1/*
2 * Copyright (c) 2000-2022 Apple Inc. All rights reserved.
3 *
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
5 *
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
14 *
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
17 *
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
25 *
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27 */
28/*
29 * Copyright (c) 1982, 1986, 1993, 1994, 1995
30 * The Regents of the University of California. All rights reserved.
31 *
32 * Redistribution and use in source and binary forms, with or without
33 * modification, are permitted provided that the following conditions
34 * are met:
35 * 1. Redistributions of source code must retain the above copyright
36 * notice, this list of conditions and the following disclaimer.
37 * 2. Redistributions in binary form must reproduce the above copyright
38 * notice, this list of conditions and the following disclaimer in the
39 * documentation and/or other materials provided with the distribution.
40 * 3. All advertising materials mentioning features or use of this software
41 * must display the following acknowledgement:
42 * This product includes software developed by the University of
43 * California, Berkeley and its contributors.
44 * 4. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
47 *
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58 * SUCH DAMAGE.
59 *
60 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
61 * $FreeBSD: src/sys/netinet/tcp_var.h,v 1.56.2.8 2001/08/22 00:59:13 silby Exp $
62 */
63
64#ifndef _NETINET_TCP_VAR_H_
65#define _NETINET_TCP_VAR_H_
66#include <sys/types.h>
67#include <sys/appleapiopts.h>
68#include <sys/queue.h>
69#include <netinet/in_pcb.h>
70#include <netinet/tcp.h>
71#include <netinet/tcp_timer.h>
72#if !KERNEL
73#include <TargetConditionals.h>
74#endif
75
76#if defined(__LP64__)
77#define _TCPCB_PTR(x) u_int32_t
78#define _TCPCB_LIST_HEAD(name, type) \
79struct name { \
80 u_int32_t lh_first; \
81}
82#else
83#define _TCPCB_PTR(x) x
84#define _TCPCB_LIST_HEAD(name, type) LIST_HEAD(name, type)
85#endif
86
87#ifdef KERNEL_PRIVATE
88
89#define TCP_RETRANSHZ 1000 /* granularity of TCP timestamps, 1ms */
90/* Minimum time quantum within which the timers are coalesced */
91#define TCP_TIMER_10MS_QUANTUM (TCP_RETRANSHZ/100) /* every 10ms */
92#define TCP_TIMER_100MS_QUANTUM (TCP_RETRANSHZ/10) /* every 100ms */
93#define TCP_TIMER_500MS_QUANTUM (TCP_RETRANSHZ/2) /* every 500ms */
94
95#define TCP_RETRANSHZ_TO_USEC 1000
96
97#define N_TIME_WAIT_SLOTS 128 /* must be power of 2 */
98
99/* Always allow at least 16 packets worth of recv window when adjusting
100 * recv window using inter-packet arrival jitter.
101 */
102#define MIN_IAJ_WIN 16
103
104/* A variation in delay of this many milliseconds is tolerable. This limit has to
105 * be low but greater than zero. We also use standard deviation on jitter to adjust
106 * this limit for different link and connection types.
107 */
108#define ALLOWED_IAJ 5
109
110/* Ignore the first few packets on a connection until the ACK clock gets going
111 */
112#define IAJ_IGNORE_PKTCNT 40
113
114/* Let the accumulated IAJ value increase by this threshold at most. This limit
115 * will control how many ALLOWED_IAJ measurements a receiver will have to see
116 * before opening the receive window
117 */
118#define ACC_IAJ_HIGH_THRESH 100
119
120/* When accumulated IAJ reaches this value, the receiver starts to react by
121 * closing the window
122 */
123#define ACC_IAJ_REACT_LIMIT 200
124
125/* If the number of small packets (smaller than IAJ packet size) seen on a
126 * connection is more than this threshold, reset the size and learn it again.
127 * This is needed because the sender might send smaller segments after PMTU
128 * discovery and the receiver has to learn the new size.
129 */
130#define RESET_IAJ_SIZE_THRESH 20
131
132/*
133 * Adaptive timeout is a read/write timeout specified by the application to
134 * get a socket event when the transport layer detects a stall in data
135 * transfer. The value specified is the number of probes that can be sent
136 * to the peer before generating an event. Since it is not specified as
137 * a time value, the timeout will adjust based on the RTT seen on the link.
138 * The timeout will start only when there is an indication that the read/write
139 * operation is not making progress.
140 *
141 * If a write operation stalls, the probe will be retransmission of data.
142 * If a read operation stalls, the probe will be a keep-alive packet.
143 *
144 * The maximum value of adaptive timeout is set to 10 which will allow
145 * transmission of enough number of probes to the peer.
146 */
147#define TCP_ADAPTIVE_TIMEOUT_MAX 10
148
149#define TCP_CONNECTIVITY_PROBES_MAX 5
150
151/* 3 bit mask used for Accurate ECN ACE field */
152#define TCP_ACE_MASK (0x7)
153/* Divisor used for Accurate ECN ACE field */
154#define TCP_ACE_DIV (1 << 3)
155
156/* 24 bit mask used for Accurate ECN option */
157#define TCP_ACO_MASK (0xFFFFFF)
158
159/*
160 * Kernel variables for tcp.
161 */
162
163/* TCP segment queue entry */
164struct tseg_qent {
165 LIST_ENTRY(tseg_qent) tqe_q;
166 int tqe_len; /* TCP segment data length */
167 struct tcphdr *tqe_th; /* a pointer to tcp header */
168 struct mbuf *tqe_m; /* mbuf contains packet */
169};
170LIST_HEAD(tsegqe_head, tseg_qent);
171
172struct sackblk {
173 tcp_seq start; /* start seq no. of sack block */
174 tcp_seq end; /* end seq no. */
175};
176
177struct sackhole {
178 tcp_seq start; /* start seq no. of hole */
179 tcp_seq end; /* end seq no. */
180 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */
181 u_int32_t rxmit_start; /* timestamp of first retransmission */
182 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */
183};
184
185struct sackhint {
186 struct sackhole *nexthole;
187 int sack_bytes_rexmit;
188 int sack_bytes_acked;
189};
190
191struct tcp_rxt_seg {
192 tcp_seq rx_start;
193 tcp_seq rx_end;
194 u_int16_t rx_count;
195 u_int16_t rx_flags;
196#define TCP_RXT_SPURIOUS 0x1 /* received DSACK notification */
197 SLIST_ENTRY(tcp_rxt_seg) rx_link;
198};
199
200struct tcp_notify_ack_marker {
201 tcp_seq notify_snd_una; /* Notify when snd_una crosses this seq */
202 tcp_notify_ack_id_t notify_id;
203 SLIST_ENTRY(tcp_notify_ack_marker) notify_next;
204};
205
206struct tcptemp {
207 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
208 struct tcphdr tt_t;
209};
210
211struct bwmeas {
212 tcp_seq bw_start; /* start of bw measurement */
213 uint32_t bw_ts; /* timestamp when bw measurement started */
214 uint32_t bw_size; /* burst size in bytes for this bw measurement */
215 uint32_t bw_minsizepkts; /* Min burst size as segments */
216 uint32_t bw_maxsizepkts; /* Max burst size as segments */
217 uint32_t bw_minsize; /* Min size in bytes */
218 uint32_t bw_maxsize; /* Max size in bytes */
219 uint32_t bw_sndbw; /* Measured send bandwidth */
220 uint32_t bw_sndbw_max; /* Max measured bandwidth */
221 uint32_t bw_rcvbw_max; /* Max receive bandwidth measured */
222};
223
224/* MPTCP Data sequence map entry */
225struct mpt_dsn_map {
226 uint64_t mpt_dsn; /* data seq num recvd */
227 uint32_t mpt_sseq; /* relative subflow # */
228 uint16_t mpt_len; /* length of mapping */
229 uint16_t mpt_csum; /* checksum value if on */
230 uint8_t mpt_dfin; /* It's a DATA_FIN */
231};
232#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */
233
234struct tcp_ccstate {
235 union {
236 struct tcp_cubic_state {
237 u_int32_t tc_last_max; /* cwnd at last loss */
238 u_int32_t tc_epoch_start; /* TS of last loss */
239 u_int32_t tc_origin_point; /* window at the start of an epoch */
240 u_int32_t tc_tcp_win; /* computed tcp win */
241 u_int32_t tc_tcp_bytes_acked; /* bytes acked */
242 u_int32_t tc_avg_lastmax; /* Average of last max */
243 u_int32_t tc_mean_deviation; /* Mean absolute deviation */
244 float tc_epoch_period; /* K parameter */
245 } _cubic_state_;
246#define cub_last_max __u__._cubic_state_.tc_last_max
247#define cub_epoch_start __u__._cubic_state_.tc_epoch_start
248#define cub_origin_point __u__._cubic_state_.tc_origin_point
249#define cub_tcp_win __u__._cubic_state_.tc_tcp_win
250#define cub_tcp_bytes_acked __u__._cubic_state_.tc_tcp_bytes_acked
251#define cub_epoch_period __u__._cubic_state_.tc_epoch_period
252#define cub_avg_lastmax __u__._cubic_state_.tc_avg_lastmax
253#define cub_mean_dev __u__._cubic_state_.tc_mean_deviation
254 struct tcp_ledbat_state {
255 uint32_t num_slowdown_events;
256 uint32_t slowdown_ts;
257 uint32_t slowdown_begin;
258 uint32_t md_bytes_acked;
259 } _ledbat_state_;
260#define ledbat_slowdown_events __u__._ledbat_state_.num_slowdown_events
261#define ledbat_slowdown_ts __u__._ledbat_state_.slowdown_ts
262#define ledbat_slowdown_begin __u__._ledbat_state_.slowdown_begin
263#define ledbat_md_bytes_acked __u__._ledbat_state_.md_bytes_acked
264 } __u__;
265};
266
267struct tcp_rledbat_state {
268 uint32_t num_slowdown_events; /* Number of slowdown events until reset */
269 uint32_t slowdown_ts; /* Next slowdown timestamp */
270 uint32_t slowdown_begin; /* Slowdown begin time */
271 uint32_t reduction_end; /* Used to track rl_win reduction once per RTT */
272 uint32_t rcvd_bytes; /* bytes received used for byte counting */
273 uint32_t md_rcvd_bytes; /* bytes received during MD used for byte counting */
274 uint32_t win; /* receive Ledbat window size */
275 uint32_t ssthresh; /* receive Ledbat ssthresh */
276 uint32_t drained_bytes; /* bytes drained from the flight-size */
277 uint32_t win_ws; /* receive Ledbat window after avoiding window shrinking */
278};
279
280/*
281 * Tcp control block, one per tcp; fields:
282 * Organized for 16 byte cacheline efficiency.
