xref: /openbmc/linux/net/sched/sch_gred.c (revision c21b37f6) 
1 /*
2  * net/sched/sch_gred.c	Generic Random Early Detection queue.
3  *
4  *
5  *              This program is free software; you can redistribute it and/or
6  *              modify it under the terms of the GNU General Public License
7  *              as published by the Free Software Foundation; either version
8  *              2 of the License, or (at your option) any later version.
9  *
10  * Authors:    J Hadi Salim (hadi@cyberus.ca) 1998-2002
11  *
12  *             991129: -  Bug fix with grio mode
13  *		       - a better sing. AvgQ mode with Grio(WRED)
14  *		       - A finer grained VQ dequeue based on sugestion
15  *		         from Ren Liu
16  *		       - More error checks
17  *
18  *  For all the glorious comments look at include/net/red.h
19  */
20 
21 #include <linux/module.h>
22 #include <linux/types.h>
23 #include <linux/kernel.h>
24 #include <linux/skbuff.h>
25 #include <net/pkt_sched.h>
26 #include <net/red.h>
27 
28 #define GRED_DEF_PRIO (MAX_DPs / 2)
29 #define GRED_VQ_MASK (MAX_DPs - 1)
30 
31 struct gred_sched_data;
32 struct gred_sched;
33 
34 struct gred_sched_data
35 {
36 	u32		limit;		/* HARD maximal queue length	*/
37 	u32      	DP;		/* the drop pramaters */
38 	u32		bytesin;	/* bytes seen on virtualQ so far*/
39 	u32		packetsin;	/* packets seen on virtualQ so far*/
40 	u32		backlog;	/* bytes on the virtualQ */
41 	u8		prio;		/* the prio of this vq */
42 
43 	struct red_parms parms;
44 	struct red_stats stats;
45 };
46 
47 enum {
48 	GRED_WRED_MODE = 1,
49 	GRED_RIO_MODE,
50 };
51 
52 struct gred_sched
53 {
54 	struct gred_sched_data *tab[MAX_DPs];
55 	unsigned long	flags;
56 	u32		red_flags;
57 	u32 		DPs;
58 	u32 		def;
59 	struct red_parms wred_set;
60 };
61 
62 static inline int gred_wred_mode(struct gred_sched *table)
63 {
64 	return test_bit(GRED_WRED_MODE, &table->flags);
65 }
66 
67 static inline void gred_enable_wred_mode(struct gred_sched *table)
68 {
69 	__set_bit(GRED_WRED_MODE, &table->flags);
70 }
71 
72 static inline void gred_disable_wred_mode(struct gred_sched *table)
73 {
74 	__clear_bit(GRED_WRED_MODE, &table->flags);
75 }
76 
77 static inline int gred_rio_mode(struct gred_sched *table)
78 {
79 	return test_bit(GRED_RIO_MODE, &table->flags);
80 }
81 
82 static inline void gred_enable_rio_mode(struct gred_sched *table)
83 {
84 	__set_bit(GRED_RIO_MODE, &table->flags);
85 }
86 
87 static inline void gred_disable_rio_mode(struct gred_sched *table)
88 {
89 	__clear_bit(GRED_RIO_MODE, &table->flags);
90 }
91 
92 static inline int gred_wred_mode_check(struct Qdisc *sch)
93 {
94 	struct gred_sched *table = qdisc_priv(sch);
95 	int i;
96 
97 	/* Really ugly O(n^2) but shouldn't be necessary too frequent. */
98 	for (i = 0; i < table->DPs; i++) {
99 		struct gred_sched_data *q = table->tab[i];
100 		int n;
101 
102 		if (q == NULL)
103 			continue;
104 
105 		for (n = 0; n < table->DPs; n++)
106 			if (table->tab[n] && table->tab[n] != q &&
107 			    table->tab[n]->prio == q->prio)
108 				return 1;
109 	}
110 
111 	return 0;
112 }
113 
114 static inline unsigned int gred_backlog(struct gred_sched *table,
115 					struct gred_sched_data *q,
116 					struct Qdisc *sch)
117 {
118 	if (gred_wred_mode(table))
119 		return sch->qstats.backlog;
120 	else
121 		return q->backlog;
122 }
