]> git.saurik.com Git - apple/xnu.git/blob - bsd/netinet/tcp_var.h
80bf16e9aac51ed963df1226c0645b8c988e26cf
[apple/xnu.git] / bsd / netinet / tcp_var.h
1 /*
2 * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
3 *
4 * @APPLE_LICENSE_HEADER_START@
5 *
6 * The contents of this file constitute Original Code as defined in and
7 * are subject to the Apple Public Source License Version 1.1 (the
8 * "License"). You may not use this file except in compliance with the
9 * License. Please obtain a copy of the License at
10 * http://www.apple.com/publicsource and read it before using this file.
11 *
12 * This Original Code and all software distributed under the License are
13 * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
14 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
15 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT. Please see the
17 * License for the specific language governing rights and limitations
18 * under the License.
19 *
20 * @APPLE_LICENSE_HEADER_END@
21 */
22 /*
23 * Copyright (c) 1982, 1986, 1993, 1994, 1995
24 * The Regents of the University of California. All rights reserved.
25 *
26 * Redistribution and use in source and binary forms, with or without
27 * modification, are permitted provided that the following conditions
28 * are met:
29 * 1. Redistributions of source code must retain the above copyright
30 * notice, this list of conditions and the following disclaimer.
31 * 2. Redistributions in binary form must reproduce the above copyright
32 * notice, this list of conditions and the following disclaimer in the
33 * documentation and/or other materials provided with the distribution.
34 * 3. All advertising materials mentioning features or use of this software
35 * must display the following acknowledgement:
36 * This product includes software developed by the University of
37 * California, Berkeley and its contributors.
38 * 4. Neither the name of the University nor the names of its contributors
39 * may be used to endorse or promote products derived from this software
40 * without specific prior written permission.
41 *
42 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
43 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
44 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
45 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
46 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
47 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
48 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
49 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
50 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
51 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
52 * SUCH DAMAGE.
53 *
54 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
55 * $FreeBSD: src/sys/netinet/tcp_var.h,v 1.56.2.8 2001/08/22 00:59:13 silby Exp $
56 */
57
58 #ifndef _NETINET_TCP_VAR_H_
59 #define _NETINET_TCP_VAR_H_
60 #include <sys/appleapiopts.h>
61 #include <netinet/tcp_timer.h>
62 #ifdef __APPLE_API_PRIVATE
63
64 #define N_TIME_WAIT_SLOTS 128 /* must be power of 2 */
65
66 /*
67 * Kernel variables for tcp.
68 */
69
70 /* TCP segment queue entry */
71 struct tseg_qent {
72 LIST_ENTRY(tseg_qent) tqe_q;
73 int tqe_len; /* TCP segment data length */
74 struct tcphdr *tqe_th; /* a pointer to tcp header */
75 struct mbuf *tqe_m; /* mbuf contains packet */
76 };
77 LIST_HEAD(tsegqe_head, tseg_qent);
78 #ifdef MALLOC_DECLARE
79 MALLOC_DECLARE(M_TSEGQ);
80 #endif
81
82 struct tcptemp {
83 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
84 struct tcphdr tt_t;
85 };
86
87 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */
88
89 #ifdef __APPLE__
90 #define TCP_DELACK_BITSET(hash_elem)\
91 delack_bitmask[((hash_elem) >> 5)] |= 1 << ((hash_elem) & 0x1F)
92
93 #define DELACK_BITMASK_ON 1
94 #define DELACK_BITMASK_THRESH 300
95 #endif
96
97 /*
98 * Tcp control block, one per tcp; fields:
99 * Organized for 16 byte cacheline efficiency.
