summaryrefslogtreecommitdiff
path: root/sys/altq/altq_rio.c
blob: b9a6d5575d05a9c25b6f7399a92db9d30d670622 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
/*	$OpenBSD: altq_rio.c,v 1.2 2001/08/09 14:32:59 deraadt Exp $	*/
/*	$KAME: altq_rio.c,v 1.8 2000/12/14 08:12:46 thorpej Exp $	*/

/*
 * Copyright (C) 1998-2000
 *	Sony Computer Science Laboratories Inc.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY SONY CSL AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL SONY CSL OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */
/*
 * Copyright (c) 1990-1994 Regents of the University of California.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the Computer Systems
 *	Engineering Group at Lawrence Berkeley Laboratory.
 * 4. Neither the name of the University nor of the Laboratory may be used
 *    to endorse or promote products derived from this software without
 *    specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

#include <sys/param.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/sockio.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/errno.h>
#include <sys/kernel.h>

#include <net/if.h>
#include <net/if_types.h>

#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#ifdef INET6
#include <netinet/ip6.h>
#endif

#include <altq/altq.h>
#include <altq/altq_conf.h>
#include <altq/altq_cdnr.h>
#include <altq/altq_red.h>
#include <altq/altq_rio.h>

/*
 * RIO: RED with IN/OUT bit
 *   described in
 *	"Explicit Allocation of Best Effort Packet Delivery Service"
 *	David D. Clark and Wenjia Fang, MIT Lab for Computer Science
 *	http://diffserv.lcs.mit.edu/Papers/exp-alloc-ddc-wf.{ps,pdf}
 *
 * this implementation is extended to support more than 2 drop precedence
 * values as described in RFC2597 (Assured Forwarding PHB Group).
 *
 */
/*
 * AF DS (differentiated service) codepoints.
 * (classes can be mapped to CBQ or H-FSC classes.)
 * 
 *      0   1   2   3   4   5   6   7
 *    +---+---+---+---+---+---+---+---+
 *    |   CLASS   |DropPre| 0 |  CU   |
 *    +---+---+---+---+---+---+---+---+
 *
 *    class 1: 001
 *    class 2: 010
 *    class 3: 011
 *    class 4: 100
 *
 *    low drop prec:    01
 *    medium drop prec: 10
 *    high drop prec:   01
 */

/* normal red parameters */
#define	W_WEIGHT	512	/* inverse of weight of EWMA (511/512) */
				/* q_weight = 0.00195 */

/* red parameters for a slow link */
#define	W_WEIGHT_1	128	/* inverse of weight of EWMA (127/128) */
				/* q_weight = 0.0078125 */

/* red parameters for a very slow link (e.g., dialup) */
#define	W_WEIGHT_2	64	/* inverse of weight of EWMA (63/64) */
				/* q_weight = 0.015625 */

/* fixed-point uses 12-bit decimal places */
#define	FP_SHIFT	12	/* fixed-point shift */

/* red parameters for drop probability */
#define	INV_P_MAX	10	/* inverse of max drop probability */
#define	TH_MIN		 5	/* min threshold */
#define	TH_MAX		15	/* max threshold */

#define	RIO_LIMIT	60	/* default max queue lenght */
#define	RIO_STATS		/* collect statistics */

#define	TV_DELTA(a, b, delta) {					\
	register int	xxs;					\
								\
	delta = (a)->tv_usec - (b)->tv_usec; 			\
	if ((xxs = (a)->tv_sec - (b)->tv_sec) != 0) { 		\
		if (xxs < 0) { 					\
			printf("rm_class: bogus time values");	\
			delta = 60000000;			\
		} else if (xxs > 4)  {				\
			if (xxs > 60)				\
				delta = 60000000;		\
			else					\
				delta += xxs * 1000000;		\
		} else while (xxs > 0) {			\
			delta += 1000000;			\
			xxs--;					\
		}						\
	}							\
}

