xref: /freebsd/sys/netinet/tcp_sack.c (revision 87b759f0fa1f7554d50ce640c40138512bbded44)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995
5  *	The Regents of the University of California.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the University nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 /*-
34  *
35  * NRL grants permission for redistribution and use in source and binary
36  * forms, with or without modification, of the software and documentation
37  * created at NRL provided that the following conditions are met:
38  *
39  * 1. Redistributions of source code must retain the above copyright
40  *    notice, this list of conditions and the following disclaimer.
41  * 2. Redistributions in binary form must reproduce the above copyright
42  *    notice, this list of conditions and the following disclaimer in the
43  *    documentation and/or other materials provided with the distribution.
44  * 3. All advertising materials mentioning features or use of this software
45  *    must display the following acknowledgements:
46  *	This product includes software developed by the University of
47  *	California, Berkeley and its contributors.
48  *	This product includes software developed at the Information
49  *	Technology Division, US Naval Research Laboratory.
50  * 4. Neither the name of the NRL nor the names of its contributors
51  *    may be used to endorse or promote products derived from this software
52  *    without specific prior written permission.
53  *
54  * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS
55  * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
56  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
57  * PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL NRL OR
58  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
59  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
60  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
61  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
62  * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
63  * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
64  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
65  *
66  * The views and conclusions contained in the software and documentation
67  * are those of the authors and should not be interpreted as representing
68  * official policies, either expressed or implied, of the US Naval
69  * Research Laboratory (NRL).
70  */
71 
72 #include <sys/cdefs.h>
73 #include "opt_inet.h"
74 #include "opt_inet6.h"
75 
76 #include <sys/param.h>
77 #include <sys/systm.h>
78 #include <sys/kernel.h>
79 #include <sys/sysctl.h>
80 #include <sys/malloc.h>
81 #include <sys/mbuf.h>
82 #include <sys/proc.h>		/* for proc0 declaration */
83 #include <sys/protosw.h>
84 #include <sys/socket.h>
85 #include <sys/socketvar.h>
86 #include <sys/syslog.h>
87 #include <sys/systm.h>
88 
89 #include <machine/cpu.h>	/* before tcp_seq.h, for tcp_random18() */
90 
91 #include <vm/uma.h>
92 
93 #include <net/if.h>
94 #include <net/if_var.h>
95 #include <net/route.h>
96 #include <net/vnet.h>
97 
98 #include <netinet/in.h>
99 #include <netinet/in_systm.h>
100 #include <netinet/ip.h>
101 #include <netinet/in_var.h>
102 #include <netinet/in_pcb.h>
103 #include <netinet/ip_var.h>
104 #include <netinet/ip6.h>
105 #include <netinet/icmp6.h>
106 #include <netinet6/nd6.h>
107 #include <netinet6/ip6_var.h>
108 #include <netinet6/in6_pcb.h>
109 #include <netinet/tcp.h>
110 #include <netinet/tcp_fsm.h>
111 #include <netinet/tcp_seq.h>
112 #include <netinet/tcp_timer.h>
113 #include <netinet/tcp_var.h>
114 #include <netinet/tcpip.h>
115 #include <netinet/cc/cc.h>
116 
117 #include <machine/in_cksum.h>
118 
119 VNET_DECLARE(struct uma_zone *, sack_hole_zone);
120 #define	V_sack_hole_zone		VNET(sack_hole_zone)
121 
122 SYSCTL_NODE(_net_inet_tcp, OID_AUTO, sack, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
123     "TCP SACK");
124 
125 VNET_DEFINE(int, tcp_do_sack) = 1;
126 SYSCTL_INT(_net_inet_tcp_sack, OID_AUTO, enable, CTLFLAG_VNET | CTLFLAG_RW,
127     &VNET_NAME(tcp_do_sack), 0,
128     "Enable/Disable TCP SACK support");
129 
130 VNET_DEFINE(int, tcp_do_newsack) = 1;
131 SYSCTL_INT(_net_inet_tcp_sack, OID_AUTO, revised, CTLFLAG_VNET | CTLFLAG_RW,
132     &VNET_NAME(tcp_do_newsack), 0,
133     "Use revised SACK loss recovery per RFC 6675");
134 
135 VNET_DEFINE(int, tcp_do_lrd) = 1;
136 SYSCTL_INT(_net_inet_tcp_sack, OID_AUTO, lrd, CTLFLAG_VNET | CTLFLAG_RW,
137     &VNET_NAME(tcp_do_lrd), 1,
138     "Perform Lost Retransmission Detection");
139 
140 VNET_DEFINE(int, tcp_sack_tso) = 0;
141 SYSCTL_INT(_net_inet_tcp_sack, OID_AUTO, tso, CTLFLAG_VNET | CTLFLAG_RW,
142     &VNET_NAME(tcp_sack_tso), 0,
143     "Allow TSO during SACK loss recovery");
144 
145 VNET_DEFINE(int, tcp_sack_maxholes) = 128;
146 SYSCTL_INT(_net_inet_tcp_sack, OID_AUTO, maxholes, CTLFLAG_VNET | CTLFLAG_RW,
147     &VNET_NAME(tcp_sack_maxholes), 0,
148     "Maximum number of TCP SACK holes allowed per connection");
149 
150 VNET_DEFINE(int, tcp_sack_globalmaxholes) = 65536;
151 SYSCTL_INT(_net_inet_tcp_sack, OID_AUTO, globalmaxholes, CTLFLAG_VNET | CTLFLAG_RW,
152     &VNET_NAME(tcp_sack_globalmaxholes), 0,
153     "Global maximum number of TCP SACK holes");
