summaryrefslogtreecommitdiff
path: root/sys/netinet/tcp_output.c
blob: 8b3c5c576834ec2d357d81e15bd1662d8ccadb8d (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
/*	$OpenBSD: tcp_output.c,v 1.16 1999/01/11 02:01:36 deraadt Exp $	*/
/*	$NetBSD: tcp_output.c,v 1.16 1997/06/03 16:17:09 kml Exp $	*/

/*
 * Copyright (c) 1982, 1986, 1988, 1990, 1993
 *	The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	@(#)tcp_output.c	8.3 (Berkeley) 12/30/93
 */

/*
%%% portions-copyright-nrl-95
Portions of this software are Copyright 1995-1998 by Randall Atkinson,
Ronald Lee, Daniel McDonald, Bao Phan, and Chris Winters. All Rights
Reserved. All rights under this copyright have been assigned to the US
Naval Research Laboratory (NRL). The NRL Copyright Notice and License
Agreement Version 1.1 (January 17, 1995) applies to these portions of the
software.
You should have received a copy of the license with this software. If you
didn't get a copy, you may request one from <license@ipv6.nrl.navy.mil>.
*/

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/errno.h>
#include <sys/domain.h>

#include <net/route.h>

#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/in_pcb.h>
#include <netinet/ip_var.h>
#include <netinet/tcp.h>
#define	TCPOUTFLAGS
#include <netinet/tcp_fsm.h>
#include <netinet/tcp_seq.h>
#include <netinet/tcp_timer.h>
#include <netinet/tcp_var.h>
#include <netinet/tcpip.h>
#include <netinet/tcp_debug.h>

#ifdef TUBA
#include <netiso/iso.h>
#include <netiso/tuba_table.h>
#endif

#ifdef INET6
#include <netinet6/tcpipv6.h>
#endif /* INET6 */

#ifdef notyet
extern struct mbuf *m_copypack();
#endif

#ifdef TCP_SACK
extern int tcprexmtthresh;
#endif

#ifdef TCP_SACK
#define MAX_TCPOPTLEN	40	/* need 40 at least for 3 SACKs + TIMESTAMP */
#else
#define MAX_TCPOPTLEN	32	/* max # bytes that go in options */
#endif

#ifdef TCP_SACK
#ifdef TCP_SACK_DEBUG
void
tcp_print_holes(tp)
struct tcpcb *tp;
{
	struct sackhole *p = tp->snd_holes;
	if (p == 0)
		return;
	printf("Hole report: start--end dups rxmit\n");
	while (p) {
		printf("%x--%x d %d r %x\n",  p->start, p->end, p->dups,
                    p->rxmit);
		p = p->next;
	}
	printf("\n");
}
#endif /* TCP_SACK_DEBUG */

/*
 * Returns pointer to a sackhole if there are any pending retransmissions;
 * NULL otherwise.
 */
struct sackhole *
tcp_sack_output(tp)
register struct tcpcb *tp;
{
	struct sackhole *p;
	if (tp->sack_disable)
		return 0;
	p = tp->snd_holes;
	while (p) {
		if (p->dups >= tcprexmtthresh && SEQ_LT(p->rxmit, p->end)) {
			if (SEQ_LT(p->rxmit, tp->snd_una)) {/* old SACK hole */
				p = p->next;
				continue;
			}
#ifdef TCP_SACK_DEBUG
			if (p)
				tcp_print_holes(tp);
#endif
			return p;
		}
        	p = p->next;
	}
	return 0;
}

/*
 * After a timeout, the SACK list may be rebuilt.  This SACK information
 * should be used to avoid retransmitting SACKed data.  This function
 * traverses the SACK list to see if snd_nxt should be moved forward.
 */
void
tcp_sack_adjust(tp)
	struct tcpcb *tp;
{
	int i;

	for (i = 0; i < tp->rcv_numsacks; i++) {
		if (SEQ_LT(tp->snd_nxt, tp->sackblks[i].start))
			break;
		if (SEQ_LEQ(tp->sackblks[i].end, tp->snd_nxt))
			continue;
		if (tp->sackblks[i].start == 0 && tp->sackblks[i].end == 0)
			continue;
		/* snd_nxt must be in middle of block of SACKed data */
		tp->snd_nxt = tp->sackblks[i].end;
		break;
	}
}
#endif /* TCP_SACK */