/* rio_list keeps all rio_queue_t's allocated. */
static rio_queue_t *rio_list = NULL;
/* default rio parameter values */
static struct redparams default_rio_params[RIO_NDROPPREC] = {
  /* th_min,		 th_max,     inv_pmax */
  { TH_MAX * 2 + TH_MIN, TH_MAX * 3, INV_P_MAX }, /* low drop precedence */
  { TH_MAX + TH_MIN,	 TH_MAX * 2, INV_P_MAX }, /* medium drop precedence */
  { TH_MIN,		 TH_MAX,     INV_P_MAX }  /* high drop precedence */
};

/* internal function prototypes */
static int rio_enqueue __P((struct ifaltq *, struct mbuf *,
			    struct altq_pktattr *));
static struct mbuf *rio_dequeue __P((struct ifaltq *, int));
static int rio_request __P((struct ifaltq *, int, void *));
static int rio_detach __P((rio_queue_t *));
static int dscp2index __P((u_int8_t));

/*
 * rio device interface
 */
altqdev_decl(rio);

int
rioopen(dev, flag, fmt, p)
	dev_t dev;
	int flag, fmt;
	struct proc *p;
{
	/* everything will be done when the queueing scheme is attached. */
	return 0;
}

int
rioclose(dev, flag, fmt, p)
	dev_t dev;
	int flag, fmt;
	struct proc *p;
{
	rio_queue_t *rqp;
	int err, error = 0;

	while ((rqp = rio_list) != NULL) {
		/* destroy all */
		err = rio_detach(rqp);
		if (err != 0 && error == 0)
			error = err;
	}

	return error;
}

int
rioioctl(dev, cmd, addr, flag, p)
	dev_t dev;
	ioctlcmd_t cmd;
	caddr_t addr;
	int flag;
	struct proc *p;
{
	rio_queue_t *rqp;
	struct rio_interface *ifacep;
	struct ifnet *ifp;
	int	error = 0;

	/* check super-user privilege */
	switch (cmd) {
	case RIO_GETSTATS:
		break;
	default:
#if (__FreeBSD_version > 400000)
		if ((error = suser(p)) != 0)
			return (error);
#else
		if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
			return (error);
#endif
		break;
	}
    
	switch (cmd) {

	case RIO_ENABLE:
		ifacep = (struct rio_interface *)addr;
		if ((rqp = altq_lookup(ifacep->rio_ifname, ALTQT_RIO)) == NULL) {
			error = EBADF;
			break;
		}
		error = altq_enable(rqp->rq_ifq);
		break;

	case RIO_DISABLE:
		ifacep = (struct rio_interface *)addr;
		if ((rqp = altq_lookup(ifacep->rio_ifname, ALTQT_RIO)) == NULL) {
			error = EBADF;
			break;
		}
		error = altq_disable(rqp->rq_ifq);
		break;

	case RIO_IF_ATTACH:
		ifp = ifunit(((struct rio_interface *)addr)->rio_ifname);
		if (ifp == NULL) {
			error = ENXIO;
			break;
		}

		/* allocate and initialize rio_queue_t */
		MALLOC(rqp, rio_queue_t *, sizeof(rio_queue_t), M_DEVBUF, M_WAITOK);
		if (rqp == NULL) {
			error = ENOMEM;
			break;
		}
		bzero(rqp, sizeof(rio_queue_t));

		MALLOC(rqp->rq_q, class_queue_t *, sizeof(class_queue_t),
		       M_DEVBUF, M_WAITOK);
		if (rqp->rq_q == NULL) {
			FREE(rqp, M_DEVBUF);
			error = ENOMEM;
			break;
		}
		bzero(rqp->rq_q, sizeof(class_queue_t));

		rqp->rq_rio = rio_alloc(0, NULL, 0, 0);
		if (rqp->rq_rio == NULL) {
			FREE(rqp->rq_q, M_DEVBUF);
			FREE(rqp, M_DEVBUF);
			error = ENOMEM;
			break;
		}

		rqp->rq_ifq = &ifp->if_snd;
		qtail(rqp->rq_q) = NULL;
		qlen(rqp->rq_q) = 0;
		qlimit(rqp->rq_q) = RIO_LIMIT;
		qtype(rqp->rq_q) = Q_RIO;

