xref: /linux/net/sched/sch_tbf.c (revision cc1e6315e83db0e517dd9279050b88adc83a7eba)
1 /*
2  * net/sched/sch_tbf.c	Token Bucket Filter queue.
3  *
4  *		This program is free software; you can redistribute it and/or
5  *		modify it under the terms of the GNU General Public License
6  *		as published by the Free Software Foundation; either version
7  *		2 of the License, or (at your option) any later version.
8  *
9  * Authors:	Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
10  *		Dmitry Torokhov <dtor@mail.ru> - allow attaching inner qdiscs -
11  *						 original idea by Martin Devera
12  *
13  */
14 
15 #include <linux/module.h>
16 #include <linux/types.h>
17 #include <linux/kernel.h>
18 #include <linux/string.h>
19 #include <linux/errno.h>
20 #include <linux/skbuff.h>
21 #include <net/netlink.h>
22 #include <net/sch_generic.h>
23 #include <net/pkt_sched.h>
24 
25 
26 /*	Simple Token Bucket Filter.
27 	=======================================
28 
29 	SOURCE.
30 	-------
31 
32 	None.
33 
34 	Description.
35 	------------
36 
37 	A data flow obeys TBF with rate R and depth B, if for any
38 	time interval t_i...t_f the number of transmitted bits
39 	does not exceed B + R*(t_f-t_i).
40 
41 	Packetized version of this definition:
42 	The sequence of packets of sizes s_i served at moments t_i
43 	obeys TBF, if for any i<=k:
44 
45 	s_i+....+s_k <= B + R*(t_k - t_i)
46 
47 	Algorithm.
48 	----------
49 
50 	Let N(t_i) be B/R initially and N(t) grow continuously with time as:
51 
52 	N(t+delta) = min{B/R, N(t) + delta}
53 
54 	If the first packet in queue has length S, it may be
55 	transmitted only at the time t_* when S/R <= N(t_*),
56 	and in this case N(t) jumps:
57 
58 	N(t_* + 0) = N(t_* - 0) - S/R.
59 
60 
61 
62 	Actually, QoS requires two TBF to be applied to a data stream.
63 	One of them controls steady state burst size, another
64 	one with rate P (peak rate) and depth M (equal to link MTU)
65 	limits bursts at a smaller time scale.
66 
67 	It is easy to see that P>R, and B>M. If P is infinity, this double
68 	TBF is equivalent to a single one.
69 
70 	When TBF works in reshaping mode, latency is estimated as:
71 
72 	lat = max ((L-B)/R, (L-M)/P)
73 
74 
75 	NOTES.
76 	------
77 
78 	If TBF throttles, it starts a watchdog timer, which will wake it up
79 	when it is ready to transmit.
80 	Note that the minimal timer resolution is 1/HZ.
81 	If no new packets arrive during this period,
82 	or if the device is not awaken by EOI for some previous packet,
83 	TBF can stop its activity for 1/HZ.
84 
85 
86 	This means, that with depth B, the maximal rate is
87 
88 	R_crit = B*HZ
89 
90 	F.e. for 10Mbit ethernet and HZ=100 the minimal allowed B is ~10Kbytes.
91 
92 	Note that the peak rate TBF is much more tough: with MTU 1500
93 	P_crit = 150Kbytes/sec. So, if you need greater peak
94 	rates, use alpha with HZ=1000 :-)
95 
96 	With classful TBF, limit is just kept for backwards compatibility.
97 	It is passed to the default bfifo qdisc - if the inner qdisc is
98 	changed the limit is not effective anymore.
99 */
100 
101 struct tbf_sched_data {
102 /* Parameters */
103 	u32		limit;		/* Maximal length of backlog: bytes */
104 	u32		max_size;
105 	s64		buffer;		/* Token bucket depth/rate: MUST BE >= MTU/B */
106 	s64		mtu;
107 	struct psched_ratecfg rate;
108 	struct psched_ratecfg peak;
109 
110 /* Variables */
111 	s64	tokens;			/* Current number of B tokens */
112 	s64	ptokens;		/* Current number of P tokens */
113 	s64	t_c;			/* Time check-point */
114 	struct Qdisc	*qdisc;		/* Inner qdisc, default - bfifo queue */
115 	struct qdisc_watchdog watchdog;	/* Watchdog timer */
116 };
117 
118 
119 /* Time to Length, convert time in ns to length in bytes
120  * to determinate how many bytes can be sent in given time.