/*
 * Tcp output routine: figure out what should be sent and send it.
 */
int
tcp_output(tp)
	register struct tcpcb *tp;
{
	register struct socket *so = tp->t_inpcb->inp_socket;
	register long len, win;
	int off, flags, error;
	register struct mbuf *m;
	register struct tcpiphdr *ti;
	register struct tcphdr *th;
#ifdef INET6
	register struct tcpipv6hdr *ti6 = NULL;
#endif /* INET6 */
	u_char opt[MAX_TCPOPTLEN];
	unsigned int optlen, hdrlen;
	int idle, sendalot;
#ifdef TCP_SACK
	int i, sack_rxmit = 0;
	struct sackhole *p;
#endif
#if defined(TCP_SACK) || defined(TCP_NEWRENO)
	int maxburst = TCP_MAXBURST;
#endif

	/*
	 * Determine length of data that should be transmitted,
	 * and flags that will be used.
	 * If there is some data or critical controls (SYN, RST)
	 * to send, then transmit; otherwise, investigate further.
	 */
	idle = (tp->snd_max == tp->snd_una);
	if (idle && tp->t_idle >= tp->t_rxtcur)
		/*
		 * We have been idle for "a while" and no acks are
		 * expected to clock out any data we send --
		 * slow start to get ack "clock" running again.
		 */
		tp->snd_cwnd = tp->t_maxseg;
again:
	sendalot = 0;
#ifdef TCP_SACK
	/*
	 * If we've recently taken a timeout, snd_max will be greater than
	 * snd_nxt.  There may be SACK information that allows us to avoid
	 * resending already delivered data.  Adjust snd_nxt accordingly.
	 */
	if (!tp->sack_disable && SEQ_LT(tp->snd_nxt, tp->snd_max))
		tcp_sack_adjust(tp);
#endif
	off = tp->snd_nxt - tp->snd_una;
	win = ulmin(tp->snd_wnd, tp->snd_cwnd);

	flags = tcp_outflags[tp->t_state];
	/*
	 * If in persist timeout with window of 0, send 1 byte.
	 * Otherwise, if window is small but nonzero
	 * and timer expired, we will send what we can
	 * and go to transmit state.
	 */

#ifdef TCP_SACK
	/* 
	 * Send any SACK-generated retransmissions.  If we're explicitly trying
	 * to send out new data (when sendalot is 1), bypass this function.
	 * If we retransmit in fast recovery mode, decrement snd_cwnd, since
	 * we're replacing a (future) new transmission with a retransmission 
	 * now, and we previously incremented snd_cwnd in tcp_input().
	 */
	if (!tp->sack_disable && !sendalot) {
		if ((p = tcp_sack_output(tp))) {
			off = p->rxmit - tp->snd_una;
			sack_rxmit = 1;
#if 0
			/* Coalesce holes into a single retransmission */
#endif
			len = min(tp->t_maxseg, p->end - p->rxmit);
#ifndef TCP_FACK
			/* in FACK, hold snd_cwnd constant during recovery */
			if (SEQ_LT(tp->snd_una, tp->snd_last))
				tp->snd_cwnd -= tp->t_maxseg;
#endif
    		}
	}
#endif /* TCP_SACK */

	if (tp->t_force) {
		if (win == 0) {
			/*
			 * If we still have some data to send, then
			 * clear the FIN bit.  Usually this would
			 * happen below when it realizes that we
			 * aren't sending all the data.  However,
			 * if we have exactly 1 byte of unset data,
			 * then it won't clear the FIN bit below,
			 * and if we are in persist state, we wind
			 * up sending the packet without recording
			 * that we sent the FIN bit.
			 *
			 * We can't just blindly clear the FIN bit,
			 * because if we don't have any more data
			 * to send then the probe will be the FIN
			 * itself.
			 */
			if (off < so->so_snd.sb_cc)
				flags &= ~TH_FIN;
			win = 1;
		} else {
			tp->t_timer[TCPT_PERSIST] = 0;
			tp->t_rxtshift = 0;
		}
	}