100 */
101 struct tcpcb {
102 struct tsegqe_head t_segq;
103 int t_dupacks; /* consecutive dup acks recd */
104 struct tcptemp *unused; /* unused now: was t_template */
105
106 int t_timer[TCPT_NTIMERS]; /* tcp timers */
107
108 struct inpcb *t_inpcb; /* back pointer to internet pcb */
109 int t_state; /* state of this connection */
110 u_int t_flags;
111 #define TF_ACKNOW 0x00001 /* ack peer immediately */
112 #define TF_DELACK 0x00002 /* ack, but try to delay it */
113 #define TF_NODELAY 0x00004 /* don't delay packets to coalesce */
114 #define TF_NOOPT 0x00008 /* don't use tcp options */
115 #define TF_SENTFIN 0x00010 /* have sent FIN */
116 #define TF_REQ_SCALE 0x00020 /* have/will request window scaling */
117 #define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */
118 #define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */
119 #define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */
120 #define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */
121 #define TF_NEEDSYN 0x00400 /* send SYN (implicit state) */
122 #define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */
123 #define TF_NOPUSH 0x01000 /* don't push */
124 #define TF_REQ_CC 0x02000 /* have/will request CC */
125 #define TF_RCVD_CC 0x04000 /* a CC was received in SYN */
126 #define TF_SENDCCNEW 0x08000 /* send CCnew instead of CC in SYN */
127 #define TF_MORETOCOME 0x10000 /* More data to be appended to sock */
128 #define TF_LQ_OVERFLOW 0x20000 /* listen queue overflow */
129 #define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */
130 #define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */
131
132 int t_force; /* 1 if forcing out a byte */
133
134 tcp_seq snd_una; /* send unacknowledged */
135 tcp_seq snd_max; /* highest sequence number sent;
136 * used to recognize retransmits
137 */
138 tcp_seq snd_nxt; /* send next */
139 tcp_seq snd_up; /* send urgent pointer */
140
141 tcp_seq snd_wl1; /* window update seg seq number */
142 tcp_seq snd_wl2; /* window update seg ack number */
143 tcp_seq iss; /* initial send sequence number */
144 tcp_seq irs; /* initial receive sequence number */
145
146 tcp_seq rcv_nxt; /* receive next */
147 tcp_seq rcv_adv; /* advertised window */
148 u_long rcv_wnd; /* receive window */
149 tcp_seq rcv_up; /* receive urgent pointer */
150
151 u_long snd_wnd; /* send window */
152 u_long snd_cwnd; /* congestion-controlled window */
153 u_long snd_ssthresh; /* snd_cwnd size threshold for
154 * for slow start exponential to
155 * linear switch
156 */
157 u_int t_maxopd; /* mss plus options */
158
159 u_long t_rcvtime; /* inactivity time */
160 u_long t_starttime; /* time connection was established */
161 int t_rtttime; /* round trip time */
162 tcp_seq t_rtseq; /* sequence number being timed */
163
164 int t_rxtcur; /* current retransmit value (ticks) */
165 u_int t_maxseg; /* maximum segment size */
166 int t_srtt; /* smoothed round-trip time */
167 int t_rttvar; /* variance in round-trip time */
168
169 int t_rxtshift; /* log(2) of rexmt exp. backoff */
170 u_int t_rttmin; /* minimum rtt allowed */
171 u_long t_rttupdated; /* number of times rtt sampled */
172 u_long max_sndwnd; /* largest window peer has offered */
173
174 int t_softerror; /* possible error not yet reported */
175 /* out-of-band data */
176 char t_oobflags; /* have some */
177 char t_iobc; /* input character */
178 #define TCPOOB_HAVEDATA 0x01
179 #define TCPOOB_HADDATA 0x02
180 /* RFC 1323 variables */
181 u_char snd_scale; /* window scaling for send window */
182 u_char rcv_scale; /* window scaling for recv window */
183 u_char request_r_scale; /* pending window scaling */
184 u_char requested_s_scale;
185 u_long ts_recent; /* timestamp echo data */
186
187 u_long ts_recent_age; /* when last updated */
188 tcp_seq last_ack_sent;
189 /* RFC 1644 variables */
190 tcp_cc cc_send; /* send connection count */
191 tcp_cc cc_recv; /* receive connection count */
192 tcp_seq snd_recover; /* for use in fast recovery */
193 /* experimental */
194 u_long snd_cwnd_prev; /* cwnd prior to retransmit */
195 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */
196 u_long t_badrxtwin; /* window for retransmit recovery */
197 };
198
199 /*
200 * Structure to hold TCP options that are only used during segment
201 * processing (in tcp_input), but not held in the tcpcb.