123 
124 static inline u16 tc_index_to_dp(struct sk_buff *skb)
125 {
126 	return skb->tc_index & GRED_VQ_MASK;
127 }
128 
129 static inline void gred_load_wred_set(struct gred_sched *table,
130 				      struct gred_sched_data *q)
131 {
132 	q->parms.qavg = table->wred_set.qavg;
133 	q->parms.qidlestart = table->wred_set.qidlestart;
134 }
135 
136 static inline void gred_store_wred_set(struct gred_sched *table,
137 				       struct gred_sched_data *q)
138 {
139 	table->wred_set.qavg = q->parms.qavg;
140 }
141 
142 static inline int gred_use_ecn(struct gred_sched *t)
143 {
144 	return t->red_flags & TC_RED_ECN;
145 }
146 
147 static inline int gred_use_harddrop(struct gred_sched *t)
148 {
149 	return t->red_flags & TC_RED_HARDDROP;
150 }
151 
152 static int gred_enqueue(struct sk_buff *skb, struct Qdisc* sch)
153 {
154 	struct gred_sched_data *q=NULL;
155 	struct gred_sched *t= qdisc_priv(sch);
156 	unsigned long qavg = 0;
157 	u16 dp = tc_index_to_dp(skb);
158 
159 	if (dp >= t->DPs  || (q = t->tab[dp]) == NULL) {
160 		dp = t->def;
161 
162 		if ((q = t->tab[dp]) == NULL) {
163 			/* Pass through packets not assigned to a DP
164 			 * if no default DP has been configured. This
165 			 * allows for DP flows to be left untouched.
166 			 */
167 			if (skb_queue_len(&sch->q) < sch->dev->tx_queue_len)
168 				return qdisc_enqueue_tail(skb, sch);
169 			else
170 				goto drop;
171 		}
172 
173 		/* fix tc_index? --could be controvesial but needed for
174 		   requeueing */
175 		skb->tc_index = (skb->tc_index & ~GRED_VQ_MASK) | dp;
176 	}
177 
178 	/* sum up all the qaves of prios <= to ours to get the new qave */
179 	if (!gred_wred_mode(t) && gred_rio_mode(t)) {
180 		int i;
181 
182 		for (i = 0; i < t->DPs; i++) {
183 			if (t->tab[i] && t->tab[i]->prio < q->prio &&
184 			    !red_is_idling(&t->tab[i]->parms))
185 				qavg +=t->tab[i]->parms.qavg;
186 		}
187 
188 	}
189 
190 	q->packetsin++;
191 	q->bytesin += skb->len;
192 
193 	if (gred_wred_mode(t))
194 		gred_load_wred_set(t, q);
195 
196 	q->parms.qavg = red_calc_qavg(&q->parms, gred_backlog(t, q, sch));
197 
198 	if (red_is_idling(&q->parms))
199 		red_end_of_idle_period(&q->parms);
200 
201 	if (gred_wred_mode(t))
202 		gred_store_wred_set(t, q);
203 
204 	switch (red_action(&q->parms, q->parms.qavg + qavg)) {
205 		case RED_DONT_MARK:
206 			break;
207 
208 		case RED_PROB_MARK:
209 			sch->qstats.overlimits++;
210 			if (!gred_use_ecn(t) || !INET_ECN_set_ce(skb)) {
211 				q->stats.prob_drop++;
212 				goto congestion_drop;
213 			}
214 
215 			q->stats.prob_mark++;
216 			break;
217 
218 		case RED_HARD_MARK:
219 			sch->qstats.overlimits++;
220 			if (gred_use_harddrop(t) || !gred_use_ecn(t) ||
221 			    !INET_ECN_set_ce(skb)) {
222 				q->stats.forced_drop++;
223 				goto congestion_drop;
224 			}
225 			q->stats.forced_mark++;
226 			break;
227 	}
228 
229 	if (q->backlog + skb->len <= q->limit) {
230 		q->backlog += skb->len;
231 		return qdisc_enqueue_tail(skb, sch);
232 	}
233 
234 	q->stats.pdrop++;
235 drop:
236 	return qdisc_drop(skb, sch);
237 
238 congestion_drop:
239 	qdisc_drop(skb, sch);
240 	return NET_XMIT_CN;
241 }
242 
243 static int gred_requeue(struct sk_buff *skb, struct Qdisc* sch)
244 {
245 	struct gred_sched *t = qdisc_priv(sch);
246 	struct gred_sched_data *q;
247 	u16 dp = tc_index_to_dp(skb);
