1/*-
2 * Copyright (c) 2001-2007, by Cisco Systems, Inc. All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions are met:
6 *
7 * a) Redistributions of source code must retain the above copyright notice,
8 * this list of conditions and the following disclaimer.
9 *
10 * b) Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in
12 * the documentation and/or other materials provided with the distribution.
13 *
14 * c) Neither the name of Cisco Systems, Inc. nor the names of its
15 * contributors may be used to endorse or promote products derived
16 * from this software without specific prior written permission.
17 *
18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
20 * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
22 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
28 * THE POSSIBILITY OF SUCH DAMAGE.
29 */
30
31#include <netinet/sctp_os.h>
32#include <netinet/sctp_var.h>
33#include <netinet/sctp_sysctl.h>
34#include <netinet/sctp_pcb.h>
35#include <netinet/sctp_header.h>
36#include <netinet/sctputil.h>
37#include <netinet/sctp_output.h>
38#include <netinet/sctp_input.h>
39#include <netinet/sctp_indata.h>
40#include <netinet/sctp_uio.h>
41#include <netinet/sctp_timer.h>
42#include <netinet/sctp_auth.h>
43#include <netinet/sctp_asconf.h>
44#include <netinet/sctp_cc_functions.h>
45#include <sys/cdefs.h>
46__FBSDID("$FreeBSD: head/sys/netinet/sctp_cc_functions.c 179157 2008-05-20 13:47:46Z rrs $");
47void
48sctp_set_initial_cc_param(struct sctp_tcb *stcb, struct sctp_nets *net)
49{
50 /*
51 * We take the max of the burst limit times a MTU or the
52 * INITIAL_CWND. We then limit this to 4 MTU's of sending. cwnd must
53 * be at least 2 MTU.
54 */
55 net->cwnd = min((net->mtu * 4), max((2 * net->mtu), SCTP_INITIAL_CWND));
56 net->ssthresh = stcb->asoc.peers_rwnd;
57
58 if (sctp_logging_level & (SCTP_CWND_MONITOR_ENABLE | SCTP_CWND_LOGGING_ENABLE)) {
59 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_INITIALIZATION);
60 }
61}
62
63void
64sctp_cwnd_update_after_fr(struct sctp_tcb *stcb,
65 struct sctp_association *asoc)
66{
67 struct sctp_nets *net;
68
69 /*-
70 * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off == 1) &&
71 * (net->fast_retran_loss_recovery == 0)))
72 */
73 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
74 if ((asoc->fast_retran_loss_recovery == 0) || (sctp_cmt_on_off == 1)) {
75 /* out of a RFC2582 Fast recovery window? */
76 if (net->net_ack > 0) {
77 /*
78 * per section 7.2.3, are there any
79 * destinations that had a fast retransmit
80 * to them. If so what we need to do is
81 * adjust ssthresh and cwnd.
82 */
83 struct sctp_tmit_chunk *lchk;
84 int old_cwnd = net->cwnd;
85
86 net->ssthresh = net->cwnd / 2;
87 if (net->ssthresh < (net->mtu * 2)) {
88 net->ssthresh = 2 * net->mtu;
89 }
90 net->cwnd = net->ssthresh;
91 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
92 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd),
93 SCTP_CWND_LOG_FROM_FR);
94 }
95 lchk = TAILQ_FIRST(&asoc->send_queue);
96
97 net->partial_bytes_acked = 0;
98 /* Turn on fast recovery window */
99 asoc->fast_retran_loss_recovery = 1;
100 if (lchk == NULL) {
101 /* Mark end of the window */
102 asoc->fast_recovery_tsn = asoc->sending_seq - 1;
103 } else {
104 asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
105 }
106
107 /*
108 * CMT fast recovery -- per destination
109 * recovery variable.
110 */
111 net->fast_retran_loss_recovery = 1;
112
113 if (lchk == NULL) {
114 /* Mark end of the window */
115 net->fast_recovery_tsn = asoc->sending_seq - 1;
116 } else {
117 net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
118 }
119
120 /*
121 * Disable Nonce Sum Checking and store the
122 * resync tsn
123 */
124 asoc->nonce_sum_check = 0;
125 asoc->nonce_resync_tsn = asoc->fast_recovery_tsn + 1;
126
127 sctp_timer_stop(SCTP_TIMER_TYPE_SEND,
128 stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32);
129 sctp_timer_start(SCTP_TIMER_TYPE_SEND,
130 stcb->sctp_ep, stcb, net);
131 }
132 } else if (net->net_ack > 0) {
133 /*
134 * Mark a peg that we WOULD have done a cwnd
135 * reduction but RFC2582 prevented this action.
136 */
137 SCTP_STAT_INCR(sctps_fastretransinrtt);
138 }
139 }
140}
141
142void
143sctp_cwnd_update_after_sack(struct sctp_tcb *stcb,
144 struct sctp_association *asoc,
145 int accum_moved, int reneged_all, int will_exit)
146{
147 struct sctp_nets *net;
148
149 /******************************/
150 /* update cwnd and Early FR */
151 /******************************/
152 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
153
154#ifdef JANA_CMT_FAST_RECOVERY
155 /*
156 * CMT fast recovery code. Need to debug.
157 */
158 if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) {
159 if (compare_with_wrap(asoc->last_acked_seq,
160 net->fast_recovery_tsn, MAX_TSN) ||
161 (asoc->last_acked_seq == net->fast_recovery_tsn) ||
162 compare_with_wrap(net->pseudo_cumack, net->fast_recovery_tsn, MAX_TSN) ||
163 (net->pseudo_cumack == net->fast_recovery_tsn)) {
164 net->will_exit_fast_recovery = 1;
165 }
166 }
167#endif
168 if (sctp_early_fr) {
169 /*
170 * So, first of all do we need to have a Early FR
171 * timer running?
172 */
173 if (((TAILQ_FIRST(&asoc->sent_queue)) &&
174 (net->ref_count > 1) &&
175 (net->flight_size < net->cwnd)) ||
176 (reneged_all)) {
177 /*
178 * yes, so in this case stop it if its
179 * running, and then restart it. Reneging
180 * all is a special case where we want to
181 * run the Early FR timer and then force the
182 * last few unacked to be sent, causing us
183 * to illicit a sack with gaps to force out
184 * the others.
185 */
186 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
187 SCTP_STAT_INCR(sctps_earlyfrstpidsck2);
188 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
189 SCTP_FROM_SCTP_INDATA + SCTP_LOC_20);
190 }
191 SCTP_STAT_INCR(sctps_earlyfrstrid);
192 sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net);
193 } else {
194 /* No, stop it if its running */
195 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
196 SCTP_STAT_INCR(sctps_earlyfrstpidsck3);
197 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
198 SCTP_FROM_SCTP_INDATA + SCTP_LOC_21);
199 }
200 }
201 }
202 /* if nothing was acked on this destination skip it */
203 if (net->net_ack == 0) {
204 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
205 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK);
206 }
207 continue;
208 }
209 if (net->net_ack2 > 0) {
210 /*
211 * Karn's rule applies to clearing error count, this
212 * is optional.
213 */
214 net->error_count = 0;
215 if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) ==
216 SCTP_ADDR_NOT_REACHABLE) {
217 /* addr came good */
218 net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE;
219 net->dest_state |= SCTP_ADDR_REACHABLE;
220 sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb,
221 SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED);
222 /* now was it the primary? if so restore */
223 if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) {
224 (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net);
225 }
226 }
227 /*
228 * JRS 5/14/07 - If CMT PF is on and the destination
229 * is in PF state, set the destination to active
230 * state and set the cwnd to one or two MTU's based
231 * on whether PF1 or PF2 is being used.
232 *
233 * Should we stop any running T3 timer here?
234 */
235 if (sctp_cmt_on_off && sctp_cmt_pf && ((net->dest_state & SCTP_ADDR_PF) ==
236 SCTP_ADDR_PF)) {
237 net->dest_state &= ~SCTP_ADDR_PF;
238 net->cwnd = net->mtu * sctp_cmt_pf;
239 SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n",
240 net, net->cwnd);
241 /*
242 * Since the cwnd value is explicitly set,
243 * skip the code that updates the cwnd
244 * value.
245 */
246 goto skip_cwnd_update;
247 }
248 }
249#ifdef JANA_CMT_FAST_RECOVERY
250 /*
251 * CMT fast recovery code
252 */
253 /*
254 * if (sctp_cmt_on_off == 1 &&
255 * net->fast_retran_loss_recovery &&
256 * net->will_exit_fast_recovery == 0) { @@@ Do something }
257 * else if (sctp_cmt_on_off == 0 &&
258 * asoc->fast_retran_loss_recovery && will_exit == 0) {
259 */
260#endif
261
262 if (asoc->fast_retran_loss_recovery && will_exit == 0 && sctp_cmt_on_off == 0) {
263 /*
264 * If we are in loss recovery we skip any cwnd
265 * update
266 */
267 goto skip_cwnd_update;
268 }
269 /*
270 * CMT: CUC algorithm. Update cwnd if pseudo-cumack has
271 * moved.
272 */
273 if (accum_moved || (sctp_cmt_on_off && net->new_pseudo_cumack)) {
274 /* If the cumulative ack moved we can proceed */
275 if (net->cwnd <= net->ssthresh) {
276 /* We are in slow start */
277 if (net->flight_size + net->net_ack >= net->cwnd) {
278 if (net->net_ack > (net->mtu * sctp_L2_abc_variable)) {
279 net->cwnd += (net->mtu * sctp_L2_abc_variable);
280 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
281 sctp_log_cwnd(stcb, net, net->mtu,
282 SCTP_CWND_LOG_FROM_SS);
283 }
284 } else {
285 net->cwnd += net->net_ack;
286 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
287 sctp_log_cwnd(stcb, net, net->net_ack,
288 SCTP_CWND_LOG_FROM_SS);
289 }
290 }
291 } else {
292 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
293 sctp_log_cwnd(stcb, net, net->net_ack,
294 SCTP_CWND_LOG_NOADV_SS);
295 }
296 }
297 } else {
298 /* We are in congestion avoidance */
299 /*
300 * Add to pba
301 */
302 net->partial_bytes_acked += net->net_ack;
303
304 if ((net->flight_size + net->net_ack >= net->cwnd) &&
305 (net->partial_bytes_acked >= net->cwnd)) {
306 net->partial_bytes_acked -= net->cwnd;
307 net->cwnd += net->mtu;
308 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
309 sctp_log_cwnd(stcb, net, net->mtu,
310 SCTP_CWND_LOG_FROM_CA);
311 }
312 } else {
313 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
314 sctp_log_cwnd(stcb, net, net->net_ack,
315 SCTP_CWND_LOG_NOADV_CA);
316 }
317 }
318 }
319 } else {
320 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
321 sctp_log_cwnd(stcb, net, net->mtu,
322 SCTP_CWND_LOG_NO_CUMACK);
323 }
324 }
325skip_cwnd_update:
326 /*
327 * NOW, according to Karn's rule do we need to restore the
328 * RTO timer back? Check our net_ack2. If not set then we
329 * have a ambiguity.. i.e. all data ack'd was sent to more
330 * than one place.
331 */
332 if (net->net_ack2) {
333 /* restore any doubled timers */
334 net->RTO = ((net->lastsa >> 2) + net->lastsv) >> 1;
335 if (net->RTO < stcb->asoc.minrto) {
336 net->RTO = stcb->asoc.minrto;
337 }
338 if (net->RTO > stcb->asoc.maxrto) {
339 net->RTO = stcb->asoc.maxrto;
340 }
341 }
342 }
343}
344
345void
346sctp_cwnd_update_after_timeout(struct sctp_tcb *stcb, struct sctp_nets *net)
347{
348 int old_cwnd = net->cwnd;
349
350 net->ssthresh = max(net->cwnd / 2, 2 * net->mtu);
351 net->cwnd = net->mtu;
352 net->partial_bytes_acked = 0;
353
354 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
355 sctp_log_cwnd(stcb, net, net->cwnd - old_cwnd, SCTP_CWND_LOG_FROM_RTX);
356 }
357}
358
359void
360sctp_cwnd_update_after_ecn_echo(struct sctp_tcb *stcb, struct sctp_nets *net)
361{
362 int old_cwnd = net->cwnd;
363
364 SCTP_STAT_INCR(sctps_ecnereducedcwnd);
365 net->ssthresh = net->cwnd / 2;
366 if (net->ssthresh < net->mtu) {
367 net->ssthresh = net->mtu;
368 /* here back off the timer as well, to slow us down */
369 net->RTO <<= 1;
370 }
371 net->cwnd = net->ssthresh;
372 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
373 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT);
374 }
375}
376
377void
378sctp_cwnd_update_after_packet_dropped(struct sctp_tcb *stcb,
379 struct sctp_nets *net, struct sctp_pktdrop_chunk *cp,
380 uint32_t * bottle_bw, uint32_t * on_queue)
381{
382 uint32_t bw_avail;
383 int rtt, incr;
384 int old_cwnd = net->cwnd;
385
386 /* need real RTT for this calc */
387 rtt = ((net->lastsa >> 2) + net->lastsv) >> 1;
388 /* get bottle neck bw */
389 *bottle_bw = ntohl(cp->bottle_bw);
390 /* and whats on queue */
391 *on_queue = ntohl(cp->current_onq);
392 /*
393 * adjust the on-queue if our flight is more it could be that the
394 * router has not yet gotten data "in-flight" to it
395 */
396 if (*on_queue < net->flight_size)
397 *on_queue = net->flight_size;
398 /* calculate the available space */
399 bw_avail = (*bottle_bw * rtt) / 1000;
400 if (bw_avail > *bottle_bw) {
401 /*
402 * Cap the growth to no more than the bottle neck. This can
403 * happen as RTT slides up due to queues. It also means if
404 * you have more than a 1 second RTT with a empty queue you
405 * will be limited to the bottle_bw per second no matter if
406 * other points have 1/2 the RTT and you could get more
407 * out...
