec) limited by receive window */ 276 __u64 tcpi_sndbuf_limited; /* Time (usec) limited by send buffer */ 277 278 __u32 tcpi_delivered; 279 __u32 tcpi_delivered_ce; 280 281 __u64 tcpi_bytes_sent; /* RFC4898 tcpEStatsPerfHCDataOctetsOut */ 282 __u64 tcpi_bytes_retrans; /* RFC4898 tcpEStatsPerfOctetsRetrans */ 283 __u32 tcpi_dsack_dups; /* RFC4898 tcpEStatsStackDSACKDups */ 284 __u32 tcpi_reord_seen; /* reordering events seen */ 285 286 __u32 tcpi_rcv_ooopack; /* Out-of-order packets received */ 287 288 __u32 tcpi_snd_wnd; /* peer's advertised receive window after 289 * scaling (bytes) 290 */ 291 __u32 tcpi_rcv_wnd; /* local advertised receive window after 292 * scaling (bytes) 293 */ 294 295 __u32 tcpi_rehash; /* PLB or timeout triggered rehash attempts */ 296 297 __u16 tcpi_total_rto; /* Total number of RTO timeouts, including 298 * SYN/SYN-ACK and recurring timeouts. 299 */ 300 __u16 tcpi_total_rto_recoveries; /* Total number of RTO 301 * recoveries, including any 302 * unfinished recovery. 303 */ 304 __u32 tcpi_total_rto_time; /* Total time spent in RTO recoveries 305 * in milliseconds, including any 306 * unfinished recovery. 307 */ 308 }; 309 310 /* netlink attributes types for SCM_TIMESTAMPING_OPT_STATS */ 311 enum { 312 TCP_NLA_PAD, 313 TCP_NLA_BUSY, /* Time (usec) busy sending data */ 314 TCP_NLA_RWND_LIMITED, /* Time (usec) limited by receive window */ 315 TCP_NLA_SNDBUF_LIMITED, /* Time (usec) limited by send buffer */ 316 TCP_NLA_DATA_SEGS_OUT, /* Data pkts sent including retransmission */ 317 TCP_NLA_TOTAL_RETRANS, /* Data pkts retransmitted */ 318 TCP_NLA_PACING_RATE, /* Pacing rate in bytes per second */ 319 TCP_NLA_DELIVERY_RATE, /* Delivery rate in bytes per second */ 320 TCP_NLA_SND_CWND, /* Sending congestion window */ 321 TCP_NLA_REORDERING, /* Reordering metric */ 322 TCP_NLA_MIN_RTT, /* minimum RTT */ 323 TCP_NLA_RECUR_RETRANS, /* Recurring retransmits for the current pkt */ 324 TCP_NLA_DELIVERY_RATE_APP_LMT, /* delivery rate application limited ? */ 325 TCP_NLA_SNDQ_SIZE, /* Data (bytes) pending in send queue */ 326 TCP_NLA_CA_STATE, /* ca_state of socket */ 327 TCP_NLA_SND_SSTHRESH, /* Slow start size threshold */ 328 TCP_NLA_DELIVERED, /* Data pkts delivered incl. out-of-order */ 329 TCP_NLA_DELIVERED_CE, /* Like above but only ones w/ CE marks */ 330 TCP_NLA_BYTES_SENT, /* Data bytes sent including retransmission */ 331 TCP_NLA_BYTES_RETRANS, /* Data bytes retransmitted */ 332 TCP_NLA_DSACK_DUPS, /* DSACK blocks received */ 333 TCP_NLA_REORD_SEEN, /* reordering events seen */ 334 TCP_NLA_SRTT, /* smoothed RTT in usecs */ 335 TCP_NLA_TIMEOUT_REHASH, /* Timeout-triggered rehash attempts */ 336 TCP_NLA_BYTES_NOTSENT, /* Bytes in write queue not yet sent */ 337 TCP_NLA_EDT, /* Earliest departure time (CLOCK_MONOTONIC) */ 338 TCP_NLA_TTL, /* TTL or hop limit of a packet received */ 339 TCP_NLA_REHASH, /* PLB and timeout triggered rehash attempts */ 340 }; 341 342 /* for TCP_MD5SIG socket option */ 343 #define TCP_MD5SIG_MAXKEYLEN 80 344 345 /* tcp_md5sig extension flags for TCP_MD5SIG_EXT */ 346 #define TCP_MD5SIG_FLAG_PREFIX 0x1 /* address prefix length */ 347 #define TCP_MD5SIG_FLAG_IFINDEX 0x2 /* ifindex set */ 348 349 struct tcp_md5sig { 350 struct __kernel_sockaddr_storage tcpm_addr; /* address associated */ 351 __u8 tcpm_flags; /* extension flags */ 352 __u8 tcpm_prefixlen; /* address prefix */ 353 __u16 tcpm_keylen; /* key length */ 354 