1 /* 2 * Copyright (c) 1982, 1986, 1993, 1994, 1995 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 34 * $FreeBSD$ 35 */ 36 37 #ifndef _NETINET_TCP_VAR_H_ 38 #define _NETINET_TCP_VAR_H_ 39 /* 40 * Kernel variables for tcp. 41 */ 42 43 /* TCP segment queue entry */ 44 struct tseg_qent { 45 LIST_ENTRY(tseg_qent) tqe_q; 46 int tqe_len; /* TCP segment data length */ 47 struct tcphdr *tqe_th; /* a pointer to tcp header */ 48 struct mbuf *tqe_m; /* mbuf contains packet */ 49 }; 50 LIST_HEAD(tsegqe_head, tseg_qent); 51 #ifdef MALLOC_DECLARE 52 MALLOC_DECLARE(M_TSEGQ); 53 #endif 54 55 struct tcptemp { 56 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 57 struct tcphdr tt_t; 58 }; 59 60 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 61 62 /* 63 * Tcp control block, one per tcp; fields: 64 * Organized for 16 byte cacheline efficiency. 65 */ 66 struct tcpcb { 67 struct tsegqe_head t_segq; 68 int t_dupacks; /* consecutive dup acks recd */ 69 struct tcptemp *t_template; /* skeletal packet for transmit */ 70 71 struct callout *tt_rexmt; /* retransmit timer */ 72 struct callout *tt_persist; /* retransmit persistence */ 73 struct callout *tt_keep; /* keepalive */ 74 struct callout *tt_2msl; /* 2*msl TIME_WAIT timer */ 75 struct callout *tt_delack; /* delayed ACK timer */ 76 77 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 78 int t_state; /* state of this connection */ 79 u_int t_flags; 80 #define TF_ACKNOW 0x00001 /* ack peer immediately */ 81 #define TF_DELACK 0x00002 /* ack, but try to delay it */ 82 #define TF_NODELAY 0x00004 /* don't delay packets to coalesce */ 83 #define TF_NOOPT 0x00008 /* don't use tcp options */ 84 #define TF_SENTFIN 0x00010 /* have sent FIN */ 85 #define TF_REQ_SCALE 0x00020 /* have/will request window scaling */ 86 #define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */ 87 #define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */ 88 #define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */ 89 #define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */ 90 #define TF_NEEDSYN 0x00400 /* send SYN (implicit state) */ 91 #define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */ 92 #define TF_NOPUSH 0x01000 /* don't push */ 93 #define TF_REQ_CC 0x02000 /* have/will request CC */ 94 #define TF_RCVD_CC 0x04000 /* a CC was received in SYN */ 95 #define TF_SENDCCNEW 0x08000 /* send CCnew instead of CC in SYN */ 96 #define TF_MORETOCOME 0x10000 /* More data to be appended to sock */ 97 int t_force; /* 1 if forcing out a byte */ 98 99 tcp_seq snd_una; /* send unacknowledged */ 100 tcp_seq snd_max; /* highest sequence number sent; 101 * used to recognize retransmits 102 */ 103 tcp_seq snd_nxt; /* send next */ 104 tcp_seq snd_up; /* send urgent pointer */ 105 106 tcp_seq snd_wl1; /* window update seg seq number */ 107 tcp_seq snd_wl2; /* window update seg ack number */ 108 tcp_seq iss; /* initial send sequence number */ 109 tcp_seq irs; /* initial receive sequence number */ 110 111 tcp_seq rcv_nxt; /* receive next */ 112 tcp_seq rcv_adv; /* advertised window */ 113 u_long rcv_wnd; /* receive window */ 114 tcp_seq rcv_up; /* receive urgent pointer */ 115 116 u_long snd_wnd; /* send window */ 117 u_long snd_cwnd; /* congestion-controlled window */ 118 u_long snd_ssthresh; /* snd_cwnd size threshold for 119 * for slow start exponential to 120 * linear switch 121 */ 122 u_int t_maxopd; /* mss plus options */ 123 124 u_long t_rcvtime; /* inactivity time */ 125 u_long t_starttime; /* time connection was established */ 126 int t_rtttime; /* round trip time */ 127 tcp_seq t_rtseq; /* sequence number being timed */ 128 129 int t_rxtcur; /* current retransmit value (ticks) */ 130 u_int t_maxseg; /* maximum segment size */ 131 int t_srtt; /* smoothed round-trip time */ 132 int t_rttvar; /* variance in round-trip time */ 133 134 