408 */
409 bw_avail = *bottle_bw;
410 }
411 if (*on_queue > bw_avail) {
412 /*
413 * No room for anything else don't allow anything else to be
414 * "added to the fire".
415 */
416 int seg_inflight, seg_onqueue, my_portion;
417
418 net->partial_bytes_acked = 0;
419
420 /* how much are we over queue size? */
421 incr = *on_queue - bw_avail;
422 if (stcb->asoc.seen_a_sack_this_pkt) {
423 /*
424 * undo any cwnd adjustment that the sack might have
425 * made
426 */
427 net->cwnd = net->prev_cwnd;
428 }
429 /* Now how much of that is mine? */
430 seg_inflight = net->flight_size / net->mtu;
431 seg_onqueue = *on_queue / net->mtu;
432 my_portion = (incr * seg_inflight) / seg_onqueue;
433
434 /* Have I made an adjustment already */
435 if (net->cwnd > net->flight_size) {
436 /*
437 * for this flight I made an adjustment we need to
438 * decrease the portion by a share our previous
439 * adjustment.
440 */
441 int diff_adj;
442
443 diff_adj = net->cwnd - net->flight_size;
444 if (diff_adj > my_portion)
445 my_portion = 0;
446 else
447 my_portion -= diff_adj;
448 }
449 /*
450 * back down to the previous cwnd (assume we have had a sack
451 * before this packet). minus what ever portion of the
452 * overage is my fault.
453 */
454 net->cwnd -= my_portion;
455
456 /* we will NOT back down more than 1 MTU */
457 if (net->cwnd <= net->mtu) {
458 net->cwnd = net->mtu;
459 }
460 /* force into CA */
461 net->ssthresh = net->cwnd - 1;
462 } else {
463 /*
464 * Take 1/4 of the space left or max burst up .. whichever
465 * is less.
466 */
467 incr = min((bw_avail - *on_queue) >> 2,
468 stcb->asoc.max_burst * net->mtu);
469 net->cwnd += incr;
470 }
471 if (net->cwnd > bw_avail) {
472 /* We can't exceed the pipe size */
473 net->cwnd = bw_avail;
474 }
475 if (net->cwnd < net->mtu) {
476 /* We always have 1 MTU */
477 net->cwnd = net->mtu;
478 }
479 if (net->cwnd - old_cwnd != 0) {
480 /* log only changes */
481 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
482 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd),
483 SCTP_CWND_LOG_FROM_SAT);
484 }
485 }
486}
487
488void
489sctp_cwnd_update_after_output(struct sctp_tcb *stcb,
490 struct sctp_nets *net, int burst_limit)
491{
492 int old_cwnd = net->cwnd;
493
494 if (net->ssthresh < net->cwnd)
495 net->ssthresh = net->cwnd;
496 net->cwnd = (net->flight_size + (burst_limit * net->mtu));
497
498 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
499 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_BRST);
500 }
501}
502
503void
504sctp_cwnd_update_after_fr_timer(struct sctp_inpcb *inp,
505 struct sctp_tcb *stcb, struct sctp_nets *net)
506{
507 int old_cwnd = net->cwnd;
508
509 sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_EARLY_FR_TMR, SCTP_SO_NOT_LOCKED);
510 /*
511 * make a small adjustment to cwnd and force to CA.
512 */
513 if (net->cwnd > net->mtu)
514 /* drop down one MTU after sending */
515 net->cwnd -= net->mtu;
516 if (net->cwnd < net->ssthresh)
517 /* still in SS move to CA */
518 net->ssthresh = net->cwnd - 1;
519 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
520 sctp_log_cwnd(stcb, net, (old_cwnd - net->cwnd), SCTP_CWND_LOG_FROM_FR);
521 }
522}
523
524struct sctp_hs_raise_drop {
525 int32_t cwnd;
526 int32_t increase;
527 int32_t drop_percent;
528};
529
530#define SCTP_HS_TABLE_SIZE 73
531
532struct sctp_hs_raise_drop sctp_cwnd_adjust[SCTP_HS_TABLE_SIZE] = {
533 {38, 1, 50}, /* 0 */
534 {118, 2, 44}, /* 1 */
535 {221, 3, 41}, /* 2 */
536 {347, 4, 38}, /* 3 */
537 {495, 5, 37}, /* 4 */
538 {663, 6, 35}, /* 5 */
539 {851, 7, 34}, /* 6 */
540 {1058, 8, 33}, /* 7 */
541 {1284, 9, 32}, /* 8 */
542 {1529, 10, 31}, /* 9 */
543 {1793, 11, 30}, /* 10 */
544 {2076, 12, 29}, /* 11 */
545 {2378, 13, 28}, /* 12 */
546 {2699, 14, 28}, /* 13 */
547 {3039, 15, 27}, /* 14 */
548 {3399, 16, 27}, /* 15 */
549 {3778, 17, 26}, /* 16 */
550 {4177, 18, 26}, /* 17 */
551 {4596, 19, 25}, /* 18 */
552 {5036, 20, 25}, /* 19 */
553 {5497, 21, 24}, /* 20 */
554 {5979, 22, 24}, /* 21 */
555 {6483, 23, 23}, /* 22 */
556 {7009, 24, 23}, /* 23 */
557 {7558, 25, 22}, /* 24 */
558 {8130, 26, 22}, /* 25 */
559 {8726, 27, 22}, /* 26 */
560 {9346, 28, 21}, /* 27 */
561 {9991, 29, 21}, /* 28 */
562 {10661, 30, 21}, /* 29 */
563 {11358, 31, 20}, /* 30 */
564 {12082, 32, 20}, /* 31 */
565 {12834, 33, 20}, /* 32 */
566 {13614, 34, 19}, /* 33 */
567 {14424, 35, 19}, /* 34 */
568 {15265, 36, 19}, /* 35 */
569 {16137, 37, 19}, /* 36 */
570 {17042, 38, 18}, /* 37 */
571 {17981, 39, 18}, /* 38 */
572 {18955, 40, 18}, /* 39 */
573 {19965, 41, 17}, /* 40 */
574 {21013, 42, 17}, /* 41 */
575 {22101, 43, 17}, /* 42 */
576 {23230, 44, 17}, /* 43 */
577 {24402, 45, 16}, /* 44 */
578 {25618, 46, 16}, /* 45 */
579 {26881, 47, 16}, /* 46 */
580 {28193, 48, 16}, /* 47 */
581 {29557, 49, 15}, /* 48 */
582 {30975, 50, 15}, /* 49 */
583 {32450, 51, 15}, /* 50 */
584 {33986, 52, 15}, /* 51 */
585 {35586, 53, 14}, /* 52 */
586 {37253, 54, 14}, /* 53 */
587 {38992, 55, 14}, /* 54 */
588 {40808, 56, 14}, /* 55 */
589 {42707, 57, 13}, /* 56 */
590 {44694, 58, 13}, /* 57 */
591 {46776, 59, 13}, /* 58 */
592 {48961, 60, 13}, /* 59 */
593 {51258, 61, 13}, /* 60 */
594 {53677, 62, 12}, /* 61 */
595 {56230, 63, 12}, /* 62 */
596 {58932, 64, 12}, /* 63 */
597 {61799, 65, 12}, /* 64 */
598 {64851, 66, 11}, /* 65 */
599 {68113, 67, 11}, /* 66 */
600 {71617, 68, 11}, /* 67 */
601 {75401, 69, 10}, /* 68 */
602 {79517, 70, 10}, /* 69 */
603 {84035, 71, 10}, /* 70 */
604 {89053, 72, 10}, /* 71 */
605 {94717, 73, 9} /* 72 */
606};
607
608static void
609sctp_hs_cwnd_increase(struct sctp_tcb *stcb, struct sctp_nets *net)
610{
611 int cur_val, i, indx, incr;
612
613 cur_val = net->cwnd >> 10;
614 indx = SCTP_HS_TABLE_SIZE - 1;
615#ifdef SCTP_DEBUG
616 printf("HS CC CAlled.\n");
617#endif
618 if (cur_val < sctp_cwnd_adjust[0].cwnd) {
619 /* normal mode */
620 if (net->net_ack > net->mtu) {
621 net->cwnd += net->mtu;
622 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
623 sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_FROM_SS);
624 }
625 } else {
626 net->cwnd += net->net_ack;
627 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
628 sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_FROM_SS);
629 }
630 }
631 } else {
632 for (i = net->last_hs_used; i < SCTP_HS_TABLE_SIZE; i++) {
633 if (cur_val < sctp_cwnd_adjust[i].cwnd) {
634 indx = i;
635 break;
636 }
637 }
638 net->last_hs_used = indx;
639 incr = ((sctp_cwnd_adjust[indx].increase) << 10);
640 net->cwnd += incr;
641 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
642 sctp_log_cwnd(stcb, net, incr, SCTP_CWND_LOG_FROM_SS);
643 }
644 }
645}
646
647static void
648sctp_hs_cwnd_decrease(struct sctp_tcb *stcb, struct sctp_nets *net)
649{
650 int cur_val, i, indx;
651 int old_cwnd = net->cwnd;
652
653 cur_val = net->cwnd >> 10;
654 indx = net->last_hs_used;
655 if (cur_val < sctp_cwnd_adjust[0].cwnd) {
656 /* normal mode */
657 net->ssthresh = net->cwnd / 2;
658 if (net->ssthresh < (net->mtu * 2)) {
659 net->ssthresh = 2 * net->mtu;
660 }
661 net->cwnd = net->ssthresh;
662 } else {
663 /* drop by the proper amount */
664 net->ssthresh = net->cwnd - (int)((net->cwnd / 100) *
665 sctp_cwnd_adjust[net->last_hs_used].drop_percent);
666 net->cwnd = net->ssthresh;
667 /* now where are we */
668 indx = net->last_hs_used;
669 cur_val = net->cwnd >> 10;
670 /* reset where we are in the table */
671 if (cur_val < sctp_cwnd_adjust[0].cwnd) {
672 /* feel out of hs */
673 net->last_hs_used = 0;
674 } else {
675 for (i = indx; i >= 1; i--) {
676 if (cur_val > sctp_cwnd_adjust[i - 1].cwnd) {
677 break;
678 }
679 }
680 net->last_hs_used = indx;
681 }
682 }
683 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
684 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_FR);
685 }
686}
687
688void
689sctp_hs_cwnd_update_after_fr(struct sctp_tcb *stcb,
690 struct sctp_association *asoc)
691{
692 struct sctp_nets *net;
693
694 /*
695 * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off == 1) &&
696 * (net->fast_retran_loss_recovery == 0)))
697 */
698 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
699 if ((asoc->fast_retran_loss_recovery == 0) || (sctp_cmt_on_off == 1)) {
700 /* out of a RFC2582 Fast recovery window? */
701 if (net->net_ack > 0) {
702 /*
703 * per section 7.2.3, are there any
704 * destinations that had a fast retransmit
705 * to them. If so what we need to do is
706 * adjust ssthresh and cwnd.
707 */
708 struct sctp_tmit_chunk *lchk;
709
710 sctp_hs_cwnd_decrease(stcb, net);
711
712 lchk = TAILQ_FIRST(&asoc->send_queue);
713
714 net->partial_bytes_acked = 0;
715 /* Turn on fast recovery window */
716 asoc->fast_retran_loss_recovery = 1;
717 if (lchk == NULL) {
718 /* Mark end of the window */
719 asoc->fast_recovery_tsn = asoc->sending_seq - 1;
720 } else {
721 asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
722 }
723
724 /*
725 * CMT fast recovery -- per destination
726 * recovery variable.
727 */
728 net->fast_retran_loss_recovery = 1;
729
730 if (lchk == NULL) {
731 /* Mark end of the window */
732 net->fast_recovery_tsn = asoc->sending_seq - 1;
733 } else {
734 net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
735 }
736
737 /*
738 * Disable Nonce Sum Checking and store the
739 * resync tsn
740 */
741 asoc->nonce_sum_check = 0;
742 asoc->nonce_resync_tsn = asoc->fast_recovery_tsn + 1;
743
744 sctp_timer_stop(SCTP_TIMER_TYPE_SEND,
745 stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32);
746 sctp_timer_start(SCTP_TIMER_TYPE_SEND,
747 stcb->sctp_ep, stcb, net);
748 }
749 } else if (net->net_ack > 0) {
750 /*
751 * Mark a peg that we WOULD have done a cwnd
752 * reduction but RFC2582 prevented this action.
753 */
754 SCTP_STAT_INCR(sctps_fastretransinrtt);
755 }
756 }
757}
758
759void
760sctp_hs_cwnd_update_after_sack(struct sctp_tcb *stcb,
761 struct sctp_association *asoc,
762 int accum_moved, int reneged_all, int will_exit)
763{
764 struct sctp_nets *net;
765
766 /******************************/
767 /* update cwnd and Early FR */
768 /******************************/
769 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
770
771#ifdef JANA_CMT_FAST_RECOVERY
772 /*
773 * CMT fast recovery code. Need to debug.
774 */
775 if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) {
776 if (compare_with_wrap(asoc->last_acked_seq,
777 net->fast_recovery_tsn, MAX_TSN) ||
778 (asoc->last_acked_seq == net->fast_recovery_tsn) ||
779 compare_with_wrap(net->pseudo_cumack, net->fast_recovery_tsn, MAX_TSN) ||
780 (net->pseudo_cumack == net->fast_recovery_tsn)) {
781 net->will_exit_fast_recovery = 1;
782 }
783 }
784#endif
785 if (sctp_early_fr) {
786 /*
787 * So, first of all do we need to have a Early FR
788 * timer running?
789 */
790 if (((TAILQ_FIRST(&asoc->sent_queue)) &&
791 (net->ref_count > 1) &&
792 (net->flight_size < net->cwnd)) ||
793 (reneged_all)) {
794 /*
795 * yes, so in this case stop it if its
796 * running, and then restart it. Reneging
797 * all is a special case where we want to
798 * run the Early FR timer and then force the
799 * last few unacked to be sent, causing us
800 * to illicit a sack with gaps to force out
801 * the others.