202 * It's basically used to reduce the number of parameters
203 * to tcp_dooptions.
204 */
205 struct tcpopt {
206 u_long to_flag; /* which options are present */
207 #define TOF_TS 0x0001 /* timestamp */
208 #define TOF_CC 0x0002 /* CC and CCnew are exclusive */
209 #define TOF_CCNEW 0x0004
210 #define TOF_CCECHO 0x0008
211 u_long to_tsval;
212 u_long to_tsecr;
213 tcp_cc to_cc; /* holds CC or CCnew */
214 tcp_cc to_ccecho;
215 u_short reserved; /* unused now: was to_maxseg */
216 };
217
218 /*
219 * The TAO cache entry which is stored in the protocol family specific
220 * portion of the route metrics.
221 */
222 struct rmxp_tao {
223 tcp_cc tao_cc; /* latest CC in valid SYN */
224 tcp_cc tao_ccsent; /* latest CC sent to peer */
225 u_short tao_mssopt; /* peer's cached MSS */
226 #ifdef notyet
227 u_short tao_flags; /* cache status flags */
228 #define TAOF_DONT 0x0001 /* peer doesn't understand rfc1644 */
229 #define TAOF_OK 0x0002 /* peer does understand rfc1644 */
230 #define TAOF_UNDEF 0 /* we don't know yet */
231 #endif /* notyet */
232 };
233 #define rmx_taop(r) ((struct rmxp_tao *)(r).rmx_filler)
234
235 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
236 #define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
237
238 /*
239 * The smoothed round-trip time and estimated variance
240 * are stored as fixed point numbers scaled by the values below.
241 * For convenience, these scales are also used in smoothing the average
242 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
243 * With these scales, srtt has 3 bits to the right of the binary point,
244 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
245 * binary point, and is smoothed with an ALPHA of 0.75.
246 */
247 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
248 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */
249 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */
250 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */
251 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
252
253 /*
254 * The initial retransmission should happen at rtt + 4 * rttvar.
255 * Because of the way we do the smoothing, srtt and rttvar
256 * will each average +1/2 tick of bias. When we compute
257 * the retransmit timer, we want 1/2 tick of rounding and
258 * 1 extra tick because of +-1/2 tick uncertainty in the
259 * firing of the timer. The bias will give us exactly the
260 * 1.5 tick we need. But, because the bias is
261 * statistical, we have to test that we don't drop below
262 * the minimum feasible timer (which is 2 ticks).
263 * This version of the macro adapted from a paper by Lawrence
264 * Brakmo and Larry Peterson which outlines a problem caused
265 * by insufficient precision in the original implementation,
266 * which results in inappropriately large RTO values for very
267 * fast networks.
268 */
269 #define TCP_REXMTVAL(tp) \
270 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
271 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
272 #endif /* __APPLE_API_PRIVATE */
273
274 #ifdef __APPLE_API_UNSTABLE
275 /*
276 * TCP statistics.
277 * Many of these should be kept per connection,
278 * but that's inconvenient at the moment.