#ifdef TCP_SACK
	if (!sack_rxmit) {
#endif
	len = ulmin(so->so_snd.sb_cc, win) - off;

#if defined(TCP_SACK) && defined(TCP_FACK)
	/* 
	 * If we're in fast recovery (SEQ_GT(tp->snd_last, tp->snd_una)), and 
	 * amount of outstanding data (snd_awnd) is >= snd_cwnd, then
	 * do not send data (like zero window conditions)
	 */
	if (!tp->sack_disable && len && SEQ_GT(tp->snd_last, tp->snd_una) && 
	    (tp->snd_awnd >= tp->snd_cwnd)) 
		len = 0;
#endif /* TCP_FACK */
#ifdef TCP_SACK
	}
#endif

	if (len < 0) {
		/*
		 * If FIN has been sent but not acked,
		 * but we haven't been called to retransmit,
		 * len will be -1.  Otherwise, window shrank
		 * after we sent into it.  If window shrank to 0,
		 * cancel pending retransmit and pull snd_nxt
		 * back to (closed) window.  We will enter persist
		 * state below.  If the window didn't close completely,
		 * just wait for an ACK.
		 */
		len = 0;
		if (win == 0) {
			tp->t_timer[TCPT_REXMT] = 0;
			tp->snd_nxt = tp->snd_una;
		}
	}
	if (len > tp->t_maxseg) {
		len = tp->t_maxseg;
		sendalot = 1;
	}
	if (SEQ_LT(tp->snd_nxt + len, tp->snd_una + so->so_snd.sb_cc))
		flags &= ~TH_FIN;

	win = sbspace(&so->so_rcv);

	/*
	 * Sender silly window avoidance.  If connection is idle
	 * and can send all data, a maximum segment,
	 * at least a maximum default-size segment do it,
	 * or are forced, do it; otherwise don't bother.
	 * If peer's buffer is tiny, then send
	 * when window is at least half open.
	 * If retransmitting (possibly after persist timer forced us
	 * to send into a small window), then must resend.
	 */
	if (len) {
		if (len == tp->t_maxseg)
			goto send;
		if ((idle || tp->t_flags & TF_NODELAY) &&
		    len + off >= so->so_snd.sb_cc)
			goto send;
		if (tp->t_force)
			goto send;
		if (len >= tp->max_sndwnd / 2 && tp->max_sndwnd > 0)
			goto send;
		if (SEQ_LT(tp->snd_nxt, tp->snd_max))
			goto send;
#ifdef TCP_SACK
		if (sack_rxmit)
			goto send;
#endif
	}

	/*
	 * Compare available window to amount of window
	 * known to peer (as advertised window less
	 * next expected input).  If the difference is at least two
	 * max size segments, or at least 50% of the maximum possible
	 * window, then want to send a window update to peer.
	 */
	if (win > 0) {
		/* 
		 * "adv" is the amount we can increase the window,
		 * taking into account that we are limited by
		 * TCP_MAXWIN << tp->rcv_scale.
		 */
		long adv = lmin(win, (long)TCP_MAXWIN << tp->rcv_scale) -
			(tp->rcv_adv - tp->rcv_nxt);

		if (adv >= (long) (2 * tp->t_maxseg))
			goto send;
		if (2 * adv >= (long) so->so_rcv.sb_hiwat)
			goto send;
	}

	/*
	 * Send if we owe peer an ACK.
	 */
	if (tp->t_flags & TF_ACKNOW)
		goto send;
	if (flags & (TH_SYN|TH_RST))
		goto send;
	if (SEQ_GT(tp->snd_up, tp->snd_una))
		goto send;
	/*
	 * If our state indicates that FIN should be sent
	 * and we have not yet done so, or we're retransmitting the FIN,
	 * then we need to send.
	 */
	if (flags & TH_FIN &&
	    ((tp->t_flags & TF_SENTFIN) == 0 || tp->snd_nxt == tp->snd_una))
		goto send;