802 */
803 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
804 SCTP_STAT_INCR(sctps_earlyfrstpidsck2);
805 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
806 SCTP_FROM_SCTP_INDATA + SCTP_LOC_20);
807 }
808 SCTP_STAT_INCR(sctps_earlyfrstrid);
809 sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net);
810 } else {
811 /* No, stop it if its running */
812 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
813 SCTP_STAT_INCR(sctps_earlyfrstpidsck3);
814 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
815 SCTP_FROM_SCTP_INDATA + SCTP_LOC_21);
816 }
817 }
818 }
819 /* if nothing was acked on this destination skip it */
820 if (net->net_ack == 0) {
821 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
822 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK);
823 }
824 continue;
825 }
826 if (net->net_ack2 > 0) {
827 /*
828 * Karn's rule applies to clearing error count, this
829 * is optional.
830 */
831 net->error_count = 0;
832 if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) ==
833 SCTP_ADDR_NOT_REACHABLE) {
834 /* addr came good */
835 net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE;
836 net->dest_state |= SCTP_ADDR_REACHABLE;
837 sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb,
838 SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED);
839 /* now was it the primary? if so restore */
840 if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) {
841 (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net);
842 }
843 }
844 /*
845 * JRS 5/14/07 - If CMT PF is on and the destination
846 * is in PF state, set the destination to active
847 * state and set the cwnd to one or two MTU's based
848 * on whether PF1 or PF2 is being used.
849 *
850 * Should we stop any running T3 timer here?
851 */
852 if (sctp_cmt_on_off && sctp_cmt_pf && ((net->dest_state & SCTP_ADDR_PF) ==
853 SCTP_ADDR_PF)) {
854 net->dest_state &= ~SCTP_ADDR_PF;
855 net->cwnd = net->mtu * sctp_cmt_pf;
856 SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n",
857 net, net->cwnd);
858 /*
859 * Since the cwnd value is explicitly set,
860 * skip the code that updates the cwnd
861 * value.
862 */
863 goto skip_cwnd_update;
864 }
865 }
866#ifdef JANA_CMT_FAST_RECOVERY
867 /*
868 * CMT fast recovery code
869 */
870 /*
871 * if (sctp_cmt_on_off == 1 &&
872 * net->fast_retran_loss_recovery &&
873 * net->will_exit_fast_recovery == 0) { @@@ Do something }
874 * else if (sctp_cmt_on_off == 0 &&
875 * asoc->fast_retran_loss_recovery && will_exit == 0) {
876 */
877#endif
878
879 if (asoc->fast_retran_loss_recovery && will_exit == 0 && sctp_cmt_on_off == 0) {
880 /*
881 * If we are in loss recovery we skip any cwnd
882 * update
883 */
884 goto skip_cwnd_update;
885 }
886 /*
887 * CMT: CUC algorithm. Update cwnd if pseudo-cumack has
888 * moved.
889 */
890 if (accum_moved || (sctp_cmt_on_off && net->new_pseudo_cumack)) {
891 /* If the cumulative ack moved we can proceed */
892 if (net->cwnd <= net->ssthresh) {
893 /* We are in slow start */
894 if (net->flight_size + net->net_ack >= net->cwnd) {
895
896 sctp_hs_cwnd_increase(stcb, net);
897
898 } else {
899 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
900 sctp_log_cwnd(stcb, net, net->net_ack,
901 SCTP_CWND_LOG_NOADV_SS);
902 }
903 }
904 } else {
905 /* We are in congestion avoidance */
906 net->partial_bytes_acked += net->net_ack;
907 if ((net->flight_size + net->net_ack >= net->cwnd) &&
908 (net->partial_bytes_acked >= net->cwnd)) {
909 net->partial_bytes_acked -= net->cwnd;
910 net->cwnd += net->mtu;
911 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
912 sctp_log_cwnd(stcb, net, net->mtu,
913 SCTP_CWND_LOG_FROM_CA);
914 }
915 } else {
916 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
917 sctp_log_cwnd(stcb, net, net->net_ack,
918 SCTP_CWND_LOG_NOADV_CA);
919 }
920 }
921 }
922 } else {
923 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
924 sctp_log_cwnd(stcb, net, net->mtu,
925 SCTP_CWND_LOG_NO_CUMACK);
926 }
927 }
928skip_cwnd_update:
929 /*
930 * NOW, according to Karn's rule do we need to restore the
931 * RTO timer back? Check our net_ack2. If not set then we
932 * have a ambiguity.. i.e. all data ack'd was sent to more
933 * than one place.
934 */
935 if (net->net_ack2) {
936 /* restore any doubled timers */
937 net->RTO = ((net->lastsa >> 2) + net->lastsv) >> 1;
938 if (net->RTO < stcb->asoc.minrto) {
939 net->RTO = stcb->asoc.minrto;
940 }
941 if (net->RTO > stcb->asoc.maxrto) {
942 net->RTO = stcb->asoc.maxrto;
943 }
944 }
945 }
946}
947
948
949/*
950 * H-TCP congestion control. The algorithm is detailed in:
951 * R.N.Shorten, D.J.Leith:
952 * "H-TCP: TCP for high-speed and long-distance networks"
953 * Proc. PFLDnet, Argonne, 2004.
954 * http://www.hamilton.ie/net/htcp3.pdf
955 */
956
957
958static int use_rtt_scaling = 1;
959static int use_bandwidth_switch = 1;
960
961static inline int
962between(uint32_t seq1, uint32_t seq2, uint32_t seq3)
963{
964 return seq3 - seq2 >= seq1 - seq2;
965}
966
967static inline uint32_t
968htcp_cong_time(struct htcp *ca)
969{
970 return sctp_get_tick_count() - ca->last_cong;
971}
972
973static inline uint32_t
974htcp_ccount(struct htcp *ca)
975{
976 return htcp_cong_time(ca) / ca->minRTT;
977}
978
979static inline void
980htcp_reset(struct htcp *ca)
981{
982 ca->undo_last_cong = ca->last_cong;
983 ca->undo_maxRTT = ca->maxRTT;
984 ca->undo_old_maxB = ca->old_maxB;
985 ca->last_cong = sctp_get_tick_count();
986}
987
988#ifdef SCTP_NOT_USED
989
990static uint32_t
991htcp_cwnd_undo(struct sctp_tcb *stcb, struct sctp_nets *net)
992{
993 net->htcp_ca.last_cong = net->htcp_ca.undo_last_cong;
994 net->htcp_ca.maxRTT = net->htcp_ca.undo_maxRTT;
995 net->htcp_ca.old_maxB = net->htcp_ca.undo_old_maxB;
996 return max(net->cwnd, ((net->ssthresh / net->mtu << 7) / net->htcp_ca.beta) * net->mtu);
997}
998
999#endif
1000
1001static inline void
1002measure_rtt(struct sctp_tcb *stcb, struct sctp_nets *net)
1003{
1004 uint32_t srtt = net->lastsa >> 3;
1005
1006 /* keep track of minimum RTT seen so far, minRTT is zero at first */
1007 if (net->htcp_ca.minRTT > srtt || !net->htcp_ca.minRTT)
1008 net->htcp_ca.minRTT = srtt;
1009
1010 /* max RTT */
1011 if (net->fast_retran_ip == 0 && net->ssthresh < 0xFFFF && htcp_ccount(&net->htcp_ca) > 3) {
1012 if (net->htcp_ca.maxRTT < net->htcp_ca.minRTT)
1013 net->htcp_ca.maxRTT = net->htcp_ca.minRTT;
1014 if (net->htcp_ca.maxRTT < srtt && srtt <= net->htcp_ca.maxRTT + MSEC_TO_TICKS(20))
1015 net->htcp_ca.maxRTT = srtt;
1016 }
1017}
1018
1019static void
1020measure_achieved_throughput(struct sctp_tcb *stcb, struct sctp_nets *net)
1021{
1022 uint32_t now = sctp_get_tick_count();
1023
1024 if (net->fast_retran_ip == 0)
1025 net->htcp_ca.bytes_acked = net->net_ack;
1026
1027 if (!use_bandwidth_switch)
1028 return;
1029
1030 /* achieved throughput calculations */
1031 /* JRS - not 100% sure of this statement */
1032 if (net->fast_retran_ip == 1) {
1033 net->htcp_ca.bytecount = 0;
1034 net->htcp_ca.lasttime = now;
1035 return;
1036 }
1037 net->htcp_ca.bytecount += net->net_ack;
1038
1039 if (net->htcp_ca.bytecount >= net->cwnd - ((net->htcp_ca.alpha >> 7 ? : 1) * net->mtu)
1040 && now - net->htcp_ca.lasttime >= net->htcp_ca.minRTT
1041 && net->htcp_ca.minRTT > 0) {
1042 uint32_t cur_Bi = net->htcp_ca.bytecount / net->mtu * hz / (now - net->htcp_ca.lasttime);
1043
1044 if (htcp_ccount(&net->htcp_ca) <= 3) {
1045 /* just after backoff */
1046 net->htcp_ca.minB = net->htcp_ca.maxB = net->htcp_ca.Bi = cur_Bi;
1047 } else {
1048 net->htcp_ca.Bi = (3 * net->htcp_ca.Bi + cur_Bi) / 4;
1049 if (net->htcp_ca.Bi > net->htcp_ca.maxB)
1050 net->htcp_ca.maxB = net->htcp_ca.Bi;
1051 if (net->htcp_ca.minB > net->htcp_ca.maxB)
1052 net->htcp_ca.minB = net->htcp_ca.maxB;
1053 }
1054 net->htcp_ca.bytecount = 0;
1055 net->htcp_ca.lasttime = now;
1056 }
1057}
1058
1059static inline void
1060htcp_beta_update(struct htcp *ca, uint32_t minRTT, uint32_t maxRTT)
1061{
1062 if (use_bandwidth_switch) {
1063 uint32_t maxB = ca->maxB;
1064 uint32_t old_maxB = ca->old_maxB;
1065
1066 ca->old_maxB = ca->maxB;
1067
1068 if (!between(5 * maxB, 4 * old_maxB, 6 * old_maxB)) {
1069 ca->beta = BETA_MIN;
1070 ca->modeswitch = 0;
1071 return;
1072 }
1073 }
1074 if (ca->modeswitch && minRTT > (uint32_t) MSEC_TO_TICKS(10) && maxRTT) {
1075 ca->beta = (minRTT << 7) / maxRTT;
1076 if (ca->beta < BETA_MIN)
1077 ca->beta = BETA_MIN;
1078 else if (ca->beta > BETA_MAX)
1079 ca->beta = BETA_MAX;
1080 } else {
1081 ca->beta = BETA_MIN;
1082 ca->modeswitch = 1;
1083 }
1084}
1085
1086static inline void
1087htcp_alpha_update(struct htcp *ca)
1088{
1089 uint32_t minRTT = ca->minRTT;
1090 uint32_t factor = 1;
1091 uint32_t diff = htcp_cong_time(ca);
1092
1093 if (diff > (uint32_t) hz) {
1094 diff -= hz;
1095 factor = 1 + (10 * diff + ((diff / 2) * (diff / 2) / hz)) / hz;
1096 }
1097 if (use_rtt_scaling && minRTT) {
1098 uint32_t scale = (hz << 3) / (10 * minRTT);
1099
1100 scale = min(max(scale, 1U << 2), 10U << 3); /* clamping ratio to
1101 * interval [0.5,10]<<3 */
1102 factor = (factor << 3) / scale;
1103 if (!factor)
1104 factor = 1;
1105 }
1106 ca->alpha = 2 * factor * ((1 << 7) - ca->beta);
1107 if (!ca->alpha)
1108 ca->alpha = ALPHA_BASE;
1109}
1110
1111/* After we have the rtt data to calculate beta, we'd still prefer to wait one
1112 * rtt before we adjust our beta to ensure we are working from a consistent
1113 * data.
1114 *
1115 * This function should be called when we hit a congestion event since only at
1116 * that point do we really have a real sense of maxRTT (the queues en route
1117 * were getting just too full now).
1118 */
1119static void
1120htcp_param_update(struct sctp_tcb *stcb, struct sctp_nets *net)
1121{
1122 uint32_t minRTT = net->htcp_ca.minRTT;
1123 uint32_t maxRTT = net->htcp_ca.maxRTT;
1124
1125 htcp_beta_update(&net->htcp_ca, minRTT, maxRTT);
1126 htcp_alpha_update(&net->htcp_ca);
1127
1128 /*
1129 * add slowly fading memory for maxRTT to accommodate routing
1130 * changes etc
1131 */
1132 if (minRTT > 0 && maxRTT > minRTT)
1133 net->htcp_ca.maxRTT = minRTT + ((maxRTT - minRTT) * 95) / 100;
1134}
1135
1136static uint32_t
1137htcp_recalc_ssthresh(struct sctp_tcb *stcb, struct sctp_nets *net)
1138{
1139 htcp_param_update(stcb, net);
1140 return max(((net->cwnd / net->mtu * net->htcp_ca.beta) >> 7) * net->mtu, 2U * net->mtu);
1141}
1142
1143static void
1144htcp_cong_avoid(struct sctp_tcb *stcb, struct sctp_nets *net)
1145{
1146 /*-
1147 * How to handle these functions?
1148 * if (!tcp_is_cwnd_limited(sk, in_flight)) RRS - good question.
1149 * return;
1150 */
1151 if (net->cwnd <= net->ssthresh) {
1152 /* We are in slow start */
1153 if (net->flight_size + net->net_ack >= net->cwnd) {
1154 if (net->net_ack > (net->mtu * sctp_L2_abc_variable)) {
1155 net->cwnd += (net->mtu * sctp_L2_abc_variable);
1156 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1157 sctp_log_cwnd(stcb, net, net->mtu,
1158 SCTP_CWND_LOG_FROM_SS);
1159 }
1160 } else {
1161 net->cwnd += net->net_ack;
1162 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1163 sctp_log_cwnd(stcb, net, net->net_ack,
1164 SCTP_CWND_LOG_FROM_SS);
1165 }
1166 }
1167 } else {
1168 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
1169 sctp_log_cwnd(stcb, net, net->net_ack,
1170 SCTP_CWND_LOG_NOADV_SS);
1171 }
1172 }
1173 } else {
1174 measure_rtt(stcb, net);
1175
1176 /*
1177 * In dangerous area, increase slowly. In theory this is
1178 * net->cwnd += alpha / net->cwnd
1179 */
1180 /* What is snd_cwnd_cnt?? */
1181 if (((net->partial_bytes_acked / net->mtu * net->htcp_ca.alpha) >> 7) * net->mtu >= net->cwnd) {
1182 /*-
1183 * Does SCTP have a cwnd clamp?