279 */
280 struct tcpstat {
281 u_long tcps_connattempt; /* connections initiated */
282 u_long tcps_accepts; /* connections accepted */
283 u_long tcps_connects; /* connections established */
284 u_long tcps_drops; /* connections dropped */
285 u_long tcps_conndrops; /* embryonic connections dropped */
286 u_long tcps_closed; /* conn. closed (includes drops) */
287 u_long tcps_segstimed; /* segs where we tried to get rtt */
288 u_long tcps_rttupdated; /* times we succeeded */
289 u_long tcps_delack; /* delayed acks sent */
290 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
291 u_long tcps_rexmttimeo; /* retransmit timeouts */
292 u_long tcps_persisttimeo; /* persist timeouts */
293 u_long tcps_keeptimeo; /* keepalive timeouts */
294 u_long tcps_keepprobe; /* keepalive probes sent */
295 u_long tcps_keepdrops; /* connections dropped in keepalive */
296
297 u_long tcps_sndtotal; /* total packets sent */
298 u_long tcps_sndpack; /* data packets sent */
299 u_long tcps_sndbyte; /* data bytes sent */
300 u_long tcps_sndrexmitpack; /* data packets retransmitted */
301 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */
302 u_long tcps_sndacks; /* ack-only packets sent */
303 u_long tcps_sndprobe; /* window probes sent */
304 u_long tcps_sndurg; /* packets sent with URG only */
305 u_long tcps_sndwinup; /* window update-only packets sent */
306 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
307
308 u_long tcps_rcvtotal; /* total packets received */
309 u_long tcps_rcvpack; /* packets received in sequence */
310 u_long tcps_rcvbyte; /* bytes received in sequence */
311 u_long tcps_rcvbadsum; /* packets received with ccksum errs */
312 u_long tcps_rcvbadoff; /* packets received with bad offset */
313 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */
314 u_long tcps_rcvshort; /* packets received too short */
315 u_long tcps_rcvduppack; /* duplicate-only packets received */
316 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */
317 u_long tcps_rcvpartduppack; /* packets with some duplicate data */
318 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
319 u_long tcps_rcvoopack; /* out-of-order packets received */
320 u_long tcps_rcvoobyte; /* out-of-order bytes received */
321 u_long tcps_rcvpackafterwin; /* packets with data after window */
322 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */
323 u_long tcps_rcvafterclose; /* packets rcvd after "close" */
324 u_long tcps_rcvwinprobe; /* rcvd window probe packets */
325 u_long tcps_rcvdupack; /* rcvd duplicate acks */
326 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */
327 u_long tcps_rcvackpack; /* rcvd ack packets */
328 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */
329 u_long tcps_rcvwinupd; /* rcvd window update packets */
330 u_long tcps_pawsdrop; /* segments dropped due to PAWS */
331 u_long tcps_predack; /* times hdr predict ok for acks */
332 u_long tcps_preddat; /* times hdr predict ok for data pkts */
333 u_long tcps_pcbcachemiss;
334 u_long tcps_cachedrtt; /* times cached RTT in route updated */
335 u_long tcps_cachedrttvar; /* times cached rttvar updated */
336 u_long tcps_cachedssthresh; /* times cached ssthresh updated */
337 u_long tcps_usedrtt; /* times RTT initialized from route */
338 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */
339 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/
340 u_long tcps_persistdrop; /* timeout in persist state */
341 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */
342 u_long tcps_mturesent; /* resends due to MTU discovery */
343 u_long tcps_listendrop; /* listen queue overflows */
344 };
345 #endif /* __APPLE_API_UNSTABLE */
346
347 /*
348 * TCB structure exported to user-land via sysctl(3).
349 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
350 * included. Not all of our clients do.
