[linux.git] / net / sched / sch_tbf.c

/*
 * net/sched/sch_tbf.c	Token Bucket Filter queue.
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 *
 * Authors:	Alexey Kuznetsov, <[email protected]>
 *		Dmitry Torokhov <[email protected]> - allow attaching inner qdiscs -
 *						 original idea by Martin Devera
 *
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>


/*	Simple Token Bucket Filter.
	=======================================

	SOURCE.
	-------

	None.

	Description.
	------------

	A data flow obeys TBF with rate R and depth B, if for any
	time interval t_i...t_f the number of transmitted bits
	does not exceed B + R*(t_f-t_i).

	Packetized version of this definition:
	The sequence of packets of sizes s_i served at moments t_i
	obeys TBF, if for any i<=k:

	s_i+....+s_k <= B + R*(t_k - t_i)

	Algorithm.
	----------

	Let N(t_i) be B/R initially and N(t) grow continuously with time as:

	N(t+delta) = min{B/R, N(t) + delta}

	If the first packet in queue has length S, it may be
	transmitted only at the time t_* when S/R <= N(t_*),
	and in this case N(t) jumps:

	N(t_* + 0) = N(t_* - 0) - S/R.


	Actually, QoS requires two TBF to be applied to a data stream.
	One of them controls steady state burst size, another
	one with rate P (peak rate) and depth M (equal to link MTU)
	limits bursts at a smaller time scale.

	It is easy to see that P>R, and B>M. If P is infinity, this double
	TBF is equivalent to a single one.

	When TBF works in reshaping mode, latency is estimated as:

	lat = max ((L-B)/R, (L-M)/P)


	NOTES.
	------

	If TBF throttles, it starts a watchdog timer, which will wake it up
	when it is ready to transmit.
	Note that the minimal timer resolution is 1/HZ.
	If no new packets arrive during this period,
	or if the device is not awaken by EOI for some previous packet,
	TBF can stop its activity for 1/HZ.


	This means, that with depth B, the maximal rate is

	R_crit = B*HZ

	F.e. for 10Mbit ethernet and HZ=100 the minimal allowed B is ~10Kbytes.

	Note that the peak rate TBF is much more tough: with MTU 1500
	P_crit = 150Kbytes/sec. So, if you need greater peak
	rates, use alpha with HZ=1000 :-)

	With classful TBF, limit is just kept for backwards compatibility.
	It is passed to the default bfifo qdisc - if the inner qdisc is
	changed the limit is not effective anymore.
*/

struct tbf_sched_data {
/* Parameters */
	u32		limit;		/* Maximal length of backlog: bytes */
	u32		buffer;		/* Token bucket depth/rate: MUST BE >= MTU/B */
	u32		mtu;
	u32		max_size;
	struct qdisc_rate_table	*R_tab;
	struct qdisc_rate_table	*P_tab;

/* Variables */
	long	tokens;			/* Current number of B tokens */
	long	ptokens;		/* Current number of P tokens */
	psched_time_t	t_c;		/* Time check-point */
	struct Qdisc	*qdisc;		/* Inner qdisc, default - bfifo queue */
	struct qdisc_watchdog watchdog;	/* Watchdog timer */
};

#define L2T(q, L)   qdisc_l2t((q)->R_tab, L)
#define L2T_P(q, L) qdisc_l2t((q)->P_tab, L)

static int tbf_enqueue(struct sk_buff *skb, struct Qdisc *sch)
{
	struct tbf_sched_data *q = qdisc_priv(sch);
	int ret;

	if (qdisc_pkt_len(skb) > q->max_size)
		return qdisc_reshape_fail(skb, sch);

	ret = qdisc_enqueue(skb, q->qdisc);
	if (ret != NET_XMIT_SUCCESS) {
		if (net_xmit_drop_count(ret))
			sch->qstats.drops++;
		return ret;
	}

	sch->q.qlen++;
	return NET_XMIT_SUCCESS;
}

static unsigned int tbf_drop(struct Qdisc *sch)
{
	struct tbf_sched_data *q = qdisc_priv(sch);
	unsigned int len = 0;