154 
155 VNET_DEFINE(int, tcp_sack_globalholes) = 0;
156 SYSCTL_INT(_net_inet_tcp_sack, OID_AUTO, globalholes, CTLFLAG_VNET | CTLFLAG_RD,
157     &VNET_NAME(tcp_sack_globalholes), 0,
158     "Global number of TCP SACK holes currently allocated");
159 
160 int
161 tcp_dsack_block_exists(struct tcpcb *tp)
162 {
163 	/* Return true if a DSACK block exists */
164 	if (tp->rcv_numsacks == 0)
165 		return (0);
166 	if (SEQ_LEQ(tp->sackblks[0].end, tp->rcv_nxt))
167 		return(1);
168 	return (0);
169 }
170 
171 /*
172  * This function will find overlaps with the currently stored sackblocks
173  * and add any overlap as a dsack block upfront
174  */
175 void
176 tcp_update_dsack_list(struct tcpcb *tp, tcp_seq rcv_start, tcp_seq rcv_end)
177 {
178 	struct sackblk head_blk,mid_blk,saved_blks[MAX_SACK_BLKS];
179 	int i, j, n, identical;
180 	tcp_seq start, end;
181 
182 	INP_WLOCK_ASSERT(tptoinpcb(tp));
183 
184 	KASSERT(SEQ_LT(rcv_start, rcv_end), ("rcv_start < rcv_end"));
185 
186 	if (SEQ_LT(rcv_end, tp->rcv_nxt) ||
187 	    ((rcv_end == tp->rcv_nxt) &&
188 	     (tp->rcv_numsacks > 0 ) &&
189 	     (tp->sackblks[0].end == tp->rcv_nxt))) {
190 		saved_blks[0].start = rcv_start;
191 		saved_blks[0].end = rcv_end;
192 	} else {
193 		saved_blks[0].start = saved_blks[0].end = 0;
194 	}
195 
196 	head_blk.start = head_blk.end = 0;
197 	mid_blk.start = rcv_start;
198 	mid_blk.end = rcv_end;
199 	identical = 0;
200 
201 	for (i = 0; i < tp->rcv_numsacks; i++) {
202 		start = tp->sackblks[i].start;
203 		end = tp->sackblks[i].end;
204 		if (SEQ_LT(rcv_end, start)) {
205 			/* pkt left to sack blk */
206 			continue;
207 		}
208 		if (SEQ_GT(rcv_start, end)) {
209 			/* pkt right to sack blk */
210 			continue;
211 		}
212 		if (SEQ_GT(tp->rcv_nxt, end)) {
213 			if ((SEQ_MAX(rcv_start, start) != SEQ_MIN(rcv_end, end)) &&
214 			    (SEQ_GT(head_blk.start, SEQ_MAX(rcv_start, start)) ||
215 			    (head_blk.start == head_blk.end))) {
216 				head_blk.start = SEQ_MAX(rcv_start, start);
217 				head_blk.end = SEQ_MIN(rcv_end, end);
218 			}
219 			continue;
220 		}
221 		if (((head_blk.start == head_blk.end) ||
222 		     SEQ_LT(start, head_blk.start)) &&
223 		     (SEQ_GT(end, rcv_start) &&
224 		      SEQ_LEQ(start, rcv_end))) {
225 			head_blk.start = start;
226 			head_blk.end = end;
227 		}
228 		mid_blk.start = SEQ_MIN(mid_blk.start, start);
229 		mid_blk.end = SEQ_MAX(mid_blk.end, end);
230 		if ((mid_blk.start == start) &&
231 		    (mid_blk.end == end))
232 			identical = 1;
233 	}
234 	if (SEQ_LT(head_blk.start, head_blk.end)) {
235 		/* store overlapping range */
236 		saved_blks[0].start = SEQ_MAX(rcv_start, head_blk.start);
237 		saved_blks[0].end   = SEQ_MIN(rcv_end, head_blk.end);
238 	}
239 	n = 1;
240 	/*
241 	 * Second, if not ACKed, store the SACK block that
242 	 * overlaps with the DSACK block unless it is identical
243 	 */
244 	if ((SEQ_LT(tp->rcv_nxt, mid_blk.end) &&
245 	    !((mid_blk.start == saved_blks[0].start) &&
246 	    (mid_blk.end == saved_blks[0].end))) ||
247 	    identical == 1) {
248 		saved_blks[n].start = mid_blk.start;
249 		saved_blks[n++].end = mid_blk.end;
250 	}
251 	for (j = 0; (j < tp->rcv_numsacks) && (n < MAX_SACK_BLKS); j++) {
252 		if (((SEQ_LT(tp->sackblks[j].end, mid_blk.start) ||
253 		      SEQ_GT(tp->sackblks[j].start, mid_blk.end)) &&
254 		    (SEQ_GT(tp->sackblks[j].start, tp->rcv_nxt))))
255 		saved_blks[n++] = tp->sackblks[j];
256 	}
257 	j = 0;
258 	for (i = 0; i < n; i++) {
259 		/* we can end up with a stale initial entry */
260 		if (SEQ_LT(saved_blks[i].start, saved_blks[i].end)) {
261 			tp->sackblks[j++] = saved_blks[i];
262 		}
263 	}
264 	tp->rcv_numsacks = j;
265 }
266 
267 /*
268  * This function is called upon receipt of new valid data (while not in
269  * header prediction mode), and it updates the ordered list of sacks.
270  */
271 void
272 tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_start, tcp_seq rcv_end)
273 {
274 	/*
275 	 * First reported block MUST be the most recent one.  Subsequent
276 	 * blocks SHOULD be in the order in which they arrived at the
277 	 * receiver.  These two conditions make the implementation fully
278 	 * compliant with RFC 2018.
279 	 */
280 	struct sackblk head_blk, saved_blks[MAX_SACK_BLKS];
281 	int num_head, num_saved, i;
282 
283 	INP_WLOCK_ASSERT(tptoinpcb(tp));
284 
285 	/* Check arguments. */
286 	KASSERT(SEQ_LEQ(rcv_start, rcv_end), ("rcv_start <= rcv_end"));
287 
288 	if ((rcv_start == rcv_end) &&
289 	    (tp->rcv_numsacks >= 1) &&
290 	    (rcv_end == tp->sackblks[0].end)) {
291 		/* retaining DSACK block below rcv_nxt (todrop) */
292 		head_blk = tp->sackblks[0];
293 	} else {
294 		/* SACK block for the received segment. */
295 		head_blk.start = rcv_start;
296 		head_blk.end = rcv_end;
297 	}
298 
299 	/*
300 	 * Merge updated SACK blocks into head_blk, and save unchanged SACK
301 	 * blocks into saved_blks[].  num_saved will have the number of the
302 	 * saved SACK blocks.
303 	 */
304 	num_saved = 0;
305 	for (i = 0; i < tp->rcv_numsacks; i++) {
306 		tcp_seq start = tp->sackblks[i].start;
307 		tcp_seq end = tp->sackblks[i].end;
308 		if (SEQ_GEQ(start, end) || SEQ_LEQ(start, tp->rcv_nxt)) {
309 			/*
310 			 * Discard this SACK block.