		/*
		 * set RIO to this ifnet structure.
		 */
		error = altq_attach(rqp->rq_ifq, ALTQT_RIO, rqp,
				    rio_enqueue, rio_dequeue, rio_request,
				    NULL, NULL);
		if (error) {
			rio_destroy(rqp->rq_rio);
			FREE(rqp->rq_q, M_DEVBUF);
			FREE(rqp, M_DEVBUF);
			break;
		}

		/* add this state to the rio list */
		rqp->rq_next = rio_list;
		rio_list = rqp;
		break;

	case RIO_IF_DETACH:
		ifacep = (struct rio_interface *)addr;
		if ((rqp = altq_lookup(ifacep->rio_ifname, ALTQT_RIO)) == NULL) {
			error = EBADF;
			break;
		}
		error = rio_detach(rqp);
		break;

	case RIO_GETSTATS:
		do {
			struct rio_stats *q_stats;
			rio_t *rp;
			int i;

			q_stats = (struct rio_stats *)addr;
			if ((rqp = altq_lookup(q_stats->iface.rio_ifname,
					       ALTQT_RIO)) == NULL) {
				error = EBADF;
				break;
			}

			rp = rqp->rq_rio;

			q_stats->q_limit = qlimit(rqp->rq_q);
			q_stats->weight	= rp->rio_weight;
			q_stats->flags = rp->rio_flags;

			for (i = 0; i < RIO_NDROPPREC; i++) {
				q_stats->q_len[i] = rp->rio_precstate[i].qlen;
				bcopy(&rp->q_stats[i], &q_stats->q_stats[i],
				      sizeof(struct redstats));
				q_stats->q_stats[i].q_avg =
				    rp->rio_precstate[i].avg >> rp->rio_wshift;

				q_stats->q_params[i].inv_pmax
					= rp->rio_precstate[i].inv_pmax;
				q_stats->q_params[i].th_min
					= rp->rio_precstate[i].th_min;
				q_stats->q_params[i].th_max
					= rp->rio_precstate[i].th_max;
			}
		} while (0);
		break;

	case RIO_CONFIG:
		do {
			struct rio_conf *fc;
			rio_t	*new;
			int s, limit, i;

			fc = (struct rio_conf *)addr;
			if ((rqp = altq_lookup(fc->iface.rio_ifname,
					       ALTQT_RIO)) == NULL) {
				error = EBADF;
				break;
			}

			new = rio_alloc(fc->rio_weight, &fc->q_params[0],
					fc->rio_flags, fc->rio_pkttime);
			if (new == NULL) {
				error = ENOMEM;
				break;
			}

			s = splimp();
			_flushq(rqp->rq_q);
			limit = fc->rio_limit;
			if (limit < fc->q_params[RIO_NDROPPREC-1].th_max)
				limit = fc->q_params[RIO_NDROPPREC-1].th_max;
			qlimit(rqp->rq_q) = limit;

			rio_destroy(rqp->rq_rio);
			rqp->rq_rio = new;

			splx(s);

			/* write back new values */
			fc->rio_limit = limit;
			for (i = 0; i < RIO_NDROPPREC; i++) {
				fc->q_params[i].inv_pmax =
					rqp->rq_rio->rio_precstate[i].inv_pmax;
				fc->q_params[i].th_min =
					rqp->rq_rio->rio_precstate[i].th_min;
				fc->q_params[i].th_max =
					rqp->rq_rio->rio_precstate[i].th_max;
			}
		} while (0);
		break;

	case RIO_SETDEFAULTS:
		do {
			struct redparams *rp;
			int i;

			rp = (struct redparams *)addr;
			for (i = 0; i < RIO_NDROPPREC; i++)
				default_rio_params[i] = rp[i];
		} while (0);
		break;