1184 * if (net->snd_cwnd < net->snd_cwnd_clamp) - Nope (RRS).
1185 */
1186 net->cwnd += net->mtu;
1187 net->partial_bytes_acked = 0;
1188 htcp_alpha_update(&net->htcp_ca);
1189 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1190 sctp_log_cwnd(stcb, net, net->mtu,
1191 SCTP_CWND_LOG_FROM_CA);
1192 }
1193 } else {
1194 net->partial_bytes_acked += net->net_ack;
1195 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
1196 sctp_log_cwnd(stcb, net, net->net_ack,
1197 SCTP_CWND_LOG_NOADV_CA);
1198 }
1199 }
1200
1201 net->htcp_ca.bytes_acked = net->mtu;
1202 }
1203}
1204
1205#ifdef SCTP_NOT_USED
1206/* Lower bound on congestion window. */
1207static uint32_t
1208htcp_min_cwnd(struct sctp_tcb *stcb, struct sctp_nets *net)
1209{
1210 return net->ssthresh;
1211}
1212
1213#endif
1214
1215static void
1216htcp_init(struct sctp_tcb *stcb, struct sctp_nets *net)
1217{
1218 memset(&net->htcp_ca, 0, sizeof(struct htcp));
1219 net->htcp_ca.alpha = ALPHA_BASE;
1220 net->htcp_ca.beta = BETA_MIN;
1221 net->htcp_ca.bytes_acked = net->mtu;
1222 net->htcp_ca.last_cong = sctp_get_tick_count();
1223}
1224
1225void
1226sctp_htcp_set_initial_cc_param(struct sctp_tcb *stcb, struct sctp_nets *net)
1227{
1228 /*
1229 * We take the max of the burst limit times a MTU or the
1230 * INITIAL_CWND. We then limit this to 4 MTU's of sending.
1231 */
1232 net->cwnd = min((net->mtu * 4), max((2 * net->mtu), SCTP_INITIAL_CWND));
1233 net->ssthresh = stcb->asoc.peers_rwnd;
1234 htcp_init(stcb, net);
1235
1236 if (sctp_logging_level & (SCTP_CWND_MONITOR_ENABLE | SCTP_CWND_LOGGING_ENABLE)) {
1237 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_INITIALIZATION);
1238 }
1239}
1240
1241void
1242sctp_htcp_cwnd_update_after_sack(struct sctp_tcb *stcb,
1243 struct sctp_association *asoc,
1244 int accum_moved, int reneged_all, int will_exit)
1245{
1246 struct sctp_nets *net;
1247
1248 /******************************/
1249 /* update cwnd and Early FR */
1250 /******************************/
1251 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
1252
1253#ifdef JANA_CMT_FAST_RECOVERY
1254 /*
1255 * CMT fast recovery code. Need to debug.
1256 */
1257 if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) {
1258 if (compare_with_wrap(asoc->last_acked_seq,
1259 net->fast_recovery_tsn, MAX_TSN) ||
1260 (asoc->last_acked_seq == net->fast_recovery_tsn) ||
1261 compare_with_wrap(net->pseudo_cumack, net->fast_recovery_tsn, MAX_TSN) ||
1262 (net->pseudo_cumack == net->fast_recovery_tsn)) {
1263 net->will_exit_fast_recovery = 1;
1264 }
1265 }
1266#endif
1267 if (sctp_early_fr) {
1268 /*
1269 * So, first of all do we need to have a Early FR
1270 * timer running?
1271 */
1272 if (((TAILQ_FIRST(&asoc->sent_queue)) &&
1273 (net->ref_count > 1) &&
1274 (net->flight_size < net->cwnd)) ||
1275 (reneged_all)) {
1276 /*
1277 * yes, so in this case stop it if its
1278 * running, and then restart it. Reneging
1279 * all is a special case where we want to
1280 * run the Early FR timer and then force the
1281 * last few unacked to be sent, causing us
1282 * to illicit a sack with gaps to force out
1283 * the others.
1284 */
1285 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
1286 SCTP_STAT_INCR(sctps_earlyfrstpidsck2);
1287 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
1288 SCTP_FROM_SCTP_INDATA + SCTP_LOC_20);
1289 }
1290 SCTP_STAT_INCR(sctps_earlyfrstrid);
1291 sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net);
1292 } else {
1293 /* No, stop it if its running */
1294 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
1295 SCTP_STAT_INCR(sctps_earlyfrstpidsck3);
1296 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
1297 SCTP_FROM_SCTP_INDATA + SCTP_LOC_21);
1298 }
1299 }
1300 }
1301 /* if nothing was acked on this destination skip it */
1302 if (net->net_ack == 0) {
1303 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
1304 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK);
1305 }
1306 continue;
1307 }
1308 if (net->net_ack2 > 0) {
1309 /*
1310 * Karn's rule applies to clearing error count, this
1311 * is optional.
1312 */
1313 net->error_count = 0;
1314 if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) ==
1315 SCTP_ADDR_NOT_REACHABLE) {
1316 /* addr came good */
1317 net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE;
1318 net->dest_state |= SCTP_ADDR_REACHABLE;
1319 sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb,
1320 SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED);
1321 /* now was it the primary? if so restore */
1322 if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) {
1323 (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net);
1324 }
1325 }
1326 /*
1327 * JRS 5/14/07 - If CMT PF is on and the destination
1328 * is in PF state, set the destination to active
1329 * state and set the cwnd to one or two MTU's based
1330 * on whether PF1 or PF2 is being used.
1331 *
1332 * Should we stop any running T3 timer here?
1333 */
1334 if (sctp_cmt_on_off && sctp_cmt_pf && ((net->dest_state & SCTP_ADDR_PF) ==
1335 SCTP_ADDR_PF)) {
1336 net->dest_state &= ~SCTP_ADDR_PF;
1337 net->cwnd = net->mtu * sctp_cmt_pf;
1338 SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n",
1339 net, net->cwnd);
1340 /*
1341 * Since the cwnd value is explicitly set,
1342 * skip the code that updates the cwnd
1343 * value.
1344 */
1345 goto skip_cwnd_update;
1346 }
1347 }
1348#ifdef JANA_CMT_FAST_RECOVERY
1349 /*
1350 * CMT fast recovery code
1351 */
1352 /*
1353 * if (sctp_cmt_on_off == 1 &&
1354 * net->fast_retran_loss_recovery &&
1355 * net->will_exit_fast_recovery == 0) { @@@ Do something }
1356 * else if (sctp_cmt_on_off == 0 &&
1357 * asoc->fast_retran_loss_recovery && will_exit == 0) {
1358 */
1359#endif
1360
1361 if (asoc->fast_retran_loss_recovery && will_exit == 0 && sctp_cmt_on_off == 0) {
1362 /*
1363 * If we are in loss recovery we skip any cwnd
1364 * update
1365 */
1366 goto skip_cwnd_update;
1367 }
1368 /*
1369 * CMT: CUC algorithm. Update cwnd if pseudo-cumack has
1370 * moved.
1371 */
1372 if (accum_moved || (sctp_cmt_on_off && net->new_pseudo_cumack)) {
1373 htcp_cong_avoid(stcb, net);
1374 measure_achieved_throughput(stcb, net);
1375 } else {
1376 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
1377 sctp_log_cwnd(stcb, net, net->mtu,
1378 SCTP_CWND_LOG_NO_CUMACK);
1379 }
1380 }
1381skip_cwnd_update:
1382 /*
1383 * NOW, according to Karn's rule do we need to restore the
1384 * RTO timer back? Check our net_ack2. If not set then we
1385 * have a ambiguity.. i.e. all data ack'd was sent to more
1386 * than one place.
1387 */
1388 if (net->net_ack2) {
1389 /* restore any doubled timers */
1390 net->RTO = ((net->lastsa >> 2) + net->lastsv) >> 1;
1391 if (net->RTO < stcb->asoc.minrto) {
1392 net->RTO = stcb->asoc.minrto;
1393 }
1394 if (net->RTO > stcb->asoc.maxrto) {
1395 net->RTO = stcb->asoc.maxrto;
1396 }
1397 }
1398 }
1399}
1400
1401void
1402sctp_htcp_cwnd_update_after_fr(struct sctp_tcb *stcb,
1403 struct sctp_association *asoc)
1404{
1405 struct sctp_nets *net;
1406
1407 /*
1408 * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off == 1) &&
1409 * (net->fast_retran_loss_recovery == 0)))
1410 */
1411 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
1412 if ((asoc->fast_retran_loss_recovery == 0) || (sctp_cmt_on_off == 1)) {
1413 /* out of a RFC2582 Fast recovery window? */
1414 if (net->net_ack > 0) {
1415 /*
1416 * per section 7.2.3, are there any
1417 * destinations that had a fast retransmit
1418 * to them. If so what we need to do is
1419 * adjust ssthresh and cwnd.
1420 */
1421 struct sctp_tmit_chunk *lchk;
1422 int old_cwnd = net->cwnd;
1423
1424 /* JRS - reset as if state were changed */
1425 htcp_reset(&net->htcp_ca);
1426 net->ssthresh = htcp_recalc_ssthresh(stcb, net);
1427 net->cwnd = net->ssthresh;
1428 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1429 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd),
1430 SCTP_CWND_LOG_FROM_FR);
1431 }
1432 lchk = TAILQ_FIRST(&asoc->send_queue);
1433
1434 net->partial_bytes_acked = 0;
1435 /* Turn on fast recovery window */
1436 asoc->fast_retran_loss_recovery = 1;
1437 if (lchk == NULL) {
1438 /* Mark end of the window */
1439 asoc->fast_recovery_tsn = asoc->sending_seq - 1;
1440 } else {
1441 asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
1442 }
1443
1444 /*
1445 * CMT fast recovery -- per destination
1446 * recovery variable.
1447 */
1448 net->fast_retran_loss_recovery = 1;
1449
1450 if (lchk == NULL) {
1451 /* Mark end of the window */
1452 net->fast_recovery_tsn = asoc->sending_seq - 1;
1453 } else {
1454 net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
1455 }
1456
1457 /*
1458 * Disable Nonce Sum Checking and store the
1459 * resync tsn
1460 */
1461 asoc->nonce_sum_check = 0;
1462 asoc->nonce_resync_tsn = asoc->fast_recovery_tsn + 1;
1463
1464 sctp_timer_stop(SCTP_TIMER_TYPE_SEND,
1465 stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32);
1466 sctp_timer_start(SCTP_TIMER_TYPE_SEND,
1467 stcb->sctp_ep, stcb, net);
1468 }
1469 } else if (net->net_ack > 0) {
1470 /*
1471 * Mark a peg that we WOULD have done a cwnd
1472 * reduction but RFC2582 prevented this action.
1473 */
1474 SCTP_STAT_INCR(sctps_fastretransinrtt);
1475 }
1476 }
1477}
1478
1479void
1480sctp_htcp_cwnd_update_after_timeout(struct sctp_tcb *stcb,
1481 struct sctp_nets *net)
1482{
1483 int old_cwnd = net->cwnd;
1484
1485 /* JRS - reset as if the state were being changed to timeout */
1486 htcp_reset(&net->htcp_ca);
1487 net->ssthresh = htcp_recalc_ssthresh(stcb, net);
1488 net->cwnd = net->mtu;
1489 net->partial_bytes_acked = 0;
1490 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1491 sctp_log_cwnd(stcb, net, net->cwnd - old_cwnd, SCTP_CWND_LOG_FROM_RTX);
1492 }
1493}
1494
1495void
1496sctp_htcp_cwnd_update_after_fr_timer(struct sctp_inpcb *inp,
1497 struct sctp_tcb *stcb, struct sctp_nets *net)
1498{
1499 int old_cwnd;
1500
1501 old_cwnd = net->cwnd;
1502
1503 sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_EARLY_FR_TMR, SCTP_SO_NOT_LOCKED);
1504 net->htcp_ca.last_cong = sctp_get_tick_count();
1505 /*
1506 * make a small adjustment to cwnd and force to CA.
1507 */
1508 if (net->cwnd > net->mtu)
1509 /* drop down one MTU after sending */
1510 net->cwnd -= net->mtu;
1511 if (net->cwnd < net->ssthresh)
1512 /* still in SS move to CA */
1513 net->ssthresh = net->cwnd - 1;
1514 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1515 sctp_log_cwnd(stcb, net, (old_cwnd - net->cwnd), SCTP_CWND_LOG_FROM_FR);
1516 }
1517}
1518
1519void
1520sctp_htcp_cwnd_update_after_ecn_echo(struct sctp_tcb *stcb,
1521 struct sctp_nets *net)
1522{
1523 int old_cwnd;
1524
1525 old_cwnd = net->cwnd;
1526
1527 /* JRS - reset hctp as if state changed */
1528 htcp_reset(&net->htcp_ca);
1529 SCTP_STAT_INCR(sctps_ecnereducedcwnd);
1530 net->ssthresh = htcp_recalc_ssthresh(stcb, net);
1531 if (net->ssthresh < net->mtu) {
1532 net->ssthresh = net->mtu;
1533 /* here back off the timer as well, to slow us down */
1534 net->RTO <<= 1;
1535 }
1536 net->cwnd = net->ssthresh;
1537 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1538 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT);
1539 }
1540}
2 * Copyright (c) 2001-2007, by Cisco Systems, Inc. All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions are met:
6 *
7 * a) Redistributions of source code must retain the above copyright notice,
8 * this list of conditions and the following disclaimer.