311 			 */
312 		} else if (SEQ_LEQ(head_blk.start, end) &&
313 			   SEQ_GEQ(head_blk.end, start)) {
314 			/*
315 			 * Merge this SACK block into head_blk.  This SACK
316 			 * block itself will be discarded.
317 			 */
318 			/*
319 			 * |-|
320 			 *   |---|  merge
321 			 *
322 			 *     |-|
323 			 * |---|    merge
324 			 *
325 			 * |-----|
326 			 *   |-|    DSACK smaller
327 			 *
328 			 *   |-|
329 			 * |-----|  DSACK smaller
330 			 */
331 			if (head_blk.start == end)
332 				head_blk.start = start;
333 			else if (head_blk.end == start)
334 				head_blk.end = end;
335 			else {
336 				if (SEQ_LT(head_blk.start, start)) {
337 					tcp_seq temp = start;
338 					start = head_blk.start;
339 					head_blk.start = temp;
340 				}
341 				if (SEQ_GT(head_blk.end, end)) {
342 					tcp_seq temp = end;
343 					end = head_blk.end;
344 					head_blk.end = temp;
345 				}
346 				if ((head_blk.start != start) ||
347 				    (head_blk.end != end)) {
348 					if ((num_saved >= 1) &&
349 					   SEQ_GEQ(saved_blks[num_saved-1].start, start) &&
350 					   SEQ_LEQ(saved_blks[num_saved-1].end, end))
351 						num_saved--;
352 					saved_blks[num_saved].start = start;
353 					saved_blks[num_saved].end = end;
354 					num_saved++;
355 				}
356 			}
357 		} else {
358 			/*
359 			 * This block supercedes the prior block
360 			 */
361 			if ((num_saved >= 1) &&
362 			   SEQ_GEQ(saved_blks[num_saved-1].start, start) &&
363 			   SEQ_LEQ(saved_blks[num_saved-1].end, end))
364 				num_saved--;
365 			/*
366 			 * Save this SACK block.
367 			 */
368 			saved_blks[num_saved].start = start;
369 			saved_blks[num_saved].end = end;
370 			num_saved++;
371 		}
372 	}
373 
374 	/*
375 	 * Update SACK list in tp->sackblks[].
376 	 */
377 	num_head = 0;
378 	if (SEQ_LT(rcv_start, rcv_end)) {
379 		/*
380 		 * The received data segment is an out-of-order segment.  Put
381 		 * head_blk at the top of SACK list.
382 		 */
383 		tp->sackblks[0] = head_blk;
384 		num_head = 1;
385 		/*
386 		 * If the number of saved SACK blocks exceeds its limit,
387 		 * discard the last SACK block.
388 		 */
389 		if (num_saved >= MAX_SACK_BLKS)
390 			num_saved--;
391 	}
392 	if ((rcv_start == rcv_end) &&
393 	    (rcv_start == tp->sackblks[0].end)) {
394 		num_head = 1;
395 	}
396 	if (num_saved > 0) {
397 		/*
398 		 * Copy the saved SACK blocks back.
399 		 */
400 		bcopy(saved_blks, &tp->sackblks[num_head],
401 		      sizeof(struct sackblk) * num_saved);
402 	}
403 
404 	/* Save the number of SACK blocks. */
405 	tp->rcv_numsacks = num_head + num_saved;
406 }
407 
408 void
409 tcp_clean_dsack_blocks(struct tcpcb *tp)
410 {
411 	struct sackblk saved_blks[MAX_SACK_BLKS];
412 	int num_saved, i;
413 
414 	INP_WLOCK_ASSERT(tptoinpcb(tp));
415 	/*
416 	 * Clean up any DSACK blocks that
417 	 * are in our queue of sack blocks.
418 	 *
419 	 */
420 	num_saved = 0;
421 	for (i = 0; i < tp->rcv_numsacks; i++) {
422 		tcp_seq start = tp->sackblks[i].start;
423 		tcp_seq end = tp->sackblks[i].end;
424 		if (SEQ_GEQ(start, end) || SEQ_LEQ(start, tp->rcv_nxt)) {
425 			/*
426 			 * Discard this D-SACK block.
427 			 */
428 			continue;
429 		}
430 		/*
431 		 * Save this SACK block.
432 		 */
433 		saved_blks[num_saved].start = start;
434 		saved_blks[num_saved].end = end;
435 		num_saved++;
436 	}
437 	if (num_saved > 0) {
438 		/*
439 		 * Copy the saved SACK blocks back.
440 		 */
441 		bcopy(saved_blks, &tp->sackblks[0],
442 		      sizeof(struct sackblk) * num_saved);
443 	}
444 	tp->rcv_numsacks = num_saved;
445 }
446 
447 /*
448  * Delete all receiver-side SACK information.
449  */
450 void
451 tcp_clean_sackreport(struct tcpcb *tp)
452 {
453 	int i;
454 
455 	INP_WLOCK_ASSERT(tptoinpcb(tp));
456 	tp->rcv_numsacks = 0;
457 	for (i = 0; i < MAX_SACK_BLKS; i++)
458 		tp->sackblks[i].start = tp->sackblks[i].end=0;
459 }
460 
461 /*
462  * Allocate struct sackhole.
463  */
464 static struct sackhole *
465 tcp_sackhole_alloc(struct tcpcb *tp, tcp_seq start, tcp_seq end)
466 {
467 	struct sackhole *hole;
468 
469 	if (tp->snd_numholes >= V_tcp_sack_maxholes ||
470 	    V_tcp_sack_globalholes >= V_tcp_sack_globalmaxholes) {
471 		TCPSTAT_INC(tcps_sack_sboverflow);
472 		return NULL;
473 	}
474 
475 	hole = (struct sackhole *)uma_zalloc(V_sack_hole_zone, M_NOWAIT);
476 	if (hole == NULL)
477 		return NULL;
478 
479 	hole->start = start;
480 	hole->end = end;
481 	hole->rxmit = start;
482 
483 	tp->snd_numholes++;
484 	atomic_add_int(&V_tcp_sack_globalholes, 1);
485 
486 	return hole;
487 }
488 
489 /*
490  * Free struct sackhole.