	default:
		error = EINVAL;
		break;
	}

	return error;
}

static int
rio_detach(rqp)
	rio_queue_t *rqp;
{
	rio_queue_t *tmp;
	int error = 0;

	if (ALTQ_IS_ENABLED(rqp->rq_ifq))
		altq_disable(rqp->rq_ifq);

	if ((error = altq_detach(rqp->rq_ifq)))
		return (error);

	if (rio_list == rqp)
		rio_list = rqp->rq_next;
	else {
		for (tmp = rio_list; tmp != NULL; tmp = tmp->rq_next)
			if (tmp->rq_next == rqp) {
				tmp->rq_next = rqp->rq_next;
				break;
			}
		if (tmp == NULL)
			printf("rio_detach: no state found in rio_list!\n");
	}

	rio_destroy(rqp->rq_rio);
	FREE(rqp->rq_q, M_DEVBUF);
	FREE(rqp, M_DEVBUF);
	return (error);
}

/*
 * rio support routines
 */
static int
rio_request(ifq, req, arg)
	struct ifaltq *ifq;
	int req;
	void *arg;
{
	rio_queue_t *rqp = (rio_queue_t *)ifq->altq_disc;

	switch (req) {
	case ALTRQ_PURGE:
		_flushq(rqp->rq_q);
		if (ALTQ_IS_ENABLED(ifq))
			ifq->ifq_len = 0;
		break;
	}
	return (0);
}


rio_t *
rio_alloc(weight, params, flags, pkttime)
	int	weight;
	struct redparams *params;
	int	flags, pkttime;
{
	rio_t 	*rp;
	int	w, i;
	int	npkts_per_sec;
	
	MALLOC(rp, rio_t *, sizeof(rio_t), M_DEVBUF, M_WAITOK);
	if (rp == NULL)
		return (NULL);
	bzero(rp, sizeof(rio_t));

	rp->rio_flags = flags;
	if (pkttime == 0)
		/* default packet time: 1000 bytes / 10Mbps * 8 * 1000000 */
		rp->rio_pkttime = 800;
	else 
		rp->rio_pkttime = pkttime;

	if (weight != 0)
		rp->rio_weight = weight;
	else {
		/* use derfault */
		rp->rio_weight = W_WEIGHT;

		/* when the link is very slow, adjust red parameters */
		npkts_per_sec = 1000000 / rp->rio_pkttime;
		if (npkts_per_sec < 50) {
			/* up to about 400Kbps */
			rp->rio_weight = W_WEIGHT_2;
		} else if (npkts_per_sec < 300) {
			/* up to about 2.4Mbps */
			rp->rio_weight = W_WEIGHT_1;
		}
	}

	/* calculate wshift.  weight must be power of 2 */
	w = rp->rio_weight;
	for (i = 0; w > 1; i++)
		w = w >> 1;
	rp->rio_wshift = i;
	w = 1 << rp->rio_wshift;
	if (w != rp->rio_weight) {
		printf("invalid weight value %d for red! use %d\n",
		       rp->rio_weight, w);
		rp->rio_weight = w;
	}

	/* allocate weight table */
	rp->rio_wtab = wtab_alloc(rp->rio_weight);

	for (i = 0; i < RIO_NDROPPREC; i++) {
		struct dropprec_state *prec = &rp->rio_precstate[i];

		prec->avg = 0;
		prec->idle = 1;

		if (params == NULL || params[i].inv_pmax == 0)
			prec->inv_pmax = default_rio_params[i].inv_pmax;
		else
			prec->inv_pmax = params[i].inv_pmax;
		if (params == NULL || params[i].th_min == 0)
			prec->th_min = default_rio_params[i].th_min;
		else
			prec->th_min = params[i].th_min;
		if (params == NULL || params[i].th_max == 0)
			prec->th_max = default_rio_params[i].th_max;
		else
			prec->th_max = params[i].th_max;

		/*
		 * th_min_s and th_max_s are scaled versions of th_min
		 * and th_max to be compared with avg.
		 */
		prec->th_min_s = prec->th_min << (rp->rio_wshift + FP_SHIFT);
		prec->th_max_s = prec->th_max << (rp->rio_wshift + FP_SHIFT);