283 */
284struct tcpcb {
285 struct tsegqe_head t_segq;
286 uint32_t t_dupacks; /* consecutive dup acks recd */
287 int t_state; /* state of this connection */
288 uint32_t t_timer[TCPT_NTIMERS]; /* tcp timers */
289 struct tcptimerentry tentry; /* entry in timer list */
290
291 struct inpcb *t_inpcb; /* back pointer to internet pcb */
292 uint32_t t_flags;
293#define TF_ACKNOW 0x00001 /* ack peer immediately */
294#define TF_DELACK 0x00002 /* ack, but try to delay it */
295#define TF_NODELAY 0x00004 /* don't delay packets to coalesce */
296#define TF_NOOPT 0x00008 /* don't use tcp options */
297#define TF_SENTFIN 0x00010 /* have sent FIN */
298#define TF_REQ_SCALE 0x00020 /* have/will request window scaling */
299#define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */
300#define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */
301#define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */
302#define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */
303#define TF_NEEDSYN 0x00400 /* send SYN (implicit state) - unused but needed for backwards compatibility */
304#define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */
305#define TF_NOPUSH 0x01000 /* don't push */
306#define TF_REQ_CC 0x02000 /* have/will request CC */
307#define TF_RCVD_CC 0x04000 /* a CC was received in SYN */
308#define TF_SENDCCNEW 0x08000 /* Unused */
309#define TF_MORETOCOME 0x10000 /* More data to be appended to sock */
310#define TF_LOCAL 0x20000 /* connection to a host on local link */
311#define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */
312#define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */
313#define TF_LASTIDLE 0x100000 /* connection was previously idle */
314#define TF_FASTRECOVERY 0x200000 /* in NewReno Fast Recovery */
315#define TF_WASFRECOVERY 0x400000 /* was in NewReno Fast Recovery */
316#define TF_SIGNATURE 0x800000 /* require MD5 digests (RFC2385) */
317#define TF_MAXSEGSNT 0x1000000 /* last segment sent was a full segment */
318#define TF_STREAMING_ON 0x2000000 /* Receiver detected streaming */
319#define TF_PMTUD 0x4000000 /* Perform Path MTU Discovery for this connection */
320#define TF_CLOSING 0x8000000 /* pending tcp close */
321#define TF_TSO 0x10000000 /* TCP Segment Offloading is enable on this connection */
322#define TF_BLACKHOLE 0x20000000 /* Path MTU Discovery Black Hole detection */
323#define TF_TIMER_ONLIST 0x40000000 /* pcb is on tcp_timer_list */
324#define TF_STRETCHACK 0x80000000 /* receiver is going to delay acks */
325
326 tcp_seq snd_una; /* send unacknowledged */
327 tcp_seq snd_max; /* highest sequence number sent;
328 * used to recognize retransmits
329 */
330 tcp_seq snd_nxt; /* send next */
331 tcp_seq snd_up; /* send urgent pointer */
332
333 tcp_seq snd_wl1; /* window update seg seq number */
334 tcp_seq snd_wl2; /* window update seg ack number */
335 tcp_seq iss; /* initial send sequence number */
336 tcp_seq irs; /* initial receive sequence number */
337
338 tcp_seq rcv_nxt; /* receive next */
339 tcp_seq rcv_adv; /* advertised window */
340 uint32_t rcv_wnd; /* receive window */
341 uint32_t t_last_recwin;
342 tcp_seq rcv_up; /* receive urgent pointer */
343
344 uint32_t snd_wnd; /* send window */
345 uint32_t snd_cwnd; /* congestion-controlled window */
346 uint32_t snd_ssthresh; /* snd_cwnd size threshold for
347 * for slow start exponential to
348 * linear switch
349 */
350 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */
351
352 uint32_t t_maxopd; /* mss plus options */
353 uint32_t t_rcvtime; /* time at which a packet was received */
354 uint32_t t_sndtime; /* time at which we last sent new data */
355 uint32_t t_starttime; /* time connection was established */
356 int t_rtttime; /* tcp clock when rtt calculation was started */
357 tcp_seq t_rtseq; /* sequence number being timed */
358
359 uint32_t rfbuf_ts; /* recv buffer autoscaling timestamp */
360 uint32_t rfbuf_cnt; /* recv buffer autoscaling byte count */
361 uint32_t rfbuf_space; /* Current "ideal" estimate of the space */
362
363 int t_rxtcur; /* current retransmit value (ticks) */
364 unsigned int t_maxseg; /* maximum segment size */
365 int t_srtt; /* smoothed round-trip time */
366 int t_rttvar; /* variance in round-trip time */
367
368 uint64_t t_accsleep_ms; /* accumulated sleep time since last boot */
369 uint32_t t_reassqlen; /* length of reassembly queue */
370 uint32_t t_reassq_mbcnt; /* amount in bytes of mbuf space used */
371 uint16_t t_rxtshift; /* log(2) of rexmt exp. backoff */
372 uint32_t t_rttmin; /* minimum rtt allowed */
373 uint32_t t_rttbest; /* best rtt we've seen */
374 uint32_t t_rttcur; /* most recent value of rtt */
375 uint32_t t_rttupdated; /* number of times rtt sampled */
376 uint32_t t_rxt_conndroptime; /* retxmt conn gets dropped after this time, when set */
377 uint32_t t_rxtstart; /* time at which retransmission started */
378 uint32_t max_sndwnd; /* largest window peer has offered */
379
380 int t_softerror; /* possible error not yet reported */
381/* out-of-band data */
382 char t_oobflags; /* have some */
383 char t_iobc; /* input character */
384#define TCPOOB_HAVEDATA 0x01
385#define TCPOOB_HADDATA 0x02
386/* RFC 1323 variables */
387 u_int8_t snd_scale; /* window scaling for send window */
388 u_int8_t rcv_scale; /* window scaling for recv window */
389 u_int8_t request_r_scale; /* pending window scaling */
390 u_int8_t requested_s_scale;
391 u_int8_t tcp_cc_index; /* index of congestion control algorithm */
392 u_int8_t t_adaptive_rtimo; /* Read timeout used as a multiple of RTT */
393 u_int8_t t_adaptive_wtimo; /* Write timeout used as a multiple of RTT */
394 u_int8_t t_stretchack_delayed; /* stretch ack delayed */
395
396/* State for limiting early retransmits when SACK is not enabled */
397 u_int16_t t_early_rexmt_count; /* count of early rexmts */
398 u_int32_t t_early_rexmt_win; /* window for limiting early rexmts */
399
400 u_int32_t ts_recent; /* timestamp echo data */
401
402 u_int32_t ts_recent_age; /* when last updated */
403 tcp_seq last_ack_sent;
404/* RFC 3465 variables */
405 u_int32_t t_bytes_acked; /* ABC "bytes_acked" parameter */
406
407 int t_lastchain; /* amount of packets chained last time around */
408 uint16_t t_unacksegs; /* received but unacked segments for delaying acks */
409 uint16_t t_unacksegs_ce; /* received but unacked segments/pure ACKs that were CE marked */
410
411 /*
412 * Pretty arbitrary value ;-)
413 * Goal is to make sure that some ACKs are being sent more frequently
414 * to allow the other side to ramp-up.
415 */
416#define TCP_FORCED_ACKS_COUNT 16
417 uint16_t t_forced_acks; /* count of pure ACKs that need to be forced out */
418 uint8_t t_rexmtthresh; /* duplicate ack threshold for entering fast recovery */
419 uint8_t t_rtimo_probes; /* number of adaptive rtimo probes sent */
420 uint32_t t_persist_timeout; /* ZWP persistence limit as set by PERSIST_TIMEOUT */
421 uint32_t t_persist_stop; /* persistence limit deadline if triggered by ZWP */
422 uint32_t t_notsent_lowat; /* Low water for not sent data */
423
424/* Receiver state for stretch-ack algorithm */
425 u_int32_t rcv_unackwin; /* to measure win for stretching acks */
426 u_int32_t rcv_by_unackwin; /* bytes seen during the last ack-stretching win */
427 u_int32_t rcv_by_unackhalfwin;
428 u_int32_t rcv_nostrack_ts; /* timestamp when stretch ack was disabled automatically */
429 u_int32_t rcv_nostrack_pkts; /* pkts received since strech ack was disabled */
430 u_int16_t rcv_waitforss; /* wait for packets during slow-start */
431
432/* ECN stats */
433 u_int32_t ecn_flags;
434#define TE_SETUPSENT 0x00000001 /* We have sent classic ECN-SETUP SYN or SYN-ACK */
435#define TE_SETUPRECEIVED 0x00000002 /* We have received classic ECN-SETUP SYN or SYN-ACK */
436#define TE_SENDIPECT 0x00000004 /* We haven't sent or received non-ECN-setup SYN or SYN-ACK, set IP ECT on outbound packet */
437#define TE_SENDCWR 0x00000008 /* Next non-retransmit should have TCP CWR set, only used for classic ECN */
438#define TE_SENDECE 0x00000010 /* Next packet should have TCP ECE set, only used for classic ECN */
439#define TE_INRECOVERY 0x00000020 /* connection entered recovery after receiving ECE */
440#define TE_RECV_ECN_CE 0x00000040 /* Received IPTOS_ECN_CE marking atleast once */
441#define TE_RECV_ECN_ECE 0x00000080 /* Received ECE marking atleast once */
442#define TE_LOST_SYN 0x00000100 /* Lost SYN with ECN setup */
443#define TE_LOST_SYNACK 0x00000200 /* Lost SYN-ACK with ECN setup */
444#define TE_ECN_MODE_ENABLE 0x00000400 /* Option ECN mode set to enable */
445#define TE_ECN_MODE_DISABLE 0x00000800 /* Option ECN mode set to disable */
446#define TE_ENABLE_ECN 0x00001000 /* Enable negotiation of ECN */
447#define TE_ECN_ON (TE_SETUPSENT | TE_SETUPRECEIVED) /* ECN was successfully negotiated on a connection */
448#define TE_CEHEURI_SET 0x00002000 /* We did our CE-probing at the beginning */
449#define TE_CLIENT_SETUP 0x00004000 /* setup from client side */
450#define TE_RCVD_SYN_RST 0x00008000 /* Received RST to the first ECN enabled SYN */
451#define TE_ACE_SETUP_NON_ECT 0x00010000 /* Encode received non-ECT either for SYN-ACK (server) or final ACK (client) */
452#define TE_ACE_SETUP_ECT1 0x00020000 /* Encode received ECT1 either for SYN-ACK (server) or final ACK (client) */
453#define TE_ACE_SETUP_ECT0 0x00040000 /* Encode received ECT0 either for SYN-ACK (server) or final ACK (client) */
454#define TE_ACE_SETUP_CE 0x00080000 /* Encode received CE either for SYN-ACK (server) or final ACK (client) */
455#define TE_ACE_SETUPSENT 0x00100000 /* We have sent Accurate ECN setup SYN or SYN-ACK */
456#define TE_ACE_SETUPRECEIVED 0x00200000 /* We have received Accurate ECN setup SYN or SYN-ACK */
457#define TE_ACC_ECN_ON (TE_ACE_SETUPSENT | TE_ACE_SETUPRECEIVED) /* Accurate ECN was negotiated */
458#define TE_ACE_FINAL_ACK_3WHS 0x00400000 /* Client has received SYN-ACK and will now send final ACK of 3WHS, only used for AccECN */
459#define TE_ACO_ECT1 0x00800000 /* ECT1 counter changed flag, used to decide ordering for AccECN option */
460#define TE_ACO_ECT0 0x01000000 /* ECT0 counter changed flag, used to decide ordering for AccECN option */
461#define TE_RETRY_WITHOUT_ACO 0x02000000 /* Data segment with AccECN option was not acknowledged, retry without AccECN option */
462#define TE_FORCE_ECT1 0x40000000 /* Force setting ECT1 on outgoing packets for testing purpose */
463#define TE_FORCE_ECT0 0x80000000 /* Force setting ECT0 on outgoing packets for testing purpose */
464
465 u_int32_t t_ecn_recv_ce; /* Received CE from the network */
466 u_int32_t t_ecn_recv_cwr; /* Packets received with CWR */
467 uint32_t t_client_accecn_state; /* Client's Accurate ECN state */
468 uint32_t t_server_accecn_state; /* Server's Accurate ECN state */
469 uint64_t t_ecn_capable_packets_sent; /* Packets sent with ECT */
470 uint64_t t_ecn_capable_packets_acked; /* Packets sent with ECT that were acked */
471 uint64_t t_ecn_capable_packets_marked; /* Packets sent with ECT that were marked */
472 uint64_t t_ecn_capable_packets_lost; /* Packets sent with ECT that were lost */
473
474 uint16_t t_prev_ace_flags; /* ACE flags that were sent in previous packet, used for retransmitting after timeout */
475 uint8_t t_prev_ip_ecn; /* IP ECN flag on the previous packet, used for change-triggered ACKs */
476
477/* ACE CE packet counters */
478 uint32_t t_rcv_ce_packets; /* Number of CE received packets at the receiver */
479 uint32_t t_snd_ce_packets; /* Synced number of CE received feedback at the sender */
480 uint32_t t_delta_ce_packets; /* Change in CE count between previous and current ACK */
481
482/* AccECN option byte counters, sender side counters (s_) must not be used until they are synced from TCP options */