491  */
492 static void
493 tcp_sackhole_free(struct tcpcb *tp, struct sackhole *hole)
494 {
495 
496 	uma_zfree(V_sack_hole_zone, hole);
497 
498 	tp->snd_numholes--;
499 	atomic_subtract_int(&V_tcp_sack_globalholes, 1);
500 
501 	KASSERT(tp->snd_numholes >= 0, ("tp->snd_numholes >= 0"));
502 	KASSERT(V_tcp_sack_globalholes >= 0, ("tcp_sack_globalholes >= 0"));
503 }
504 
505 /*
506  * Insert new SACK hole into scoreboard.
507  */
508 static struct sackhole *
509 tcp_sackhole_insert(struct tcpcb *tp, tcp_seq start, tcp_seq end,
510     struct sackhole *after)
511 {
512 	struct sackhole *hole;
513 
514 	/* Allocate a new SACK hole. */
515 	hole = tcp_sackhole_alloc(tp, start, end);
516 	if (hole == NULL)
517 		return NULL;
518 
519 	/* Insert the new SACK hole into scoreboard. */
520 	if (after != NULL)
521 		TAILQ_INSERT_AFTER(&tp->snd_holes, after, hole, scblink);
522 	else
523 		TAILQ_INSERT_TAIL(&tp->snd_holes, hole, scblink);
524 
525 	/* Update SACK hint. */
526 	if (tp->sackhint.nexthole == NULL)
527 		tp->sackhint.nexthole = hole;
528 
529 	return hole;
530 }
531 
532 /*
533  * Remove SACK hole from scoreboard.
534  */
535 static void
536 tcp_sackhole_remove(struct tcpcb *tp, struct sackhole *hole)
537 {
538 
539 	/* Update SACK hint. */
540 	if (tp->sackhint.nexthole == hole)
541 		tp->sackhint.nexthole = TAILQ_NEXT(hole, scblink);
542 
543 	/* Remove this SACK hole. */
544 	TAILQ_REMOVE(&tp->snd_holes, hole, scblink);
545 
546 	/* Free this SACK hole. */
547 	tcp_sackhole_free(tp, hole);
548 }
549 
550 /*
551  * Process cumulative ACK and the TCP SACK option to update the scoreboard.
552  * tp->snd_holes is an ordered list of holes (oldest to newest, in terms of
553  * the sequence space).
554  * Returns SACK_NEWLOSS if incoming ACK indicates ongoing loss (hole split, new hole),
555  * SACK_CHANGE if incoming ACK has previously unknown SACK information,
556  * SACK_NOCHANGE otherwise.
557  */
558 sackstatus_t
559 tcp_sack_doack(struct tcpcb *tp, struct tcpopt *to, tcp_seq th_ack)
560 {
561 	struct sackhole *cur, *temp;
562 	struct sackblk sack, sack_blocks[TCP_MAX_SACK + 1], *sblkp;
563 	int i, j, num_sack_blks;
564 	sackstatus_t sack_changed;
565 	int delivered_data, left_edge_delta;
566 	int maxseg = tp->t_maxseg - MAX_TCPOPTLEN;
567 
568 	tcp_seq loss_hiack = 0;
569 	int loss_thresh = 0;
570 	int loss_sblks = 0;
571 	int notlost_bytes = 0;
572 
573 	INP_WLOCK_ASSERT(tptoinpcb(tp));
574 
575 	num_sack_blks = 0;
576 	sack_changed = SACK_NOCHANGE;
577 	delivered_data = 0;
578 	left_edge_delta = 0;
579 	/*
580 	 * If SND.UNA will be advanced by SEG.ACK, and if SACK holes exist,
581 	 * treat [SND.UNA, SEG.ACK) as if it is a SACK block.
582 	 * Account changes to SND.UNA always in delivered data.
583 	 */
584 	if (SEQ_LT(tp->snd_una, th_ack) && !TAILQ_EMPTY(&tp->snd_holes)) {
585 		left_edge_delta = th_ack - tp->snd_una;
586 		sack_blocks[num_sack_blks].start = tp->snd_una;
587 		sack_blocks[num_sack_blks++].end = th_ack;
588 		/*
589 		 * Pulling snd_fack forward if we got here
590 		 * due to DSACK blocks
591 		 */
592 		if (SEQ_LT(tp->snd_fack, th_ack)) {
593 			delivered_data += th_ack - tp->snd_una;
594 			tp->snd_fack = th_ack;
595 			sack_changed = SACK_CHANGE;
596 		}
597 	}
598 	/*
599 	 * Append received valid SACK blocks to sack_blocks[], but only if we
600 	 * received new blocks from the other side.
601 	 */
602 	if (to->to_flags & TOF_SACK) {
603 		for (i = 0; i < to->to_nsacks; i++) {
604 			bcopy((to->to_sacks + i * TCPOLEN_SACK),
605 			    &sack, sizeof(sack));
606 			sack.start = ntohl(sack.start);
607 			sack.end = ntohl(sack.end);
608 			if (SEQ_GT(sack.end, sack.start) &&
609 			    SEQ_GT(sack.start, tp->snd_una) &&
610 			    SEQ_GT(sack.start, th_ack) &&
611 			    SEQ_LT(sack.start, tp->snd_max) &&
612 			    SEQ_GT(sack.end, tp->snd_una) &&
613 			    SEQ_LEQ(sack.end, tp->snd_max) &&
614 			    ((sack.end - sack.start) >= maxseg ||
615 			     SEQ_GEQ(sack.end, tp->snd_max))) {
616 				sack_blocks[num_sack_blks++] = sack;
617 			} else if (SEQ_LEQ(sack.start, th_ack) &&
618 			    SEQ_LEQ(sack.end, th_ack)) {
619 				/*
620 				 * Its a D-SACK block.
621 				 */
622 				tcp_record_dsack(tp, sack.start, sack.end, 0);
623 			}
624 		}
625 	}
626 	/*
627 	 * Return if SND.UNA is not advanced and no valid SACK block is
628 	 * received.
629 	 */
630 	if (num_sack_blks == 0)
631 		return (sack_changed);
632 
633 	/*
634 	 * Sort the SACK blocks so we can update the scoreboard with just one
635 	 * pass. The overhead of sorting up to 4+1 elements is less than
636 	 * making up to 4+1 passes over the scoreboard.