		/*
		 * precompute probability denominator
		 *  probd = (2 * (TH_MAX-TH_MIN) / pmax) in fixed-point
		 */
		prec->probd = (2 * (prec->th_max - prec->th_min)
			       * prec->inv_pmax) << FP_SHIFT;

		microtime(&prec->last);
	}

	return (rp);
}

void
rio_destroy(rp)
	rio_t *rp;
{
	wtab_destroy(rp->rio_wtab);
	FREE(rp, M_DEVBUF);
}

void 
rio_getstats(rp, sp)
	rio_t *rp;
	struct redstats *sp;
{
	int i;
	
	for (i = 0; i < RIO_NDROPPREC; i++) {
		bcopy(&rp->q_stats[i], sp, sizeof(struct redstats));
		sp->q_avg = rp->rio_precstate[i].avg >> rp->rio_wshift;
		sp++;
	}
}

/*
 * enqueue routine:
 *
 *	returns: 0 when successfully queued.
 *		 ENOBUFS when drop occurs.
 */
static int
rio_enqueue(ifq, m, pktattr)
	struct ifaltq *ifq;
	struct mbuf *m;
	struct altq_pktattr *pktattr;
{
	rio_queue_t *rqp = (rio_queue_t *)ifq->altq_disc;
	int error = 0;

	if (rio_addq(rqp->rq_rio, rqp->rq_q, m, pktattr) == 0)
		ifq->ifq_len++;
	else
		error = ENOBUFS;
	return error;
}

#if (RIO_NDROPPREC == 3)
/*
 * internally, a drop precedence value is converted to an index
 * starting from 0.
 */
static int
dscp2index(u_int8_t dscp)
{
	int dpindex = dscp & AF_DROPPRECMASK;

	if (dpindex == 0)
		return (0);
	return ((dpindex >> 3) - 1);
}
#endif

#if 1
/*
 * kludge: when a packet is dequeued, we need to know its drop precedence
 * in order to keep the queue length of each drop precedence.
 * use m_pkthdr.rcvif to pass this info.
 */
#define	RIOM_SET_PRECINDEX(m, idx)	\
	do { (m)->m_pkthdr.rcvif = (struct ifnet *)((long)(idx)); } while (0)
#define	RIOM_GET_PRECINDEX(m)	\
	({ long idx; idx = (long)((m)->m_pkthdr.rcvif); \
	(m)->m_pkthdr.rcvif = NULL; idx; })
#endif

int
rio_addq(rp, q, m, pktattr)
	rio_t *rp;
	class_queue_t *q;
	struct mbuf *m;
	struct altq_pktattr *pktattr;
{
	int avg, droptype;
	u_int8_t dsfield, odsfield;
	int dpindex, i, n, t;
	struct timeval now;
	struct dropprec_state *prec;

	dsfield = odsfield = read_dsfield(m, pktattr);
	dpindex = dscp2index(dsfield);

	/*
	 * update avg of the precedence states whose drop precedence
	 * is larger than or equal to the drop precedence of the packet
	 */
	now.tv_sec = 0;
	for (i = dpindex; i < RIO_NDROPPREC; i++) {
		prec = &rp->rio_precstate[i];
		avg = prec->avg;
		if (prec->idle) {
			prec->idle = 0;
			if (now.tv_sec == 0)
				microtime(&now);
			t = (now.tv_sec - prec->last.tv_sec);
			if (t > 60)
				avg = 0;
			else {
				t = t * 1000000 +
					(now.tv_usec - prec->last.tv_usec);
				n = t / rp->rio_pkttime;
				/* calculate (avg = (1 - Wq)^n * avg) */
				if (n > 0)
					avg = (avg >> FP_SHIFT) *
						pow_w(rp->rio_wtab, n);
			}
		}