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 135 u_int t_rttmin; /* minimum rtt allowed */ 136 u_long t_rttupdated; /* number of times rtt sampled */ 137 u_long max_sndwnd; /* largest window peer has offered */ 138 139 int t_softerror; /* possible error not yet reported */ 140 /* out-of-band data */ 141 char t_oobflags; /* have some */ 142 char t_iobc; /* input character */ 143 #define TCPOOB_HAVEDATA 0x01 144 #define TCPOOB_HADDATA 0x02 145 /* RFC 1323 variables */ 146 u_char snd_scale; /* window scaling for send window */ 147 u_char rcv_scale; /* window scaling for recv window */ 148 u_char request_r_scale; /* pending window scaling */ 149 u_char requested_s_scale; 150 u_long ts_recent; /* timestamp echo data */ 151 152 u_long ts_recent_age; /* when last updated */ 153 tcp_seq last_ack_sent; 154 /* RFC 1644 variables */ 155 tcp_cc cc_send; /* send connection count */ 156 tcp_cc cc_recv; /* receive connection count */ 157 /* experimental */ 158 u_long snd_cwnd_prev; /* cwnd prior to retransmit */ 159 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ 160 u_long t_badrxtwin; /* window for retransmit recovery */ 161 }; 162 163 /* 164 * Structure to hold TCP options that are only used during segment 165 * processing (in tcp_input), but not held in the tcpcb. 166 * It's basically used to reduce the number of parameters 167 * to tcp_dooptions. 168 */ 169 struct tcpopt { 170 u_long to_flag; /* which options are present */ 171 #define TOF_TS 0x0001 /* timestamp */ 172 #define TOF_CC 0x0002 /* CC and CCnew are exclusive */ 173 #define TOF_CCNEW 0x0004 174 #define TOF_CCECHO 0x0008 175 u_long to_tsval; 176 u_long to_tsecr; 177 tcp_cc to_cc; /* holds CC or CCnew */ 178 tcp_cc to_ccecho; 179 }; 180 181 /* 182 * The TAO cache entry which is stored in the protocol family specific 183 * portion of the route metrics. 184 */ 185 struct rmxp_tao { 186 tcp_cc tao_cc; /* latest CC in valid SYN */ 187 tcp_cc tao_ccsent; /* latest CC sent to peer */ 188 u_short tao_mssopt; /* peer's cached MSS */ 189 #ifdef notyet 190 u_short tao_flags; /* cache status flags */ 191 #define TAOF_DONT 0x0001 /* peer doesn't understand rfc1644 */ 192 #define TAOF_OK 0x0002 /* peer does understand rfc1644 */ 193 #define TAOF_UNDEF 0 /* we don't know yet */ 194 #endif /* notyet */ 195 }; 196 #define rmx_taop(r) ((struct rmxp_tao *)(r).rmx_filler) 197 198 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 199 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 200 201 /* 202 * The smoothed round-trip time and estimated variance 203 * are stored as fixed point numbers scaled by the values below. 204 * For convenience, these scales are also used in smoothing the average 205 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 206 * With these scales, srtt has 3 bits to the right of the binary point, 207 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 208 * binary point, and is smoothed with an ALPHA of 0.75. 209 */ 210 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 211 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 212 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 213 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 214 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 215 216 /* 217 * The initial retransmission should happen at rtt + 4 * rttvar. 218 * Because of the way we do the smoothing, srtt and rttvar 219 * will each average +1/2 tick of bias. When we compute 220 * the retransmit timer, we want 1/2 tick of rounding and 221 * 1 extra tick because of +-1/2 tick uncertainty in the 222 * firing of the timer. The bias will give us exactly the 223 * 1.5 tick we need. But, because the bias is 224 * statistical, we have to test that we don't drop below 225 * the minimum feasible timer (which is 2 ticks). 