	/*
	 * TCP window updates are not reliable, rather a polling protocol
	 * using ``persist'' packets is used to insure receipt of window
	 * updates.  The three ``states'' for the output side are:
	 *	idle			not doing retransmits or persists
	 *	persisting		to move a small or zero window
	 *	(re)transmitting	and thereby not persisting
	 *
	 * tp->t_timer[TCPT_PERSIST]
	 *	is set when we are in persist state.
	 * tp->t_force
	 *	is set when we are called to send a persist packet.
	 * tp->t_timer[TCPT_REXMT]
	 *	is set when we are retransmitting
	 * The output side is idle when both timers are zero.
	 *
	 * If send window is too small, there is data to transmit, and no
	 * retransmit or persist is pending, then go to persist state.
	 * If nothing happens soon, send when timer expires:
	 * if window is nonzero, transmit what we can,
	 * otherwise force out a byte.
	 */
	if (so->so_snd.sb_cc && tp->t_timer[TCPT_REXMT] == 0 &&
	    tp->t_timer[TCPT_PERSIST] == 0) {
		tp->t_rxtshift = 0;
		tcp_setpersist(tp);
	}

	/*
	 * No reason to send a segment, just return.
	 */
	return (0);

send:
	/*
	 * Before ESTABLISHED, force sending of initial options
	 * unless TCP set not to do any options.
	 * NOTE: we assume that the IP/TCP header plus TCP options
	 * always fit in a single mbuf, leaving room for a maximum
	 * link header, i.e.
	 *	max_linkhdr + sizeof (struct tcpiphdr) + optlen <= MHLEN
	 */
#ifdef INET6
        /*
	 * For IPv6, this has changed to be:
	 *      max_linkhdr + sizeof(struct tcphdr) + optlen + 
	 *           sizeof(struct ipv6)  <= MHLEN
	 * This MIGHT be harder...
	 */
#endif /* INET6 */
	optlen = 0;
#ifdef INET6
	if (tp->pf == PF_INET6)  /* if tp->pf is 0, then assume IPv4. */
	  hdrlen = sizeof(struct tcphdr) + sizeof(struct ipv6);
	else
#endif /* INET6 */
	hdrlen = sizeof (struct tcpiphdr);
	if (flags & TH_SYN) {
		tp->snd_nxt = tp->iss;
		if ((tp->t_flags & TF_NOOPT) == 0) {
			u_int16_t mss;

			opt[0] = TCPOPT_MAXSEG;
			opt[1] = 4;
			mss = htons((u_int16_t) tcp_mss(tp, 0));
			bcopy((caddr_t)&mss, (caddr_t)(opt + 2), sizeof(mss));
			optlen = 4;
#ifdef TCP_SACK
			/* 
			 * If this is the first SYN of connection (not a SYN 
			 * ACK), include SACK_PERMIT_HDR option.  If this is a 
			 * SYN ACK, include SACK_PERMIT_HDR option if peer has 
			 * already done so.
			 */
			if (!tp->sack_disable && ((flags & TH_ACK) == 0 ||
			    (tp->t_flags & TF_SACK_PERMIT))) {
				*((u_int32_t *) (opt + optlen)) =
				    htonl(TCPOPT_SACK_PERMIT_HDR);
				optlen += 4;
			}
#endif
	 
			if ((tp->t_flags & TF_REQ_SCALE) &&
			    ((flags & TH_ACK) == 0 ||
			    (tp->t_flags & TF_RCVD_SCALE))) {
				*((u_int32_t *) (opt + optlen)) = htonl(
					TCPOPT_NOP << 24 |
					TCPOPT_WINDOW << 16 |
					TCPOLEN_WINDOW << 8 |
					tp->request_r_scale);
				optlen += 4;
			}
		}
	}
 