637 	 */
638 	for (i = 0; i < num_sack_blks; i++) {
639 		for (j = i + 1; j < num_sack_blks; j++) {
640 			if (SEQ_GT(sack_blocks[i].end, sack_blocks[j].end)) {
641 				sack = sack_blocks[i];
642 				sack_blocks[i] = sack_blocks[j];
643 				sack_blocks[j] = sack;
644 			}
645 		}
646 	}
647 	if (TAILQ_EMPTY(&tp->snd_holes)) {
648 		/*
649 		 * Empty scoreboard. Need to initialize snd_fack (it may be
650 		 * uninitialized or have a bogus value). Scoreboard holes
651 		 * (from the sack blocks received) are created later below
652 		 * (in the logic that adds holes to the tail of the
653 		 * scoreboard).
654 		 */
655 		tp->snd_fack = SEQ_MAX(tp->snd_una, th_ack);
656 		tp->sackhint.sacked_bytes = 0;	/* reset */
657 		tp->sackhint.hole_bytes = 0;
658 	}
659 	/*
660 	 * In the while-loop below, incoming SACK blocks (sack_blocks[]) and
661 	 * SACK holes (snd_holes) are traversed from their tails with just
662 	 * one pass in order to reduce the number of compares especially when
663 	 * the bandwidth-delay product is large.
664 	 *
665 	 * Note: Typically, in the first RTT of SACK recovery, the highest
666 	 * three or four SACK blocks with the same ack number are received.
667 	 * In the second RTT, if retransmitted data segments are not lost,
668 	 * the highest three or four SACK blocks with ack number advancing
669 	 * are received.
670 	 */
671 	sblkp = &sack_blocks[num_sack_blks - 1];	/* Last SACK block */
672 	tp->sackhint.last_sack_ack = sblkp->end;
673 	if (SEQ_LT(tp->snd_fack, sblkp->start)) {
674 		/*
675 		 * The highest SACK block is beyond fack.  First,
676 		 * check if there was a successful Rescue Retransmission,
677 		 * and move this hole left. With normal holes, snd_fack
678 		 * is always to the right of the end.
679 		 */
680 		if (((temp = TAILQ_LAST(&tp->snd_holes, sackhole_head)) != NULL) &&
681 		    SEQ_LEQ(tp->snd_fack,temp->end)) {
682 			tp->sackhint.hole_bytes -= temp->end - temp->start;
683 			temp->start = SEQ_MAX(tp->snd_fack, SEQ_MAX(tp->snd_una, th_ack));
684 			temp->end = sblkp->start;
685 			temp->rxmit = temp->start;
686 			delivered_data += sblkp->end - sblkp->start;
687 			tp->sackhint.hole_bytes += temp->end - temp->start;
688 			KASSERT(tp->sackhint.hole_bytes >= 0,
689 			    ("sackhint hole bytes >= 0"));
690 			tp->snd_fack = sblkp->end;
691 			sblkp--;
692 			sack_changed = SACK_NEWLOSS;
693 		} else {
694 			/*
695 			 * Append a new SACK hole at the tail.  If the
696 			 * second or later highest SACK blocks are also
697 			 * beyond the current fack, they will be inserted
698 			 * by way of hole splitting in the while-loop below.
699 			 */
700 			temp = tcp_sackhole_insert(tp, tp->snd_fack,sblkp->start,NULL);
701 			if (temp != NULL) {
702 				delivered_data += sblkp->end - sblkp->start;
703 				tp->sackhint.hole_bytes += temp->end - temp->start;
704 				tp->snd_fack = sblkp->end;
705 				/* Go to the previous sack block. */
706 				sblkp--;
707 				sack_changed = SACK_CHANGE;
708 			} else {
709 				/*
710 				 * We failed to add a new hole based on the current
711 				 * sack block.  Skip over all the sack blocks that
712 				 * fall completely to the right of snd_fack and
713 				 * proceed to trim the scoreboard based on the
714 				 * remaining sack blocks.  This also trims the
715 				 * scoreboard for th_ack (which is sack_blocks[0]).
716 				 */
717 				while (sblkp >= sack_blocks &&
718 				       SEQ_LT(tp->snd_fack, sblkp->start))
719 					sblkp--;
720 				if (sblkp >= sack_blocks &&
721 				    SEQ_LT(tp->snd_fack, sblkp->end)) {
722 					delivered_data += sblkp->end - tp->snd_fack;
723 					tp->snd_fack = sblkp->end;
724 					/*
725 					 * While the Scoreboard didn't change in
726 					 * size, we only ended up here because
727 					 * some SACK data had to be dismissed.
728 					 */
729 					sack_changed = SACK_NEWLOSS;
730 				}
731 			}
732 		}
733 	} else if (SEQ_LT(tp->snd_fack, sblkp->end)) {
734 		/* fack is advanced. */
735 		delivered_data += sblkp->end - tp->snd_fack;
736 		tp->snd_fack = sblkp->end;
737 		sack_changed = SACK_CHANGE;
738 	}
739 	cur = TAILQ_LAST(&tp->snd_holes, sackhole_head); /* Last SACK hole. */
740 	loss_hiack = tp->snd_fack;
741 
742 	/*
743 	 * Since the incoming sack blocks are sorted, we can process them
744 	 * making one sweep of the scoreboard.
745 	 */
746 	while (cur != NULL) {
747 		if (!(sblkp >= sack_blocks)) {
748 			if (((loss_sblks >= tcprexmtthresh) ||
749 			    (loss_thresh > (tcprexmtthresh-1)*tp->t_maxseg)))
750 				break;
751 			loss_thresh += loss_hiack - cur->end;
752 			loss_hiack = cur->start;
753 			loss_sblks++;
754 			if (!((loss_sblks >= tcprexmtthresh) ||
755 			    (loss_thresh > (tcprexmtthresh-1)*tp->t_maxseg))) {
756 				notlost_bytes += cur->end - cur->start;
757 			} else {
758 				break;
759 			}
760 			cur = TAILQ_PREV(cur, sackhole_head, scblink);
761 			continue;
762 		}
763 		if (SEQ_GEQ(sblkp->start, cur->end)) {
764 			/*
765 			 * SACKs data beyond the current hole.  Go to the
766 			 * previous sack block.
767 			 */
768 			sblkp--;
769 			continue;
770 		}
771 		if (SEQ_LEQ(sblkp->end, cur->start)) {
772 			/*
773 			 * SACKs data before the current hole.  Go to the
774 			 * previous hole.