9 *
10 * b) Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in
12 * the documentation and/or other materials provided with the distribution.
13 *
14 * c) Neither the name of Cisco Systems, Inc. nor the names of its
15 * contributors may be used to endorse or promote products derived
16 * from this software without specific prior written permission.
17 *
18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
20 * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
22 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
28 * THE POSSIBILITY OF SUCH DAMAGE.
29 */
30
31#include <netinet/sctp_os.h>
32#include <netinet/sctp_var.h>
33#include <netinet/sctp_sysctl.h>
34#include <netinet/sctp_pcb.h>
35#include <netinet/sctp_header.h>
36#include <netinet/sctputil.h>
37#include <netinet/sctp_output.h>
38#include <netinet/sctp_input.h>
39#include <netinet/sctp_indata.h>
40#include <netinet/sctp_uio.h>
41#include <netinet/sctp_timer.h>
42#include <netinet/sctp_auth.h>
43#include <netinet/sctp_asconf.h>
44#include <netinet/sctp_cc_functions.h>
45#include <sys/cdefs.h>
46__FBSDID("$FreeBSD: head/sys/netinet/sctp_cc_functions.c 179157 2008-05-20 13:47:46Z rrs $");
47void
48sctp_set_initial_cc_param(struct sctp_tcb *stcb, struct sctp_nets *net)
49{
50 /*
51 * We take the max of the burst limit times a MTU or the
52 * INITIAL_CWND. We then limit this to 4 MTU's of sending. cwnd must
53 * be at least 2 MTU.
54 */
55 net->cwnd = min((net->mtu * 4), max((2 * net->mtu), SCTP_INITIAL_CWND));
56 net->ssthresh = stcb->asoc.peers_rwnd;
57
58 if (sctp_logging_level & (SCTP_CWND_MONITOR_ENABLE | SCTP_CWND_LOGGING_ENABLE)) {
59 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_INITIALIZATION);
60 }
61}
62
63void
64sctp_cwnd_update_after_fr(struct sctp_tcb *stcb,
65 struct sctp_association *asoc)
66{
67 struct sctp_nets *net;
68
69 /*-
70 * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off == 1) &&
71 * (net->fast_retran_loss_recovery == 0)))
72 */
73 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
74 if ((asoc->fast_retran_loss_recovery == 0) || (sctp_cmt_on_off == 1)) {
75 /* out of a RFC2582 Fast recovery window? */
76 if (net->net_ack > 0) {
77 /*
78 * per section 7.2.3, are there any
79 * destinations that had a fast retransmit
80 * to them. If so what we need to do is
81 * adjust ssthresh and cwnd.
82 */
83 struct sctp_tmit_chunk *lchk;
84 int old_cwnd = net->cwnd;
85
86 net->ssthresh = net->cwnd / 2;
87 if (net->ssthresh < (net->mtu * 2)) {
88 net->ssthresh = 2 * net->mtu;
89 }
90 net->cwnd = net->ssthresh;
91 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
92 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd),
93 SCTP_CWND_LOG_FROM_FR);
94 }
95 lchk = TAILQ_FIRST(&asoc->send_queue);
96
97 net->partial_bytes_acked = 0;
98 /* Turn on fast recovery window */
99 asoc->fast_retran_loss_recovery = 1;
100 if (lchk == NULL) {
101 /* Mark end of the window */
102 asoc->fast_recovery_tsn = asoc->sending_seq - 1;
103 } else {
104 asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
105 }
106
107 /*
108 * CMT fast recovery -- per destination
109 * recovery variable.
110 */
111 net->fast_retran_loss_recovery = 1;
112
113 if (lchk == NULL) {
114 /* Mark end of the window */
115 net->fast_recovery_tsn = asoc->sending_seq - 1;
116 } else {
117 net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
118 }
119
120 /*
121 * Disable Nonce Sum Checking and store the
122 * resync tsn
123 */
124 asoc->nonce_sum_check = 0;
125 asoc->nonce_resync_tsn = asoc->fast_recovery_tsn + 1;
126
127 sctp_timer_stop(SCTP_TIMER_TYPE_SEND,
128 stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32);
129 sctp_timer_start(SCTP_TIMER_TYPE_SEND,
130 stcb->sctp_ep, stcb, net);
131 }
132 } else if (net->net_ack > 0) {
133 /*
134 * Mark a peg that we WOULD have done a cwnd
135 * reduction but RFC2582 prevented this action.
136 */
137 SCTP_STAT_INCR(sctps_fastretransinrtt);
138 }
139 }
140}
141
142void
143sctp_cwnd_update_after_sack(struct sctp_tcb *stcb,
144 struct sctp_association *asoc,
145 int accum_moved, int reneged_all, int will_exit)
146{
147 struct sctp_nets *net;
148
149 /******************************/
150 /* update cwnd and Early FR */
151 /******************************/
152 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
153
154#ifdef JANA_CMT_FAST_RECOVERY
155 /*
156 * CMT fast recovery code. Need to debug.
157 */
158 if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) {
159 if (compare_with_wrap(asoc->last_acked_seq,
160 net->fast_recovery_tsn, MAX_TSN) ||
161 (asoc->last_acked_seq == net->fast_recovery_tsn) ||
162 compare_with_wrap(net->pseudo_cumack, net->fast_recovery_tsn, MAX_TSN) ||
163 (net->pseudo_cumack == net->fast_recovery_tsn)) {
164 net->will_exit_fast_recovery = 1;
165 }
166 }
167#endif
168 if (sctp_early_fr) {
169 /*
170 * So, first of all do we need to have a Early FR
171 * timer running?
172 */
173 if (((TAILQ_FIRST(&asoc->sent_queue)) &&
174 (net->ref_count > 1) &&
175 (net->flight_size < net->cwnd)) ||
176 (reneged_all)) {
177 /*
178 * yes, so in this case stop it if its
179 * running, and then restart it. Reneging
180 * all is a special case where we want to
181 * run the Early FR timer and then force the
182 * last few unacked to be sent, causing us
183 * to illicit a sack with gaps to force out
184 * the others.
185 */
186 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
187 SCTP_STAT_INCR(sctps_earlyfrstpidsck2);
188 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
189 SCTP_FROM_SCTP_INDATA + SCTP_LOC_20);
190 }
191 SCTP_STAT_INCR(sctps_earlyfrstrid);
192 sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net);
193 } else {
194 /* No, stop it if its running */
195 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
196 SCTP_STAT_INCR(sctps_earlyfrstpidsck3);
197 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
198 SCTP_FROM_SCTP_INDATA + SCTP_LOC_21);
199 }
200 }
201 }
202 /* if nothing was acked on this destination skip it */
203 if (net->net_ack == 0) {
204 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
205 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK);
206 }
207 continue;
208 }
209 if (net->net_ack2 > 0) {
210 /*
211 * Karn's rule applies to clearing error count, this
212 * is optional.
213 */
214 net->error_count = 0;
215 if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) ==
216 SCTP_ADDR_NOT_REACHABLE) {
217 /* addr came good */
218 net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE;
219 net->dest_state |= SCTP_ADDR_REACHABLE;
220 sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb,
221 SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED);
222 /* now was it the primary? if so restore */
223 if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) {
224 (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net);
225 }
226 }
227 /*
228 * JRS 5/14/07 - If CMT PF is on and the destination
229 * is in PF state, set the destination to active
230 * state and set the cwnd to one or two MTU's based
231 * on whether PF1 or PF2 is being used.
232 *
233 * Should we stop any running T3 timer here?
234 */
235 if (sctp_cmt_on_off && sctp_cmt_pf && ((net->dest_state & SCTP_ADDR_PF) ==
236 SCTP_ADDR_PF)) {
237 net->dest_state &= ~SCTP_ADDR_PF;
238 net->cwnd = net->mtu * sctp_cmt_pf;
239 SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n",
240 net, net->cwnd);
241 /*
242 * Since the cwnd value is explicitly set,
243 * skip the code that updates the cwnd
244 * value.
245 */
246 goto skip_cwnd_update;
247 }
248 }
249#ifdef JANA_CMT_FAST_RECOVERY
250 /*
251 * CMT fast recovery code
252 */
253 /*
254 * if (sctp_cmt_on_off == 1 &&
255 * net->fast_retran_loss_recovery &&
256 * net->will_exit_fast_recovery == 0) { @@@ Do something }
257 * else if (sctp_cmt_on_off == 0 &&
258 * asoc->fast_retran_loss_recovery && will_exit == 0) {
259 */
260#endif
261
262 if (asoc->fast_retran_loss_recovery && will_exit == 0 && sctp_cmt_on_off == 0) {
263 /*
264 * If we are in loss recovery we skip any cwnd
265 * update
266 */
267 goto skip_cwnd_update;
268 }
269 /*
270 * CMT: CUC algorithm. Update cwnd if pseudo-cumack has
271 * moved.
272 */
273 if (accum_moved || (sctp_cmt_on_off && net->new_pseudo_cumack)) {
274 /* If the cumulative ack moved we can proceed */
275 if (net->cwnd <= net->ssthresh) {
276 /* We are in slow start */
277 if (net->flight_size + net->net_ack >= net->cwnd) {
278 if (net->net_ack > (net->mtu * sctp_L2_abc_variable)) {
279 net->cwnd += (net->mtu * sctp_L2_abc_variable);
280 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
281 sctp_log_cwnd(stcb, net, net->mtu,
282 SCTP_CWND_LOG_FROM_SS);
283 }
284 } else {
285 net->cwnd += net->net_ack;
286 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
287 sctp_log_cwnd(stcb, net, net->net_ack,
288 SCTP_CWND_LOG_FROM_SS);
289 }
290 }
291 } else {
292 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
293 sctp_log_cwnd(stcb, net, net->net_ack,
294 SCTP_CWND_LOG_NOADV_SS);
295 }
296 }
297 } else {
298 /* We are in congestion avoidance */
299 /*
300 * Add to pba
301 */
302 net->partial_bytes_acked += net->net_ack;
303
304 if ((net->flight_size + net->net_ack >= net->cwnd) &&
305 (net->partial_bytes_acked >= net->cwnd)) {
306 net->partial_bytes_acked -= net->cwnd;
307 net->cwnd += net->mtu;
308 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
309 sctp_log_cwnd(stcb, net, net->mtu,
310 SCTP_CWND_LOG_FROM_CA);
311 }
312 } else {
313 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
314 sctp_log_cwnd(stcb, net, net->net_ack,
315 SCTP_CWND_LOG_NOADV_CA);
316 }
317 }
318 }
319 } else {
320 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
321 sctp_log_cwnd(stcb, net, net->mtu,
322 SCTP_CWND_LOG_NO_CUMACK);
323 }
324 }
325skip_cwnd_update:
326 /*
327 * NOW, according to Karn's rule do we need to restore the
328 * RTO timer back? Check our net_ack2. If not set then we
329 * have a ambiguity.. i.e. all data ack'd was sent to more
330 * than one place.
331 */
332 if (net->net_ack2) {
333 /* restore any doubled timers */
334 net->RTO = ((net->lastsa >> 2) + net->lastsv) >> 1;
335 if (net->RTO < stcb->asoc.minrto) {
336 net->RTO = stcb->asoc.minrto;
337 }
338 if (net->RTO > stcb->asoc.maxrto) {
339 net->RTO = stcb->asoc.maxrto;
340 }
341 }
342 }
343}
344
345void
346sctp_cwnd_update_after_timeout(struct sctp_tcb *stcb, struct sctp_nets *net)
347{
348 int old_cwnd = net->cwnd;
349
350 net->ssthresh = max(net->cwnd / 2, 2 * net->mtu);
351 net->cwnd = net->mtu;
352 net->partial_bytes_acked = 0;
353
354 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
355 sctp_log_cwnd(stcb, net, net->cwnd - old_cwnd, SCTP_CWND_LOG_FROM_RTX);
356 }
357}
358
359void
360sctp_cwnd_update_after_ecn_echo(struct sctp_tcb *stcb, struct sctp_nets *net)
361{
362 int old_cwnd = net->cwnd;
363
364 SCTP_STAT_INCR(sctps_ecnereducedcwnd);
365 net->ssthresh = net->cwnd / 2;
366 if (net->ssthresh < net->mtu) {
367 net->ssthresh = net->mtu;
368 /* here back off the timer as well, to slow us down */
369 net->RTO <<= 1;
370 }
371 net->cwnd = net->ssthresh;
372 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
373 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT);
374 }
375}
376
377void
378sctp_cwnd_update_after_packet_dropped(struct sctp_tcb *stcb,
379 struct sctp_nets *net, struct sctp_pktdrop_chunk *cp,
380 uint32_t * bottle_bw, uint32_t * on_queue)
381{
382 uint32_t bw_avail;
383 int rtt, incr;
384 int old_cwnd = net->cwnd;
385
386 /* need real RTT for this calc */
387 rtt = ((net->lastsa >> 2) + net->lastsv) >> 1;
388 /* get bottle neck bw */
389 *bottle_bw = ntohl(cp->bottle_bw);
390 /* and whats on queue */
391 *on_queue = ntohl(cp->current_onq);
392 /*
393 * adjust the on-queue if our flight is more it could be that the
394 * router has not yet gotten data "in-flight" to it
395 */
396 if (*on_queue < net->flight_size)
397 *on_queue = net->flight_size;
398 /* calculate the available space */
399 bw_avail = (*bottle_bw * rtt) / 1000;
400 if (bw_avail > *bottle_bw) {
401 /*
402 * Cap the growth to no more than the bottle neck. This can
403 * happen as RTT slides up due to queues. It also means if
404 * you have more than a 1 second RTT with a empty queue you
405 * will be limited to the bottle_bw per second no matter if
406 * other points have 1/2 the RTT and you could get more
407 * out...
408 */
409 bw_avail = *bottle_bw;
410 }
411 if (*on_queue > bw_avail) {
412 /*
413 * No room for anything else don't allow anything else to be
414 * "added to the fire".