248 
249 	if (dp >= t->DPs || (q = t->tab[dp]) == NULL) {
250 		if (net_ratelimit())
251 			printk(KERN_WARNING "GRED: Unable to relocate VQ 0x%x "
252 			       "for requeue, screwing up backlog.\n",
253 			       tc_index_to_dp(skb));
254 	} else {
255 		if (red_is_idling(&q->parms))
256 			red_end_of_idle_period(&q->parms);
257 		q->backlog += skb->len;
258 	}
259 
260 	return qdisc_requeue(skb, sch);
261 }
262 
263 static struct sk_buff *gred_dequeue(struct Qdisc* sch)
264 {
265 	struct sk_buff *skb;
266 	struct gred_sched *t = qdisc_priv(sch);
267 
268 	skb = qdisc_dequeue_head(sch);
269 
270 	if (skb) {
271 		struct gred_sched_data *q;
272 		u16 dp = tc_index_to_dp(skb);
273 
274 		if (dp >= t->DPs || (q = t->tab[dp]) == NULL) {
275 			if (net_ratelimit())
276 				printk(KERN_WARNING "GRED: Unable to relocate "
277 				       "VQ 0x%x after dequeue, screwing up "
278 				       "backlog.\n", tc_index_to_dp(skb));
279 		} else {
280 			q->backlog -= skb->len;
281 
282 			if (!q->backlog && !gred_wred_mode(t))
283 				red_start_of_idle_period(&q->parms);
284 		}
285 
286 		return skb;
287 	}
288 
289 	if (gred_wred_mode(t) && !red_is_idling(&t->wred_set))
290 		red_start_of_idle_period(&t->wred_set);
291 
292 	return NULL;
293 }
294 
295 static unsigned int gred_drop(struct Qdisc* sch)
296 {
297 	struct sk_buff *skb;
298 	struct gred_sched *t = qdisc_priv(sch);
299 
300 	skb = qdisc_dequeue_tail(sch);
301 	if (skb) {
302 		unsigned int len = skb->len;
303 		struct gred_sched_data *q;
304 		u16 dp = tc_index_to_dp(skb);
305 
306 		if (dp >= t->DPs || (q = t->tab[dp]) == NULL) {
307 			if (net_ratelimit())
308 				printk(KERN_WARNING "GRED: Unable to relocate "
309 				       "VQ 0x%x while dropping, screwing up "
310 				       "backlog.\n", tc_index_to_dp(skb));
311 		} else {
312 			q->backlog -= len;
313 			q->stats.other++;
314 
315 			if (!q->backlog && !gred_wred_mode(t))
316 				red_start_of_idle_period(&q->parms);
317 		}
318 
319 		qdisc_drop(skb, sch);
320 		return len;
321 	}
322 
323 	if (gred_wred_mode(t) && !red_is_idling(&t->wred_set))
324 		red_start_of_idle_period(&t->wred_set);
325 
326 	return 0;
327 
328 }
329 
330 static void gred_reset(struct Qdisc* sch)
331 {
332 	int i;
333 	struct gred_sched *t = qdisc_priv(sch);
334 
335 	qdisc_reset_queue(sch);
336 
337 	for (i = 0; i < t->DPs; i++) {
338 		struct gred_sched_data *q = t->tab[i];
339 
340 		if (!q)
341 			continue;
342 
343 		red_restart(&q->parms);
344 		q->backlog = 0;
345 	}
346 }
347 
348 static inline void gred_destroy_vq(struct gred_sched_data *q)
349 {
350 	kfree(q);
351 }
352 
353 static inline int gred_change_table_def(struct Qdisc *sch, struct rtattr *dps)
354 {
355 	struct gred_sched *table = qdisc_priv(sch);
356 	struct tc_gred_sopt *sopt;
357 	int i;
358 
359 	if (dps == NULL || RTA_PAYLOAD(dps) < sizeof(*sopt))
360 		return -EINVAL;
361 
362 	sopt = RTA_DATA(dps);
363 
364 	if (sopt->DPs > MAX_DPs || sopt->DPs == 0 || sopt->def_DP >= sopt->DPs)
365 		return -EINVAL;
366 
367 	sch_tree_lock(sch);
368 	table->DPs = sopt->DPs;
369 	table->def = sopt->def_DP;
370 	table->red_flags = sopt->flags;
371 
372 	/*
373 	 * Every entry point to GRED is synchronized with the above code
374 	 * and the DP is checked against DPs, i.e. shadowed VQs can no
375 	 * longer be found so we can unlock right here.