	if (q->qdisc->ops->drop && (len = q->qdisc->ops->drop(q->qdisc)) != 0) {
		sch->q.qlen--;
		sch->qstats.drops++;
	}
	return len;
}

static struct sk_buff *tbf_dequeue(struct Qdisc *sch)
{
	struct tbf_sched_data *q = qdisc_priv(sch);
	struct sk_buff *skb;

	skb = q->qdisc->ops->peek(q->qdisc);

	if (skb) {
		psched_time_t now;
		long toks;
		long ptoks = 0;
		unsigned int len = qdisc_pkt_len(skb);

		now = psched_get_time();
		toks = psched_tdiff_bounded(now, q->t_c, q->buffer);

		if (q->P_tab) {
			ptoks = toks + q->ptokens;
			if (ptoks > (long)q->mtu)
				ptoks = q->mtu;
			ptoks -= L2T_P(q, len);
		}
		toks += q->tokens;
		if (toks > (long)q->buffer)
			toks = q->buffer;
		toks -= L2T(q, len);

		if ((toks|ptoks) >= 0) {
			skb = qdisc_dequeue_peeked(q->qdisc);
			if (unlikely(!skb))
				return NULL;

			q->t_c = now;
			q->tokens = toks;
			q->ptokens = ptoks;
			sch->q.qlen--;
			qdisc_unthrottled(sch);
			qdisc_bstats_update(sch, skb);
			return skb;
		}

		qdisc_watchdog_schedule(&q->watchdog,
					now + max_t(long, -toks, -ptoks));

		/* Maybe we have a shorter packet in the queue,
		   which can be sent now. It sounds cool,
		   but, however, this is wrong in principle.
		   We MUST NOT reorder packets under these circumstances.

		   Really, if we split the flow into independent
		   subflows, it would be a very good solution.
		   This is the main idea of all FQ algorithms
		   (cf. CSZ, HPFQ, HFSC)
		 */

		sch->qstats.overlimits++;
	}
	return NULL;
}

static void tbf_reset(struct Qdisc *sch)
{
	struct tbf_sched_data *q = qdisc_priv(sch);

	qdisc_reset(q->qdisc);
	sch->q.qlen = 0;
	q->t_c = psched_get_time();
	q->tokens = q->buffer;
	q->ptokens = q->mtu;
	qdisc_watchdog_cancel(&q->watchdog);
}

static const struct nla_policy tbf_policy[TCA_TBF_MAX + 1] = {
	[TCA_TBF_PARMS]	= { .len = sizeof(struct tc_tbf_qopt) },
	[TCA_TBF_RTAB]	= { .type = NLA_BINARY, .len = TC_RTAB_SIZE },
	[TCA_TBF_PTAB]	= { .type = NLA_BINARY, .len = TC_RTAB_SIZE },
};

static int tbf_change(struct Qdisc *sch, struct nlattr *opt)
{
	int err;
	struct tbf_sched_data *q = qdisc_priv(sch);
	struct nlattr *tb[TCA_TBF_PTAB + 1];
	struct tc_tbf_qopt *qopt;
	struct qdisc_rate_table *rtab = NULL;
	struct qdisc_rate_table *ptab = NULL;
	struct Qdisc *child = NULL;
	int max_size, n;

	err = nla_parse_nested(tb, TCA_TBF_PTAB, opt, tbf_policy);
	if (err < 0)
		return err;

	err = -EINVAL;
	if (tb[TCA_TBF_PARMS] == NULL)
		goto done;

	qopt = nla_data(tb[TCA_TBF_PARMS]);
	rtab = qdisc_get_rtab(&qopt->rate, tb[TCA_TBF_RTAB]);
	if (rtab == NULL)
		goto done;

	if (qopt->peakrate.rate) {
		if (qopt->peakrate.rate > qopt->rate.rate)
			ptab = qdisc_get_rtab(&qopt->peakrate, tb[TCA_TBF_PTAB]);
		if (ptab == NULL)
			goto done;
	}

	for (n = 0; n < 256; n++)
		if (rtab->data[n] > qopt->buffer)
			break;
	max_size = (n << qopt->rate.cell_log) - 1;
	if (ptab) {
		int size;