	/*
	 * Send a timestamp and echo-reply if this is a SYN and our side 
	 * wants to use timestamps (TF_REQ_TSTMP is set) or both our side
	 * and our peer have sent timestamps in our SYN's.
	 */
	if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP &&
	     (flags & TH_RST) == 0 &&
	    ((flags & (TH_SYN|TH_ACK)) == TH_SYN ||
	     (tp->t_flags & TF_RCVD_TSTMP))) {
		u_int32_t *lp = (u_int32_t *)(opt + optlen);
 
		/* Form timestamp option as shown in appendix A of RFC 1323. */
		*lp++ = htonl(TCPOPT_TSTAMP_HDR);
		*lp++ = htonl(tcp_now);
		*lp   = htonl(tp->ts_recent);
		optlen += TCPOLEN_TSTAMP_APPA;
	}

#ifdef TCP_SACK
	/*
	 * Send SACKs if necessary.  This should be the last option processed.
	 * Only as many SACKs are sent as are permitted by the maximum options
	 * size.  No more than three SACKs are sent.
	 */
	if (!tp->sack_disable && tp->t_state == TCPS_ESTABLISHED &&
	    (tp->t_flags & (TF_SACK_PERMIT|TF_NOOPT)) == TF_SACK_PERMIT &&
	    tp->rcv_numsacks) {
		u_int32_t *lp = (u_int32_t *)(opt + optlen);
		u_int32_t *olp = lp++;
		int count = 0;  /* actual number of SACKs inserted */
		int maxsack = (MAX_TCPOPTLEN - (optlen + 4))/TCPOLEN_SACK;

		maxsack = min(maxsack, TCP_MAX_SACK);
		for (i = 0; (i < tp->rcv_numsacks && count < maxsack); i++) {
			struct sackblk sack = tp->sackblks[i];
			if (sack.start == 0 && sack.end == 0)
				continue;
			*lp++ = htonl(sack.start);
			*lp++ = htonl(sack.end);
			count++;
		}
		*olp = htonl(TCPOPT_SACK_HDR|(TCPOLEN_SACK*count+2));
		optlen += TCPOLEN_SACK*count + 4; /* including leading NOPs */
	}
#endif /* TCP_SACK */

	hdrlen += optlen;
 
	/*
	 * Adjust data length if insertion of options will
	 * bump the packet length beyond the t_maxopd length.
	 */
	if (len > tp->t_maxopd - optlen) {
		len = tp->t_maxopd - optlen;
		sendalot = 1;
		flags &= ~TH_FIN;
	 }

#ifdef DIAGNOSTIC
	if (max_linkhdr + hdrlen > MHLEN)
		panic("tcphdr too big");
#endif

	/*
	 * Grab a header mbuf, attaching a copy of data to
	 * be transmitted, and initialize the header from
	 * the template for sends on this connection.
	 */
	if (len) {
		if (tp->t_force && len == 1)
			tcpstat.tcps_sndprobe++;
		else if (SEQ_LT(tp->snd_nxt, tp->snd_max)) {
			tcpstat.tcps_sndrexmitpack++;
			tcpstat.tcps_sndrexmitbyte += len;
		} else {
			tcpstat.tcps_sndpack++;
			tcpstat.tcps_sndbyte += len;
		}
#ifdef notyet
		if ((m = m_copypack(so->so_snd.sb_mb, off,
		    (int)len, max_linkhdr + hdrlen)) == 0) {
			error = ENOBUFS;
			goto out;
		}
		/*
		 * m_copypack left space for our hdr; use it.
		 */
		m->m_len += hdrlen;
		m->m_data -= hdrlen;
#else
		MGETHDR(m, M_DONTWAIT, MT_HEADER);
		if (m == NULL) {
			error = ENOBUFS;
			goto out;
		}
		m->m_data += max_linkhdr;
		m->m_len = hdrlen;
		if (len <= MHLEN - hdrlen - max_linkhdr) {
			m_copydata(so->so_snd.sb_mb, off, (int) len,
			    mtod(m, caddr_t) + hdrlen);
			m->m_len += len;
		} else {
			m->m_next = m_copy(so->so_snd.sb_mb, off, (int) len);
			if (m->m_next == 0) {
				(void) m_free(m);
				error = ENOBUFS;
				goto out;
			}
		}
#endif
		/*
		 * If we're sending everything we've got, set PUSH.
		 * (This will keep happy those implementations which only
		 * give data to the user when a buffer fills or
		 * a PUSH comes in.)
		 */
		if (off + len == so->so_snd.sb_cc)
			flags |= TH_PUSH;
	} else {
		if (tp->t_flags & TF_ACKNOW)
			tcpstat.tcps_sndacks++;
		else if (flags & (TH_SYN|TH_FIN|TH_RST))
			tcpstat.tcps_sndctrl++;
		else if (SEQ_GT(tp->snd_up, tp->snd_una))
			tcpstat.tcps_sndurg++;
		else
			tcpstat.tcps_sndwinup++;