775 			 */
776 			loss_thresh += loss_hiack - cur->end;
777 			loss_hiack = cur->start;
778 			loss_sblks++;
779 			if (!((loss_sblks >= tcprexmtthresh) ||
780 			    (loss_thresh > (tcprexmtthresh-1)*tp->t_maxseg)))
781 				notlost_bytes += cur->end - cur->start;
782 			cur = TAILQ_PREV(cur, sackhole_head, scblink);
783 			continue;
784 		}
785 		tp->sackhint.sack_bytes_rexmit -=
786 		    (SEQ_MIN(cur->rxmit, cur->end) - cur->start);
787 		KASSERT(tp->sackhint.sack_bytes_rexmit >= 0,
788 		    ("sackhint bytes rtx >= 0"));
789 		sack_changed = SACK_CHANGE;
790 		if (SEQ_LEQ(sblkp->start, cur->start)) {
791 			/* Data acks at least the beginning of hole. */
792 			if (SEQ_GEQ(sblkp->end, cur->end)) {
793 				/* Acks entire hole, so delete hole. */
794 				delivered_data += (cur->end - cur->start);
795 				temp = cur;
796 				cur = TAILQ_PREV(cur, sackhole_head, scblink);
797 				tp->sackhint.hole_bytes -= temp->end - temp->start;
798 				tcp_sackhole_remove(tp, temp);
799 				/*
800 				 * The sack block may ack all or part of the
801 				 * next hole too, so continue onto the next
802 				 * hole.
803 				 */
804 				continue;
805 			} else {
806 				/* Move start of hole forward. */
807 				delivered_data += (sblkp->end - cur->start);
808 				tp->sackhint.hole_bytes -= sblkp->end - cur->start;
809 				cur->start = sblkp->end;
810 				cur->rxmit = SEQ_MAX(cur->rxmit, cur->start);
811 			}
812 		} else {
813 			/* Data acks at least the end of hole. */
814 			if (SEQ_GEQ(sblkp->end, cur->end)) {
815 				/* Move end of hole backward. */
816 				delivered_data += (cur->end - sblkp->start);
817 				tp->sackhint.hole_bytes -= cur->end - sblkp->start;
818 				cur->end = sblkp->start;
819 				cur->rxmit = SEQ_MIN(cur->rxmit, cur->end);
820 				if ((tp->t_flags & TF_LRD) && SEQ_GEQ(cur->rxmit, cur->end))
821 					cur->rxmit = tp->snd_recover;
822 			} else {
823 				/*
824 				 * ACKs some data in middle of a hole; need
825 				 * to split current hole
826 				 */
827 				temp = tcp_sackhole_insert(tp, sblkp->end,
828 				    cur->end, cur);
829 				sack_changed = SACK_NEWLOSS;
830 				if (temp != NULL) {
831 					if (SEQ_GT(cur->rxmit, temp->rxmit)) {
832 						temp->rxmit = cur->rxmit;
833 						tp->sackhint.sack_bytes_rexmit +=
834 						    (SEQ_MIN(temp->rxmit,
835 						    temp->end) - temp->start);
836 					}
837 					tp->sackhint.hole_bytes -= sblkp->end - sblkp->start;
838 					loss_thresh += loss_hiack - temp->end;
839 					loss_hiack = temp->start;
840 					loss_sblks++;
841 					if (!((loss_sblks >= tcprexmtthresh) ||
842 					    (loss_thresh > (tcprexmtthresh-1)*tp->t_maxseg)))
843 						notlost_bytes += temp->end - temp->start;
844 					cur->end = sblkp->start;
845 					cur->rxmit = SEQ_MIN(cur->rxmit,
846 					    cur->end);
847 					if ((tp->t_flags & TF_LRD) && SEQ_GEQ(cur->rxmit, cur->end))
848 						cur->rxmit = tp->snd_recover;
849 					delivered_data += (sblkp->end - sblkp->start);
850 				}
851 			}
852 		}
853 		tp->sackhint.sack_bytes_rexmit +=
854 		    (SEQ_MIN(cur->rxmit, cur->end) - cur->start);
855 		/*
856 		 * Testing sblkp->start against cur->start tells us whether
857 		 * we're done with the sack block or the sack hole.
858 		 * Accordingly, we advance one or the other.
859 		 */
860 		if (SEQ_LEQ(sblkp->start, cur->start)) {
861 			loss_thresh += loss_hiack - cur->end;
862 			loss_hiack = cur->start;
863 			loss_sblks++;
864 			if (!((loss_sblks >= tcprexmtthresh) ||
865 			    (loss_thresh > (tcprexmtthresh-1)*tp->t_maxseg)))
866 				notlost_bytes += cur->end - cur->start;
867 			cur = TAILQ_PREV(cur, sackhole_head, scblink);
868 		} else {
869 			sblkp--;
870 		}
871 	}
872 
873 	KASSERT(!(TAILQ_EMPTY(&tp->snd_holes) && (tp->sackhint.hole_bytes != 0)),
874 	    ("SACK scoreboard empty, but accounting non-zero\n"));
875 
876 	KASSERT(notlost_bytes <= tp->sackhint.hole_bytes,
877 	    ("SACK: more bytes marked notlost than in scoreboard holes"));
878 
879 	if (!(to->to_flags & TOF_SACK))
880 		/*
881 		 * If this ACK did not contain any
882 		 * SACK blocks, any only moved the
883 		 * left edge right, it is a pure
884 		 * cumulative ACK. Do not count
885 		 * DupAck for this. Also required
886 		 * for RFC6675 rescue retransmission.
887 		 */
888 		sack_changed = SACK_NOCHANGE;
889 	tp->sackhint.delivered_data = delivered_data;
890 	tp->sackhint.sacked_bytes += delivered_data - left_edge_delta;
891 	tp->sackhint.lost_bytes = tp->sackhint.hole_bytes - notlost_bytes;
892 	KASSERT((delivered_data >= 0), ("delivered_data < 0"));
893 	KASSERT((tp->sackhint.sacked_bytes >= 0), ("sacked_bytes < 0"));
894 	return (sack_changed);
895 }
896 
897 /*
898  * Free all SACK holes to clear the scoreboard.