		for (n = 0; n < 256; n++)
			if (ptab->data[n] > qopt->mtu)
				break;
		size = (n << qopt->peakrate.cell_log) - 1;
		if (size < max_size)
			max_size = size;
	}
	if (max_size < 0)
		goto done;

	if (q->qdisc != &noop_qdisc) {
		err = fifo_set_limit(q->qdisc, qopt->limit);
		if (err)
			goto done;
	} else if (qopt->limit > 0) {
		child = fifo_create_dflt(sch, &bfifo_qdisc_ops, qopt->limit);
		if (IS_ERR(child)) {
			err = PTR_ERR(child);
			goto done;
		}
	}

	sch_tree_lock(sch);
	if (child) {
		qdisc_tree_decrease_qlen(q->qdisc, q->qdisc->q.qlen);
		qdisc_destroy(q->qdisc);
		q->qdisc = child;
	}
	q->limit = qopt->limit;
	q->mtu = qopt->mtu;
	q->max_size = max_size;
	q->buffer = qopt->buffer;
	q->tokens = q->buffer;
	q->ptokens = q->mtu;

	swap(q->R_tab, rtab);
	swap(q->P_tab, ptab);

	sch_tree_unlock(sch);
	err = 0;
done:
	if (rtab)
		qdisc_put_rtab(rtab);
	if (ptab)
		qdisc_put_rtab(ptab);
	return err;
}

static int tbf_init(struct Qdisc *sch, struct nlattr *opt)
{
	struct tbf_sched_data *q = qdisc_priv(sch);

	if (opt == NULL)
		return -EINVAL;

	q->t_c = psched_get_time();
	qdisc_watchdog_init(&q->watchdog, sch);
	q->qdisc = &noop_qdisc;

	return tbf_change(sch, opt);
}

static void tbf_destroy(struct Qdisc *sch)
{
	struct tbf_sched_data *q = qdisc_priv(sch);

	qdisc_watchdog_cancel(&q->watchdog);

	if (q->P_tab)
		qdisc_put_rtab(q->P_tab);
	if (q->R_tab)
		qdisc_put_rtab(q->R_tab);

	qdisc_destroy(q->qdisc);
}

static int tbf_dump(struct Qdisc *sch, struct sk_buff *skb)
{
	struct tbf_sched_data *q = qdisc_priv(sch);
	struct nlattr *nest;
	struct tc_tbf_qopt opt;

	sch->qstats.backlog = q->qdisc->qstats.backlog;
	nest = nla_nest_start(skb, TCA_OPTIONS);
	if (nest == NULL)
		goto nla_put_failure;

	opt.limit = q->limit;
	opt.rate = q->R_tab->rate;
	if (q->P_tab)
		opt.peakrate = q->P_tab->rate;
	else
		memset(&opt.peakrate, 0, sizeof(opt.peakrate));
	opt.mtu = q->mtu;
	opt.buffer = q->buffer;
	if (nla_put(skb, TCA_TBF_PARMS, sizeof(opt), &opt))
		goto nla_put_failure;

	nla_nest_end(skb, nest);
	return skb->len;

nla_put_failure:
	nla_nest_cancel(skb, nest);
	return -1;
}

static int tbf_dump_class(struct Qdisc *sch, unsigned long cl,
			  struct sk_buff *skb, struct tcmsg *tcm)
{
	struct tbf_sched_data *q = qdisc_priv(sch);

	tcm->tcm_handle |= TC_H_MIN(1);
	tcm->tcm_info = q->qdisc->handle;

	return 0;
}

static int tbf_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new,
		     struct Qdisc **old)
{
	struct tbf_sched_data *q = qdisc_priv(sch);

	if (new == NULL)
		new = &noop_qdisc;

	sch_tree_lock(sch);
	*old = q->qdisc;
	q->qdisc = new;
	qdisc_tree_decrease_qlen(*old, (*old)->q.qlen);
	qdisc_reset(*old);
	sch_tree_unlock(sch);

	return 0;
}

static struct Qdisc *tbf_leaf(struct Qdisc *sch, unsigned long arg)
{
	struct tbf_sched_data *q = qdisc_priv(sch);
	return q->qdisc;
}