int tcpm_ifindex; /* device index for scope */ 355 __u8 tcpm_key[TCP_MD5SIG_MAXKEYLEN]; /* key (binary) */ 356 }; 357 358 /* INET_DIAG_MD5SIG */ 359 struct tcp_diag_md5sig { 360 __u8 tcpm_family; 361 __u8 tcpm_prefixlen; 362 __u16 tcpm_keylen; 363 __be32 tcpm_addr[4]; 364 __u8 tcpm_key[TCP_MD5SIG_MAXKEYLEN]; 365 }; 366 367 #define TCP_AO_MAXKEYLEN 80 368 369 #define TCP_AO_KEYF_IFINDEX (1 << 0) /* L3 ifindex for VRF */ 370 371 struct tcp_ao_add { /* setsockopt(TCP_AO_ADD_KEY) */ 372 struct __kernel_sockaddr_storage addr; /* peer's address for the key */ 373 char alg_name[64]; /* crypto hash algorithm to use */ 374 __s32 ifindex; /* L3 dev index for VRF */ 375 __u32 set_current :1, /* set key as Current_key at once */ 376 set_rnext :1, /* request it from peer with RNext_key */ 377 reserved :30; /* must be 0 */ 378 __u16 reserved2; /* padding, must be 0 */ 379 __u8 prefix; /* peer's address prefix */ 380 __u8 sndid; /* SendID for outgoing segments */ 381 __u8 rcvid; /* RecvID to match for incoming seg */ 382 __u8 maclen; /* length of authentication code (hash) */ 383 __u8 keyflags; /* see TCP_AO_KEYF_ */ 384 __u8 keylen; /* length of ::key */ 385 __u8 key[TCP_AO_MAXKEYLEN]; 386 } __attribute__((aligned(8))); 387 388 struct tcp_ao_del { /* setsockopt(TCP_AO_DEL_KEY) */ 389 struct __kernel_sockaddr_storage addr; /* peer's address for the key */ 390 __s32 ifindex; /* L3 dev index for VRF */ 391 __u32 set_current :1, /* corresponding ::current_key */ 392 set_rnext :1, /* corresponding ::rnext */ 393 reserved :30; /* must be 0 */ 394 __u16 reserved2; /* padding, must be 0 */ 395 __u8 prefix; /* peer's address prefix */ 396 __u8 sndid; /* SendID for outgoing segments */ 397 __u8 rcvid; /* RecvID to match for incoming seg */ 398 __u8 current_key; /* KeyID to set as Current_key */ 399 __u8 rnext; /* KeyID to set as Rnext_key */ 400 __u8 keyflags; /* see TCP_AO_KEYF_ */ 401 } __attribute__((aligned(8))); 402 403 struct tcp_ao_info_opt { /* setsockopt(TCP_AO_INFO) */ 404 __u32 set_current :1, /* corresponding ::current_key */ 405 set_rnext :1, /* corresponding ::rnext */ 406 ao_required :1, /* don't accept non-AO connects */ 407 set_counters :1, /* set/clear ::pkt_* counters */ 408 reserved :28; /* must be 0 */ 409 __u16 reserved2; /* padding, must be 0 */ 410 __u8 current_key; /* KeyID to set as Current_key */ 411 __u8 rnext; /* KeyID to set as Rnext_key */ 412 __u64 pkt_good; /* verified segments */ 413 __u64 pkt_bad; /* failed verification */ 414 __u64 pkt_key_not_found; /* could not find a key to verify */ 415 __u64 pkt_ao_required; /* segments missing TCP-AO sign */ 416 } __attribute__((aligned(8))); 417 418 /* setsockopt(fd, IPPROTO_TCP, TCP_ZEROCOPY_RECEIVE, ...) */ 419 420 #define TCP_RECEIVE_ZEROCOPY_FLAG_TLB_CLEAN_HINT 0x1 421 struct tcp_zerocopy_receive { 422 __u64 address; /* in: address of mapping */ 423 __u32 length; /* in/out: number of bytes to map/mapped */ 424 __u32 recv_skip_hint; /* out: amount of bytes to skip */ 425 __u32 inq; /* out: amount of bytes in read queue */ 426 __s32 err; /* out: socket error */ 427 __u64 copybuf_address; /* in: copybuf address (small reads) */ 428 __s32 copybuf_len; /* in/out: copybuf bytes avail/used or error */ 429 __u32 flags; /* in: flags */ 430 __u64 msg_control; /* ancillary data */ 431 __u64 msg_controllen; 432 __u32 msg_flags; 433 __u32 reserved; /* set to 0 for now */ 434 }; 435 #endif /* _UAPI_LINUX_TCP_H */ 436