415 */
416 int seg_inflight, seg_onqueue, my_portion;
417
418 net->partial_bytes_acked = 0;
419
420 /* how much are we over queue size? */
421 incr = *on_queue - bw_avail;
422 if (stcb->asoc.seen_a_sack_this_pkt) {
423 /*
424 * undo any cwnd adjustment that the sack might have
425 * made
426 */
427 net->cwnd = net->prev_cwnd;
428 }
429 /* Now how much of that is mine? */
430 seg_inflight = net->flight_size / net->mtu;
431 seg_onqueue = *on_queue / net->mtu;
432 my_portion = (incr * seg_inflight) / seg_onqueue;
433
434 /* Have I made an adjustment already */
435 if (net->cwnd > net->flight_size) {
436 /*
437 * for this flight I made an adjustment we need to
438 * decrease the portion by a share our previous
439 * adjustment.
440 */
441 int diff_adj;
442
443 diff_adj = net->cwnd - net->flight_size;
444 if (diff_adj > my_portion)
445 my_portion = 0;
446 else
447 my_portion -= diff_adj;
448 }
449 /*
450 * back down to the previous cwnd (assume we have had a sack
451 * before this packet). minus what ever portion of the
452 * overage is my fault.
453 */
454 net->cwnd -= my_portion;
455
456 /* we will NOT back down more than 1 MTU */
457 if (net->cwnd <= net->mtu) {
458 net->cwnd = net->mtu;
459 }
460 /* force into CA */
461 net->ssthresh = net->cwnd - 1;
462 } else {
463 /*
464 * Take 1/4 of the space left or max burst up .. whichever
465 * is less.
466 */
467 incr = min((bw_avail - *on_queue) >> 2,
468 stcb->asoc.max_burst * net->mtu);
469 net->cwnd += incr;
470 }
471 if (net->cwnd > bw_avail) {
472 /* We can't exceed the pipe size */
473 net->cwnd = bw_avail;
474 }
475 if (net->cwnd < net->mtu) {
476 /* We always have 1 MTU */
477 net->cwnd = net->mtu;
478 }
479 if (net->cwnd - old_cwnd != 0) {
480 /* log only changes */
481 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
482 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd),
483 SCTP_CWND_LOG_FROM_SAT);
484 }
485 }
486}
487
488void
489sctp_cwnd_update_after_output(struct sctp_tcb *stcb,
490 struct sctp_nets *net, int burst_limit)
491{
492 int old_cwnd = net->cwnd;
493
494 if (net->ssthresh < net->cwnd)
495 net->ssthresh = net->cwnd;
496 net->cwnd = (net->flight_size + (burst_limit * net->mtu));
497
498 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
499 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_BRST);
500 }
501}
502
503void
504sctp_cwnd_update_after_fr_timer(struct sctp_inpcb *inp,
505 struct sctp_tcb *stcb, struct sctp_nets *net)
506{
507 int old_cwnd = net->cwnd;
508
509 sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_EARLY_FR_TMR, SCTP_SO_NOT_LOCKED);
510 /*
511 * make a small adjustment to cwnd and force to CA.
512 */
513 if (net->cwnd > net->mtu)
514 /* drop down one MTU after sending */
515 net->cwnd -= net->mtu;
516 if (net->cwnd < net->ssthresh)
517 /* still in SS move to CA */
518 net->ssthresh = net->cwnd - 1;
519 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
520 sctp_log_cwnd(stcb, net, (old_cwnd - net->cwnd), SCTP_CWND_LOG_FROM_FR);
521 }
522}
523
524struct sctp_hs_raise_drop {
525 int32_t cwnd;
526 int32_t increase;
527 int32_t drop_percent;
528};
529
530#define SCTP_HS_TABLE_SIZE 73
531
532struct sctp_hs_raise_drop sctp_cwnd_adjust[SCTP_HS_TABLE_SIZE] = {
533 {38, 1, 50}, /* 0 */
534 {118, 2, 44}, /* 1 */
535 {221, 3, 41}, /* 2 */
536 {347, 4, 38}, /* 3 */
537 {495, 5, 37}, /* 4 */
538 {663, 6, 35}, /* 5 */
539 {851, 7, 34}, /* 6 */
540 {1058, 8, 33}, /* 7 */
541 {1284, 9, 32}, /* 8 */
542 {1529, 10, 31}, /* 9 */
543 {1793, 11, 30}, /* 10 */
544 {2076, 12, 29}, /* 11 */
545 {2378, 13, 28}, /* 12 */
546 {2699, 14, 28}, /* 13 */
547 {3039, 15, 27}, /* 14 */
548 {3399, 16, 27}, /* 15 */
549 {3778, 17, 26}, /* 16 */
550 {4177, 18, 26}, /* 17 */
551 {4596, 19, 25}, /* 18 */
552 {5036, 20, 25}, /* 19 */
553 {5497, 21, 24}, /* 20 */
554 {5979, 22, 24}, /* 21 */
555 {6483, 23, 23}, /* 22 */
556 {7009, 24, 23}, /* 23 */
557 {7558, 25, 22}, /* 24 */
558 {8130, 26, 22}, /* 25 */
559 {8726, 27, 22}, /* 26 */
560 {9346, 28, 21}, /* 27 */
561 {9991, 29, 21}, /* 28 */
562 {10661, 30, 21}, /* 29 */
563 {11358, 31, 20}, /* 30 */
564 {12082, 32, 20}, /* 31 */
565 {12834, 33, 20}, /* 32 */
566 {13614, 34, 19}, /* 33 */
567 {14424, 35, 19}, /* 34 */
568 {15265, 36, 19}, /* 35 */
569 {16137, 37, 19}, /* 36 */
570 {17042, 38, 18}, /* 37 */
571 {17981, 39, 18}, /* 38 */
572 {18955, 40, 18}, /* 39 */
573 {19965, 41, 17}, /* 40 */
574 {21013, 42, 17}, /* 41 */
575 {22101, 43, 17}, /* 42 */
576 {23230, 44, 17}, /* 43 */
577 {24402, 45, 16}, /* 44 */
578 {25618, 46, 16}, /* 45 */
579 {26881, 47, 16}, /* 46 */
580 {28193, 48, 16}, /* 47 */
581 {29557, 49, 15}, /* 48 */
582 {30975, 50, 15}, /* 49 */
583 {32450, 51, 15}, /* 50 */
584 {33986, 52, 15}, /* 51 */
585 {35586, 53, 14}, /* 52 */
586 {37253, 54, 14}, /* 53 */
587 {38992, 55, 14}, /* 54 */
588 {40808, 56, 14}, /* 55 */
589 {42707, 57, 13}, /* 56 */
590 {44694, 58, 13}, /* 57 */
591 {46776, 59, 13}, /* 58 */
592 {48961, 60, 13}, /* 59 */
593 {51258, 61, 13}, /* 60 */
594 {53677, 62, 12}, /* 61 */
595 {56230, 63, 12}, /* 62 */
596 {58932, 64, 12}, /* 63 */
597 {61799, 65, 12}, /* 64 */
598 {64851, 66, 11}, /* 65 */
599 {68113, 67, 11}, /* 66 */
600 {71617, 68, 11}, /* 67 */
601 {75401, 69, 10}, /* 68 */
602 {79517, 70, 10}, /* 69 */
603 {84035, 71, 10}, /* 70 */
604 {89053, 72, 10}, /* 71 */
605 {94717, 73, 9} /* 72 */
606};
607
608static void
609sctp_hs_cwnd_increase(struct sctp_tcb *stcb, struct sctp_nets *net)
610{
611 int cur_val, i, indx, incr;
612
613 cur_val = net->cwnd >> 10;
614 indx = SCTP_HS_TABLE_SIZE - 1;
615#ifdef SCTP_DEBUG
616 printf("HS CC CAlled.\n");
617#endif
618 if (cur_val < sctp_cwnd_adjust[0].cwnd) {
619 /* normal mode */
620 if (net->net_ack > net->mtu) {
621 net->cwnd += net->mtu;
622 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
623 sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_FROM_SS);
624 }
625 } else {
626 net->cwnd += net->net_ack;
627 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
628 sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_FROM_SS);
629 }
630 }
631 } else {
632 for (i = net->last_hs_used; i < SCTP_HS_TABLE_SIZE; i++) {
633 if (cur_val < sctp_cwnd_adjust[i].cwnd) {
634 indx = i;
635 break;
636 }
637 }
638 net->last_hs_used = indx;
639 incr = ((sctp_cwnd_adjust[indx].increase) << 10);
640 net->cwnd += incr;
641 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
642 sctp_log_cwnd(stcb, net, incr, SCTP_CWND_LOG_FROM_SS);
643 }
644 }
645}
646
647static void
648sctp_hs_cwnd_decrease(struct sctp_tcb *stcb, struct sctp_nets *net)
649{
650 int cur_val, i, indx;
651 int old_cwnd = net->cwnd;
652
653 cur_val = net->cwnd >> 10;
654 indx = net->last_hs_used;
655 if (cur_val < sctp_cwnd_adjust[0].cwnd) {
656 /* normal mode */
657 net->ssthresh = net->cwnd / 2;
658 if (net->ssthresh < (net->mtu * 2)) {
659 net->ssthresh = 2 * net->mtu;
660 }
661 net->cwnd = net->ssthresh;
662 } else {
663 /* drop by the proper amount */
664 net->ssthresh = net->cwnd - (int)((net->cwnd / 100) *
665 sctp_cwnd_adjust[net->last_hs_used].drop_percent);
666 net->cwnd = net->ssthresh;
667 /* now where are we */
668 indx = net->last_hs_used;
669 cur_val = net->cwnd >> 10;
670 /* reset where we are in the table */
671 if (cur_val < sctp_cwnd_adjust[0].cwnd) {
672 /* feel out of hs */
673 net->last_hs_used = 0;
674 } else {
675 for (i = indx; i >= 1; i--) {
676 if (cur_val > sctp_cwnd_adjust[i - 1].cwnd) {
677 break;
678 }
679 }
680 net->last_hs_used = indx;
681 }
682 }
683 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
684 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_FR);
685 }
686}
687
688void
689sctp_hs_cwnd_update_after_fr(struct sctp_tcb *stcb,
690 struct sctp_association *asoc)
691{
692 struct sctp_nets *net;
693
694 /*
695 * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off == 1) &&
696 * (net->fast_retran_loss_recovery == 0)))
697 */
698 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
699 if ((asoc->fast_retran_loss_recovery == 0) || (sctp_cmt_on_off == 1)) {
700 /* out of a RFC2582 Fast recovery window? */
701 if (net->net_ack > 0) {
702 /*
703 * per section 7.2.3, are there any
704 * destinations that had a fast retransmit
705 * to them. If so what we need to do is
706 * adjust ssthresh and cwnd.
707 */
708 struct sctp_tmit_chunk *lchk;
709
710 sctp_hs_cwnd_decrease(stcb, net);
711
712 lchk = TAILQ_FIRST(&asoc->send_queue);
713
714 net->partial_bytes_acked = 0;
715 /* Turn on fast recovery window */
716 asoc->fast_retran_loss_recovery = 1;
717 if (lchk == NULL) {
718 /* Mark end of the window */
719 asoc->fast_recovery_tsn = asoc->sending_seq - 1;
720 } else {
721 asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
722 }
723
724 /*
725 * CMT fast recovery -- per destination
726 * recovery variable.
727 */
728 net->fast_retran_loss_recovery = 1;
729
730 if (lchk == NULL) {
731 /* Mark end of the window */
732 net->fast_recovery_tsn = asoc->sending_seq - 1;
733 } else {
734 net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
735 }
736
737 /*
738 * Disable Nonce Sum Checking and store the
739 * resync tsn
740 */
741 asoc->nonce_sum_check = 0;
742 asoc->nonce_resync_tsn = asoc->fast_recovery_tsn + 1;
743
744 sctp_timer_stop(SCTP_TIMER_TYPE_SEND,
745 stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32);
746 sctp_timer_start(SCTP_TIMER_TYPE_SEND,
747 stcb->sctp_ep, stcb, net);
748 }
749 } else if (net->net_ack > 0) {
750 /*
751 * Mark a peg that we WOULD have done a cwnd
752 * reduction but RFC2582 prevented this action.
753 */
754 SCTP_STAT_INCR(sctps_fastretransinrtt);
755 }
756 }
757}
758
759void
760sctp_hs_cwnd_update_after_sack(struct sctp_tcb *stcb,
761 struct sctp_association *asoc,
762 int accum_moved, int reneged_all, int will_exit)
763{
764 struct sctp_nets *net;
765
766 /******************************/
767 /* update cwnd and Early FR */
768 /******************************/
769 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
770
771#ifdef JANA_CMT_FAST_RECOVERY
772 /*
773 * CMT fast recovery code. Need to debug.
774 */
775 if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) {
776 if (compare_with_wrap(asoc->last_acked_seq,
777 net->fast_recovery_tsn, MAX_TSN) ||
778 (asoc->last_acked_seq == net->fast_recovery_tsn) ||
779 compare_with_wrap(net->pseudo_cumack, net->fast_recovery_tsn, MAX_TSN) ||
780 (net->pseudo_cumack == net->fast_recovery_tsn)) {
781 net->will_exit_fast_recovery = 1;
782 }
783 }
784#endif
785 if (sctp_early_fr) {
786 /*
787 * So, first of all do we need to have a Early FR
788 * timer running?
789 */
790 if (((TAILQ_FIRST(&asoc->sent_queue)) &&
791 (net->ref_count > 1) &&
792 (net->flight_size < net->cwnd)) ||
793 (reneged_all)) {
794 /*
795 * yes, so in this case stop it if its
796 * running, and then restart it. Reneging
797 * all is a special case where we want to
798 * run the Early FR timer and then force the
799 * last few unacked to be sent, causing us
800 * to illicit a sack with gaps to force out
801 * the others.