static unsigned long tbf_get(struct Qdisc *sch, u32 classid)
{
	return 1;
}

static void tbf_put(struct Qdisc *sch, unsigned long arg)
{
}

static void tbf_walk(struct Qdisc *sch, struct qdisc_walker *walker)
{
	if (!walker->stop) {
		if (walker->count >= walker->skip)
			if (walker->fn(sch, 1, walker) < 0) {
				walker->stop = 1;
				return;
			}
		walker->count++;
	}
}

static const struct Qdisc_class_ops tbf_class_ops = {
	.graft		=	tbf_graft,
	.leaf		=	tbf_leaf,
	.get		=	tbf_get,
	.put		=	tbf_put,
	.walk		=	tbf_walk,
	.dump		=	tbf_dump_class,
};

static struct Qdisc_ops tbf_qdisc_ops __read_mostly = {
	.next		=	NULL,
	.cl_ops		=	&tbf_class_ops,
	.id		=	"tbf",
	.priv_size	=	sizeof(struct tbf_sched_data),
	.enqueue	=	tbf_enqueue,
	.dequeue	=	tbf_dequeue,
	.peek		=	qdisc_peek_dequeued,
	.drop		=	tbf_drop,
	.init		=	tbf_init,
	.reset		=	tbf_reset,
	.destroy	=	tbf_destroy,
	.change		=	tbf_change,
	.dump		=	tbf_dump,
	.owner		=	THIS_MODULE,
};

static int __init tbf_module_init(void)
{
	return register_qdisc(&tbf_qdisc_ops);
}