121  */
122 static u64 psched_ns_t2l(const struct psched_ratecfg *r,
123 			 u64 time_in_ns)
124 {
125 	/* The formula is :
126 	 * len = (time_in_ns * r->rate_bytes_ps) / NSEC_PER_SEC
127 	 */
128 	u64 len = time_in_ns * r->rate_bytes_ps;
129 
130 	do_div(len, NSEC_PER_SEC);
131 
132 	if (unlikely(r->linklayer == TC_LINKLAYER_ATM)) {
133 		do_div(len, 53);
134 		len = len * 48;
135 	}
136 
137 	if (len > r->overhead)
138 		len -= r->overhead;
139 	else
140 		len = 0;
141 
142 	return len;
143 }
144 
145 /* GSO packet is too big, segment it so that tbf can transmit
146  * each segment in time
147  */
148 static int tbf_segment(struct sk_buff *skb, struct Qdisc *sch,
149 		       struct sk_buff **to_free)
150 {
151 	struct tbf_sched_data *q = qdisc_priv(sch);
152 	struct sk_buff *segs, *nskb;
153 	netdev_features_t features = netif_skb_features(skb);
154 	unsigned int len = 0, prev_len = qdisc_pkt_len(skb);
155 	int ret, nb;
156 
157 	segs = skb_gso_segment(skb, features & ~NETIF_F_GSO_MASK);
158 
159 	if (IS_ERR_OR_NULL(segs))
160 		return qdisc_drop(skb, sch, to_free);
161 
162 	nb = 0;
163 	while (segs) {
164 		nskb = segs->next;
165 		segs->next = NULL;
166 		qdisc_skb_cb(segs)->pkt_len = segs->len;
167 		len += segs->len;
168 		ret = qdisc_enqueue(segs, q->qdisc, to_free);
169 		if (ret != NET_XMIT_SUCCESS) {
170 			if (net_xmit_drop_count(ret))
171 				qdisc_qstats_drop(sch);
172 		} else {
173 			nb++;
174 		}
175 		segs = nskb;
176 	}
177 	sch->q.qlen += nb;
178 	if (nb > 1)
179 		qdisc_tree_reduce_backlog(sch, 1 - nb, prev_len - len);
180 	consume_skb(skb);
181 	return nb > 0 ? NET_XMIT_SUCCESS : NET_XMIT_DROP;
182 }
183 
184 static int tbf_enqueue(struct sk_buff *skb, struct Qdisc *sch,
185 		       struct sk_buff **to_free)
186 {
187 	struct tbf_sched_data *q = qdisc_priv(sch);
188 	int ret;
189 
190 	if (qdisc_pkt_len(skb) > q->max_size) {
191 		if (skb_is_gso(skb) &&
192 		    skb_gso_validate_mac_len(skb, q->max_size))
193 			return tbf_segment(skb, sch, to_free);
194 		return qdisc_drop(skb, sch, to_free);
195 	}
196 	ret = qdisc_enqueue(skb, q->qdisc, to_free);
197 	if (ret != NET_XMIT_SUCCESS) {
198 		if (net_xmit_drop_count(ret))
199 			qdisc_qstats_drop(sch);
200 		return ret;
201 	}
202 
203 	qdisc_qstats_backlog_inc(sch, skb);
204 	sch->q.qlen++;
205 	return NET_XMIT_SUCCESS;
206 }
207 
208 static bool tbf_peak_present(const struct tbf_sched_data *q)
209 {
210 	return q->peak.rate_bytes_ps;
211 }
212 
213 static struct sk_buff *tbf_dequeue(struct Qdisc *sch)
214 {
215 	struct tbf_sched_data *q = qdisc_priv(sch);
216 	struct sk_buff *skb;
217 
218 	skb = q->qdisc->ops->peek(q->qdisc);
219 
220 	if (skb) {
221 		s64 now;
222 		s64 toks;
223 		s64 ptoks = 0;
224 		unsigned int len = qdisc_pkt_len(skb);
225 
226 		now = ktime_get_ns();
227 		toks = min_t(s64, now - q->t_c, q->buffer);
228 
229 		if (tbf_peak_present(q)) {
230 			ptoks = toks + q->ptokens;
231 			if (ptoks > q->mtu)
232 				ptoks = q->mtu;
233 			ptoks -= (s64) psched_l2t_ns(&q->peak, len);
234 		}
235 		toks += q->tokens;
236 		if (toks > q->buffer)
237 			toks = q->buffer;
238 		toks -= (s64) psched_l2t_ns(&q->rate, len);
239 
240 		if ((toks|ptoks) >= 0) {
241 			skb = qdisc_dequeue_peeked(q->qdisc);
242 			if (unlikely(!skb))
243 				return NULL;
244 
245 			q->t_c = now;
246 			q->tokens = toks;
247 			q->ptokens = ptoks;
248 			qdisc_qstats_backlog_dec(sch, skb);
249 			sch->q.qlen--;
250 			qdisc_bstats_update(sch, skb);
251 			return skb;
252 		}
253 
254 		qdisc_watchdog_schedule_ns(&q->watchdog,
255 					   now + max_t(long, -toks, -ptoks));
256 
257 		/* Maybe we have a shorter packet in the queue,
258 		   which can be sent now. It sounds cool,
259 		   but, however, this is wrong in principle.