483 uint64_t t_rcv_ect1_bytes; /* ECT1 byte counter at the receiver, used for AccECN option feedback */
484 uint64_t t_rcv_ect0_bytes; /* ECT0 byte counter at the receiver, used for AccECN option feedback */
485 uint64_t t_rcv_ce_bytes; /* CE byte counter at the receiver, used for AccECN option feedback */
486 uint64_t t_snd_ect1_bytes; /* Synced ECT1 byte counter at the sender */
487 uint64_t t_snd_ect0_bytes; /* Synced ECT0 byte counter at the sender */
488 uint64_t t_snd_ce_bytes; /* Synced CE byte counter at the sender */
489
490/* state for bad retransmit recovery */
491 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */
492 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */
493 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */
494 int t_srtt_prev; /* srtt prior to retransmit */
495 int t_rttvar_prev; /* rttvar prior to retransmit */
496 u_int32_t t_badrexmt_time; /* bad rexmt detection time */
497
498/* Packet reordering metric */
499 u_int32_t t_reorderwin; /* Reordering late time offset */
500
501/* SACK related state */
502 int16_t snd_numholes; /* number of holes seen by sender */
503 tcp_seq sack_newdata; /* New data xmitted in this recovery
504 * episode starts at this seq number */
505 TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
506 /* SACK scoreboard (sorted) */
507 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/
508 int rcv_numsacks; /* # distinct sack blks present */
509 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
510 struct sackhint sackhint; /* SACK scoreboard hint */
511 tcp_seq send_highest_sack; /* Sequence number of fresh data sent after the most recent fast-retransmit */
512 int t_new_dupacks; /* Dupacks received above send_highest_sack */
513
514 struct mbuf *t_pktlist_head; /* First packet in transmit chain */
515 struct mbuf *t_pktlist_tail; /* Last packet in transmit chain */
516 u_int32_t t_pktlist_sentlen; /* total bytes in transmit chain */
517
518 u_int32_t t_keepidle; /* keepalive idle timer (override global if > 0) */
519 u_int32_t t_keepinit; /* connection timeout, i.e. idle time
520 * in SYN_SENT or SYN_RECV state */
521 u_int32_t t_keepintvl; /* interval between keepalives */
522 u_int32_t t_keepcnt; /* number of keepalives before close */
523
524 u_int32_t tso_max_segment_size; /* TSO maximum segment unit for NIC */
525 u_int16_t t_pmtud_lastseg_size; /* size of the last sent segment */
526 u_int32_t t_pmtud_saved_maxopd; /* MSS saved before performing PMTU-D BlackHole detection */
527 u_int32_t t_pmtud_start_ts; /* Time of PMTUD blackhole detection */
528
529 struct{
530 u_int32_t rxduplicatebytes;
531 u_int32_t rxoutoforderbytes;
532 u_int32_t txretransmitbytes;
533 u_int16_t synrxtshift;
534 u_int16_t rxmitsyns;
535 u_int16_t unused_pad_to_8;
536 u_int32_t rxmitpkts;
537 uint32_t delayed_acks_sent;
538 uint32_t acks_delayed;
539 } t_stat;
540 u_int8_t t_syn_sent;
541 u_int8_t t_syn_rcvd;
542 u_int8_t t_notify_ack_count;
543 u_int8_t t_ecn_recv_ce_pkt; /* Received packet with CE-bit set (independent from last_ack_sent) */
544 u_int32_t t_cached_maxopd; /* default for MSS adjustment using link status report */
545
546 uint32_t bg_ssthresh; /* Slow start threshold until delay increases */
547 uint32_t t_flagsext; /* Another field to accommodate more flags */
548#define TF_RXTFINDROP 0x1 /* Drop conn after retransmitting FIN 3 times */
549#define TF_RCVUNACK_WAITSS 0x2 /* set when the receiver should not stretch acks */
550#define TF_BWMEAS_INPROGRESS 0x4 /* Indicate BW meas is happening */
551#define TF_MEASURESNDBW 0x8 /* Measure send bw on this connection */
552#define TF_LAST_IS_PSH 0x10 /* Indicates whether the last packet in the rcv socket buffer had the PUSH-flag set */
553#define TF_SACK_ENABLE 0x20 /* SACK is enabled */
554#define TF_RECOMPUTE_RTT 0x40 /* recompute RTT after spurious retransmit */
555#define TF_DETECT_READSTALL 0x80 /* Used to detect a stall during read operation */
556#define TF_RECV_THROTTLE 0x100 /* Input throttling active */
557#define TF_NOSTRETCHACK 0x200 /* ack every other packet */
558#define TF_NOTIMEWAIT 0x800 /* Avoid going into time-wait */
559#define TF_SENT_TLPROBE 0x1000 /* Sent data in PTO */
560#define TF_PKTS_REORDERED 0x2000 /* Detected reordering */
561#define TF_DELAY_RECOVERY 0x4000 /* delay fast recovery */
562#define TF_FORCE 0x8000 /* force 1 byte out */
563#define TF_DISABLE_STRETCHACK 0x10000 /* auto-disable stretch ack */
564#define TF_NOBLACKHOLE_DETECTION 0x20000 /* Disable PMTU blackhole detection */
565#define TF_RESCUE_RXT 0x80000 /* SACK rescue retransmit */
566#define TF_CWND_NONVALIDATED 0x100000 /* cwnd non validated */
567#define TF_IF_PROBING 0x200000 /* Trigger interface probe timeout */
568#define TF_FASTOPEN 0x400000 /* TCP Fastopen is enabled */
569#define TF_REASS_INPROG 0x800000 /* Reassembly is in progress */
570#define TF_FASTOPEN_FORCE_ENABLE 0x1000000 /* Force-enable TCP Fastopen */
571#define TF_USR_OUTPUT 0x2000000 /* In connect() or send() so tcp_output() can log */
572#define TF_L4S_ENABLED 0x8000000 /* L4S was force enabled */
573#define TF_L4S_DISABLED 0x10000000 /* L4S was force disabled */
574
575#if TRAFFIC_MGT
576 /* Inter-arrival jitter related state */
577 uint32_t iaj_rcv_ts; /* tcp clock when the first packet was received */
578 int iaj_size; /* Size of packet for iaj measurement */
579 uint8_t iaj_small_pkt; /* Count of packets smaller than iaj_size */
580 uint8_t t_pipeack_ind; /* index for next pipeack sample */
581 uint16_t iaj_pktcnt; /* packet count, to avoid throttling initially */
582 uint32_t acc_iaj; /* Accumulated iaj */
583 uint32_t avg_iaj; /* Mean */
584 uint32_t std_dev_iaj; /* Standard deviation */
585#endif /* TRAFFIC_MGT */
586 struct bwmeas *t_bwmeas; /* State for bandwidth measurement */
587 tcp_seq t_idleat; /* rcv_nxt at idle time */
588 uint8_t t_fin_sent;
589 uint8_t t_fin_rcvd;
590 uint8_t t_rst_sent;
591 uint8_t t_rst_rcvd;
592 TAILQ_ENTRY(tcpcb) t_twentry; /* link for time wait queue */
593 struct tcp_ccstate *t_ccstate; /* congestion control related state */
594 struct tcp_ccstate _t_ccstate; /* congestion control related state, non-allocated */
595/* Tail loss probe related state */
596 tcp_seq t_tlphighrxt; /* snd_nxt after PTO */
597 u_int32_t t_tlpstart; /* timestamp at PTO */
598/* DSACK data receiver state */
599 tcp_seq t_dsack_lseq; /* DSACK left sequence */
600 tcp_seq t_dsack_rseq; /* DSACK right sequence */
601/* DSACK data sender state */
602 SLIST_HEAD(tcp_rxt_seghead, tcp_rxt_seg) t_rxt_segments;
603 uint32_t t_rxt_seg_count;
604 uint32_t t_rxt_seg_drop;
605 tcp_seq t_dsack_lastuna; /* snd_una when last recovery episode started */
606/* state for congestion window validation (draft-ietf-tcpm-newcwv-07) */
607#define TCP_PIPEACK_SAMPLE_COUNT 3
608 u_int32_t t_pipeack_sample[TCP_PIPEACK_SAMPLE_COUNT]; /* pipeack, bytes acked within RTT */
609 tcp_seq t_pipeack_lastuna; /* una when pipeack measurement started */
610 u_int32_t t_pipeack;
611 u_int32_t t_lossflightsize;
612
613#if MPTCP
614 u_int32_t t_mpflags; /* flags for multipath TCP */
615
616#define TMPF_PREESTABLISHED 0x00000001 /* conn in pre-established state */
617#define TMPF_SND_KEYS 0x00000002 /* indicates that keys should be send */
618#define TMPF_MPTCP_TRUE 0x00000004 /* negotiated MPTCP successfully */
619/* UNUSED */
620#define TMPF_SND_MPPRIO 0x00000010 /* send priority of subflow */
621#define TMPF_SND_REM_ADDR 0x00000020 /* initiate address removal */
622#define TMPF_RCVD_DACK 0x00000040 /* received a data-ack */
623#define TMPF_JOINED_FLOW 0x00000080 /* Indicates additional flow */
624#define TMPF_BACKUP_PATH 0x00000100 /* Indicates backup path */
625#define TMPF_MPTCP_ACKNOW 0x00000200 /* Send Data ACK */
626#define TMPF_SEND_DSN 0x00000400 /* Send DSN mapping */
627#define TMPF_SEND_DFIN 0x00000800 /* Send Data FIN */
628#define TMPF_RECV_DFIN 0x00001000 /* Recv Data FIN */
629#define TMPF_SENT_JOIN 0x00002000 /* Sent Join */
630#define TMPF_RECVD_JOIN 0x00004000 /* Received Join */
631#define TMPF_RESET 0x00008000 /* Send RST */
632#define TMPF_TCP_FALLBACK 0x00010000 /* Fallback to TCP */
633#define TMPF_FASTCLOSERCV 0x00020000 /* Received Fastclose option */
634#define TMPF_EMBED_DSN 0x00040000 /* tp has DSN mapping */
635#define TMPF_MPTCP_READY 0x00080000 /* Can send DSS options on data */
636#define TMPF_INFIN_SENT 0x00100000 /* Sent infinite mapping */
637#define TMPF_SND_MPFAIL 0x00200000 /* Received mapping csum failure */
638#define TMPF_SND_JACK 0x00400000 /* Send a Join-ACK */
639#define TMPF_TFO_REQUEST 0x00800000 /* TFO Requested */
640#define TMPF_MPTCP_ECHO_ADDR 0x01000000 /* MPTCP echoes add_addr */
641
642#define TMPF_MPTCP_SIGNALS (TMPF_SND_MPPRIO | TMPF_SND_REM_ADDR | TMPF_SND_MPFAIL | TMPF_SND_JACK | TMPF_MPTCP_ECHO_ADDR)
643
644 tcp_seq t_mpuna; /* unacknowledged sequence */
645 struct mptcb *t_mptcb; /* pointer to MPTCP TCB */
646 struct mptsub *t_mpsub; /* pointer to the MPTCP subflow */
647 struct mpt_dsn_map t_rcv_map; /* Receive mapping list */
648 u_int8_t t_local_aid; /* Addr Id for authentication */
649 u_int8_t t_rem_aid; /* Addr ID of another subflow */
650 u_int8_t t_mprxtshift; /* join retransmission */
651#endif /* MPTCP */
652
653#define TFO_F_OFFER_COOKIE 0x01 /* We will offer a cookie */
654#define TFO_F_COOKIE_VALID 0x02 /* The received cookie is valid */
655#define TFO_F_COOKIE_REQ 0x04 /* Client requested a new cookie */
656#define TFO_F_COOKIE_SENT 0x08 /* Client did send a cookie in the SYN */
657#define TFO_F_SYN_LOSS 0x10 /* A SYN-loss triggered a fallback to regular TCP on the client-side */
658#define TFO_F_NO_SNDPROBING 0x20 /* This network is guaranteed to support TFO in the upstream direction */
659#define TFO_F_HEURISTIC_DONE 0x40 /* We have already marked this network as bad */
660 u_int8_t t_tfo_flags;
661#define TFO_S_SYNDATA_RCV 0x01 /* SYN+data has been received */
662#define TFO_S_COOKIEREQ_RECV 0x02 /* TFO-cookie request received */
663#define TFO_S_COOKIE_SENT 0x04 /* TFO-cookie announced in SYN/ACK */
664#define TFO_S_COOKIE_INVALID 0x08 /* Received TFO-cookie is invalid */
665#define TFO_S_COOKIE_REQ 0x10 /* TFO-cookie requested within the SYN */
666#define TFO_S_COOKIE_RCV 0x20 /* TFO-cookie received in SYN/ACK */
667#define TFO_S_SYN_DATA_SENT 0x40 /* SYN+data sent */
668#define TFO_S_SYN_DATA_ACKED 0x80 /* SYN+data has been acknowledged in SYN/ACK */
669#define TFO_S_SYN_LOSS 0x0100 /* SYN+TFO has been lost - fallback to regular TCP */
670#define TFO_S_COOKIE_WRONG 0x0200 /* Cookie we sent in the SYN was wrong */
671#define TFO_S_NO_COOKIE_RCV 0x0400 /* We asked for a cookie but didn't get one */
672#define TFO_S_HEURISTICS_DISABLE 0x0800 /* TFO-heuristics disabled it for this connection */
673#define TFO_S_SEND_BLACKHOLE 0x1000 /* TFO got blackholed in the send direction */
674#define TFO_S_RECV_BLACKHOLE 0x2000 /* TFO got blackholed in the recv direction */
675#define TFO_S_ONE_BYTE_PROXY 0x4000 /* TFO failed because of a proxy acknowledging just one byte */
676 u_int16_t t_tfo_stats;
677
678 u_int8_t t_tfo_probes; /* TFO-probes we did send */
679/*
680 * This here is the TFO-probing state-machine. Transitions are as follows:
681 *
682 * Current state: PROBE_NONE
683 * Event: SYN+DATA acknowledged
684 * Action: Transition to PROBE_PROBING and set keepalive-timer
685 *
686 * Current state: PROBE_PROBING (initial state)
687 * Event: Receive data
688 * Action: Transition to PROBE_NONE and cancel keepalive-timer
689 * Event: Receive ACK that does not indicate a hole
690 * Action: Transition to PROBE_NONE and cancel keepalive-timer
691 * Event: Receive ACK that indicates a hole
692 * Action: Transition to PROBE_WAIT_DATA and set a short timer
693 * to wait for the final segment.