static void __exit tbf_module_exit(void)
{
	unregister_qdisc(&tbf_qdisc_ops);
}
module_init(tbf_module_init)
module_exit(tbf_module_exit)
MODULE_LICENSE("GPL");
Commit	Line	Data
1da177e4 LT	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, <[email protected]>
	10	* Dmitry Torokhov <[email protected]> - allow attaching inner qdiscs -
	11	* original idea by Martin Devera
	12	*
	13	*/
	14
1da177e4	15	#include <linux/module.h>
1da177e4 LT	16	#include <linux/types.h>
1da177e4 LT	17	#include <linux/kernel.h>
1da177e4	18	#include <linux/string.h>
1da177e4	19	#include <linux/errno.h>
1da177e4	20	#include <linux/skbuff.h>
0ba48053	21	#include <net/netlink.h>
1da177e4 LT	22	#include <net/pkt_sched.h>
	23
	24
	25	/* Simple Token Bucket Filter.
	26	=======================================
	27
	28	SOURCE.
	29	-------
	30
	31	None.
	32
	33	Description.
	34	------------
	35
	36	A data flow obeys TBF with rate R and depth B, if for any
	37	time interval t_i...t_f the number of transmitted bits
	38	does not exceed B + R*(t_f-t_i).
	39
	40	Packetized version of this definition:
	41	The sequence of packets of sizes s_i served at moments t_i
	42	obeys TBF, if for any i<=k:
	43
	44	s_i+....+s_k <= B + R*(t_k - t_i)
	45
	46	Algorithm.
	47	----------
	48
	49	Let N(t_i) be B/R initially and N(t) grow continuously with time as:
	50
	51	N(t+delta) = min{B/R, N(t) + delta}
	52
	53	If the first packet in queue has length S, it may be
	54	transmitted only at the time t_* when S/R <= N(t_*),
	55	and in this case N(t) jumps:
	56
	57	N(t_* + 0) = N(t_* - 0) - S/R.
	58
	59
	60
	61	Actually, QoS requires two TBF to be applied to a data stream.
	62	One of them controls steady state burst size, another
	63	one with rate P (peak rate) and depth M (equal to link MTU)
	64	limits bursts at a smaller time scale.
	65
	66	It is easy to see that P>R, and B>M. If P is infinity, this double
	67	TBF is equivalent to a single one.
	68
	69	When TBF works in reshaping mode, latency is estimated as:
	70
	71	lat = max ((L-B)/R, (L-M)/P)
	72
	73
	74	NOTES.
	75	------
	76
	77	If TBF throttles, it starts a watchdog timer, which will wake it up
	78	when it is ready to transmit.
	79	Note that the minimal timer resolution is 1/HZ.
	80	If no new packets arrive during this period,
	81	or if the device is not awaken by EOI for some previous packet,
	82	TBF can stop its activity for 1/HZ.
	83
	84
	85	This means, that with depth B, the maximal rate is
86
87	R_crit = B*HZ
88
89	F.e. for 10Mbit ethernet and HZ=100 the minimal allowed B is ~10Kbytes.
90
91	Note that the peak rate TBF is much more tough: with MTU 1500
92	P_crit = 150Kbytes/sec. So, if you need greater peak
93	rates, use alpha with HZ=1000 :-)
94
95	With classful TBF, limit is just kept for backwards compatibility.
96	It is passed to the default bfifo qdisc - if the inner qdisc is
97	changed the limit is not effective anymore.
98	*/
99
cc7ec456	100	struct tbf_sched_data {
1da177e4 LT	101	/* Parameters */
	102	u32 limit; /* Maximal length of backlog: bytes */
	103	u32 buffer; /* Token bucket depth/rate: MUST BE >= MTU/B */
	104	u32 mtu;
	105	u32 max_size;
	106	struct qdisc_rate_table *R_tab;
	107	struct qdisc_rate_table *P_tab;
	108
	109	/* Variables */
	110	long tokens; /* Current number of B tokens */
	111	long ptokens; /* Current number of P tokens */
	112	psched_time_t t_c; /* Time check-point */