376 	 */
377 	sch_tree_unlock(sch);
378 
379 	if (sopt->grio) {
380 		gred_enable_rio_mode(table);
381 		gred_disable_wred_mode(table);
382 		if (gred_wred_mode_check(sch))
383 			gred_enable_wred_mode(table);
384 	} else {
385 		gred_disable_rio_mode(table);
386 		gred_disable_wred_mode(table);
387 	}
388 
389 	for (i = table->DPs; i < MAX_DPs; i++) {
390 		if (table->tab[i]) {
391 			printk(KERN_WARNING "GRED: Warning: Destroying "
392 			       "shadowed VQ 0x%x\n", i);
393 			gred_destroy_vq(table->tab[i]);
394 			table->tab[i] = NULL;
395 		}
396 	}
397 
398 	return 0;
399 }
400 
401 static inline int gred_change_vq(struct Qdisc *sch, int dp,
402 				 struct tc_gred_qopt *ctl, int prio, u8 *stab)
403 {
404 	struct gred_sched *table = qdisc_priv(sch);
405 	struct gred_sched_data *q;
406 
407 	if (table->tab[dp] == NULL) {
408 		table->tab[dp] = kzalloc(sizeof(*q), GFP_KERNEL);
409 		if (table->tab[dp] == NULL)
410 			return -ENOMEM;
411 	}
412 
413 	q = table->tab[dp];
414 	q->DP = dp;
415 	q->prio = prio;
416 	q->limit = ctl->limit;
417 
418 	if (q->backlog == 0)
419 		red_end_of_idle_period(&q->parms);
420 
421 	red_set_parms(&q->parms,
422 		      ctl->qth_min, ctl->qth_max, ctl->Wlog, ctl->Plog,
423 		      ctl->Scell_log, stab);
424 
425 	return 0;
426 }
427 
428 static int gred_change(struct Qdisc *sch, struct rtattr *opt)
429 {
430 	struct gred_sched *table = qdisc_priv(sch);
431 	struct tc_gred_qopt *ctl;
432 	struct rtattr *tb[TCA_GRED_MAX];
433 	int err = -EINVAL, prio = GRED_DEF_PRIO;
434 	u8 *stab;
435 
436 	if (opt == NULL || rtattr_parse_nested(tb, TCA_GRED_MAX, opt))
437 		return -EINVAL;
438 
439 	if (tb[TCA_GRED_PARMS-1] == NULL && tb[TCA_GRED_STAB-1] == NULL)
440 		return gred_change_table_def(sch, opt);
441 
442 	if (tb[TCA_GRED_PARMS-1] == NULL ||
443 	    RTA_PAYLOAD(tb[TCA_GRED_PARMS-1]) < sizeof(*ctl) ||
444 	    tb[TCA_GRED_STAB-1] == NULL ||
445 	    RTA_PAYLOAD(tb[TCA_GRED_STAB-1]) < 256)
446 		return -EINVAL;
447 
448 	ctl = RTA_DATA(tb[TCA_GRED_PARMS-1]);
449 	stab = RTA_DATA(tb[TCA_GRED_STAB-1]);
450 
451 	if (ctl->DP >= table->DPs)
452 		goto errout;
453 
454 	if (gred_rio_mode(table)) {
455 		if (ctl->prio == 0) {
456 			int def_prio = GRED_DEF_PRIO;
457 
458 			if (table->tab[table->def])
459 				def_prio = table->tab[table->def]->prio;
460 
461 			printk(KERN_DEBUG "GRED: DP %u does not have a prio "
462 			       "setting default to %d\n", ctl->DP, def_prio);
463 
464 			prio = def_prio;
465 		} else
466 			prio = ctl->prio;
467 	}
468 
469 	sch_tree_lock(sch);
470 
471 	err = gred_change_vq(sch, ctl->DP, ctl, prio, stab);
472 	if (err < 0)
473 		goto errout_locked;
474 
475 	if (gred_rio_mode(table)) {
476 		gred_disable_wred_mode(table);
477 		if (gred_wred_mode_check(sch))
478 			gred_enable_wred_mode(table);
479 	}
480 
481 	err = 0;
482 
483 errout_locked:
484 	sch_tree_unlock(sch);
485 errout:
486 	return err;
487 }
488 
489 static int gred_init(struct Qdisc *sch, struct rtattr *opt)
490 {
491 	struct rtattr *tb[TCA_GRED_MAX];
492 