		/* run estimator. (avg is scaled by WEIGHT in fixed-point) */
		avg += (prec->qlen << FP_SHIFT) - (avg >> rp->rio_wshift);
		prec->avg = avg;		/* save the new value */
		/*
		 * count keeps a tally of arriving traffic that has not
		 * been dropped.
		 */
		prec->count++;
	}

	prec = &rp->rio_precstate[dpindex];
	avg = prec->avg;
    
	/* see if we drop early */
	droptype = DTYPE_NODROP;
	if (avg >= prec->th_min_s && prec->qlen > 1) {
		if (avg >= prec->th_max_s) {
			/* avg >= th_max: forced drop */
			droptype = DTYPE_FORCED;
		} else if (prec->old == 0) {
			/* first exceeds th_min */
			prec->count = 1;
			prec->old = 1;
		} else if (drop_early((avg - prec->th_min_s) >> rp->rio_wshift,
				      prec->probd, prec->count)) {
			/* unforced drop by red */
			droptype = DTYPE_EARLY;
		}
	} else {
		/* avg < th_min */
		prec->old = 0;
	}

	/*
	 * if the queue length hits the hard limit, it's a forced drop.
	 */
	if (droptype == DTYPE_NODROP && qlen(q) >= qlimit(q))
		droptype = DTYPE_FORCED;

	if (droptype != DTYPE_NODROP) {
		/* always drop incoming packet (as opposed to randomdrop) */
		for (i = dpindex; i < RIO_NDROPPREC; i++)
			rp->rio_precstate[i].count = 0;
#ifdef RIO_STATS
		if (droptype == DTYPE_EARLY)
			rp->q_stats[dpindex].drop_unforced++;
		else
			rp->q_stats[dpindex].drop_forced++;
		PKTCNTR_ADD(&rp->q_stats[dpindex].drop_cnt, m_pktlen(m));
#endif
		m_freem(m);
		return (-1);
	}

	for (i = dpindex; i < RIO_NDROPPREC; i++)
		rp->rio_precstate[i].qlen++;

	/* save drop precedence index in mbuf hdr */
	RIOM_SET_PRECINDEX(m, dpindex);

	if (rp->rio_flags & RIOF_CLEARDSCP)
		dsfield &= ~DSCP_MASK;

	if (dsfield != odsfield)
		write_dsfield(m, pktattr, dsfield);

	_addq(q, m);

#ifdef RIO_STATS
	PKTCNTR_ADD(&rp->q_stats[dpindex].xmit_cnt, m_pktlen(m));
#endif
	return (0);
}

/*
 * dequeue routine:
 *	must be called in splimp.
 *
 *	returns: mbuf dequeued.
 *		 NULL when no packet is available in the queue.
 */

static struct mbuf *
rio_dequeue(ifq, op)
	struct ifaltq *ifq;
	int op;
{
	rio_queue_t *rqp = (rio_queue_t *)ifq->altq_disc;
	struct mbuf *m = NULL;

	if (op == ALTDQ_POLL)
		return qhead(rqp->rq_q);

	m = rio_getq(rqp->rq_rio, rqp->rq_q);
	if (m != NULL)
		ifq->ifq_len--;
	return m;
}

struct mbuf *
rio_getq(rp, q)
	rio_t *rp;
	class_queue_t *q;
{
	struct mbuf *m;
	int dpindex, i;

	if ((m = _getq(q)) == NULL)
		return NULL;

	dpindex = RIOM_GET_PRECINDEX(m);
	for (i = dpindex; i < RIO_NDROPPREC; i++) {
		if (--rp->rio_precstate[i].qlen == 0) {
			if (rp->rio_precstate[i].idle == 0) {
				rp->rio_precstate[i].idle = 1;
				microtime(&rp->rio_precstate[i].last);
			}
		}
	}
	return (m);
}

#ifdef KLD_MODULE

static struct altqsw rio_sw =
	{"rio", rioopen, rioclose, rioioctl};

ALTQ_MODULE(altq_rio, ALTQT_RIO, &rio_sw);

#endif /* KLD_MODULE */