694 * Event: Keepalive-timeout (did not receive any segment)
695 * Action: Signal ETIMEDOUT as with regular keepalive-timers
696 *
697 * Current state: PROBE_WAIT_DATA
698 * Event: Receive data
699 * Action: Transition to PROBE_NONE and cancel keepalive-timer
700 * Event: Data-timeout (did not receive the expected data)
701 * Action: Signal ENODATA up to the app and close everything.
702 */
703#define TFO_PROBE_NONE 0 /* Not probing now */
704#define TFO_PROBE_PROBING 1 /* Sending out TCP-keepalives waiting for reply */
705#define TFO_PROBE_WAIT_DATA 2 /* Received reply, waiting for data */
706 u_int8_t t_tfo_probe_state;
707
708 u_int32_t t_rcvoopack; /* out-of-order packets received */
709 u_int32_t t_pawsdrop; /* segments dropped due to PAWS */
710 u_int32_t t_sack_recovery_episode; /* SACK recovery episodes */
711 u_int32_t t_reordered_pkts; /* packets reorderd */
712 u_int32_t t_dsack_sent; /* Sent DSACK notification */
713 u_int32_t t_dsack_recvd; /* Received a valid DSACK option */
714 SLIST_HEAD(, tcp_notify_ack_marker) t_notify_ack; /* state for notifying data acknowledgements */
715 u_int32_t t_recv_throttle_ts; /* TS for start of recv throttle */
716 u_int32_t t_rxt_minimum_timeout; /* minimum retransmit timeout in ms */
717 uint32_t t_challengeack_last; /* last time challenge ACK was sent per sec */
718 uint32_t t_challengeack_count; /* # of challenge ACKs already sent per sec */
719
720 u_int32_t t_connect_time; /* time when the connection started */
721
722 uint64_t t_rcvwnd_limited_total_time;
723 uint64_t t_rcvwnd_limited_start_time;
724
725 uint32_t t_comp_gencnt; /* Current compression generation-count */
726 uint32_t t_comp_lastinc; /* Last time the gen-count was changed - should change every TCP_COMP_CHANGE_RATE ms */
727#define TCP_COMP_CHANGE_RATE 5 /* Intervals at which we change the gencnt. Means that worst-case we send one ACK every TCP_COMP_CHANGE_RATE ms */
728
729 uint32_t t_ts_offset; /* Randomized timestamp offset to hide on-the-wire timestamp */
730
731#define NCURR_RTT_HIST 4 /* Number of current RTT samples (k) */
732 uint32_t curr_rtt_hist[NCURR_RTT_HIST]; /* last k current RTT samples */
733 uint32_t curr_rtt_min; /* Minimum current RTT from last k samples */
734 uint32_t curr_rtt_index; /* Index for current RTT samples */
735
736 tcp_seq rcv_high; /* highest sequence number received */
737 uint32_t tsv_high; /* timestamp value of highest received sequence number */
738 struct tcp_rledbat_state t_rlstate; /* State used by rLedbat */
739
740 uint32_t rcv_srtt; /* receiver's SRTT, coarse when Timestamp not supported */
741 uint32_t rcv_rtt_est_ts; /* start of measurement for estimating RTT when Timestamp not supported */
742 uint32_t rcv_rtt_est_seq; /* expected sequence number for completion of 1RTT when Timestamp not supported */
743
744 uuid_t t_fsw_uuid;
745 uuid_t t_flow_uuid;
746};
747
748#define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY)
749#define SACK_ENABLED(tp) (tp->t_flagsext & TF_SACK_ENABLE)
750
751/*
752 * If the connection is in a throttled state due to advisory feedback from
753 * the interface output queue, reset that state. We do this in favor
754 * of entering recovery because the data transfer during recovery
755 * should be just a trickle and it will help to improve performance.
756 * We also do not want to back off twice in the same RTT.
757 */
758#define ENTER_FASTRECOVERY(_tp_) do { \
759 (_tp_)->t_flags |= TF_FASTRECOVERY; \
760 if (INP_IS_FLOW_CONTROLLED((_tp_)->t_inpcb)) \
761 inp_reset_fc_state((_tp_)->t_inpcb); \
762 if (!SLIST_EMPTY(&tp->t_rxt_segments)) \
763 tcp_rxtseg_clean(tp); \
764 (_tp_)->t_new_dupacks = 0; \
765} while(0)
766
767#define EXIT_FASTRECOVERY(_tp_) do { \
768 (_tp_)->t_flags &= ~TF_FASTRECOVERY; \
769 (_tp_)->t_dupacks = 0; \
770 (_tp_)->t_new_dupacks = 0; \
771 (_tp_)->t_rexmtthresh = (uint8_t)tcprexmtthresh; \
772 (_tp_)->t_bytes_acked = 0; \
773 (_tp_)->ecn_flags &= ~TE_INRECOVERY; \
774 (_tp_)->t_timer[TCPT_PTO] = 0; \
775 (_tp_)->t_flagsext &= ~TF_RESCUE_RXT; \
776 (_tp_)->t_lossflightsize = 0; \
777 (_tp_)->sackhint.sack_bytes_acked = 0; \
778} while(0)
779
780/*
781 * When the number of duplicate acks received is less than
782 * the retransmit threshold, use Limited Transmit algorithm
783 */
784extern uint8_t tcprexmtthresh;
785#define ALLOW_LIMITED_TRANSMIT(_tp_) \
786 ((_tp_)->t_dupacks > 0 && \
787 (_tp_)->t_dupacks < (_tp_)->t_rexmtthresh && \
788 ((_tp_)->t_flagsext & (TF_PKTS_REORDERED|TF_DELAY_RECOVERY)) \
789 != (TF_PKTS_REORDERED|TF_DELAY_RECOVERY))
790
791/*
792 * This condition is true if timestamp option is supported
793 * on a connection.
794 */
795#define TSTMP_SUPPORTED(_tp_) \
796 (((_tp_)->t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP)) == \
797 (TF_REQ_TSTMP|TF_RCVD_TSTMP))
798
799/*
800 * This condition is true if window scale option is supported
801 * on a connection
802 */
803#define TCP_WINDOW_SCALE_ENABLED(_tp_) \
804 (((_tp_)->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == \
805 (TF_RCVD_SCALE|TF_REQ_SCALE))
806
807/* Is ECN negotiated end-to-end */
808#define TCP_ECN_ENABLED(_tp_) \
809 (((_tp_)->ecn_flags & (TE_ECN_ON)) == (TE_ECN_ON))
810
811extern int tcp_acc_ecn;
812/*
813 * Accurate ECN feedback is enabled if
814 * 1. It is not disabled explicitly by tcp options
815 * 2. It is enabled either by sysctl or L4S toggle or A/B deployment or tcp_options,
816 * It supports ACE field as well as AccECN option for ECN feedback
817 */
818#define TCP_ACC_ECN_ENABLED(_tp_) \
819 (((_tp_)->t_flagsext & TF_L4S_DISABLED) == 0 && \
820 (tcp_acc_ecn == 1 || ((_tp_)->t_flagsext & TF_L4S_ENABLED)))
821
822/* Accurate ECN is enabled and negotiated end-to-end */
823#define TCP_ACC_ECN_ON(_tp_) \
824 (TCP_ACC_ECN_ENABLED(_tp_) && \
825 (((_tp_)->ecn_flags & (TE_ACC_ECN_ON)) == (TE_ACC_ECN_ON)))
826
827/*
828 * Gives number of bytes acked by this ack
829 */
830#define BYTES_ACKED(_th_, _tp_) \
831 ((_th_)->th_ack - (_tp_)->snd_una)
832
833/* Returns true if a DSACK option should be sent */
834#define TCP_SEND_DSACK_OPT(_tp_) \
835 ((_tp_)->t_dsack_lseq > 0 && (_tp_)->t_dsack_rseq > 0)
836
837/*
838 * Returns true if a DSACK sequence is within the max send window that will
839 * be accepted. In order to set a window to validate sequence numbers, the
840 * max send window within which a DSACK option is processed is limited.
841 *
842 * We need to choose a maximum window to check if the sequence number is
843 * within the window. One arbitrary choice is 256 * MSS because if the
844 * window is as large as 256 segments it might be big enough to ignore the
845 * DSACK option. Choosing a much larger limit means that the memory for
846 * retransmit segments can be held for a longer time.
847 */
848#define TCP_DSACK_MAX_SEND_WINDOW(_tp_) (MIN((_tp_)->snd_wnd, tcp_autosndbuf_max))
849#define TCP_DSACK_SEQ_IN_WINDOW(_tp_, _seq_, _una_) \
850 (SEQ_LEQ((_seq_), (_tp_)->snd_max) && \
851 SEQ_GEQ((_seq_), ((_una_) - TCP_DSACK_MAX_SEND_WINDOW(_tp_))))
852
853#define TCP_RESET_REXMT_STATE(_tp_) do { \
854 (_tp_)->t_rxtshift = 0; \
855 (_tp_)->t_rxtstart = 0; \
856 mptcp_reset_rexmit_state((_tp_)); \
857} while(0);
858
859#define TCP_IF_STATE_CHANGED(tp, probe_if_index) \
860 (probe_if_index > 0 && tp->t_inpcb->inp_last_outifp != NULL && \
861 probe_if_index == tp->t_inpcb->inp_last_outifp->if_index)
862
863#define TCP_RLEDBAT_ENABLED(_tp) \
864 (tcp_rledbat == 1 && TSTMP_SUPPORTED(_tp))
865
866#define TCP_RECV_BG(_so) \
867 (tcp_recv_bg == 1 || IS_TCP_RECV_BG(_so))
868
869#define TCP_USE_RLEDBAT(_tp, _so) \
870 TCP_RLEDBAT_ENABLED(_tp) && TCP_RECV_BG(_so)
871
872/*
873 * Structure to hold TCP options that are only used during segment
874 * processing (in tcp_input), but not held in the tcpcb.
875 * It's basically used to reduce the number of parameters
876 * to tcp_dooptions.
877 */
878struct tcpopt {
879 uint32_t to_flags; /* which options are present */
880#define TOF_TS 0x0001 /* timestamp */
881#define TOF_MSS 0x0010
882#define TOF_SCALE 0x0020
883#define TOF_SIGNATURE 0x0040 /* signature option present */
884#define TOF_SIGLEN 0x0080 /* signature length valid (RFC2385) */
885#define TOF_SACK 0x0100 /* Peer sent SACK option */
886#define TOF_MPTCP 0x0200 /* MPTCP options to be dropped */
887#define TOF_TFO 0x0400 /* TFO cookie option present */
888#define TOF_TFOREQ 0x0800 /* TFO cookie request present */
889 uint32_t to_tsval;
890 uint32_t to_tsecr;
891 uint16_t to_mss;
892 uint8_t to_requested_s_scale;
893 uint8_t to_nsacks; /* number of SACK blocks */
894 u_char *to_sacks; /* pointer to the first SACK blocks */
895 u_char *to_tfo; /* pointer to the TFO cookie */
896};
897
898#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
899#define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
900
901/* TFO-specific defines */
902#define TFO_COOKIE_LEN_MIN 4
903#define TFO_COOKIE_LEN_DEFAULT 8
904#define TFO_COOKIE_LEN_MAX 16
905
906/*
907 * The initial retransmission should happen at rtt + 4 * rttvar.
908 * Because of the way we do the smoothing, srtt and rttvar
909 * will each average +1/2 tick of bias. When we compute
910 * the retransmit timer, we want 1/2 tick of rounding and
911 * 1 extra tick because of +-1/2 tick uncertainty in the
912 * firing of the timer. The bias will give us exactly the
913 * 1.5 tick we need. But, because the bias is
914 * statistical, we have to test that we don't drop below
915 * the minimum feasible timer (which is 2 ticks).
916 * This version of the macro adapted from a paper by Lawrence
917 * Brakmo and Larry Peterson which outlines a problem caused
918 * by insufficient precision in the original implementation,
919 * which results in inappropriately large RTO values for very
920 * fast networks.