226 * This version of the macro adapted from a paper by Lawrence 227 * Brakmo and Larry Peterson which outlines a problem caused 228 * by insufficient precision in the original implementation, 229 * which results in inappropriately large RTO values for very 230 * fast networks. 231 */ 232 #define TCP_REXMTVAL(tp) \ 233 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 234 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 235 236 /* 237 * TCP statistics. 238 * Many of these should be kept per connection, 239 * but that's inconvenient at the moment. 240 */ 241 struct tcpstat { 242 u_long tcps_connattempt; /* connections initiated */ 243 u_long tcps_accepts; /* connections accepted */ 244 u_long tcps_connects; /* connections established */ 245 u_long tcps_drops; /* connections dropped */ 246 u_long tcps_conndrops; /* embryonic connections dropped */ 247 u_long tcps_closed; /* conn. closed (includes drops) */ 248 u_long tcps_segstimed; /* segs where we tried to get rtt */ 249 u_long tcps_rttupdated; /* times we succeeded */ 250 u_long tcps_delack; /* delayed acks sent */ 251 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 252 u_long tcps_rexmttimeo; /* retransmit timeouts */ 253 u_long tcps_persisttimeo; /* persist timeouts */ 254 u_long tcps_keeptimeo; /* keepalive timeouts */ 255 u_long tcps_keepprobe; /* keepalive probes sent */ 256 u_long tcps_keepdrops; /* connections dropped in keepalive */ 257 258 u_long tcps_sndtotal; /* total packets sent */ 259 u_long tcps_sndpack; /* data packets sent */ 260 u_long tcps_sndbyte; /* data bytes sent */ 261 u_long tcps_sndrexmitpack; /* data packets retransmitted */ 262 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ 263 u_long tcps_sndacks; /* ack-only packets sent */ 264 u_long tcps_sndprobe; /* window probes sent */ 265 u_long tcps_sndurg; /* packets sent with URG only */ 266 u_long tcps_sndwinup; /* window update-only packets sent */ 267 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 268 269 u_long tcps_rcvtotal; /* total packets received */ 270 u_long tcps_rcvpack; /* packets received in sequence */ 271 u_long tcps_rcvbyte; /* bytes received in sequence */ 272 u_long tcps_rcvbadsum; /* packets received with ccksum errs */ 273 u_long tcps_rcvbadoff; /* packets received with bad offset */ 274 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ 275 u_long tcps_rcvshort; /* packets received too short */ 276 u_long tcps_rcvduppack; /* duplicate-only packets received */ 277 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ 278 u_long tcps_rcvpartduppack; /* packets with some duplicate data */ 279 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 280 u_long tcps_rcvoopack; /* out-of-order packets received */ 281 u_long tcps_rcvoobyte; /* out-of-order bytes received */ 282 u_long tcps_rcvpackafterwin; /* packets with data after window */ 283 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ 284 u_long tcps_rcvafterclose; /* packets rcvd after "close" */ 285 u_long tcps_rcvwinprobe; /* rcvd window probe packets */ 286 u_long tcps_rcvdupack; /* rcvd duplicate acks */ 287 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 288 u_long tcps_rcvackpack; /* rcvd ack packets */ 289 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ 290 u_long tcps_rcvwinupd; /* rcvd window update packets */ 291 u_long tcps_pawsdrop; /* segments dropped due to PAWS */ 292 u_long tcps_predack; /* times hdr predict ok for acks */ 293 u_long tcps_preddat; /* times hdr predict ok for data pkts */ 294 u_long tcps_pcbcachemiss; 295 u_long tcps_cachedrtt; /* times cached RTT in route updated */ 296 u_long tcps_cachedrttvar; /* times cached rttvar updated */ 297 u_long tcps_cachedssthresh; /* times cached ssthresh updated */ 298 u_long tcps_usedrtt; /* times RTT initialized from route */ 299 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */ 300 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/ 301 u_long tcps_persistdrop; /* timeout in persist state */ 302 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 303 u_long tcps_mturesent; /* resends due to MTU discovery */ 304 u_long tcps_listendrop; /* listen queue overflows */ 305 }; 306 307 /* 308 * TCB structure exported to user-land via sysctl(3). 