1da177e4	113	struct Qdisc qdisc; / Inner qdisc, default - bfifo queue */
f7f593e3	114	struct qdisc_watchdog watchdog; /* Watchdog timer */
1da177e4 LT	115	};
1da177e4 LT	116
cc7ec456 ED	117	#define L2T(q, L) qdisc_l2t((q)->R_tab, L)
cc7ec456 ED	118	#define L2T_P(q, L) qdisc_l2t((q)->P_tab, L)
1da177e4	119
cc7ec456	120	static int tbf_enqueue(struct sk_buff skb, struct Qdisc sch)
1da177e4 LT	121	{
	122	struct tbf_sched_data *q = qdisc_priv(sch);
	123	int ret;
	124
69747650 DM	125	if (qdisc_pkt_len(skb) > q->max_size)
69747650 DM	126	return qdisc_reshape_fail(skb, sch);
1da177e4	127
5f86173b	128	ret = qdisc_enqueue(skb, q->qdisc);
9871e50e	129	if (ret != NET_XMIT_SUCCESS) {
378a2f09 JP	130	if (net_xmit_drop_count(ret))
378a2f09 JP	131	sch->qstats.drops++;
1da177e4 LT	132	return ret;
	133	}
	134
	135	sch->q.qlen++;
9871e50e	136	return NET_XMIT_SUCCESS;
1da177e4 LT	137	}
1da177e4 LT	138
cc7ec456	139	static unsigned int tbf_drop(struct Qdisc *sch)
1da177e4 LT	140	{
1da177e4 LT	141	struct tbf_sched_data *q = qdisc_priv(sch);
6d037a26	142	unsigned int len = 0;
1da177e4	143
6d037a26	144	if (q->qdisc->ops->drop && (len = q->qdisc->ops->drop(q->qdisc)) != 0) {
1da177e4 LT	145	sch->q.qlen--;
	146	sch->qstats.drops++;
	147	}
	148	return len;
	149	}
	150
cc7ec456	151	static struct sk_buff tbf_dequeue(struct Qdisc sch)
1da177e4 LT	152	{
	153	struct tbf_sched_data *q = qdisc_priv(sch);
	154	struct sk_buff *skb;
	155
03c05f0d	156	skb = q->qdisc->ops->peek(q->qdisc);
1da177e4 LT	157
	158	if (skb) {
	159	psched_time_t now;
f7f593e3	160	long toks;
1da177e4	161	long ptoks = 0;
0abf77e5	162	unsigned int len = qdisc_pkt_len(skb);
1da177e4	163
3bebcda2	164	now = psched_get_time();
03cc45c0	165	toks = psched_tdiff_bounded(now, q->t_c, q->buffer);
1da177e4 LT	166
	167	if (q->P_tab) {
	168	ptoks = toks + q->ptokens;
	169	if (ptoks > (long)q->mtu)
	170	ptoks = q->mtu;
	171	ptoks -= L2T_P(q, len);
	172	}
	173	toks += q->tokens;
	174	if (toks > (long)q->buffer)
	175	toks = q->buffer;
	176	toks -= L2T(q, len);
	177
	178	if ((toks\|ptoks) >= 0) {
77be155c	179	skb = qdisc_dequeue_peeked(q->qdisc);
03c05f0d JP	180	if (unlikely(!skb))
	181	return NULL;
	182
1da177e4 LT	183	q->t_c = now;
	184	q->tokens = toks;
	185	q->ptokens = ptoks;
	186	sch->q.qlen--;
fd245a4a	187	qdisc_unthrottled(sch);
9190b3b3	188	qdisc_bstats_update(sch, skb);
1da177e4 LT	189	return skb;
	190	}
	191
f7f593e3 PM	192	qdisc_watchdog_schedule(&q->watchdog,
f7f593e3 PM	193	now + max_t(long, -toks, -ptoks));
1da177e4 LT	194
	195	/* Maybe we have a shorter packet in the queue,
	196	which can be sent now. It sounds cool,
	197	but, however, this is wrong in principle.
	198	We MUST NOT reorder packets under these circumstances.
	199
	200	Really, if we split the flow into independent
	201	subflows, it would be a very good solution.
	202	This is the main idea of all FQ algorithms
	203	(cf. CSZ, HPFQ, HFSC)
	204	*/
	205
1da177e4 LT	206	sch->qstats.overlimits++;
	207	}
	208	return NULL;
	209	}
	210
cc7ec456	211	static void tbf_reset(struct Qdisc *sch)
1da177e4 LT	212	{
	213	struct tbf_sched_data *q = qdisc_priv(sch);
	214
	215	qdisc_reset(q->qdisc);
	216	sch->q.qlen = 0;
3bebcda2	217	q->t_c = psched_get_time();
1da177e4 LT	218	q->tokens = q->buffer;
1da177e4 LT	219	q->ptokens = q->mtu;
f7f593e3	220	qdisc_watchdog_cancel(&q->watchdog);
1da177e4 LT	221	}
1da177e4 LT	222
27a3421e PM	223	static const struct nla_policy tbf_policy[TCA_TBF_MAX + 1] = {