493 	if (opt == NULL || rtattr_parse_nested(tb, TCA_GRED_MAX, opt))
494 		return -EINVAL;
495 
496 	if (tb[TCA_GRED_PARMS-1] || tb[TCA_GRED_STAB-1])
497 		return -EINVAL;
498 
499 	return gred_change_table_def(sch, tb[TCA_GRED_DPS-1]);
500 }
501 
502 static int gred_dump(struct Qdisc *sch, struct sk_buff *skb)
503 {
504 	struct gred_sched *table = qdisc_priv(sch);
505 	struct rtattr *parms, *opts = NULL;
506 	int i;
507 	struct tc_gred_sopt sopt = {
508 		.DPs	= table->DPs,
509 		.def_DP	= table->def,
510 		.grio	= gred_rio_mode(table),
511 		.flags	= table->red_flags,
512 	};
513 
514 	opts = RTA_NEST(skb, TCA_OPTIONS);
515 	RTA_PUT(skb, TCA_GRED_DPS, sizeof(sopt), &sopt);
516 	parms = RTA_NEST(skb, TCA_GRED_PARMS);
517 
518 	for (i = 0; i < MAX_DPs; i++) {
519 		struct gred_sched_data *q = table->tab[i];
520 		struct tc_gred_qopt opt;
521 
522 		memset(&opt, 0, sizeof(opt));
523 
524 		if (!q) {
525 			/* hack -- fix at some point with proper message
526 			   This is how we indicate to tc that there is no VQ
527 			   at this DP */
528 
529 			opt.DP = MAX_DPs + i;
530 			goto append_opt;
531 		}
532 
533 		opt.limit	= q->limit;
534 		opt.DP		= q->DP;
535 		opt.backlog	= q->backlog;
536 		opt.prio	= q->prio;
537 		opt.qth_min	= q->parms.qth_min >> q->parms.Wlog;
538 		opt.qth_max	= q->parms.qth_max >> q->parms.Wlog;
539 		opt.Wlog	= q->parms.Wlog;
540 		opt.Plog	= q->parms.Plog;
541 		opt.Scell_log	= q->parms.Scell_log;
542 		opt.other	= q->stats.other;
543 		opt.early	= q->stats.prob_drop;
544 		opt.forced	= q->stats.forced_drop;
545 		opt.pdrop	= q->stats.pdrop;
546 		opt.packets	= q->packetsin;
547 		opt.bytesin	= q->bytesin;
548 
549 		if (gred_wred_mode(table)) {
550 			q->parms.qidlestart =
551 				table->tab[table->def]->parms.qidlestart;
552 			q->parms.qavg = table->tab[table->def]->parms.qavg;
553 		}
554 
555 		opt.qave = red_calc_qavg(&q->parms, q->parms.qavg);
556 
557 append_opt:
558 		RTA_APPEND(skb, sizeof(opt), &opt);
559 	}
560 
561 	RTA_NEST_END(skb, parms);
562 
563 	return RTA_NEST_END(skb, opts);
564 
565 rtattr_failure:
566 	return RTA_NEST_CANCEL(skb, opts);
567 }
568 
569 static void gred_destroy(struct Qdisc *sch)
570 {
571 	struct gred_sched *table = qdisc_priv(sch);
572 	int i;
573 
574 	for (i = 0; i < table->DPs; i++) {
575 		if (table->tab[i])
576 			gred_destroy_vq(table->tab[i]);
577 	}
578 }
579 
580 static struct Qdisc_ops gred_qdisc_ops = {
581 	.id		=	"gred",
582 	.priv_size	=	sizeof(struct gred_sched),
583 	.enqueue	=	gred_enqueue,
584 	.dequeue	=	gred_dequeue,
585 	.requeue	=	gred_requeue,
586 	.drop		=	gred_drop,
587 	.init		=	gred_init,
588 	.reset		=	gred_reset,
589 	.destroy	=	gred_destroy,
590 	.change		=	gred_change,
591 	.dump		=	gred_dump,
592 	.owner		=	THIS_MODULE,
593 };
594 
595 static int __init gred_module_init(void)
596 {
597 	return register_qdisc(&gred_qdisc_ops);
598 }
599 
600 static void __exit gred_module_exit(void)
601 {
602 	unregister_qdisc(&gred_qdisc_ops);
603 }
604 
605 module_init(gred_module_init)
606 module_exit(gred_module_exit)
607 
608 MODULE_LICENSE("GPL");
609