921 */
922#define TCP_REXMTVAL(tp) \
923 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
924 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
925
926/*
927 * Jaguar compatible TCP control block, for xtcpcb
928 * Does not have the old fields
929 */
930struct otcpcb {
931#else
932struct tseg_qent;
933_TCPCB_LIST_HEAD(tsegqe_head, tseg_qent);
934
935struct tcpcb {
936#endif /* KERNEL_PRIVATE */
937#if defined(KERNEL_PRIVATE)
938 u_int32_t t_segq;
939#else
940 struct tsegqe_head t_segq;
941#endif /* KERNEL_PRIVATE */
942 int t_dupacks; /* consecutive dup acks recd */
943 u_int32_t unused; /* unused now: was t_template */
944
945 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */
946
947 _TCPCB_PTR(struct inpcb *) t_inpcb; /* back pointer to internet pcb */
948 int t_state; /* state of this connection */
949 u_int t_flags;
950#define TF_ACKNOW 0x00001 /* ack peer immediately */
951#define TF_DELACK 0x00002 /* ack, but try to delay it */
952#define TF_NODELAY 0x00004 /* don't delay packets to coalesce */
953#define TF_NOOPT 0x00008 /* don't use tcp options */
954#define TF_SENTFIN 0x00010 /* have sent FIN */
955#define TF_REQ_SCALE 0x00020 /* have/will request window scaling */
956#define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */
957#define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */
958#define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */
959#define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */
960#define TF_NEEDSYN 0x00400 /* send SYN (implicit state) - unused but needed for backwards compatibility */
961#define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */
962#define TF_NOPUSH 0x01000 /* don't push */
963#define TF_REQ_CC 0x02000 /* have/will request CC */
964#define TF_RCVD_CC 0x04000 /* a CC was received in SYN */
965#define TF_SENDCCNEW 0x08000 /* Not implemented */
966#define TF_MORETOCOME 0x10000 /* More data to be appended to sock */
967#define TF_LQ_OVERFLOW 0x20000 /* listen queue overflow */
968#define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */
969#define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */
970
971 int t_force; /* 1 if forcing out a byte */
972
973 tcp_seq snd_una; /* send unacknowledged */
974 tcp_seq snd_max; /* highest sequence number sent;
975 * used to recognize retransmits
976 */
977 tcp_seq snd_nxt; /* send next */
978 tcp_seq snd_up; /* send urgent pointer */
979
980 tcp_seq snd_wl1; /* window update seg seq number */
981 tcp_seq snd_wl2; /* window update seg ack number */
982 tcp_seq iss; /* initial send sequence number */
983 tcp_seq irs; /* initial receive sequence number */
984
985 tcp_seq rcv_nxt; /* receive next */
986 tcp_seq rcv_adv; /* advertised window */
987 u_int32_t rcv_wnd; /* receive window */
988 tcp_seq rcv_up; /* receive urgent pointer */
989
990 u_int32_t snd_wnd; /* send window */
991 u_int32_t snd_cwnd; /* congestion-controlled window */
992 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for
993 * for slow start exponential to
994 * linear switch
995 */
996 u_int t_maxopd; /* mss plus options */
997
998 u_int32_t t_rcvtime; /* time at which a packet was received */
999 u_int32_t t_starttime; /* time connection was established */
1000 int t_rtttime; /* round trip time */
1001 tcp_seq t_rtseq; /* sequence number being timed */
1002
1003 int t_rxtcur; /* current retransmit value (ticks) */
1004 u_int t_maxseg; /* maximum segment size */
1005 int t_srtt; /* smoothed round-trip time */
1006 int t_rttvar; /* variance in round-trip time */
1007
1008 int t_rxtshift; /* log(2) of rexmt exp. backoff */
1009 u_int t_rttmin; /* minimum rtt allowed */
1010 u_int32_t t_rttupdated; /* number of times rtt sampled */
1011 u_int32_t max_sndwnd; /* largest window peer has offered */
1012
1013 int t_softerror; /* possible error not yet reported */
1014/* out-of-band data */
1015 char t_oobflags; /* have some */
1016 char t_iobc; /* input character */
1017#define TCPOOB_HAVEDATA 0x01
1018#define TCPOOB_HADDATA 0x02
1019/* RFC 1323 variables */
1020 u_char snd_scale; /* window scaling for send window */
1021 u_char rcv_scale; /* window scaling for recv window */
1022 u_char request_r_scale; /* pending window scaling */
1023 u_char requested_s_scale;
1024 u_int32_t ts_recent; /* timestamp echo data */
1025
1026 u_int32_t ts_recent_age; /* when last updated */
1027 tcp_seq last_ack_sent;
1028/* RFC 1644 variables */
1029 tcp_cc cc_send; /* send connection count */
1030 tcp_cc cc_recv; /* receive connection count */
1031 tcp_seq snd_recover; /* for use in fast recovery */
1032/* experimental */
1033 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */
1034 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */
1035 u_int32_t t_badrxtwin; /* window for retransmit recovery */
1036};
1037
1038#define tcps_ecn_setup tcps_ecn_client_success
1039#define tcps_sent_cwr tcps_ecn_recv_ece
1040#define tcps_sent_ece tcps_ecn_sent_ece
1041
1042/*
1043 * TCP statistics.
1044 * Many of these should be kept per connection,
1045 * but that's inconvenient at the moment.
1046 */
1047struct tcpstat {
1048 u_int32_t tcps_connattempt; /* connections initiated */
1049 u_int32_t tcps_accepts; /* connections accepted */
1050 u_int32_t tcps_connects; /* connections established */
1051 u_int32_t tcps_drops; /* connections dropped */
1052 u_int32_t tcps_conndrops; /* embryonic connections dropped */
1053 u_int32_t tcps_closed; /* conn. closed (includes drops) */
1054 u_int32_t tcps_segstimed; /* segs where we tried to get rtt */
1055 u_int32_t tcps_rttupdated; /* times we succeeded */
1056 u_int32_t tcps_delack; /* delayed acks sent */
1057 u_int32_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
1058 u_int32_t tcps_rexmttimeo; /* retransmit timeouts */
1059 u_int32_t tcps_persisttimeo; /* persist timeouts */
1060 u_int32_t tcps_keeptimeo; /* keepalive timeouts */
1061 u_int32_t tcps_keepprobe; /* keepalive probes sent */
1062 u_int32_t tcps_keepdrops; /* connections dropped in keepalive */
1063
1064 u_int32_t tcps_sndtotal; /* total packets sent */
1065 u_int32_t tcps_sndpack; /* data packets sent */
1066 u_int32_t tcps_sndbyte; /* data bytes sent */
1067 u_int32_t tcps_sndrexmitpack; /* data packets retransmitted */
1068 u_int32_t tcps_sndrexmitbyte; /* data bytes retransmitted */
1069 u_int32_t tcps_sndacks; /* ack-only packets sent */
1070 u_int32_t tcps_sndprobe; /* window probes sent */
1071 u_int32_t tcps_sndurg; /* packets sent with URG only */
1072 u_int32_t tcps_sndwinup; /* window update-only packets sent */
1073 u_int32_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
1074
1075 u_int32_t tcps_rcvtotal; /* total packets received */
1076 u_int32_t tcps_rcvpack; /* packets received in sequence */
1077 u_int32_t tcps_rcvbyte; /* bytes received in sequence */
1078 u_int32_t tcps_rcvbadsum; /* packets received with ccksum errs */
1079 u_int32_t tcps_rcvbadoff; /* packets received with bad offset */
1080 u_int32_t tcps_rcvmemdrop; /* packets dropped for lack of memory */
1081 u_int32_t tcps_rcvshort; /* packets received too short */
1082 u_int32_t tcps_rcvduppack; /* duplicate-only packets received */
1083 u_int32_t tcps_rcvdupbyte; /* duplicate-only bytes received */
1084 u_int32_t tcps_rcvpartduppack; /* packets with some duplicate data */
1085 u_int32_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
1086 u_int32_t tcps_rcvoopack; /* out-of-order packets received */
1087 u_int32_t tcps_rcvoobyte; /* out-of-order bytes received */
1088 u_int32_t tcps_rcvpackafterwin; /* packets with data after window */
1089 u_int32_t tcps_rcvbyteafterwin; /* bytes rcvd after window */
1090 u_int32_t tcps_rcvafterclose; /* packets rcvd after "close" */
1091 u_int32_t tcps_rcvwinprobe; /* rcvd window probe packets */
1092 u_int32_t tcps_rcvdupack; /* rcvd duplicate acks */
1093 u_int32_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */
1094 u_int32_t tcps_rcvackpack; /* rcvd ack packets */
1095 u_int32_t tcps_rcvackbyte; /* bytes acked by rcvd acks */
1096 u_int32_t tcps_rcvwinupd; /* rcvd window update packets */
1097 u_int32_t tcps_pawsdrop; /* segments dropped due to PAWS */
1098 u_int32_t tcps_predack; /* times hdr predict ok for acks */
1099 u_int32_t tcps_preddat; /* times hdr predict ok for data pkts */
1100 u_int32_t tcps_pcbcachemiss;
1101 u_int32_t tcps_cachedrtt; /* times cached RTT in route updated */
1102 u_int32_t tcps_cachedrttvar; /* times cached rttvar updated */
1103 u_int32_t tcps_cachedssthresh; /* times cached ssthresh updated */
1104 u_int32_t tcps_usedrtt; /* times RTT initialized from route */
1105 u_int32_t tcps_usedrttvar; /* times RTTVAR initialized from rt */
1106 u_int32_t tcps_usedssthresh; /* times ssthresh initialized from rt*/
1107 u_int32_t tcps_persistdrop; /* timeout in persist state */
1108 u_int32_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */
1109 u_int32_t tcps_mturesent; /* resends due to MTU discovery */
1110 u_int32_t tcps_listendrop; /* listen queue overflows */
1111 u_int32_t tcps_synchallenge; /* challenge ACK due to bad SYN */
1112 u_int32_t tcps_rstchallenge; /* challenge ACK due to bad RST */
1113
1114 /* new stats from FreeBSD 5.4 sync up */
1115 u_int32_t tcps_minmssdrops; /* average minmss too low drops */
1116
1117 u_int32_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */
1118 u_int32_t tcps_badrst; /* ignored RSTs in the window */
1119
1120 u_int32_t tcps_sc_added; /* entry added to syncache */
1121 u_int32_t tcps_sc_retransmitted; /* syncache entry was retransmitted */
1122 u_int32_t tcps_sc_dupsyn; /* duplicate SYN packet */
1123 u_int32_t tcps_sc_dropped; /* could not reply to packet */
1124 u_int32_t tcps_sc_completed; /* successful extraction of entry */
1125 u_int32_t tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */
1126 u_int32_t tcps_sc_cacheoverflow; /* syncache cache limit hit */
1127 u_int32_t tcps_sc_reset; /* RST removed entry from syncache */
1128 u_int32_t tcps_sc_stale; /* timed out or listen socket gone */
1129 u_int32_t tcps_sc_aborted; /* syncache entry aborted */
1130 u_int32_t tcps_sc_badack; /* removed due to bad ACK */
1131 u_int32_t tcps_sc_unreach; /* ICMP unreachable received */
1132 u_int32_t tcps_sc_zonefail; /* zalloc() failed */
1133 u_int32_t tcps_sc_sendcookie; /* SYN cookie sent */
1134 u_int32_t tcps_sc_recvcookie; /* SYN cookie received */
1135
1136 u_int32_t tcps_hc_added; /* entry added to hostcache */
1137 u_int32_t tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */
1138
1139 /* SACK related stats */
1140 u_int32_t tcps_sack_recovery_episode; /* SACK recovery episodes */
1141 u_int32_t tcps_sack_rexmits; /* SACK rexmit segments */
1142 u_int32_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */
1143 u_int32_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */
1144 u_int32_t tcps_sack_send_blocks; /* SACK blocks (options) sent */
1145 u_int32_t tcps_sack_sboverflow; /* SACK sendblock overflow */
1146
1147 u_int32_t tcps_bg_rcvtotal; /* total background packets received */
1148 u_int32_t tcps_rxtfindrop; /* drop conn after retransmitting FIN */
1149 u_int32_t tcps_fcholdpacket; /* packets withheld because of flow control */
1150
1151 u_int32_t tcps_limited_txt; /* Limited transmit used */
1152 u_int32_t tcps_early_rexmt; /* Early retransmit used */
1153 u_int32_t tcps_sack_ackadv; /* Cumulative ack advanced along with sack */
1154
1155 /* Checksum related stats */
1156 u_int32_t tcps_rcv_swcsum; /* tcp swcksum (inbound), packets */
1157 u_int32_t tcps_rcv_swcsum_bytes; /* tcp swcksum (inbound), bytes */
1158 u_int32_t tcps_rcv6_swcsum; /* tcp6 swcksum (inbound), packets */
1159 u_int32_t tcps_rcv6_swcsum_bytes; /* tcp6 swcksum (inbound), bytes */