		MGETHDR(m, M_DONTWAIT, MT_HEADER);
		if (m == NULL) {
			error = ENOBUFS;
			goto out;
		}
		m->m_data += max_linkhdr;
		m->m_len = hdrlen;
	}
	m->m_pkthdr.rcvif = (struct ifnet *)0;
#ifdef INET6
	if (tp->pf == PF_INET6) {
		ti6 = mtod(m, struct tcpipv6hdr *);
		ti = NULL;

		if (!tp->t_template)
			panic("tcp_output");

		bcopy((caddr_t)tp->t_template, (caddr_t)ti6,
			sizeof (struct tcpipv6hdr));

		th = &ti6->ti6_t;
	} else
#endif /* INET6 */
	{
		ti = mtod(m, struct tcpiphdr *);

		if (tp->t_template == 0)
			panic("tcp_output");

		bcopy((caddr_t)tp->t_template, (caddr_t)ti,
		      sizeof (struct tcpiphdr));

		th = &ti->ti_t;
	};

	/*
	 * Fill in fields, remembering maximum advertised
	 * window for use in delaying messages about window sizes.
	 * If resending a FIN, be sure not to use a new sequence number.
	 */
	if ((flags & TH_FIN) && (tp->t_flags & TF_SENTFIN) && 
	    (tp->snd_nxt == tp->snd_max))
		tp->snd_nxt--;
	/*
	 * If we are doing retransmissions, then snd_nxt will
	 * not reflect the first unsent octet.  For ACK only
	 * packets, we do not want the sequence number of the
	 * retransmitted packet, we want the sequence number
	 * of the next unsent octet.  So, if there is no data
	 * (and no SYN or FIN), use snd_max instead of snd_nxt
	 * when filling in ti_seq.  But if we are in persist
	 * state, snd_max might reflect one byte beyond the
	 * right edge of the window, so use snd_nxt in that
	 * case, since we know we aren't doing a retransmission.
	 * (retransmit and persist are mutually exclusive...)
	 */
	if (len || (flags & (TH_SYN|TH_FIN)) || tp->t_timer[TCPT_PERSIST])
		th->th_seq = htonl(tp->snd_nxt);
	else
		th->th_seq = htonl(tp->snd_max);

#ifdef TCP_SACK
	if (sack_rxmit) {
		/* 
		 * If sendalot was turned on (due to option stuffing), turn it 
		 * off. Properly set th_seq field.  Advance the ret'x pointer 
		 * by len.  
		 */
		if (sendalot)
			sendalot = 0;
		th->th_seq = htonl(p->rxmit);
		p->rxmit += len;
#if defined(TCP_SACK) && defined(TCP_FACK)
		tp->retran_data += len;
#endif /* TCP_FACK */
	}
#endif /* TCP_SACK */

	th->th_ack = htonl(tp->rcv_nxt);
	if (optlen) {
		bcopy((caddr_t)opt, (caddr_t)(th + 1), optlen);
		th->th_off = (sizeof (struct tcphdr) + optlen) >> 2;
	}
	th->th_flags = flags;