	224	[TCA_TBF_PARMS] = { .len = sizeof(struct tc_tbf_qopt) },
	225	[TCA_TBF_RTAB] = { .type = NLA_BINARY, .len = TC_RTAB_SIZE },
	226	[TCA_TBF_PTAB] = { .type = NLA_BINARY, .len = TC_RTAB_SIZE },
	227	};
	228
cc7ec456	229	static int tbf_change(struct Qdisc sch, struct nlattr opt)
1da177e4	230	{
cee63723	231	int err;
1da177e4	232	struct tbf_sched_data *q = qdisc_priv(sch);
1e90474c	233	struct nlattr *tb[TCA_TBF_PTAB + 1];
1da177e4 LT	234	struct tc_tbf_qopt *qopt;
	235	struct qdisc_rate_table *rtab = NULL;
	236	struct qdisc_rate_table *ptab = NULL;
	237	struct Qdisc *child = NULL;
cc7ec456	238	int max_size, n;
1da177e4	239
27a3421e	240	err = nla_parse_nested(tb, TCA_TBF_PTAB, opt, tbf_policy);
cee63723 PM	241	if (err < 0)
	242	return err;
	243
	244	err = -EINVAL;
27a3421e	245	if (tb[TCA_TBF_PARMS] == NULL)
1da177e4 LT	246	goto done;
1da177e4 LT	247
1e90474c PM	248	qopt = nla_data(tb[TCA_TBF_PARMS]);
1e90474c PM	249	rtab = qdisc_get_rtab(&qopt->rate, tb[TCA_TBF_RTAB]);
1da177e4 LT	250	if (rtab == NULL)
	251	goto done;
	252
	253	if (qopt->peakrate.rate) {
	254	if (qopt->peakrate.rate > qopt->rate.rate)
1e90474c	255	ptab = qdisc_get_rtab(&qopt->peakrate, tb[TCA_TBF_PTAB]);
1da177e4 LT	256	if (ptab == NULL)
	257	goto done;
	258	}
	259
	260	for (n = 0; n < 256; n++)
cc7ec456 ED	261	if (rtab->data[n] > qopt->buffer)
	262	break;
	263	max_size = (n << qopt->rate.cell_log) - 1;
1da177e4 LT	264	if (ptab) {
	265	int size;
	266
	267	for (n = 0; n < 256; n++)
cc7ec456 ED	268	if (ptab->data[n] > qopt->mtu)
	269	break;
	270	size = (n << qopt->peakrate.cell_log) - 1;
	271	if (size < max_size)
	272	max_size = size;
1da177e4 LT	273	}
	274	if (max_size < 0)
	275	goto done;
	276
f0cd1508	277	if (q->qdisc != &noop_qdisc) {
	278	err = fifo_set_limit(q->qdisc, qopt->limit);
	279	if (err)
	280	goto done;
	281	} else if (qopt->limit > 0) {
fb0305ce PM	282	child = fifo_create_dflt(sch, &bfifo_qdisc_ops, qopt->limit);
	283	if (IS_ERR(child)) {
	284	err = PTR_ERR(child);
1da177e4	285	goto done;
fb0305ce	286	}
1da177e4 LT	287	}
	288
	289	sch_tree_lock(sch);
5e50da01 PM	290	if (child) {
5e50da01 PM	291	qdisc_tree_decrease_qlen(q->qdisc, q->qdisc->q.qlen);
b94c8afc PM	292	qdisc_destroy(q->qdisc);
b94c8afc PM	293	q->qdisc = child;
5e50da01	294	}
1da177e4 LT	295	q->limit = qopt->limit;
	296	q->mtu = qopt->mtu;
	297	q->max_size = max_size;
	298	q->buffer = qopt->buffer;
	299	q->tokens = q->buffer;
	300	q->ptokens = q->mtu;
b94c8afc	301
a0bffffc IJ	302	swap(q->R_tab, rtab);
a0bffffc IJ	303	swap(q->P_tab, ptab);
b94c8afc	304
1da177e4 LT	305	sch_tree_unlock(sch);
	306	err = 0;
	307	done:
	308	if (rtab)
	309	qdisc_put_rtab(rtab);
	310	if (ptab)
	311	qdisc_put_rtab(ptab);
	312	return err;
	313	}
	314
cc7ec456	315	static int tbf_init(struct Qdisc sch, struct nlattr opt)
1da177e4 LT	316	{
	317	struct tbf_sched_data *q = qdisc_priv(sch);
	318
	319	if (opt == NULL)
	320	return -EINVAL;
	321
3bebcda2	322	q->t_c = psched_get_time();
f7f593e3	323	qdisc_watchdog_init(&q->watchdog, sch);
1da177e4 LT	324	q->qdisc = &noop_qdisc;
	325
	326	return tbf_change(sch, opt);
	327	}
	328
	329	static void tbf_destroy(struct Qdisc *sch)
	330	{
	331	struct tbf_sched_data *q = qdisc_priv(sch);
	332
f7f593e3	333	qdisc_watchdog_cancel(&q->watchdog);
1da177e4 LT	334
	335	if (q->P_tab)
	336	qdisc_put_rtab(q->P_tab);
	337	if (q->R_tab)
	338	qdisc_put_rtab(q->R_tab);
	339
	340	qdisc_destroy(q->qdisc);