1160 u_int32_t tcps_snd_swcsum; /* tcp swcksum (outbound), packets */
1161 u_int32_t tcps_snd_swcsum_bytes; /* tcp swcksum (outbound), bytes */
1162 u_int32_t tcps_snd6_swcsum; /* tcp6 swcksum (outbound), packets */
1163 u_int32_t tcps_snd6_swcsum_bytes; /* tcp6 swcksum (outbound), bytes */
1164 u_int32_t tcps_unused_1;
1165 u_int32_t tcps_unused_2;
1166 u_int32_t tcps_unused_3;
1167
1168 /* MPTCP Related stats */
1169 u_int32_t tcps_invalid_mpcap; /* Invalid MPTCP capable opts */
1170 u_int32_t tcps_invalid_joins; /* Invalid MPTCP joins */
1171 u_int32_t tcps_mpcap_fallback; /* TCP fallback in primary */
1172 u_int32_t tcps_join_fallback; /* No MPTCP in secondary */
1173 u_int32_t tcps_estab_fallback; /* DSS option dropped */
1174 u_int32_t tcps_invalid_opt; /* Catchall error stat */
1175 u_int32_t tcps_mp_outofwin; /* Packet lies outside the
1176 * shared recv window */
1177 u_int32_t tcps_mp_reducedwin; /* Reduced subflow window */
1178 u_int32_t tcps_mp_badcsum; /* Bad DSS csum */
1179 u_int32_t tcps_mp_oodata; /* Out of order data */
1180 u_int32_t tcps_mp_switches; /* number of subflow switch */
1181 u_int32_t tcps_mp_rcvtotal; /* number of rcvd packets */
1182 u_int32_t tcps_mp_rcvbytes; /* number of bytes received */
1183 u_int32_t tcps_mp_sndpacks; /* number of data packs sent */
1184 u_int32_t tcps_mp_sndbytes; /* number of bytes sent */
1185 u_int32_t tcps_join_rxmts; /* join ack retransmits */
1186 u_int32_t tcps_tailloss_rto; /* RTO due to tail loss */
1187 u_int32_t tcps_reordered_pkts; /* packets reorderd */
1188 u_int32_t tcps_recovered_pkts; /* recovered after loss */
1189 u_int32_t tcps_pto; /* probe timeout */
1190 u_int32_t tcps_rto_after_pto; /* RTO after a probe */
1191 u_int32_t tcps_tlp_recovery; /* TLP induced fast recovery */
1192 u_int32_t tcps_tlp_recoverlastpkt; /* TLP recoverd last pkt */
1193 u_int32_t tcps_ecn_client_success; /* client-side connection negotiated ECN */
1194 u_int32_t tcps_ecn_recv_ece; /* ECE received, sent CWR */
1195 u_int32_t tcps_ecn_sent_ece; /* Sent ECE notification */
1196 u_int32_t tcps_detect_reordering; /* Detect pkt reordering */
1197 u_int32_t tcps_delay_recovery; /* Delay fast recovery */
1198 u_int32_t tcps_avoid_rxmt; /* Retransmission was avoided */
1199 u_int32_t tcps_unnecessary_rxmt; /* Retransmission was not needed */
1200 u_int32_t tcps_nostretchack; /* disabled stretch ack algorithm on a connection */
1201 u_int32_t tcps_rescue_rxmt; /* SACK rescue retransmit */
1202 u_int32_t tcps_pto_in_recovery; /* rescue retransmit in fast recovery */
1203 u_int32_t tcps_pmtudbh_reverted; /* PMTU Blackhole detection, segment size reverted */
1204
1205 /* DSACK related statistics */
1206 u_int32_t tcps_dsack_disable; /* DSACK disabled due to n/w duplication */
1207 u_int32_t tcps_dsack_ackloss; /* ignore DSACK due to ack loss */
1208 u_int32_t tcps_dsack_badrexmt; /* DSACK based bad rexmt recovery */
1209 u_int32_t tcps_dsack_sent; /* Sent DSACK notification */
1210 u_int32_t tcps_dsack_recvd; /* Received a valid DSACK option */
1211 u_int32_t tcps_dsack_recvd_old; /* Received an out of window DSACK option */
1212
1213 /* MPTCP Subflow selection stats */
1214 u_int32_t tcps_mp_sel_symtomsd; /* By symptomsd */
1215 u_int32_t tcps_mp_sel_rtt; /* By RTT comparison */
1216 u_int32_t tcps_mp_sel_rto; /* By RTO comparison */
1217 u_int32_t tcps_mp_sel_peer; /* By peer's output pattern */
1218 u_int32_t tcps_mp_num_probes; /* Number of probes sent */
1219 u_int32_t tcps_mp_verdowngrade; /* MPTCP version downgrade */
1220 u_int32_t tcps_drop_after_sleep; /* drop after long AP sleep */
1221 u_int32_t tcps_probe_if; /* probe packets after interface availability */
1222 u_int32_t tcps_probe_if_conflict; /* Can't send probe packets for interface */
1223
1224 u_int32_t tcps_ecn_client_setup; /* Attempted ECN setup from client side */
1225 u_int32_t tcps_ecn_server_setup; /* Attempted ECN setup from server side */
1226 u_int32_t tcps_ecn_server_success; /* server-side connection negotiated ECN */
1227 u_int32_t tcps_ecn_ace_syn_not_ect; /* received AccECN SYN packet with Not-ECT */
1228 u_int32_t tcps_ecn_ace_syn_ect1; /* received AccECN SYN packet with ECT1 */
1229 u_int32_t tcps_ecn_ace_syn_ect0; /* received AccECN SYN packet with ECT0 */
1230 u_int32_t tcps_ecn_ace_syn_ce; /* received AccECN SYN packet with CE */
1231 u_int32_t tcps_ecn_lost_synack; /* Lost SYN-ACK with ECN setup */
1232 u_int32_t tcps_ecn_lost_syn; /* Lost SYN with ECN setup */
1233 u_int32_t tcps_ecn_not_supported; /* Server did not support ECN setup */
1234 u_int32_t tcps_ecn_recv_ce; /* Received CE from the network */
1235 u_int32_t tcps_ecn_ace_recv_ce; /* CE count received in ACE field */
1236 u_int32_t tcps_ecn_conn_recv_ce; /* Number of connections received CE atleast once */
1237 u_int32_t tcps_ecn_conn_recv_ece; /* Number of connections received ECE atleast once */
1238 u_int32_t tcps_ecn_conn_plnoce; /* Number of connections that received no CE and sufferred packet loss */
1239 u_int32_t tcps_ecn_conn_pl_ce; /* Number of connections that received CE and sufferred packet loss */
1240 u_int32_t tcps_ecn_conn_nopl_ce; /* Number of connections that received CE and sufferred no packet loss */
1241 u_int32_t tcps_ecn_fallback_synloss; /* Number of times we did fall back due to SYN-Loss */
1242 u_int32_t tcps_ecn_fallback_reorder; /* Number of times we fallback because we detected the PAWS-issue */
1243 u_int32_t tcps_ecn_fallback_ce; /* Number of times we fallback because we received too many CEs */
1244
1245 /* TFO-related statistics */
1246 u_int32_t tcps_tfo_syn_data_rcv; /* Received a SYN+data with valid cookie */
1247 u_int32_t tcps_tfo_cookie_req_rcv;/* Received a TFO cookie-request */
1248 u_int32_t tcps_tfo_cookie_sent; /* Offered a TFO-cookie to the client */
1249 u_int32_t tcps_tfo_cookie_invalid;/* Received an invalid TFO-cookie */
1250 u_int32_t tcps_tfo_cookie_req; /* Cookie requested with the SYN */
1251 u_int32_t tcps_tfo_cookie_rcv; /* Cookie received in a SYN/ACK */
1252 u_int32_t tcps_tfo_syn_data_sent; /* SYN+data+cookie sent */
1253 u_int32_t tcps_tfo_syn_data_acked;/* SYN+data has been acknowledged */
1254 u_int32_t tcps_tfo_syn_loss; /* SYN+TFO has been lost and we fallback */
1255 u_int32_t tcps_tfo_blackhole; /* TFO got blackholed by a middlebox. */
1256 u_int32_t tcps_tfo_cookie_wrong; /* TFO-cookie we sent was wrong */
1257 u_int32_t tcps_tfo_no_cookie_rcv; /* We asked for a cookie but didn't get one */
1258 u_int32_t tcps_tfo_heuristics_disable; /* TFO got disabled due to heuristics */
1259 u_int32_t tcps_tfo_sndblackhole; /* TFO got blackholed in the sending direction */
1260 u_int32_t tcps_mss_to_default; /* Change MSS to default using link status report */
1261 u_int32_t tcps_mss_to_medium; /* Change MSS to medium using link status report */
1262 u_int32_t tcps_mss_to_low; /* Change MSS to low using link status report */
1263 u_int32_t tcps_ecn_fallback_droprst; /* ECN fallback caused by connection drop due to RST */
1264 u_int32_t tcps_ecn_fallback_droprxmt; /* ECN fallback due to drop after multiple retransmits */
1265 u_int32_t tcps_ecn_fallback_synrst; /* ECN fallback due to rst after syn */
1266
1267 u_int32_t tcps_mptcp_rcvmemdrop; /* MPTCP packets dropped for lack of memory */
1268 u_int32_t tcps_mptcp_rcvduppack; /* MPTCP duplicate-only packets received */
1269 u_int32_t tcps_mptcp_rcvpackafterwin; /* MPTCP packets with data after window */
1270
1271 /* TCP timer statistics */
1272 u_int32_t tcps_timer_drift_le_1_ms; /* Timer drift less or equal to 1 ms */
1273 u_int32_t tcps_timer_drift_le_10_ms; /* Timer drift less or equal to 10 ms */
1274 u_int32_t tcps_timer_drift_le_20_ms; /* Timer drift less or equal to 20 ms */
1275 u_int32_t tcps_timer_drift_le_50_ms; /* Timer drift less or equal to 50 ms */
1276 u_int32_t tcps_timer_drift_le_100_ms; /* Timer drift less or equal to 100 ms */
1277 u_int32_t tcps_timer_drift_le_200_ms; /* Timer drift less or equal to 200 ms */
1278 u_int32_t tcps_timer_drift_le_500_ms; /* Timer drift less or equal to 500 ms */
1279 u_int32_t tcps_timer_drift_le_1000_ms; /* Timer drift less or equal to 1000 ms */
1280 u_int32_t tcps_timer_drift_gt_1000_ms; /* Timer drift greater than 1000 ms */
1281
1282 u_int32_t tcps_mptcp_handover_attempt; /* Total number of MPTCP-attempts using handover mode */
1283 u_int32_t tcps_mptcp_interactive_attempt; /* Total number of MPTCP-attempts using interactive mode */
1284 u_int32_t tcps_mptcp_aggregate_attempt; /* Total number of MPTCP-attempts using aggregate mode */
1285 u_int32_t tcps_mptcp_fp_handover_attempt; /* Same as previous three but only for first-party apps */
1286 u_int32_t tcps_mptcp_fp_interactive_attempt;
1287 u_int32_t tcps_mptcp_fp_aggregate_attempt;
1288 u_int32_t tcps_mptcp_heuristic_fallback; /* Total number of MPTCP-connections that fell back due to heuristics */
1289 u_int32_t tcps_mptcp_fp_heuristic_fallback; /* Same as previous but for first-party apps */
1290 u_int32_t tcps_mptcp_handover_success_wifi; /* Total number of successfull handover-mode connections that *started* on WiFi */
1291 u_int32_t tcps_mptcp_handover_success_cell; /* Total number of successfull handover-mode connections that *started* on Cell */
1292 u_int32_t tcps_mptcp_interactive_success; /* Total number of interactive-mode connections that negotiated MPTCP */
1293 u_int32_t tcps_mptcp_aggregate_success; /* Same as previous but for aggregate */
1294 u_int32_t tcps_mptcp_fp_handover_success_wifi; /* Same as previous four, but for first-party apps */
1295 u_int32_t tcps_mptcp_fp_handover_success_cell;
1296 u_int32_t tcps_mptcp_fp_interactive_success;
1297 u_int32_t tcps_mptcp_fp_aggregate_success;
1298 u_int32_t tcps_mptcp_handover_cell_from_wifi; /* Total number of connections that use cell in handover-mode (coming from WiFi) */
1299 u_int32_t tcps_mptcp_handover_wifi_from_cell; /* Total number of connections that use WiFi in handover-mode (coming from cell) */
1300 u_int32_t tcps_mptcp_interactive_cell_from_wifi; /* Total number of connections that use cell in interactive mode (coming from WiFi) */
1301 u_int64_t tcps_mptcp_handover_cell_bytes; /* Total number of bytes sent on cell in handover-mode (on new subflows, ignoring initial one) */
1302 u_int64_t tcps_mptcp_interactive_cell_bytes; /* Same as previous but for interactive */
1303 u_int64_t tcps_mptcp_aggregate_cell_bytes;
1304 u_int64_t tcps_mptcp_handover_all_bytes; /* Total number of bytes sent in handover */
1305 u_int64_t tcps_mptcp_interactive_all_bytes;
1306 u_int64_t tcps_mptcp_aggregate_all_bytes;
1307 u_int32_t tcps_mptcp_back_to_wifi; /* Total number of connections that succeed to move traffic away from cell (when starting on cell) */
1308 u_int32_t tcps_mptcp_wifi_proxy; /* Total number of new subflows that fell back to regular TCP on cell */
1309 u_int32_t tcps_mptcp_cell_proxy; /* Total number of new subflows that fell back to regular TCP on WiFi */
1310
1311 /* TCP offload statistics */
1312 u_int32_t tcps_ka_offload_drops; /* Keep alive drops for timeout reported by firmware */
1313
1314 u_int32_t tcps_mptcp_triggered_cell; /* Total number of times an MPTCP-connection triggered cell bringup */
1315
1316 u_int32_t tcps_fin_timeout_drops;
1317};
1318
1319
1320struct tcpstat_local {
1321 u_int64_t badformat;
1322 u_int64_t unspecv6;
1323 u_int64_t synfin;
1324 u_int64_t badformatipsec;
1325 u_int64_t noconnnolist;
1326 u_int64_t noconnlist;
1327 u_int64_t listbadsyn;
1328 u_int64_t icmp6unreach;
1329 u_int64_t deprecate6;
1330 u_int64_t ooopacket;
1331 u_int64_t rstinsynrcv;
1332 u_int64_t dospacket;
1333 u_int64_t cleanup;
1334 u_int64_t synwindow;
1335};
1336
1337#pragma pack(4)
1338
1339/*
1340 * TCB structure exported to user-land via sysctl(3).