	/*
	 * Calculate receive window.  Don't shrink window,
	 * but avoid silly window syndrome.
	 */
	if (win < (long)(so->so_rcv.sb_hiwat / 4) && win < (long)tp->t_maxseg)
		win = 0;
	if (win > (long)TCP_MAXWIN << tp->rcv_scale)
		win = (long)TCP_MAXWIN << tp->rcv_scale;
	if (win < (long)(tp->rcv_adv - tp->rcv_nxt))
		win = (long)(tp->rcv_adv - tp->rcv_nxt);
	if (flags & TH_RST)
		win = 0;
	th->th_win = htons((u_int16_t) (win>>tp->rcv_scale));
	if (SEQ_GT(tp->snd_up, tp->snd_nxt)) {
		u_int32_t urp = tp->snd_up - tp->snd_nxt;
		if (urp > IP_MAXPACKET)
			urp = IP_MAXPACKET;
		th->th_urp = htons((u_int16_t)urp);
		th->th_flags |= TH_URG;
	} else
		/*
		 * If no urgent pointer to send, then we pull
		 * the urgent pointer to the left edge of the send window
		 * so that it doesn't drift into the send window on sequence
		 * number wraparound.
		 */
		tp->snd_up = tp->snd_una;		/* drag it along */

	/*
	 * Put TCP length in extended header, and then
	 * checksum extended header and data.
	 */
#ifdef INET6
	if (tp->pf == PF_INET6) {
	  th->th_sum = in6_cksum(m, IPPROTO_TCP,
				 sizeof(struct tcphdr) + optlen + len,
				 sizeof(struct ipv6));
	} else
#endif /* INET6 */
	{
		if (len + optlen)
			ti->ti_len = htons((u_int16_t)(sizeof (struct tcphdr) +
				optlen + len));
		ti->ti_sum = in_cksum(m, (int)(hdrlen + len));
	}

	/*
	 * In transmit state, time the transmission and arrange for
	 * the retransmit.  In persist state, just set snd_max.
	 */
	if (tp->t_force == 0 || tp->t_timer[TCPT_PERSIST] == 0) {
		tcp_seq startseq = tp->snd_nxt;

		/*
		 * Advance snd_nxt over sequence space of this segment.
		 */
		if (flags & (TH_SYN|TH_FIN)) {
			if (flags & TH_SYN)
				tp->snd_nxt++;
			if (flags & TH_FIN) {
				tp->snd_nxt++;
				tp->t_flags |= TF_SENTFIN;
			}
		}
#ifdef TCP_SACK
		if (!tp->sack_disable) {
			if (sack_rxmit && (p->rxmit != tp->snd_nxt)) {
				goto timer;
			}
		}
#endif
		tp->snd_nxt += len;
		if (SEQ_GT(tp->snd_nxt, tp->snd_max)) {
			tp->snd_max = tp->snd_nxt;
			/*
			 * Time this transmission if not a retransmission and
			 * not currently timing anything.
			 */
			if (tp->t_rtt == 0) {
				tp->t_rtt = 1;
				tp->t_rtseq = startseq;
				tcpstat.tcps_segstimed++;
			}
		}

		/*
		 * Set retransmit timer if not currently set,
		 * and not doing an ack or a keep-alive probe.
		 * Initial value for retransmit timer is smoothed
		 * round-trip time + 2 * round-trip time variance.
		 * Initialize shift counter which is used for backoff
		 * of retransmit time.
		 */
#ifdef TCP_SACK
 timer:
		if (!tp->sack_disable && sack_rxmit &&
		    tp->t_timer[TCPT_REXMT] == 0 &&
		    tp->snd_nxt != tp->snd_max) {
			tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
			if (tp->t_timer[TCPT_PERSIST]) {
				tp->t_timer[TCPT_PERSIST] = 0;
				tp->t_rxtshift = 0;
			}
		}
#endif

		if (tp->t_timer[TCPT_REXMT] == 0 &&
		    tp->snd_nxt != tp->snd_una) {
			tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
			if (tp->t_timer[TCPT_PERSIST]) {
				tp->t_timer[TCPT_PERSIST] = 0;
				tp->t_rxtshift = 0;
			}
		}
	} else
		if (SEQ_GT(tp->snd_nxt + len, tp->snd_max))
			tp->snd_max = tp->snd_nxt + len;