899  */
900 void
901 tcp_free_sackholes(struct tcpcb *tp)
902 {
903 	struct sackhole *q;
904 
905 	INP_WLOCK_ASSERT(tptoinpcb(tp));
906 	while ((q = TAILQ_FIRST(&tp->snd_holes)) != NULL)
907 		tcp_sackhole_remove(tp, q);
908 	tp->sackhint.sack_bytes_rexmit = 0;
909 	tp->sackhint.delivered_data = 0;
910 	tp->sackhint.sacked_bytes = 0;
911 	tp->sackhint.hole_bytes = 0;
912 	tp->sackhint.lost_bytes = 0;
913 
914 	KASSERT(tp->snd_numholes == 0, ("tp->snd_numholes == 0"));
915 	KASSERT(tp->sackhint.nexthole == NULL,
916 		("tp->sackhint.nexthole == NULL"));
917 }
918 
919 /*
920  * Resend all the currently existing SACK holes of
921  * the scoreboard. This is in line with the Errata to
922  * RFC 2018, which allows the use of SACK data past
923  * an RTO to good effect typically.
924  */
925 void
926 tcp_resend_sackholes(struct tcpcb *tp)
927 {
928 	struct sackhole *p;
929 
930 	INP_WLOCK_ASSERT(tptoinpcb(tp));
931 	TAILQ_FOREACH(p, &tp->snd_holes, scblink) {
932 		p->rxmit = p->start;
933 	}
934 	tp->sackhint.nexthole = TAILQ_FIRST(&tp->snd_holes);
935 	tp->sackhint.sack_bytes_rexmit = 0;
936 }
937 
938 /*
939  * Partial ack handling within a sack recovery episode.  Keeping this very
940  * simple for now.  When a partial ack is received, force snd_cwnd to a value
941  * that will allow the sender to transmit no more than 2 segments.  If
942  * necessary, a better scheme can be adopted at a later point, but for now,
943  * the goal is to prevent the sender from bursting a large amount of data in
944  * the midst of sack recovery.
945  */
946 void
947 tcp_sack_partialack(struct tcpcb *tp, struct tcphdr *th, u_int *maxsegp)
948 {
949 	struct sackhole *temp;
950 	int num_segs = 1;
951 	u_int maxseg;
952 
953 	INP_WLOCK_ASSERT(tptoinpcb(tp));
954 
955 	if (*maxsegp == 0) {
956 		*maxsegp = tcp_maxseg(tp);
957 	}
958 	maxseg = *maxsegp;
959 	tcp_timer_activate(tp, TT_REXMT, 0);
960 	tp->t_rtttime = 0;
961 	/* Send one or 2 segments based on how much new data was acked. */
962 	if ((BYTES_THIS_ACK(tp, th) / maxseg) >= 2)
963 		num_segs = 2;
964 	if (V_tcp_do_newsack) {
965 		tp->snd_cwnd = imax(tp->snd_nxt - th->th_ack +
966 				tp->sackhint.sack_bytes_rexmit -
967 				tp->sackhint.sacked_bytes -
968 				tp->sackhint.lost_bytes, maxseg) +
969 				num_segs * maxseg;
970 	} else {
971 		tp->snd_cwnd = (tp->sackhint.sack_bytes_rexmit +
972 		    imax(0, tp->snd_nxt - tp->snd_recover) +
973 		    num_segs * maxseg);
974 	}
975 	if (tp->snd_cwnd > tp->snd_ssthresh)
976 		tp->snd_cwnd = tp->snd_ssthresh;
977 	tp->t_flags |= TF_ACKNOW;
978 	/*
979 	 * RFC6675 rescue retransmission
980 	 * Add a hole between th_ack (snd_una is not yet set) and snd_max,
981 	 * if this was a pure cumulative ACK and no data was send beyond
982 	 * recovery point. Since the data in the socket has not been freed
983 	 * at this point, we check if the scoreboard is empty, and the ACK
984 	 * delivered some new data, indicating a full ACK. Also, if the
985 	 * recovery point is still at snd_max, we are probably application
986 	 * limited. However, this inference might not always be true. The
987 	 * rescue retransmission may rarely be slightly premature
988 	 * compared to RFC6675.
989 	 * The corresponding ACK+SACK will cause any further outstanding
990 	 * segments to be retransmitted. This addresses a corner case, when
991 	 * the trailing packets of a window are lost and no further data
992 	 * is available for sending.
993 	 */
994 	if ((V_tcp_do_newsack) &&
995 	    SEQ_LT(th->th_ack, tp->snd_recover) &&
996 	    TAILQ_EMPTY(&tp->snd_holes) &&
997 	    (tp->sackhint.delivered_data > 0)) {
998 		/*
999 		 * Exclude FIN sequence space in
1000 		 * the hole for the rescue retransmission,
1001 		 * and also don't create a hole, if only
1002 		 * the ACK for a FIN is outstanding.
1003 		 */
1004 		tcp_seq highdata = tp->snd_max;
1005 		if (tp->t_flags & TF_SENTFIN)
1006 			highdata--;
1007 		highdata = SEQ_MIN(highdata, tp->snd_recover);
1008 		if (SEQ_LT(th->th_ack, highdata)) {
1009 			tp->snd_fack = SEQ_MAX(th->th_ack, tp->snd_fack);
1010 			if ((temp = tcp_sackhole_insert(tp, SEQ_MAX(th->th_ack,
1011 			    highdata - maxseg), highdata, NULL)) != NULL) {
1012 				tp->sackhint.hole_bytes +=
1013 					temp->end - temp->start;
1014 			}
1015 		}
1016 	}
1017 	(void) tcp_output(tp);
1018 }
1019 
1020 /*
1021  * Returns the next hole to retransmit and the number of retransmitted bytes
1022  * from the scoreboard.  We store both the next hole and the number of
1023  * retransmitted bytes as hints (and recompute these on the fly upon SACK/ACK
1024  * reception).  This avoids scoreboard traversals completely.
1025  *
1026  * The loop here will traverse *at most* one link.  Here's the argument.  For
1027  * the loop to traverse more than 1 link before finding the next hole to
1028  * retransmit, we would need to have at least 1 node following the current
1029  * hint with (rxmit == end).  But, for all holes following the current hint,
1030  * (start == rxmit), since we have not yet retransmitted from them.
1031  * Therefore, in order to traverse more 1 link in the loop below, we need to
1032  * have at least one node following the current hint with (start == rxmit ==
1033  * end).  But that can't happen, (start == end) means that all the data in
1034  * that hole has been sacked, in which case, the hole would have been removed
1035  * from the scoreboard.