309 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 310 * included. Not all of our clients do. 311 */ 312 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 313 struct xtcpcb { 314 size_t xt_len; 315 struct inpcb xt_inp; 316 struct tcpcb xt_tp; 317 struct xsocket xt_socket; 318 u_quad_t xt_alignment_hack; 319 }; 320 #endif 321 322 /* 323 * Names for TCP sysctl objects 324 */ 325 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 326 #define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */ 327 #define TCPCTL_MSSDFLT 3 /* MSS default */ 328 #define TCPCTL_STATS 4 /* statistics (read-only) */ 329 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 330 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 331 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 332 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 333 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 334 #define TCPCTL_KEEPINIT 10 /* receive buffer space */ 335 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 336 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 337 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 338 #define TCPCTL_MAXID 14 339 340 #define TCPCTL_NAMES { \ 341 { 0, 0 }, \ 342 { "rfc1323", CTLTYPE_INT }, \ 343 { "rfc1644", CTLTYPE_INT }, \ 344 { "mssdflt", CTLTYPE_INT }, \ 345 { "stats", CTLTYPE_STRUCT }, \ 346 { "rttdflt", CTLTYPE_INT }, \ 347 { "keepidle", CTLTYPE_INT }, \ 348 { "keepintvl", CTLTYPE_INT }, \ 349 { "sendspace", CTLTYPE_INT }, \ 350 { "recvspace", CTLTYPE_INT }, \ 351 { "keepinit", CTLTYPE_INT }, \ 352 { "pcblist", CTLTYPE_STRUCT }, \ 353 { "delacktime", CTLTYPE_INT }, \ 354 { "v6mssdflt", CTLTYPE_INT }, \ 355 } 356 357 358 #ifdef _KERNEL 359 #ifdef SYSCTL_DECL 360 SYSCTL_DECL(_net_inet_tcp); 361 #endif 362 363 extern struct inpcbhead tcb; /* head of queue of active tcpcb's */ 364 extern struct inpcbinfo tcbinfo; 365 extern struct tcpstat tcpstat; /* tcp statistics */ 366 extern int tcp_mssdflt; /* XXX */ 367 extern int tcp_delack_enabled; 368 extern int ss_fltsz; 369 extern int ss_fltsz_local; 370 371 void tcp_canceltimers __P((struct tcpcb *)); 372 struct tcpcb * 373 tcp_close __P((struct tcpcb *)); 374 void tcp_ctlinput __P((int, struct sockaddr *, void *)); 375 int tcp_ctloutput __P((struct socket *, struct sockopt *)); 376 struct tcpcb * 377 tcp_drop __P((struct tcpcb *, int)); 378 void tcp_drain __P((void)); 379 void tcp_fasttimo __P((void)); 380 struct rmxp_tao * 381 tcp_gettaocache __P((struct inpcb *)); 382 void tcp_init __P((void)); 383 void tcp_input __P((struct mbuf *, int, int)); 384 void tcp_mss __P((struct tcpcb *, int)); 385 int tcp_mssopt __P((struct tcpcb *)); 386 void tcp_mtudisc __P((struct inpcb *, int)); 387 struct tcpcb * 388 tcp_newtcpcb __P((struct inpcb *)); 389 int tcp_output __P((struct tcpcb *)); 390 void tcp_quench __P((struct inpcb *, int)); 391 void tcp_respond __P((struct tcpcb *, void *, 392 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int)); 393 struct rtentry * 394 tcp_rtlookup __P((struct inpcb *)); 395 void tcp_setpersist __P((struct tcpcb *)); 396 void tcp_slowtimo __P((void)); 397 struct tcptemp * 398 tcp_template __P((struct tcpcb *)); 399 struct tcpcb * 400 tcp_timers __P((struct tcpcb *, int)); 401 void tcp_trace __P((int, int, struct tcpcb *, void *, struct tcphdr *, 402 int)); 403 404 extern struct pr_usrreqs tcp_usrreqs; 405 extern u_long tcp_sendspace; 406 extern u_long tcp_recvspace; 407 408 #endif /* _KERNEL */ 409 410 #endif /* _NETINET_TCP_VAR_H_ */ 411