	/*
	 * Trace.
	 */
	if (so->so_options & SO_DEBUG)
#if INET6
		tcp_trace(TA_OUTPUT, tp->t_state, tp,
		    (tp->pf == AF_INET6) ? (struct tcpiphdr *)ti6 : ti, 0, len);
#else
	tcp_trace(TA_OUTPUT, tp->t_state, tp, ti, 0, len);
#endif

	/*
	 * Fill in IP length and desired time to live and
	 * send to IP level.  There should be a better way
	 * to handle ttl and tos; we could keep them in
	 * the template, but need a way to checksum without them.
	 */
	m->m_pkthdr.len = hdrlen + len;
#ifdef TUBA
	if (tp->t_tuba_pcb)
		error = tuba_output(m, tp);
	else
#endif
#ifdef INET6
	if (tp->pf == PF_INET6) {
	  ((struct ipv6 *)ti6)->ipv6_length = m->m_pkthdr.len - sizeof(struct ipv6);

	  /* Following fields are already grabbed from the tcp_template. */
	  /* ((struct ipv6 *)ti6)->ipv6_versfl   = ntohl(0x60000000);
	  ((struct ipv6 *)ti6)->ipv6_nexthdr  = IPPROTO_TCP;
	  ((struct ipv6 *)ti6)->ipv6_hoplimit = 
	    tp->t_inpcb->inp_ipv6.ipv6_hoplimit;*/

	  error = ipv6_output(m, &tp->t_inpcb->inp_route6, (so->so_options & SO_DONTROUTE), NULL, NULL, tp->t_inpcb->inp_socket);
	} else
#endif /* INET6 */
    {
	((struct ip *)ti)->ip_len = m->m_pkthdr.len;
	((struct ip *)ti)->ip_ttl = tp->t_inpcb->inp_ip.ip_ttl;	/* XXX */
	((struct ip *)ti)->ip_tos = tp->t_inpcb->inp_ip.ip_tos;	/* XXX */
	error = ip_output(m, tp->t_inpcb->inp_options, &tp->t_inpcb->inp_route,
	    so->so_options & SO_DONTROUTE, 0, tp->t_inpcb);
#if defined(TCP_SACK) && defined(TCP_FACK)
	/* Update snd_awnd to reflect the new data that was sent.  */
        tp->snd_awnd = tcp_seq_subtract(tp->snd_max, tp->snd_fack) +
            tp->retran_data;                
#endif
    }
	if (error) {
out:
		if (error == ENOBUFS) {
			tcp_quench(tp->t_inpcb, 0);
			return (0);
		}
		if ((error == EHOSTUNREACH || error == ENETDOWN)
		    && TCPS_HAVERCVDSYN(tp->t_state)) {
			tp->t_softerror = error;
			return (0);
		}
		return (error);
	}
	tcpstat.tcps_sndtotal++;

	/*
	 * Data sent (as far as we can tell).
	 * If this advertises a larger window than any other segment,
	 * then remember the size of the advertised window.
	 * Any pending ACK has now been sent.
	 */
	if (win > 0 && SEQ_GT(tp->rcv_nxt+win, tp->rcv_adv))
		tp->rcv_adv = tp->rcv_nxt + win;
	tp->last_ack_sent = tp->rcv_nxt;
	tp->t_flags &= ~(TF_ACKNOW|TF_DELACK);
#if defined(TCP_SACK) || defined(TCP_NEWRENO)
	if (sendalot && --maxburst)
#else
	if (sendalot)
#endif
		goto again;
	return (0);
}

void
tcp_setpersist(tp)
	register struct tcpcb *tp;
{
	register int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1;

	if (tp->t_timer[TCPT_REXMT])
		panic("tcp_output REXMT");
	/*
	 * Start/restart persistance timer.
	 */
	TCPT_RANGESET(tp->t_timer[TCPT_PERSIST],
	    t * tcp_backoff[tp->t_rxtshift],
	    TCPTV_PERSMIN, TCPTV_PERSMAX);
	if (tp->t_rxtshift < TCP_MAXRXTSHIFT)
		tp->t_rxtshift++;
}