1341 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
1342 * included. Not all of our clients do.
1343 */
1344
1345struct xtcpcb {
1346 u_int32_t xt_len;
1347#ifdef KERNEL_PRIVATE
1348 struct inpcb_compat xt_inp;
1349#else
1350 struct inpcb xt_inp;
1351#endif
1352#ifdef KERNEL_PRIVATE
1353 struct otcpcb xt_tp;
1354#else
1355 struct tcpcb xt_tp;
1356#endif
1357 struct xsocket xt_socket;
1358 u_quad_t xt_alignment_hack;
1359};
1360
1361#if XNU_TARGET_OS_OSX || KERNEL || !(TARGET_OS_IPHONE && !TARGET_OS_SIMULATOR)
1362
1363struct xtcpcb64 {
1364 u_int32_t xt_len;
1365 struct xinpcb64 xt_inpcb;
1366
1367 u_int64_t t_segq;
1368 int t_dupacks; /* consecutive dup acks recd */
1369
1370 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */
1371
1372 int t_state; /* state of this connection */
1373 u_int t_flags;
1374
1375 int t_force; /* 1 if forcing out a byte */
1376
1377 tcp_seq snd_una; /* send unacknowledged */
1378 tcp_seq snd_max; /* highest sequence number sent;
1379 * used to recognize retransmits
1380 */
1381 tcp_seq snd_nxt; /* send next */
1382 tcp_seq snd_up; /* send urgent pointer */
1383
1384 tcp_seq snd_wl1; /* window update seg seq number */
1385 tcp_seq snd_wl2; /* window update seg ack number */
1386 tcp_seq iss; /* initial send sequence number */
1387 tcp_seq irs; /* initial receive sequence number */
1388
1389 tcp_seq rcv_nxt; /* receive next */
1390 tcp_seq rcv_adv; /* advertised window */
1391 u_int32_t rcv_wnd; /* receive window */
1392 tcp_seq rcv_up; /* receive urgent pointer */
1393
1394 u_int32_t snd_wnd; /* send window */
1395 u_int32_t snd_cwnd; /* congestion-controlled window */
1396 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for
1397 * for slow start exponential to
1398 * linear switch
1399 */
1400 u_int t_maxopd; /* mss plus options */
1401
1402 u_int32_t t_rcvtime; /* time at which a packet was received */
1403 u_int32_t t_starttime; /* time connection was established */
1404 int t_rtttime; /* round trip time */
1405 tcp_seq t_rtseq; /* sequence number being timed */
1406
1407 int t_rxtcur; /* current retransmit value (ticks) */
1408 u_int t_maxseg; /* maximum segment size */
1409 int t_srtt; /* smoothed round-trip time */
1410 int t_rttvar; /* variance in round-trip time */
1411
1412 int t_rxtshift; /* log(2) of rexmt exp. backoff */
1413 u_int t_rttmin; /* minimum rtt allowed */
1414 u_int32_t t_rttupdated; /* number of times rtt sampled */
1415 u_int32_t max_sndwnd; /* largest window peer has offered */
1416
1417 int t_softerror; /* possible error not yet reported */
1418/* out-of-band data */
1419 char t_oobflags; /* have some */
1420 char t_iobc; /* input character */
1421/* RFC 1323 variables */
1422 u_char snd_scale; /* window scaling for send window */
1423 u_char rcv_scale; /* window scaling for recv window */
1424 u_char request_r_scale; /* pending window scaling */
1425 u_char requested_s_scale;
1426 u_int32_t ts_recent; /* timestamp echo data */
1427
1428 u_int32_t ts_recent_age; /* when last updated */
1429 tcp_seq last_ack_sent;
1430/* RFC 1644 variables */
1431 tcp_cc cc_send; /* send connection count */
1432 tcp_cc cc_recv; /* receive connection count */
1433 tcp_seq snd_recover; /* for use in fast recovery */
1434/* experimental */
1435 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */
1436 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */
1437 u_int32_t t_badrxtwin; /* window for retransmit recovery */
1438
1439 u_quad_t xt_alignment_hack;
1440};
1441
1442#endif /* XNU_TARGET_OS_OSX || KERNEL || !(TARGET_OS_IPHONE && !TARGET_OS_SIMULATOR) */
1443
1444#ifdef PRIVATE
1445
1446struct xtcpcb_n {
1447 u_int32_t xt_len;
1448 u_int32_t xt_kind; /* XSO_TCPCB */
1449
1450 u_int64_t t_segq;
1451 int t_dupacks; /* consecutive dup acks recd */
1452
1453 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */
1454
1455 int t_state; /* state of this connection */
1456 u_int t_flags;
1457
1458 int t_force; /* 1 if forcing out a byte */
1459
1460 tcp_seq snd_una; /* send unacknowledged */
1461 tcp_seq snd_max; /* highest sequence number sent;
1462 * used to recognize retransmits
1463 */
1464 tcp_seq snd_nxt; /* send next */
1465 tcp_seq snd_up; /* send urgent pointer */
1466
1467 tcp_seq snd_wl1; /* window update seg seq number */
1468 tcp_seq snd_wl2; /* window update seg ack number */
1469 tcp_seq iss; /* initial send sequence number */
1470 tcp_seq irs; /* initial receive sequence number */
1471
1472 tcp_seq rcv_nxt; /* receive next */
1473 tcp_seq rcv_adv; /* advertised window */
1474 u_int32_t rcv_wnd; /* receive window */
1475 tcp_seq rcv_up; /* receive urgent pointer */
1476
1477 u_int32_t snd_wnd; /* send window */
1478 u_int32_t snd_cwnd; /* congestion-controlled window */
1479 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for
1480 * for slow start exponential to
1481 * linear switch
1482 */
1483 u_int t_maxopd; /* mss plus options */
1484
1485 u_int32_t t_rcvtime; /* time at which a packet was received */
1486 u_int32_t t_starttime; /* time connection was established */
1487 int t_rtttime; /* round trip time */
1488 tcp_seq t_rtseq; /* sequence number being timed */
1489
1490 int t_rxtcur; /* current retransmit value (ticks) */
1491 u_int t_maxseg; /* maximum segment size */
1492 int t_srtt; /* smoothed round-trip time */
1493 int t_rttvar; /* variance in round-trip time */
1494
1495 int t_rxtshift; /* log(2) of rexmt exp. backoff */
1496 u_int t_rttmin; /* minimum rtt allowed */
1497 u_int32_t t_rttupdated; /* number of times rtt sampled */
1498 u_int32_t max_sndwnd; /* largest window peer has offered */
1499
1500 int t_softerror; /* possible error not yet reported */
1501 /* out-of-band data */
1502 char t_oobflags; /* have some */
1503 char t_iobc; /* input character */
1504 /* RFC 1323 variables */
1505 u_char snd_scale; /* window scaling for send window */
1506 u_char rcv_scale; /* window scaling for recv window */
1507 u_char request_r_scale; /* pending window scaling */
1508 u_char requested_s_scale;
1509 u_int32_t ts_recent; /* timestamp echo data */
1510
1511 u_int32_t ts_recent_age; /* when last updated */
1512 tcp_seq last_ack_sent;
1513 /* RFC 1644 variables */
1514 tcp_cc cc_send; /* send connection count */
1515 tcp_cc cc_recv; /* receive connection count */
1516 tcp_seq snd_recover; /* for use in fast recovery */
1517 /* experimental */
1518 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */
1519 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */
1520};
1521
1522/*
1523 * The rtt measured is in milliseconds as the timestamp granularity is
1524 * a millisecond. The smoothed round-trip time and estimated variance
1525 * are stored as fixed point numbers scaled by the values below.
1526 * For convenience, these scales are also used in smoothing the average
1527 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
1528 * With these scales, srtt has 5 bits to the right of the binary point,
1529 * and thus an "ALPHA" of 0.875. rttvar has 4 bits to the right of the
1530 * binary point, and is smoothed with an ALPHA of 0.75.
1531 */
1532#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
1533#define TCP_RTT_SHIFT 5 /* shift for srtt; 5 bits frac. */
1534#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 4 bits */
1535#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 4 bits */
1536#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
1537
1538
1539/*
1540 * TCP structure with information that gives insight into forward progress on an interface,
1541 * exported to user-land via sysctl(3).
1542 */
1543struct xtcpprogress_indicators {
1544 u_int32_t xp_numflows; /* Total number of flows */
1545 u_int32_t xp_conn_probe_fails; /* Count of connection failures */
1546 u_int32_t xp_read_probe_fails; /* Count of read probe failures */
1547 u_int32_t xp_write_probe_fails; /* Count of write failures */
1548 u_int32_t xp_recentflows; /* Total of "recent" flows */
1549 u_int32_t xp_recentflows_unacked; /* Total of "recent" flows with unacknowledged data */
1550 u_int64_t xp_recentflows_rxbytes; /* Total of "recent" flows received bytes */
1551 u_int64_t xp_recentflows_txbytes; /* Total of "recent" flows transmitted bytes */
1552 u_int64_t xp_recentflows_rxooo; /* Total of "recent" flows received out of order bytes */
1553 u_int64_t xp_recentflows_rxdup; /* Total of "recent" flows received duplicate bytes */
1554 u_int64_t xp_recentflows_retx; /* Total of "recent" flows retransmitted bytes */
1555 u_int64_t xp_reserved1; /* Expansion */
1556 u_int64_t xp_reserved2; /* Expansion */
1557 u_int64_t xp_reserved3; /* Expansion */
1558 u_int64_t xp_reserved4; /* Expansion */
1559};
1560
1561struct tcpprogressreq {
1562 u_int64_t ifindex; /* Interface index for progress indicators */
1563 u_int64_t recentflow_maxduration; /* In mach_absolute_time, max duration for flow to be counted as "recent" */
1564 u_int64_t filter_flags; /* Optional additional filtering, values are interface properties per ntstat.h */
1565 u_int64_t xp_reserved2; /* Expansion */
1566};
1567
1568struct tcpprobereq {
1569 u_int64_t ifindex; /* Optional interface index for TCP keep-alive probing */
1570 u_int64_t enable; /* Flag to enable or disable probing (1=on, 0=off)*/
1571 u_int64_t filter_flags; /* Optional flags for filtering interfaces per ntstat.h (NSTAT_IFNET_IS_*) */
1572 u_int32_t reserved; /* Expansion */
1573 u_int32_t reserved2; /* Expansion */
1574};
1575
1576#endif /* PRIVATE */
1577
1578#pragma pack()
1579
1580/*
1581 * Names for TCP sysctl objects
1582 */
1583#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
1584#define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */
1585#define TCPCTL_MSSDFLT 3 /* MSS default */
1586#define TCPCTL_STATS 4 /* statistics (read-only) */
1587#define TCPCTL_RTTDFLT 5 /* default RTT estimate */
1588#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
1589#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
1590#define TCPCTL_SENDSPACE 8 /* send buffer space */
1591#define TCPCTL_RECVSPACE 9 /* receive buffer space */
1592#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
1593#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
1594#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
1595#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
1596#define TCPCTL_MAXID 14
1597
1598#ifdef BSD_KERNEL_PRIVATE
1599#include <sys/bitstring.h>
1600
1601#define TCP_PKTLIST_CLEAR(tp) { \
1602 (tp)->t_pktlist_head = (tp)->t_pktlist_tail = NULL; \
1603 (tp)->t_lastchain = (tp)->t_pktlist_sentlen = 0; \
1604}
1605
1606extern int tcp_TCPTV_MIN;
1607
1608#ifdef SYSCTL_DECL
1609SYSCTL_DECL(_net_inet_tcp);
1610#endif /* SYSCTL_DECL */
1611
1612extern struct inpcbhead tcb; /* head of queue of active tcpcb's */
1613extern struct inpcbinfo tcbinfo;
1614extern struct tcpstat tcpstat; /* tcp statistics */
1615extern int tcp_mssdflt; /* XXX */
1616extern int tcp_minmss;
1617#define TCP_FASTOPEN_SERVER 0x01
1618#define TCP_FASTOPEN_CLIENT 0x02
1619
1620extern int tcp_tfo_halfcnt;
1621extern int tcp_tfo_backlog;
1622extern int tcp_fastopen;
1623extern int ss_fltsz_local;
1624extern int target_qdelay;
1625extern u_int32_t tcp_now; /* for RFC 1323 timestamps */
1626extern struct timeval tcp_uptime;
1627extern lck_spin_t tcp_uptime_lock;
1628extern int tcp_delack_enabled;
1629extern int maxseg_unacked;
1630KALLOC_TYPE_DECLARE(tcp_reass_zone);
1631KALLOC_TYPE_DECLARE(tcp_rxt_seg_zone);
1632extern int tcp_ecn_outbound;
1633extern int tcp_ecn_inbound;
1634extern uint32_t tcp_do_autorcvbuf;
1635extern uint32_t tcp_autorcvbuf_max;
1636extern int tcp_recv_bg;
1637extern int tcp_do_ack_compression;
1638extern int tcp_randomize_timestamps;
1639extern int tcp_rledbat;
1640extern int tcp_use_min_curr_rtt;
1641extern int tcp_do_timestamps;
1642/*
1643 * Dummy value used for when there is no flow and we want to ensure that compression
1644 * can happen.