802 */
803 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
804 SCTP_STAT_INCR(sctps_earlyfrstpidsck2);
805 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
806 SCTP_FROM_SCTP_INDATA + SCTP_LOC_20);
807 }
808 SCTP_STAT_INCR(sctps_earlyfrstrid);
809 sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net);
810 } else {
811 /* No, stop it if its running */
812 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
813 SCTP_STAT_INCR(sctps_earlyfrstpidsck3);
814 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
815 SCTP_FROM_SCTP_INDATA + SCTP_LOC_21);
816 }
817 }
818 }
819 /* if nothing was acked on this destination skip it */
820 if (net->net_ack == 0) {
821 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
822 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK);
823 }
824 continue;
825 }
826 if (net->net_ack2 > 0) {
827 /*
828 * Karn's rule applies to clearing error count, this
829 * is optional.
830 */
831 net->error_count = 0;
832 if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) ==
833 SCTP_ADDR_NOT_REACHABLE) {
834 /* addr came good */
835 net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE;
836 net->dest_state |= SCTP_ADDR_REACHABLE;
837 sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb,
838 SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED);
839 /* now was it the primary? if so restore */
840 if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) {
841 (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net);
842 }
843 }
844 /*
845 * JRS 5/14/07 - If CMT PF is on and the destination
846 * is in PF state, set the destination to active
847 * state and set the cwnd to one or two MTU's based
848 * on whether PF1 or PF2 is being used.
849 *
850 * Should we stop any running T3 timer here?
851 */
852 if (sctp_cmt_on_off && sctp_cmt_pf && ((net->dest_state & SCTP_ADDR_PF) ==
853 SCTP_ADDR_PF)) {
854 net->dest_state &= ~SCTP_ADDR_PF;
855 net->cwnd = net->mtu * sctp_cmt_pf;
856 SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n",
857 net, net->cwnd);
858 /*
859 * Since the cwnd value is explicitly set,
860 * skip the code that updates the cwnd
861 * value.
862 */
863 goto skip_cwnd_update;
864 }
865 }
866#ifdef JANA_CMT_FAST_RECOVERY
867 /*
868 * CMT fast recovery code
869 */
870 /*
871 * if (sctp_cmt_on_off == 1 &&
872 * net->fast_retran_loss_recovery &&
873 * net->will_exit_fast_recovery == 0) { @@@ Do something }
874 * else if (sctp_cmt_on_off == 0 &&
875 * asoc->fast_retran_loss_recovery && will_exit == 0) {
876 */
877#endif
878
879 if (asoc->fast_retran_loss_recovery && will_exit == 0 && sctp_cmt_on_off == 0) {
880 /*
881 * If we are in loss recovery we skip any cwnd
882 * update
883 */
884 goto skip_cwnd_update;
885 }
886 /*
887 * CMT: CUC algorithm. Update cwnd if pseudo-cumack has
888 * moved.
889 */
890 if (accum_moved || (sctp_cmt_on_off && net->new_pseudo_cumack)) {
891 /* If the cumulative ack moved we can proceed */
892 if (net->cwnd <= net->ssthresh) {
893 /* We are in slow start */
894 if (net->flight_size + net->net_ack >= net->cwnd) {
895
896 sctp_hs_cwnd_increase(stcb, net);
897
898 } else {
899 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
900 sctp_log_cwnd(stcb, net, net->net_ack,
901 SCTP_CWND_LOG_NOADV_SS);
902 }
903 }
904 } else {
905 /* We are in congestion avoidance */
906 net->partial_bytes_acked += net->net_ack;
907 if ((net->flight_size + net->net_ack >= net->cwnd) &&
908 (net->partial_bytes_acked >= net->cwnd)) {
909 net->partial_bytes_acked -= net->cwnd;
910 net->cwnd += net->mtu;
911 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
912 sctp_log_cwnd(stcb, net, net->mtu,
913 SCTP_CWND_LOG_FROM_CA);
914 }
915 } else {
916 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
917 sctp_log_cwnd(stcb, net, net->net_ack,
918 SCTP_CWND_LOG_NOADV_CA);
919 }
920 }
921 }
922 } else {
923 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
924 sctp_log_cwnd(stcb, net, net->mtu,
925 SCTP_CWND_LOG_NO_CUMACK);
926 }
927 }
928skip_cwnd_update:
929 /*
930 * NOW, according to Karn's rule do we need to restore the
931 * RTO timer back? Check our net_ack2. If not set then we
932 * have a ambiguity.. i.e. all data ack'd was sent to more
933 * than one place.
934 */
935 if (net->net_ack2) {
936 /* restore any doubled timers */
937 net->RTO = ((net->lastsa >> 2) + net->lastsv) >> 1;
938 if (net->RTO < stcb->asoc.minrto) {
939 net->RTO = stcb->asoc.minrto;
940 }
941 if (net->RTO > stcb->asoc.maxrto) {
942 net->RTO = stcb->asoc.maxrto;
943 }
944 }
945 }
946}
947
948
949/*
950 * H-TCP congestion control. The algorithm is detailed in:
951 * R.N.Shorten, D.J.Leith:
952 * "H-TCP: TCP for high-speed and long-distance networks"
953 * Proc. PFLDnet, Argonne, 2004.
954 * http://www.hamilton.ie/net/htcp3.pdf
955 */
956
957
958static int use_rtt_scaling = 1;
959static int use_bandwidth_switch = 1;
960
961static inline int
962between(uint32_t seq1, uint32_t seq2, uint32_t seq3)
963{
964 return seq3 - seq2 >= seq1 - seq2;
965}
966
967static inline uint32_t
968htcp_cong_time(struct htcp *ca)
969{
970 return sctp_get_tick_count() - ca->last_cong;
971}
972
973static inline uint32_t
974htcp_ccount(struct htcp *ca)
975{
976 return htcp_cong_time(ca) / ca->minRTT;
977}
978
979static inline void
980htcp_reset(struct htcp *ca)
981{
982 ca->undo_last_cong = ca->last_cong;
983 ca->undo_maxRTT = ca->maxRTT;
984 ca->undo_old_maxB = ca->old_maxB;
985 ca->last_cong = sctp_get_tick_count();
986}
987
988#ifdef SCTP_NOT_USED
989
990static uint32_t
991htcp_cwnd_undo(struct sctp_tcb *stcb, struct sctp_nets *net)
992{
993 net->htcp_ca.last_cong = net->htcp_ca.undo_last_cong;
994 net->htcp_ca.maxRTT = net->htcp_ca.undo_maxRTT;
995 net->htcp_ca.old_maxB = net->htcp_ca.undo_old_maxB;
996 return max(net->cwnd, ((net->ssthresh / net->mtu << 7) / net->htcp_ca.beta) * net->mtu);
997}
998
999#endif
1000
1001static inline void
1002measure_rtt(struct sctp_tcb *stcb, struct sctp_nets *net)
1003{
1004 uint32_t srtt = net->lastsa >> 3;
1005
1006 /* keep track of minimum RTT seen so far, minRTT is zero at first */
1007 if (net->htcp_ca.minRTT > srtt || !net->htcp_ca.minRTT)
1008 net->htcp_ca.minRTT = srtt;
1009
1010 /* max RTT */
1011 if (net->fast_retran_ip == 0 && net->ssthresh < 0xFFFF && htcp_ccount(&net->htcp_ca) > 3) {
1012 if (net->htcp_ca.maxRTT < net->htcp_ca.minRTT)
1013 net->htcp_ca.maxRTT = net->htcp_ca.minRTT;
1014 if (net->htcp_ca.maxRTT < srtt && srtt <= net->htcp_ca.maxRTT + MSEC_TO_TICKS(20))
1015 net->htcp_ca.maxRTT = srtt;
1016 }
1017}
1018
1019static void
1020measure_achieved_throughput(struct sctp_tcb *stcb, struct sctp_nets *net)
1021{
1022 uint32_t now = sctp_get_tick_count();
1023
1024 if (net->fast_retran_ip == 0)
1025 net->htcp_ca.bytes_acked = net->net_ack;
1026
1027 if (!use_bandwidth_switch)
1028 return;
1029
1030 /* achieved throughput calculations */
1031 /* JRS - not 100% sure of this statement */
1032 if (net->fast_retran_ip == 1) {
1033 net->htcp_ca.bytecount = 0;
1034 net->htcp_ca.lasttime = now;
1035 return;
1036 }
1037 net->htcp_ca.bytecount += net->net_ack;
1038
1039 if (net->htcp_ca.bytecount >= net->cwnd - ((net->htcp_ca.alpha >> 7 ? : 1) * net->mtu)
1040 && now - net->htcp_ca.lasttime >= net->htcp_ca.minRTT
1041 && net->htcp_ca.minRTT > 0) {
1042 uint32_t cur_Bi = net->htcp_ca.bytecount / net->mtu * hz / (now - net->htcp_ca.lasttime);
1043
1044 if (htcp_ccount(&net->htcp_ca) <= 3) {
1045 /* just after backoff */
1046 net->htcp_ca.minB = net->htcp_ca.maxB = net->htcp_ca.Bi = cur_Bi;
1047 } else {
1048 net->htcp_ca.Bi = (3 * net->htcp_ca.Bi + cur_Bi) / 4;
1049 if (net->htcp_ca.Bi > net->htcp_ca.maxB)
1050 net->htcp_ca.maxB = net->htcp_ca.Bi;
1051 if (net->htcp_ca.minB > net->htcp_ca.maxB)
1052 net->htcp_ca.minB = net->htcp_ca.maxB;
1053 }
1054 net->htcp_ca.bytecount = 0;
1055 net->htcp_ca.lasttime = now;
1056 }
1057}
1058
1059static inline void
1060htcp_beta_update(struct htcp *ca, uint32_t minRTT, uint32_t maxRTT)
1061{
1062 if (use_bandwidth_switch) {
1063 uint32_t maxB = ca->maxB;
1064 uint32_t old_maxB = ca->old_maxB;
1065
1066 ca->old_maxB = ca->maxB;
1067
1068 if (!between(5 * maxB, 4 * old_maxB, 6 * old_maxB)) {
1069 ca->beta = BETA_MIN;
1070 ca->modeswitch = 0;
1071 return;
1072 }
1073 }
1074 if (ca->modeswitch && minRTT > (uint32_t) MSEC_TO_TICKS(10) && maxRTT) {
1075 ca->beta = (minRTT << 7) / maxRTT;
1076 if (ca->beta < BETA_MIN)
1077 ca->beta = BETA_MIN;
1078 else if (ca->beta > BETA_MAX)
1079 ca->beta = BETA_MAX;
1080 } else {
1081 ca->beta = BETA_MIN;
1082 ca->modeswitch = 1;
1083 }
1084}
1085
1086static inline void
1087htcp_alpha_update(struct htcp *ca)
1088{
1089 uint32_t minRTT = ca->minRTT;
1090 uint32_t factor = 1;
1091 uint32_t diff = htcp_cong_time(ca);
1092
1093 if (diff > (uint32_t) hz) {
1094 diff -= hz;
1095 factor = 1 + (10 * diff + ((diff / 2) * (diff / 2) / hz)) / hz;
1096 }
1097 if (use_rtt_scaling && minRTT) {
1098 uint32_t scale = (hz << 3) / (10 * minRTT);
1099
1100 scale = min(max(scale, 1U << 2), 10U << 3); /* clamping ratio to
1101 * interval [0.5,10]<<3 */
1102 factor = (factor << 3) / scale;
1103 if (!factor)
1104 factor = 1;
1105 }
1106 ca->alpha = 2 * factor * ((1 << 7) - ca->beta);
1107 if (!ca->alpha)
1108 ca->alpha = ALPHA_BASE;
1109}
1110
1111/* After we have the rtt data to calculate beta, we'd still prefer to wait one
1112 * rtt before we adjust our beta to ensure we are working from a consistent
1113 * data.
1114 *
1115 * This function should be called when we hit a congestion event since only at
1116 * that point do we really have a real sense of maxRTT (the queues en route
1117 * were getting just too full now).
1118 */
1119static void
1120htcp_param_update(struct sctp_tcb *stcb, struct sctp_nets *net)
1121{
1122 uint32_t minRTT = net->htcp_ca.minRTT;
1123 uint32_t maxRTT = net->htcp_ca.maxRTT;
1124
1125 htcp_beta_update(&net->htcp_ca, minRTT, maxRTT);
1126 htcp_alpha_update(&net->htcp_ca);
1127
1128 /*
1129 * add slowly fading memory for maxRTT to accommodate routing
1130 * changes etc
1131 */
1132 if (minRTT > 0 && maxRTT > minRTT)
1133 net->htcp_ca.maxRTT = minRTT + ((maxRTT - minRTT) * 95) / 100;
1134}
1135
1136static uint32_t
1137htcp_recalc_ssthresh(struct sctp_tcb *stcb, struct sctp_nets *net)
1138{
1139 htcp_param_update(stcb, net);
1140 return max(((net->cwnd / net->mtu * net->htcp_ca.beta) >> 7) * net->mtu, 2U * net->mtu);
1141}
1142
1143static void
1144htcp_cong_avoid(struct sctp_tcb *stcb, struct sctp_nets *net)
1145{
1146 /*-
1147 * How to handle these functions?
1148 * if (!tcp_is_cwnd_limited(sk, in_flight)) RRS - good question.
1149 * return;
1150 */
1151 if (net->cwnd <= net->ssthresh) {
1152 /* We are in slow start */
1153 if (net->flight_size + net->net_ack >= net->cwnd) {
1154 if (net->net_ack > (net->mtu * sctp_L2_abc_variable)) {
1155 net->cwnd += (net->mtu * sctp_L2_abc_variable);
1156 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1157 sctp_log_cwnd(stcb, net, net->mtu,
1158 SCTP_CWND_LOG_FROM_SS);
1159 }
1160 } else {
1161 net->cwnd += net->net_ack;
1162 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1163 sctp_log_cwnd(stcb, net, net->net_ack,
1164 SCTP_CWND_LOG_FROM_SS);
1165 }
1166 }
1167 } else {
1168 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
1169 sctp_log_cwnd(stcb, net, net->net_ack,
1170 SCTP_CWND_LOG_NOADV_SS);
1171 }
1172 }
1173 } else {
1174 measure_rtt(stcb, net);
1175
1176 /*
1177 * In dangerous area, increase slowly. In theory this is
1178 * net->cwnd += alpha / net->cwnd
1179 */
1180 /* What is snd_cwnd_cnt?? */
1181 if (((net->partial_bytes_acked / net->mtu * net->htcp_ca.alpha) >> 7) * net->mtu >= net->cwnd) {
1182 /*-
1183 * Does SCTP have a cwnd clamp?