260 		   We MUST NOT reorder packets under these circumstances.
261 
262 		   Really, if we split the flow into independent
263 		   subflows, it would be a very good solution.
264 		   This is the main idea of all FQ algorithms
265 		   (cf. CSZ, HPFQ, HFSC)
266 		 */
267 
268 		qdisc_qstats_overlimit(sch);
269 	}
270 	return NULL;
271 }
272 
273 static void tbf_reset(struct Qdisc *sch)
274 {
275 	struct tbf_sched_data *q = qdisc_priv(sch);
276 
277 	qdisc_reset(q->qdisc);
278 	sch->qstats.backlog = 0;
279 	sch->q.qlen = 0;
280 	q->t_c = ktime_get_ns();
281 	q->tokens = q->buffer;
282 	q->ptokens = q->mtu;
283 	qdisc_watchdog_cancel(&q->watchdog);
284 }
285 
286 static const struct nla_policy tbf_policy[TCA_TBF_MAX + 1] = {
287 	[TCA_TBF_PARMS]	= { .len = sizeof(struct tc_tbf_qopt) },
288 	[TCA_TBF_RTAB]	= { .type = NLA_BINARY, .len = TC_RTAB_SIZE },
289 	[TCA_TBF_PTAB]	= { .type = NLA_BINARY, .len = TC_RTAB_SIZE },
290 	[TCA_TBF_RATE64]	= { .type = NLA_U64 },
291 	[TCA_TBF_PRATE64]	= { .type = NLA_U64 },
292 	[TCA_TBF_BURST] = { .type = NLA_U32 },
293 	[TCA_TBF_PBURST] = { .type = NLA_U32 },
294 };
295 
296 static int tbf_change(struct Qdisc *sch, struct nlattr *opt,
297 		      struct netlink_ext_ack *extack)
298 {
299 	int err;
300 	struct tbf_sched_data *q = qdisc_priv(sch);
301 	struct nlattr *tb[TCA_TBF_MAX + 1];
302 	struct tc_tbf_qopt *qopt;
303 	struct Qdisc *child = NULL;
304 	struct psched_ratecfg rate;
305 	struct psched_ratecfg peak;
306 	u64 max_size;
307 	s64 buffer, mtu;
308 	u64 rate64 = 0, prate64 = 0;
309 
310 	err = nla_parse_nested(tb, TCA_TBF_MAX, opt, tbf_policy, NULL);
311 	if (err < 0)
312 		return err;
313 
314 	err = -EINVAL;
315 	if (tb[TCA_TBF_PARMS] == NULL)
316 		goto done;
317 
318 	qopt = nla_data(tb[TCA_TBF_PARMS]);
319 	if (qopt->rate.linklayer == TC_LINKLAYER_UNAWARE)
320 		qdisc_put_rtab(qdisc_get_rtab(&qopt->rate,
321 					      tb[TCA_TBF_RTAB],
322 					      NULL));
323 
324 	if (qopt->peakrate.linklayer == TC_LINKLAYER_UNAWARE)
325 			qdisc_put_rtab(qdisc_get_rtab(&qopt->peakrate,
326 						      tb[TCA_TBF_PTAB],
327 						      NULL));
328 
329 	buffer = min_t(u64, PSCHED_TICKS2NS(qopt->buffer), ~0U);
330 	mtu = min_t(u64, PSCHED_TICKS2NS(qopt->mtu), ~0U);
331 
332 	if (tb[TCA_TBF_RATE64])
333 		rate64 = nla_get_u64(tb[TCA_TBF_RATE64]);
334 	psched_ratecfg_precompute(&rate, &qopt->rate, rate64);
335 
336 	if (tb[TCA_TBF_BURST]) {
337 		max_size = nla_get_u32(tb[TCA_TBF_BURST]);
338 		buffer = psched_l2t_ns(&rate, max_size);
339 	} else {
340 		max_size = min_t(u64, psched_ns_t2l(&rate, buffer), ~0U);
341 	}
342 
343 	if (qopt->peakrate.rate) {
344 		if (tb[TCA_TBF_PRATE64])
345 			prate64 = nla_get_u64(tb[TCA_TBF_PRATE64]);
346 		psched_ratecfg_precompute(&peak, &qopt->peakrate, prate64);
347 		if (peak.rate_bytes_ps <= rate.rate_bytes_ps) {
348 			pr_warn_ratelimited("sch_tbf: peakrate %llu is lower than or equals to rate %llu !\n",