1036  */
1037 struct sackhole *
1038 tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt)
1039 {
1040 	struct sackhole *hole = NULL;
1041 
1042 	INP_WLOCK_ASSERT(tptoinpcb(tp));
1043 	*sack_bytes_rexmt = tp->sackhint.sack_bytes_rexmit;
1044 	hole = tp->sackhint.nexthole;
1045 	if (hole == NULL)
1046 		return (hole);
1047 	if (SEQ_GEQ(hole->rxmit, hole->end)) {
1048 		for (;;) {
1049 			hole = TAILQ_NEXT(hole, scblink);
1050 			if (hole == NULL)
1051 				return (hole);
1052 			if (SEQ_LT(hole->rxmit, hole->end)) {
1053 				tp->sackhint.nexthole = hole;
1054 				break;
1055 			}
1056 		}
1057 	}
1058 	KASSERT(SEQ_LT(hole->start, hole->end), ("%s: hole.start >= hole.end", __func__));
1059 	if (!(V_tcp_do_newsack)) {
1060 		KASSERT(SEQ_LT(hole->start, tp->snd_fack), ("%s: hole.start >= snd.fack", __func__));
1061 		KASSERT(SEQ_LT(hole->end, tp->snd_fack), ("%s: hole.end >= snd.fack", __func__));
1062 		KASSERT(SEQ_LT(hole->rxmit, tp->snd_fack), ("%s: hole.rxmit >= snd.fack", __func__));
1063 		if (SEQ_GEQ(hole->start, hole->end) ||
1064 		    SEQ_GEQ(hole->start, tp->snd_fack) ||
1065 		    SEQ_GEQ(hole->end, tp->snd_fack) ||
1066 		    SEQ_GEQ(hole->rxmit, tp->snd_fack)) {
1067 			log(LOG_CRIT,"tcp: invalid SACK hole (%u-%u,%u) vs fwd ack %u, ignoring.\n",
1068 					hole->start, hole->end, hole->rxmit, tp->snd_fack);
1069 			return (NULL);
1070 		}
1071 	}
1072 	return (hole);
1073 }
1074 
1075 /*
1076  * After a timeout, the SACK list may be rebuilt.  This SACK information
1077  * should be used to avoid retransmitting SACKed data.  This function
1078  * traverses the SACK list to see if snd_nxt should be moved forward.
1079  * In addition, cwnd will be inflated by the sacked bytes traversed when
1080  * moving snd_nxt forward. This prevents a traffic burst after the final
1081  * full ACK, and also keeps ACKs coming back.
1082  */
1083 int
1084 tcp_sack_adjust(struct tcpcb *tp)
1085 {
1086 	int sacked = 0;
1087 	struct sackhole *p, *cur = TAILQ_FIRST(&tp->snd_holes);
1088 
1089 	INP_WLOCK_ASSERT(tptoinpcb(tp));
1090 	if (cur == NULL) {
1091 		/* No holes */
1092 		return (0);
1093 	}
1094 	if (SEQ_GEQ(tp->snd_nxt, tp->snd_fack)) {
1095 		/* We're already beyond any SACKed blocks */
1096 		return (tp->sackhint.sacked_bytes);
1097 	}
1098 	/*
1099 	 * Two cases for which we want to advance snd_nxt:
1100 	 * i) snd_nxt lies between end of one hole and beginning of another
1101 	 * ii) snd_nxt lies between end of last hole and snd_fack
1102 	 */
1103 	while ((p = TAILQ_NEXT(cur, scblink)) != NULL) {
1104 		if (SEQ_LT(tp->snd_nxt, cur->end)) {
1105 			return (sacked);
1106 		}
1107 		sacked += p->start - cur->end;
1108 		if (SEQ_GEQ(tp->snd_nxt, p->start)) {
1109 			cur = p;
1110 		} else {
1111 			tp->snd_nxt = p->start;
1112 			return (sacked);
1113 		}
1114 	}
1115 	if (SEQ_LT(tp->snd_nxt, cur->end)) {
1116 		return (sacked);
1117 	}
1118 	tp->snd_nxt = tp->snd_fack;
1119 	return (tp->sackhint.sacked_bytes);
1120 }
1121 
1122 /*
1123  * Lost Retransmission Detection
1124  * Check is FACK is beyond the rexmit of the leftmost hole.
1125  * If yes, we restart sending from still existing holes,
1126  * and adjust cwnd via the congestion control module.
1127  */
1128 void
1129 tcp_sack_lost_retransmission(struct tcpcb *tp, struct tcphdr *th)
1130 {
1131 	struct sackhole *temp;
1132 
1133 	if (IN_RECOVERY(tp->t_flags) &&
1134 	    SEQ_GT(tp->snd_fack, tp->snd_recover) &&
1135 	    ((temp = TAILQ_FIRST(&tp->snd_holes)) != NULL) &&
1136 	    SEQ_GEQ(temp->rxmit, temp->end) &&
1137 	    SEQ_GEQ(tp->snd_fack, temp->rxmit)) {
1138 		TCPSTAT_INC(tcps_sack_lostrexmt);
1139 		/*
1140 		 * Start retransmissions from the first hole, and
1141 		 * subsequently all other remaining holes, including
1142 		 * those, which had been sent completely before.
1143 		 */
1144 		tp->sackhint.nexthole = temp;
1145 		TAILQ_FOREACH(temp, &tp->snd_holes, scblink) {
1146 			if (SEQ_GEQ(tp->snd_fack, temp->rxmit) &&
1147 			    SEQ_GEQ(temp->rxmit, temp->end))
1148 				temp->rxmit = temp->start;
1149 		}
1150 		/*
1151 		 * Remember the old ssthresh, to deduct the beta factor used
1152 		 * by the CC module. Finally, set cwnd to ssthresh just
1153 		 * prior to invoking another cwnd reduction by the CC
1154 		 * module, to not shrink it excessively.
1155 		 */
1156 		tp->snd_cwnd = tp->snd_ssthresh;
1157 		/*
1158 		 * Formally exit recovery, and let the CC module adjust
1159 		 * ssthresh as intended.
1160 		 */
1161 		EXIT_RECOVERY(tp->t_flags);
1162 		cc_cong_signal(tp, th, CC_NDUPACK);
1163 		/*
1164 		 * For PRR, adjust recover_fs as if this new reduction
1165 		 * initialized this variable.
1166 		 * cwnd will be adjusted by SACK or PRR processing
1167 		 * subsequently, only set it to a safe value here.
1168 		 */
1169 		tp->snd_cwnd = tcp_maxseg(tp);
1170 		tp->sackhint.recover_fs = (tp->snd_max - tp->snd_una) -
1171 					    tp->sackhint.recover_fs;
1172 	}
1173 }
1174