1184 * if (net->snd_cwnd < net->snd_cwnd_clamp) - Nope (RRS).
1185 */
1186 net->cwnd += net->mtu;
1187 net->partial_bytes_acked = 0;
1188 htcp_alpha_update(&net->htcp_ca);
1189 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1190 sctp_log_cwnd(stcb, net, net->mtu,
1191 SCTP_CWND_LOG_FROM_CA);
1192 }
1193 } else {
1194 net->partial_bytes_acked += net->net_ack;
1195 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
1196 sctp_log_cwnd(stcb, net, net->net_ack,
1197 SCTP_CWND_LOG_NOADV_CA);
1198 }
1199 }
1200
1201 net->htcp_ca.bytes_acked = net->mtu;
1202 }
1203}
1204
1205#ifdef SCTP_NOT_USED
1206/* Lower bound on congestion window. */
1207static uint32_t
1208htcp_min_cwnd(struct sctp_tcb *stcb, struct sctp_nets *net)
1209{
1210 return net->ssthresh;
1211}
1212
1213#endif
1214
1215static void
1216htcp_init(struct sctp_tcb *stcb, struct sctp_nets *net)
1217{
1218 memset(&net->htcp_ca, 0, sizeof(struct htcp));
1219 net->htcp_ca.alpha = ALPHA_BASE;
1220 net->htcp_ca.beta = BETA_MIN;
1221 net->htcp_ca.bytes_acked = net->mtu;
1222 net->htcp_ca.last_cong = sctp_get_tick_count();
1223}
1224
1225void
1226sctp_htcp_set_initial_cc_param(struct sctp_tcb *stcb, struct sctp_nets *net)
1227{
1228 /*
1229 * We take the max of the burst limit times a MTU or the
1230 * INITIAL_CWND. We then limit this to 4 MTU's of sending.
1231 */
1232 net->cwnd = min((net->mtu * 4), max((2 * net->mtu), SCTP_INITIAL_CWND));
1233 net->ssthresh = stcb->asoc.peers_rwnd;
1234 htcp_init(stcb, net);
1235
1236 if (sctp_logging_level & (SCTP_CWND_MONITOR_ENABLE | SCTP_CWND_LOGGING_ENABLE)) {
1237 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_INITIALIZATION);
1238 }
1239}
1240
1241void
1242sctp_htcp_cwnd_update_after_sack(struct sctp_tcb *stcb,
1243 struct sctp_association *asoc,
1244 int accum_moved, int reneged_all, int will_exit)
1245{
1246 struct sctp_nets *net;
1247
1248 /******************************/
1249 /* update cwnd and Early FR */
1250 /******************************/
1251 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
1252
1253#ifdef JANA_CMT_FAST_RECOVERY
1254 /*
1255 * CMT fast recovery code. Need to debug.
1256 */
1257 if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) {
1258 if (compare_with_wrap(asoc->last_acked_seq,
1259 net->fast_recovery_tsn, MAX_TSN) ||
1260 (asoc->last_acked_seq == net->fast_recovery_tsn) ||
1261 compare_with_wrap(net->pseudo_cumack, net->fast_recovery_tsn, MAX_TSN) ||
1262 (net->pseudo_cumack == net->fast_recovery_tsn)) {
1263 net->will_exit_fast_recovery = 1;
1264 }
1265 }
1266#endif
1267 if (sctp_early_fr) {
1268 /*
1269 * So, first of all do we need to have a Early FR
1270 * timer running?
1271 */
1272 if (((TAILQ_FIRST(&asoc->sent_queue)) &&
1273 (net->ref_count > 1) &&
1274 (net->flight_size < net->cwnd)) ||
1275 (reneged_all)) {
1276 /*
1277 * yes, so in this case stop it if its
1278 * running, and then restart it. Reneging
1279 * all is a special case where we want to
1280 * run the Early FR timer and then force the
1281 * last few unacked to be sent, causing us
1282 * to illicit a sack with gaps to force out
1283 * the others.
1284 */
1285 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
1286 SCTP_STAT_INCR(sctps_earlyfrstpidsck2);
1287 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
1288 SCTP_FROM_SCTP_INDATA + SCTP_LOC_20);
1289 }
1290 SCTP_STAT_INCR(sctps_earlyfrstrid);
1291 sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net);
1292 } else {
1293 /* No, stop it if its running */
1294 if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) {
1295 SCTP_STAT_INCR(sctps_earlyfrstpidsck3);
1296 sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net,
1297 SCTP_FROM_SCTP_INDATA + SCTP_LOC_21);
1298 }
1299 }
1300 }
1301 /* if nothing was acked on this destination skip it */
1302 if (net->net_ack == 0) {
1303 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
1304 sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK);
1305 }
1306 continue;
1307 }
1308 if (net->net_ack2 > 0) {
1309 /*
1310 * Karn's rule applies to clearing error count, this
1311 * is optional.
1312 */
1313 net->error_count = 0;
1314 if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) ==
1315 SCTP_ADDR_NOT_REACHABLE) {
1316 /* addr came good */
1317 net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE;
1318 net->dest_state |= SCTP_ADDR_REACHABLE;
1319 sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb,
1320 SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED);
1321 /* now was it the primary? if so restore */
1322 if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) {
1323 (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net);
1324 }
1325 }
1326 /*
1327 * JRS 5/14/07 - If CMT PF is on and the destination
1328 * is in PF state, set the destination to active
1329 * state and set the cwnd to one or two MTU's based
1330 * on whether PF1 or PF2 is being used.
1331 *
1332 * Should we stop any running T3 timer here?
1333 */
1334 if (sctp_cmt_on_off && sctp_cmt_pf && ((net->dest_state & SCTP_ADDR_PF) ==
1335 SCTP_ADDR_PF)) {
1336 net->dest_state &= ~SCTP_ADDR_PF;
1337 net->cwnd = net->mtu * sctp_cmt_pf;
1338 SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n",
1339 net, net->cwnd);
1340 /*
1341 * Since the cwnd value is explicitly set,
1342 * skip the code that updates the cwnd
1343 * value.
1344 */
1345 goto skip_cwnd_update;
1346 }
1347 }
1348#ifdef JANA_CMT_FAST_RECOVERY
1349 /*
1350 * CMT fast recovery code
1351 */
1352 /*
1353 * if (sctp_cmt_on_off == 1 &&
1354 * net->fast_retran_loss_recovery &&
1355 * net->will_exit_fast_recovery == 0) { @@@ Do something }
1356 * else if (sctp_cmt_on_off == 0 &&
1357 * asoc->fast_retran_loss_recovery && will_exit == 0) {
1358 */
1359#endif
1360
1361 if (asoc->fast_retran_loss_recovery && will_exit == 0 && sctp_cmt_on_off == 0) {
1362 /*
1363 * If we are in loss recovery we skip any cwnd
1364 * update
1365 */
1366 goto skip_cwnd_update;
1367 }
1368 /*
1369 * CMT: CUC algorithm. Update cwnd if pseudo-cumack has
1370 * moved.
1371 */
1372 if (accum_moved || (sctp_cmt_on_off && net->new_pseudo_cumack)) {
1373 htcp_cong_avoid(stcb, net);
1374 measure_achieved_throughput(stcb, net);
1375 } else {
1376 if (sctp_logging_level & SCTP_CWND_LOGGING_ENABLE) {
1377 sctp_log_cwnd(stcb, net, net->mtu,
1378 SCTP_CWND_LOG_NO_CUMACK);
1379 }
1380 }
1381skip_cwnd_update:
1382 /*
1383 * NOW, according to Karn's rule do we need to restore the
1384 * RTO timer back? Check our net_ack2. If not set then we
1385 * have a ambiguity.. i.e. all data ack'd was sent to more
1386 * than one place.
1387 */
1388 if (net->net_ack2) {
1389 /* restore any doubled timers */
1390 net->RTO = ((net->lastsa >> 2) + net->lastsv) >> 1;
1391 if (net->RTO < stcb->asoc.minrto) {
1392 net->RTO = stcb->asoc.minrto;
1393 }
1394 if (net->RTO > stcb->asoc.maxrto) {
1395 net->RTO = stcb->asoc.maxrto;
1396 }
1397 }
1398 }
1399}
1400
1401void
1402sctp_htcp_cwnd_update_after_fr(struct sctp_tcb *stcb,
1403 struct sctp_association *asoc)
1404{
1405 struct sctp_nets *net;
1406
1407 /*
1408 * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off == 1) &&
1409 * (net->fast_retran_loss_recovery == 0)))
1410 */
1411 TAILQ_FOREACH(net, &asoc->nets, sctp_next) {
1412 if ((asoc->fast_retran_loss_recovery == 0) || (sctp_cmt_on_off == 1)) {
1413 /* out of a RFC2582 Fast recovery window? */
1414 if (net->net_ack > 0) {
1415 /*
1416 * per section 7.2.3, are there any
1417 * destinations that had a fast retransmit
1418 * to them. If so what we need to do is
1419 * adjust ssthresh and cwnd.
1420 */
1421 struct sctp_tmit_chunk *lchk;
1422 int old_cwnd = net->cwnd;
1423
1424 /* JRS - reset as if state were changed */
1425 htcp_reset(&net->htcp_ca);
1426 net->ssthresh = htcp_recalc_ssthresh(stcb, net);
1427 net->cwnd = net->ssthresh;
1428 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1429 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd),
1430 SCTP_CWND_LOG_FROM_FR);
1431 }
1432 lchk = TAILQ_FIRST(&asoc->send_queue);
1433
1434 net->partial_bytes_acked = 0;
1435 /* Turn on fast recovery window */
1436 asoc->fast_retran_loss_recovery = 1;
1437 if (lchk == NULL) {
1438 /* Mark end of the window */
1439 asoc->fast_recovery_tsn = asoc->sending_seq - 1;
1440 } else {
1441 asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
1442 }
1443
1444 /*
1445 * CMT fast recovery -- per destination
1446 * recovery variable.
1447 */
1448 net->fast_retran_loss_recovery = 1;
1449
1450 if (lchk == NULL) {
1451 /* Mark end of the window */
1452 net->fast_recovery_tsn = asoc->sending_seq - 1;
1453 } else {
1454 net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1;
1455 }
1456
1457 /*
1458 * Disable Nonce Sum Checking and store the
1459 * resync tsn
1460 */
1461 asoc->nonce_sum_check = 0;
1462 asoc->nonce_resync_tsn = asoc->fast_recovery_tsn + 1;
1463
1464 sctp_timer_stop(SCTP_TIMER_TYPE_SEND,
1465 stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32);
1466 sctp_timer_start(SCTP_TIMER_TYPE_SEND,
1467 stcb->sctp_ep, stcb, net);
1468 }
1469 } else if (net->net_ack > 0) {
1470 /*
1471 * Mark a peg that we WOULD have done a cwnd
1472 * reduction but RFC2582 prevented this action.
1473 */
1474 SCTP_STAT_INCR(sctps_fastretransinrtt);
1475 }
1476 }
1477}
1478
1479void
1480sctp_htcp_cwnd_update_after_timeout(struct sctp_tcb *stcb,
1481 struct sctp_nets *net)
1482{
1483 int old_cwnd = net->cwnd;
1484
1485 /* JRS - reset as if the state were being changed to timeout */
1486 htcp_reset(&net->htcp_ca);
1487 net->ssthresh = htcp_recalc_ssthresh(stcb, net);
1488 net->cwnd = net->mtu;
1489 net->partial_bytes_acked = 0;
1490 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1491 sctp_log_cwnd(stcb, net, net->cwnd - old_cwnd, SCTP_CWND_LOG_FROM_RTX);
1492 }
1493}
1494
1495void
1496sctp_htcp_cwnd_update_after_fr_timer(struct sctp_inpcb *inp,
1497 struct sctp_tcb *stcb, struct sctp_nets *net)
1498{
1499 int old_cwnd;
1500
1501 old_cwnd = net->cwnd;
1502
1503 sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_EARLY_FR_TMR, SCTP_SO_NOT_LOCKED);
1504 net->htcp_ca.last_cong = sctp_get_tick_count();
1505 /*
1506 * make a small adjustment to cwnd and force to CA.
1507 */
1508 if (net->cwnd > net->mtu)
1509 /* drop down one MTU after sending */
1510 net->cwnd -= net->mtu;
1511 if (net->cwnd < net->ssthresh)
1512 /* still in SS move to CA */
1513 net->ssthresh = net->cwnd - 1;
1514 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1515 sctp_log_cwnd(stcb, net, (old_cwnd - net->cwnd), SCTP_CWND_LOG_FROM_FR);
1516 }
1517}
1518
1519void
1520sctp_htcp_cwnd_update_after_ecn_echo(struct sctp_tcb *stcb,
1521 struct sctp_nets *net)
1522{
1523 int old_cwnd;
1524
1525 old_cwnd = net->cwnd;
1526
1527 /* JRS - reset hctp as if state changed */
1528 htcp_reset(&net->htcp_ca);
1529 SCTP_STAT_INCR(sctps_ecnereducedcwnd);
1530 net->ssthresh = htcp_recalc_ssthresh(stcb, net);
1531 if (net->ssthresh < net->mtu) {
1532 net->ssthresh = net->mtu;
1533 /* here back off the timer as well, to slow us down */
1534 net->RTO <<= 1;
1535 }
1536 net->cwnd = net->ssthresh;
1537 if (sctp_logging_level & SCTP_CWND_MONITOR_ENABLE) {
1538 sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT);
1539 }
1540}