349 					peak.rate_bytes_ps, rate.rate_bytes_ps);
350 			err = -EINVAL;
351 			goto done;
352 		}
353 
354 		if (tb[TCA_TBF_PBURST]) {
355 			u32 pburst = nla_get_u32(tb[TCA_TBF_PBURST]);
356 			max_size = min_t(u32, max_size, pburst);
357 			mtu = psched_l2t_ns(&peak, pburst);
358 		} else {
359 			max_size = min_t(u64, max_size, psched_ns_t2l(&peak, mtu));
360 		}
361 	} else {
362 		memset(&peak, 0, sizeof(peak));
363 	}
364 
365 	if (max_size < psched_mtu(qdisc_dev(sch)))
366 		pr_warn_ratelimited("sch_tbf: burst %llu is lower than device %s mtu (%u) !\n",
367 				    max_size, qdisc_dev(sch)->name,
368 				    psched_mtu(qdisc_dev(sch)));
369 
370 	if (!max_size) {
371 		err = -EINVAL;
372 		goto done;
373 	}
374 
375 	if (q->qdisc != &noop_qdisc) {
376 		err = fifo_set_limit(q->qdisc, qopt->limit);
377 		if (err)
378 			goto done;
379 	} else if (qopt->limit > 0) {
380 		child = fifo_create_dflt(sch, &bfifo_qdisc_ops, qopt->limit,
381 					 extack);
382 		if (IS_ERR(child)) {
383 			err = PTR_ERR(child);
384 			goto done;
385 		}
386 
387 		/* child is fifo, no need to check for noop_qdisc */
388 		qdisc_hash_add(child, true);
389 	}
390 
391 	sch_tree_lock(sch);
392 	if (child) {
393 		qdisc_tree_reduce_backlog(q->qdisc, q->qdisc->q.qlen,
394 					  q->qdisc->qstats.backlog);
395 		qdisc_destroy(q->qdisc);
396 		q->qdisc = child;
397 	}
398 	q->limit = qopt->limit;
399 	if (tb[TCA_TBF_PBURST])
400 		q->mtu = mtu;
401 	else
402 		q->mtu = PSCHED_TICKS2NS(qopt->mtu);
403 	q->max_size = max_size;
404 	if (tb[TCA_TBF_BURST])
405 		q->buffer = buffer;
406 	else
407 		q->buffer = PSCHED_TICKS2NS(qopt->buffer);
408 	q->tokens = q->buffer;
409 	q->ptokens = q->mtu;
410 
411 	memcpy(&q->rate, &rate, sizeof(struct psched_ratecfg));
412 	memcpy(&q->peak, &peak, sizeof(struct psched_ratecfg));
413 
414 	sch_tree_unlock(sch);
415 	err = 0;
416 done:
417 	return err;
418 }
419 
420 static int tbf_init(struct Qdisc *sch, struct nlattr *opt,
421 		    struct netlink_ext_ack *extack)
422 {
423 	struct tbf_sched_data *q = qdisc_priv(sch);
424 
425 	qdisc_watchdog_init(&q->watchdog, sch);
426 	q->qdisc = &noop_qdisc;
427 
428 	if (!opt)
429 		return -EINVAL;
430 
431 	q->t_c = ktime_get_ns();
432 
433 	return tbf_change(sch, opt, extack);
434 }
435 
436 static void tbf_destroy(struct Qdisc *sch)
437 {
438 	struct tbf_sched_data *q = qdisc_priv(sch);
439 
440 	qdisc_watchdog_cancel(&q->watchdog);
441 	qdisc_destroy(q->qdisc);
442 }
443 
444 static int tbf_dump(struct Qdisc *sch, struct sk_buff *skb)
445 {
446 	struct tbf_sched_data *q = qdisc_priv(sch);
447 	struct nlattr *nest;
448 	struct tc_tbf_qopt opt;
449 
450 	sch->qstats.backlog = q->qdisc->qstats.backlog;
451 	nest = nla_nest_start(skb, TCA_OPTIONS);
452 	if (nest == NULL)
453 		goto nla_put_failure;
454 
455 	opt.limit = q->limit;
456 	psched_ratecfg_getrate(&opt.rate, &q->rate);