	341	}
	342
	343	static int tbf_dump(struct Qdisc sch, struct sk_buff skb)
	344	{
	345	struct tbf_sched_data *q = qdisc_priv(sch);
4b3550ef	346	struct nlattr *nest;
1da177e4 LT	347	struct tc_tbf_qopt opt;
1da177e4 LT	348
b0460e44	349	sch->qstats.backlog = q->qdisc->qstats.backlog;
4b3550ef PM	350	nest = nla_nest_start(skb, TCA_OPTIONS);
	351	if (nest == NULL)
	352	goto nla_put_failure;
1da177e4 LT	353
	354	opt.limit = q->limit;
	355	opt.rate = q->R_tab->rate;
	356	if (q->P_tab)
	357	opt.peakrate = q->P_tab->rate;
	358	else
	359	memset(&opt.peakrate, 0, sizeof(opt.peakrate));
	360	opt.mtu = q->mtu;
	361	opt.buffer = q->buffer;
1b34ec43 DM	362	if (nla_put(skb, TCA_TBF_PARMS, sizeof(opt), &opt))
1b34ec43 DM	363	goto nla_put_failure;
1da177e4	364
4b3550ef	365	nla_nest_end(skb, nest);
1da177e4 LT	366	return skb->len;
1da177e4 LT	367
1e90474c	368	nla_put_failure:
4b3550ef	369	nla_nest_cancel(skb, nest);
1da177e4 LT	370	return -1;
	371	}
	372
	373	static int tbf_dump_class(struct Qdisc *sch, unsigned long cl,
	374	struct sk_buff skb, struct tcmsg tcm)
	375	{
	376	struct tbf_sched_data *q = qdisc_priv(sch);
	377
1da177e4 LT	378	tcm->tcm_handle \|= TC_H_MIN(1);
	379	tcm->tcm_info = q->qdisc->handle;
	380
	381	return 0;
	382	}
	383
	384	static int tbf_graft(struct Qdisc sch, unsigned long arg, struct Qdisc new,
	385	struct Qdisc **old)
	386	{
	387	struct tbf_sched_data *q = qdisc_priv(sch);
	388
	389	if (new == NULL)
	390	new = &noop_qdisc;
	391
	392	sch_tree_lock(sch);
b94c8afc PM	393	*old = q->qdisc;
b94c8afc PM	394	q->qdisc = new;
5e50da01	395	qdisc_tree_decrease_qlen(old, (old)->q.qlen);
1da177e4	396	qdisc_reset(*old);
1da177e4 LT	397	sch_tree_unlock(sch);
	398
	399	return 0;
	400	}
	401
	402	static struct Qdisc tbf_leaf(struct Qdisc sch, unsigned long arg)
	403	{
	404	struct tbf_sched_data *q = qdisc_priv(sch);
	405	return q->qdisc;
	406	}
	407
	408	static unsigned long tbf_get(struct Qdisc *sch, u32 classid)
	409	{
	410	return 1;
	411	}
	412
	413	static void tbf_put(struct Qdisc *sch, unsigned long arg)
	414	{
	415	}
	416
1da177e4 LT	417	static void tbf_walk(struct Qdisc sch, struct qdisc_walker walker)
	418	{
	419	if (!walker->stop) {
	420	if (walker->count >= walker->skip)
	421	if (walker->fn(sch, 1, walker) < 0) {
	422	walker->stop = 1;
	423	return;
	424	}
	425	walker->count++;
	426	}
	427	}
	428
cc7ec456	429	static const struct Qdisc_class_ops tbf_class_ops = {
1da177e4 LT	430	.graft = tbf_graft,
	431	.leaf = tbf_leaf,
	432	.get = tbf_get,
	433	.put = tbf_put,
1da177e4	434	.walk = tbf_walk,
1da177e4 LT	435	.dump = tbf_dump_class,
	436	};
	437
20fea08b	438	static struct Qdisc_ops tbf_qdisc_ops __read_mostly = {
1da177e4 LT	439	.next = NULL,
	440	.cl_ops = &tbf_class_ops,
	441	.id = "tbf",
	442	.priv_size = sizeof(struct tbf_sched_data),
	443	.enqueue = tbf_enqueue,
	444	.dequeue = tbf_dequeue,
77be155c	445	.peek = qdisc_peek_dequeued,
1da177e4 LT	446	.drop = tbf_drop,
	447	.init = tbf_init,
	448	.reset = tbf_reset,
	449	.destroy = tbf_destroy,
	450	.change = tbf_change,
	451	.dump = tbf_dump,
	452	.owner = THIS_MODULE,
	453	};
	454
	455	static int __init tbf_module_init(void)
	456	{
	457	return register_qdisc(&tbf_qdisc_ops);
	458	}
	459
	460	static void __exit tbf_module_exit(void)
	461	{
	462	unregister_qdisc(&tbf_qdisc_ops);
	463	}
	464	module_init(tbf_module_init)
	465	module_exit(tbf_module_exit)
	466	MODULE_LICENSE("GPL");