1645 */
1646#define TCP_ACK_COMPRESSION_DUMMY 1
1647
1648extern int tcp_do_better_lr;
1649extern int tcp_cubic_minor_fixes;
1650extern int tcp_cubic_rfc_compliant;
1651extern int tcp_flow_control_response;
1652
1653extern int tcp_reass_total_qlen;
1654
1655struct protosw;
1656struct domain;
1657
1658struct tcp_respond_args {
1659 unsigned int ifscope;
1660 unsigned int nocell:1,
1661 noexpensive:1,
1662 awdl_unrestricted:1,
1663 intcoproc_allowed:1,
1664 keep_alive:1,
1665 noconstrained:1,
1666 management_allowed:1;
1667};
1668
1669void tcp_canceltimers(struct tcpcb *);
1670struct tcpcb *
1671tcp_close(struct tcpcb *);
1672void tcp_ctlinput(int, struct sockaddr *, void *, struct ifnet *);
1673int tcp_ctloutput(struct socket *, struct sockopt *);
1674struct tcpcb *
1675tcp_drop(struct tcpcb *, int);
1676void tcp_drain(void);
1677void tcp_getrt_rtt(struct tcpcb *tp, struct rtentry *rt);
1678void tcp_init(struct protosw *, struct domain *);
1679void tcp_input(struct mbuf *, int);
1680void tcp_mss(struct tcpcb *, int, unsigned int);
1681uint32_t tcp_ceil(double a);
1682uint32_t tcp_round_to(uint32_t val, uint32_t round);
1683uint32_t tcp_round_up(uint32_t val, uint32_t base);
1684int tcp_mssopt(struct tcpcb *);
1685void tcp_drop_syn_sent(struct inpcb *, int);
1686void tcp_mtudisc(struct inpcb *, int);
1687struct tcpcb *
1688tcp_newtcpcb(struct inpcb *);
1689int tcp_output(struct tcpcb *);
1690void tcp_respond(struct tcpcb *, void *, struct tcphdr *, struct mbuf *,
1691 tcp_seq, tcp_seq, uint8_t, struct tcp_respond_args *);
1692struct rtentry *
1693tcp_rtlookup(struct inpcb *, unsigned int);
1694void tcp_setpersist(struct tcpcb *);
1695void tcp_gc(struct inpcbinfo *);
1696void tcp_itimer(struct inpcbinfo *ipi);
1697void tcp_check_timer_state(struct tcpcb *tp);
1698void tcp_run_timerlist(void *arg1, void *arg2);
1699void tcp_sched_timers(struct tcpcb *tp);
1700
1701struct tcptemp *tcp_maketemplate(struct tcpcb *, struct mbuf **);
1702void tcp_fillheaders(struct mbuf *, struct tcpcb *, void *, void *);
1703struct tcpcb *tcp_timers(struct tcpcb *, int);
1704void tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int);
1705
1706void tcp_fill_info(struct tcpcb *, struct tcp_info *);
1707void tcp_sack_doack(struct tcpcb *, struct tcpopt *, struct tcphdr *,
1708 u_int32_t *, uint32_t *);
1709extern boolean_t tcp_sack_process_dsack(struct tcpcb *, struct tcpopt *,
1710 struct tcphdr *);
1711int tcp_detect_bad_rexmt(struct tcpcb *, struct tcphdr *, struct tcpopt *,
1712 u_int32_t rxtime);
1713void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
1714void tcp_clean_sackreport(struct tcpcb *tp);
1715uint32_t tcp_sack_adjust(struct tcpcb *tp);
1716struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
1717void tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
1718void tcp_free_sackholes(struct tcpcb *tp);
1719void tcp_sack_lost_rexmit(struct tcpcb *tp);
1720int32_t tcp_sbspace(struct tcpcb *tp);
1721void tcp_set_tso(struct tcpcb *tp, struct ifnet *ifp);
1722void tcp_set_ecn(struct tcpcb *tp, struct ifnet *ifp);
1723uint8_t tcp_get_ace(struct tcphdr *th);
1724int tcp_flight_size(struct tcpcb *tp);
1725void tcp_reset_stretch_ack(struct tcpcb *tp);
1726extern void tcp_get_ports_used(ifnet_t ifp, int, u_int32_t, bitstr_t *);
1727uint32_t tcp_count_opportunistic(unsigned int ifindex, u_int32_t flags);
1728uint32_t tcp_find_anypcb_byaddr(struct ifaddr *ifa);
1729void tcp_set_max_rwinscale(struct tcpcb *tp, struct socket *so);
1730struct bwmeas* tcp_bwmeas_alloc(struct tcpcb *tp);
1731void tcp_bwmeas_free(struct tcpcb *tp);
1732extern int32_t timer_diff(uint32_t t1, uint32_t toff1, uint32_t t2, uint32_t toff2);
1733
1734extern void tcp_set_background_cc(struct socket *);
1735extern void tcp_set_foreground_cc(struct socket *);
1736extern void tcp_set_recv_bg(struct socket *);
1737extern void tcp_clear_recv_bg(struct socket *);
1738extern boolean_t tcp_sack_byte_islost(struct tcpcb *tp);
1739#define IS_TCP_RECV_BG(_so) \
1740 ((_so)->so_flags1 & SOF1_TRAFFIC_MGT_TCP_RECVBG)
1741
1742#if TRAFFIC_MGT
1743#define CLEAR_IAJ_STATE(_tp_) (_tp_)->iaj_rcv_ts = 0
1744void reset_acc_iaj(struct tcpcb *tp);
1745#endif /* TRAFFIC_MGT */
1746
1747int tcp_lock(struct socket *, int, void *);
1748int tcp_unlock(struct socket *, int, void *);
1749void calculate_tcp_clock(void);
1750
1751extern void tcp_keepalive_reset(struct tcpcb *);
1752extern uint32_t get_base_rtt(struct tcpcb *tp);
1753
1754#ifdef _KERN_LOCKS_H_
1755lck_mtx_t * tcp_getlock(struct socket *, int);
1756#else
1757void * tcp_getlock(struct socket *, int);
1758#endif
1759
1760extern int faster_mcopy;
1761extern struct pr_usrreqs tcp_usrreqs;
1762extern u_int32_t tcp_sendspace;
1763extern u_int32_t tcp_recvspace;
1764tcp_seq tcp_new_isn(struct tcpcb *);
1765
1766extern int tcp_input_checksum(int, struct mbuf *, struct tcphdr *, int, int);
1767extern void tcp_getconninfo(struct socket *, struct conninfo_tcp *);
1768extern void add_to_time_wait(struct tcpcb *, uint32_t delay);
1769extern void add_to_time_wait_now(struct tcpcb *tp, uint32_t delay);
1770extern void tcp_pmtud_revert_segment_size(struct tcpcb *tp);
1771extern void tcp_rxtseg_insert(struct tcpcb *, tcp_seq, tcp_seq);
1772extern struct tcp_rxt_seg *tcp_rxtseg_find(struct tcpcb *, tcp_seq, tcp_seq);
1773extern void tcp_rxtseg_set_spurious(struct tcpcb *tp, tcp_seq start, tcp_seq end);
1774extern void tcp_rxtseg_clean(struct tcpcb *);
1775extern boolean_t tcp_rxtseg_detect_bad_rexmt(struct tcpcb *, tcp_seq);
1776extern boolean_t tcp_rxtseg_dsack_for_tlp(struct tcpcb *);
1777extern u_int32_t tcp_rxtseg_total_size(struct tcpcb *tp);
1778extern void tcp_rexmt_save_state(struct tcpcb *tp);
1779extern void tcp_interface_send_probe(u_int16_t if_index_available);
1780extern void tcp_probe_connectivity(struct ifnet *ifp, u_int32_t enable);
1781extern void tcp_get_connectivity_status(struct tcpcb *,
1782 struct tcp_conn_status *);
1783
1784extern void tcp_clear_keep_alive_offload(struct socket *so);
1785extern void tcp_fill_keepalive_offload_frames(struct ifnet *,
1786 struct ifnet_keepalive_offload_frame *, u_int32_t, size_t, u_int32_t *);
1787extern int tcp_notify_kao_timeout(ifnet_t ifp,
1788 struct ifnet_keepalive_offload_frame *frame);
1789
1790extern boolean_t tfo_enabled(const struct tcpcb *tp);
1791extern void tcp_disable_tfo(struct tcpcb *tp);
1792extern void tcp_tfo_gen_cookie(struct inpcb *inp, u_char *out, size_t blk_size);
1793#define TCP_FASTOPEN_KEYLEN 16
1794extern int tcp_freeq(struct tcpcb *tp);
1795extern errno_t tcp_notify_ack_id_valid(struct tcpcb *, struct socket *, u_int32_t);
1796extern errno_t tcp_add_notify_ack_marker(struct tcpcb *, u_int32_t);
1797extern void tcp_notify_ack_free(struct tcpcb *);
1798extern void tcp_notify_acknowledgement(struct tcpcb *, struct socket *);
1799extern void tcp_get_notify_ack_count(struct tcpcb *,
1800 struct tcp_notify_ack_complete *);
1801extern void tcp_get_notify_ack_ids(struct tcpcb *tp,
1802 struct tcp_notify_ack_complete *);
1803extern void tcp_update_mss_locked(struct socket *, struct ifnet *);
1804
1805extern int get_tcp_inp_list(struct inpcb **, int, inp_gen_t);
1806extern bool tcp_notify_ack_active(struct socket *so);
1807extern void tcp_set_finwait_timeout(struct tcpcb *);
1808
1809#if MPTCP
1810extern uint16_t mptcp_output_csum(struct mbuf *m, uint64_t dss_val,
1811 uint32_t sseq, uint16_t dlen);
1812extern int mptcp_adj_mss(struct tcpcb *, boolean_t);
1813extern void mptcp_insert_rmap(struct tcpcb *tp, struct mbuf *m, struct tcphdr *th);
1814extern int dump_mptcp_reass_qlen(char *, int);
1815#endif
1816
1817extern int dump_tcp_reass_qlen(char *, int);
1818extern uint32_t tcp_reass_qlen_space(struct socket *);
1819
1820__private_extern__ void tcp_update_stats_per_flow(
1821 struct ifnet_stats_per_flow *, struct ifnet *);
1822
1823#define TCP_ACK_STRATEGY_LEGACY 0
1824#define TCP_ACK_STRATEGY_MODERN 1
1825
1826extern int tcp_ack_strategy;
1827
1828#if SKYWALK
1829void tcp_add_fsw_flow(struct tcpcb *, struct ifnet *);
1830void tcp_del_fsw_flow(struct tcpcb *);
1831#else /* !SKYWALK */
1832#define tcp_add_fsw_flow(...)
1833#define tcp_del_fsw_flow(...)
1834#endif /* !SKYWALK */
1835
1836#endif /* BSD_KERNEL_PRIVATE */
1837
1838#endif /* _NETINET_TCP_VAR_H_ */
1839