457 	if (tbf_peak_present(q))
458 		psched_ratecfg_getrate(&opt.peakrate, &q->peak);
459 	else
460 		memset(&opt.peakrate, 0, sizeof(opt.peakrate));
461 	opt.mtu = PSCHED_NS2TICKS(q->mtu);
462 	opt.buffer = PSCHED_NS2TICKS(q->buffer);
463 	if (nla_put(skb, TCA_TBF_PARMS, sizeof(opt), &opt))
464 		goto nla_put_failure;
465 	if (q->rate.rate_bytes_ps >= (1ULL << 32) &&
466 	    nla_put_u64_64bit(skb, TCA_TBF_RATE64, q->rate.rate_bytes_ps,
467 			      TCA_TBF_PAD))
468 		goto nla_put_failure;
469 	if (tbf_peak_present(q) &&
470 	    q->peak.rate_bytes_ps >= (1ULL << 32) &&
471 	    nla_put_u64_64bit(skb, TCA_TBF_PRATE64, q->peak.rate_bytes_ps,
472 			      TCA_TBF_PAD))
473 		goto nla_put_failure;
474 
475 	return nla_nest_end(skb, nest);
476 
477 nla_put_failure:
478 	nla_nest_cancel(skb, nest);
479 	return -1;
480 }
481 
482 static int tbf_dump_class(struct Qdisc *sch, unsigned long cl,
483 			  struct sk_buff *skb, struct tcmsg *tcm)
484 {
485 	struct tbf_sched_data *q = qdisc_priv(sch);
486 
487 	tcm->tcm_handle |= TC_H_MIN(1);
488 	tcm->tcm_info = q->qdisc->handle;
489 
490 	return 0;
491 }
492 
493 static int tbf_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new,
494 		     struct Qdisc **old, struct netlink_ext_ack *extack)
495 {
496 	struct tbf_sched_data *q = qdisc_priv(sch);
497 
498 	if (new == NULL)
499 		new = &noop_qdisc;
500 
501 	*old = qdisc_replace(sch, new, &q->qdisc);
502 	return 0;
503 }
504 
505 static struct Qdisc *tbf_leaf(struct Qdisc *sch, unsigned long arg)
506 {
507 	struct tbf_sched_data *q = qdisc_priv(sch);
508 	return q->qdisc;
509 }
510 
511 static unsigned long tbf_find(struct Qdisc *sch, u32 classid)
512 {
513 	return 1;
514 }
515 
516 static void tbf_walk(struct Qdisc *sch, struct qdisc_walker *walker)
517 {
518 	if (!walker->stop) {
519 		if (walker->count >= walker->skip)
520 			if (walker->fn(sch, 1, walker) < 0) {
521 				walker->stop = 1;
522 				return;
523 			}
524 		walker->count++;
525 	}
526 }
527 
528 static const struct Qdisc_class_ops tbf_class_ops = {
529 	.graft		=	tbf_graft,
530 	.leaf		=	tbf_leaf,
531 	.find		=	tbf_find,
532 	.walk		=	tbf_walk,
533 	.dump		=	tbf_dump_class,
534 };
535 
536 static struct Qdisc_ops tbf_qdisc_ops __read_mostly = {
537 	.next		=	NULL,
538 	.cl_ops		=	&tbf_class_ops,
539 	.id		=	"tbf",
540 	.priv_size	=	sizeof(struct tbf_sched_data),
541 	.enqueue	=	tbf_enqueue,
542 	.dequeue	=	tbf_dequeue,
543 	.peek		=	qdisc_peek_dequeued,
544 	.init		=	tbf_init,
545 	.reset		=	tbf_reset,
546 	.destroy	=	tbf_destroy,
547 	.change		=	tbf_change,
548 	.dump		=	tbf_dump,
549 	.owner		=	THIS_MODULE,
550 };
551 
552 static int __init tbf_module_init(void)
553 {
554 	return register_qdisc(&tbf_qdisc_ops);
555 }
556 
557 static void __exit tbf_module_exit(void)
558 {
559 	unregister_qdisc(&tbf_qdisc_ops);
560 }
561 module_init(tbf_module_init)
562 module_exit(tbf_module_exit)
563 MODULE_LICENSE("GPL");
564