351 */
352 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
353 struct xtcpcb {
354 size_t xt_len;
355 struct inpcb xt_inp;
356 struct tcpcb xt_tp;
357 struct xsocket xt_socket;
358 u_quad_t xt_alignment_hack;
359 };
360 #endif
361
362 /*
363 * Names for TCP sysctl objects
364 */
365 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
366 #define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */
367 #define TCPCTL_MSSDFLT 3 /* MSS default */
368 #define TCPCTL_STATS 4 /* statistics (read-only) */
369 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */
370 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
371 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
372 #define TCPCTL_SENDSPACE 8 /* send buffer space */
373 #define TCPCTL_RECVSPACE 9 /* receive buffer space */
374 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
375 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
376 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
377 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
378 #define TCPCTL_MAXID 14
379
380 #define TCPCTL_NAMES { \
381 { 0, 0 }, \
382 { "rfc1323", CTLTYPE_INT }, \
383 { "rfc1644", CTLTYPE_INT }, \
384 { "mssdflt", CTLTYPE_INT }, \
385 { "stats", CTLTYPE_STRUCT }, \
386 { "rttdflt", CTLTYPE_INT }, \
387 { "keepidle", CTLTYPE_INT }, \
388 { "keepintvl", CTLTYPE_INT }, \
389 { "sendspace", CTLTYPE_INT }, \
390 { "recvspace", CTLTYPE_INT }, \
391 { "keepinit", CTLTYPE_INT }, \
392 { "pcblist", CTLTYPE_STRUCT }, \
393 { "delacktime", CTLTYPE_INT }, \
394 { "v6mssdflt", CTLTYPE_INT }, \
395 }
396
397 #ifdef __APPLE_API_PRIVATE
398 #ifdef KERNEL
399 #ifdef SYSCTL_DECL
400 SYSCTL_DECL(_net_inet_tcp);
401 #endif
402
403 extern struct inpcbhead tcb; /* head of queue of active tcpcb's */
404 extern struct inpcbinfo tcbinfo;
405 extern struct tcpstat tcpstat; /* tcp statistics */
406 extern int tcp_mssdflt; /* XXX */
407 extern int tcp_delack_enabled;
408 extern int tcp_do_newreno;
409 extern int ss_fltsz;
410 extern int ss_fltsz_local;
411 #ifdef __APPLE__
412 extern u_long tcp_now; /* for RFC 1323 timestamps */
413 extern int tcp_delack_enabled;
414 #endif
415
416
417 void tcp_canceltimers __P((struct tcpcb *));
418 struct tcpcb *
419 tcp_close __P((struct tcpcb *));
420 void tcp_ctlinput __P((int, struct sockaddr *, void *));
421 int tcp_ctloutput __P((struct socket *, struct sockopt *));
422 struct tcpcb *
423 tcp_drop __P((struct tcpcb *, int));
424 void tcp_drain __P((void));
425 void tcp_fasttimo __P((void));
426 struct rmxp_tao *
427 tcp_gettaocache __P((struct inpcb *));
428 void tcp_init __P((void));
429 void tcp_input __P((struct mbuf *, int));
430 void tcp_mss __P((struct tcpcb *, int));
431 int tcp_mssopt __P((struct tcpcb *));
432 void tcp_drop_syn_sent __P((struct inpcb *, int));
433 void tcp_mtudisc __P((struct inpcb *, int));
434 struct tcpcb *
435 tcp_newtcpcb __P((struct inpcb *));
436 int tcp_output __P((struct tcpcb *));
437 void tcp_quench __P((struct inpcb *, int));
438 void tcp_respond __P((struct tcpcb *, void *,
439 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int));
440 struct rtentry *
441 tcp_rtlookup __P((struct inpcb *));
442 void tcp_setpersist __P((struct tcpcb *));
443 void tcp_slowtimo __P((void));
444 struct tcptemp *
445 tcp_maketemplate __P((struct tcpcb *));
446 void tcp_fillheaders __P((struct tcpcb *, void *, void *));
447 struct tcpcb *
448 tcp_timers __P((struct tcpcb *, int));
449 void tcp_trace __P((int, int, struct tcpcb *, void *, struct tcphdr *,
450 int));
451
452 extern struct pr_usrreqs tcp_usrreqs;
453 extern u_long tcp_sendspace;
454 extern u_long tcp_recvspace;
455 tcp_seq tcp_new_isn __P((struct tcpcb *));
456
457 #endif /* KERNEL */
458 #endif /* __APPLE_API_PRIVATE */
459
460 #endif /* _NETINET_TCP_VAR_H_ */