[deliverable/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, <kuznet@ms2.inr.ac.ru>
 *		Dmitry Torokhov <dtor@mail.ru> - 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 (skb->len > q->max_size) {
		sch->qstats.drops++;
#ifdef CONFIG_NET_CLS_ACT
		if (sch->reshape_fail == NULL || sch->reshape_fail(skb, sch))
#endif
			kfree_skb(skb);

		return NET_XMIT_DROP;
	}

	if ((ret = q->qdisc->enqueue(skb, q->qdisc)) != 0) {
		sch->qstats.drops++;
		return ret;
	}

	sch->q.qlen++;
	sch->bstats.bytes += skb->len;
	sch->bstats.packets++;
	return 0;
}

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

	if ((ret = q->qdisc->ops->requeue(skb, q->qdisc)) == 0) {
		sch->q.qlen++;
		sch->qstats.requeues++;
	}

	return ret;
}

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->dequeue(q->qdisc);

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

		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) {
			q->t_c = now;
			q->tokens = toks;
			q->ptokens = ptoks;
			sch->q.qlen--;
			sch->flags &= ~TCQ_F_THROTTLED;
			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)
		 */

		if (q->qdisc->ops->requeue(skb, q->qdisc) != NET_XMIT_SUCCESS) {
			/* When requeue fails skb is dropped */
			qdisc_tree_decrease_qlen(q->qdisc, 1);
			sch->qstats.drops++;
		}

		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 struct Qdisc *tbf_create_dflt_qdisc(struct Qdisc *sch, u32 limit)
{
	struct Qdisc *q;
	struct nlattr *nla;
	int ret;

	q = qdisc_create_dflt(sch->dev, &bfifo_qdisc_ops,
			      TC_H_MAKE(sch->handle, 1));
	if (q) {
		nla = kmalloc(nla_attr_size(sizeof(struct tc_fifo_qopt)),
			      GFP_KERNEL);
		if (nla) {
			nla->nla_type = RTM_NEWQDISC;
			nla->nla_len = nla_attr_size(sizeof(struct tc_fifo_qopt));
			((struct tc_fifo_qopt *)nla_data(nla))->limit = limit;

			ret = q->ops->change(q, nla);
			kfree(nla);

			if (ret == 0)
				return q;
		}
		qdisc_destroy(q);
	}

	return NULL;
}

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, NULL);
	if (err < 0)
		return err;

	err = -EINVAL;
	if (tb[TCA_TBF_PARMS] == NULL ||
	    nla_len(tb[TCA_TBF_PARMS]) < sizeof(*qopt))
		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 (qopt->limit > 0) {
		if ((child = tbf_create_dflt_qdisc(sch, qopt->limit)) == NULL)
			goto done;
	}

	sch_tree_lock(sch);
	if (child) {
		qdisc_tree_decrease_qlen(q->qdisc, q->qdisc->q.qlen);
		qdisc_destroy(xchg(&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;
	rtab = xchg(&q->R_tab, rtab);
	ptab = xchg(&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);
	unsigned char *b = skb_tail_pointer(skb);
	struct nlattr *nla;
	struct tc_tbf_qopt opt;

	nla = (struct nlattr*)b;
	NLA_PUT(skb, TCA_OPTIONS, 0, NULL);

	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;
	NLA_PUT(skb, TCA_TBF_PARMS, sizeof(opt), &opt);
	nla->nla_len = skb_tail_pointer(skb) - b;

	return skb->len;

nla_put_failure:
	nlmsg_trim(skb, b);
	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);

	if (cl != 1) 	/* only one class */
		return -ENOENT;

	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 = xchg(&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 int tbf_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
			    struct nlattr **tca, unsigned long *arg)
{
	return -ENOSYS;
}

static int tbf_delete(struct Qdisc *sch, unsigned long arg)
{
	return -ENOSYS;
}

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 struct tcf_proto **tbf_find_tcf(struct Qdisc *sch, unsigned long cl)
{
	return NULL;
}

static const struct Qdisc_class_ops tbf_class_ops =
{
	.graft		=	tbf_graft,
	.leaf		=	tbf_leaf,
	.get		=	tbf_get,
	.put		=	tbf_put,
	.change		=	tbf_change_class,
	.delete		=	tbf_delete,
	.walk		=	tbf_walk,
	.tcf_chain	=	tbf_find_tcf,
	.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,
	.requeue	=	tbf_requeue,
	.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, <kuznet@ms2.inr.ac.ru>
	10	* Dmitry Torokhov <dtor@mail.ru> - 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
100	struct tbf_sched_data
101	{
102	/* Parameters */
103	u32 limit; /* Maximal length of backlog: bytes */
104	u32 buffer; /* Token bucket depth/rate: MUST BE >= MTU/B */
105	u32 mtu;
106	u32 max_size;
107	struct qdisc_rate_table *R_tab;
108	struct qdisc_rate_table *P_tab;
109
110	/* Variables */
111	long tokens; /* Current number of B tokens */
112	long ptokens; /* Current number of P tokens */
113	psched_time_t t_c; /* Time check-point */
1da177e4	114	struct Qdisc qdisc; / Inner qdisc, default - bfifo queue */
f7f593e3	115	struct qdisc_watchdog watchdog; /* Watchdog timer */
1da177e4 LT	116	};
1da177e4 LT	117
e9bef55d JDB	118	#define L2T(q,L) qdisc_l2t((q)->R_tab,L)
e9bef55d JDB	119	#define L2T_P(q,L) qdisc_l2t((q)->P_tab,L)
1da177e4 LT	120
	121	static int tbf_enqueue(struct sk_buff skb, struct Qdisc sch)
	122	{
	123	struct tbf_sched_data *q = qdisc_priv(sch);
	124	int ret;
	125
	126	if (skb->len > q->max_size) {
	127	sch->qstats.drops++;
c3bc7cff	128	#ifdef CONFIG_NET_CLS_ACT
1da177e4 LT	129	if (sch->reshape_fail == NULL \|\| sch->reshape_fail(skb, sch))
	130	#endif
	131	kfree_skb(skb);
	132
	133	return NET_XMIT_DROP;
	134	}
	135
	136	if ((ret = q->qdisc->enqueue(skb, q->qdisc)) != 0) {
	137	sch->qstats.drops++;
	138	return ret;
	139	}
	140
	141	sch->q.qlen++;
	142	sch->bstats.bytes += skb->len;
	143	sch->bstats.packets++;
	144	return 0;
	145	}
	146
	147	static int tbf_requeue(struct sk_buff skb, struct Qdisc sch)
	148	{
	149	struct tbf_sched_data *q = qdisc_priv(sch);
	150	int ret;
	151
	152	if ((ret = q->qdisc->ops->requeue(skb, q->qdisc)) == 0) {
	153	sch->q.qlen++;
	154	sch->qstats.requeues++;
	155	}
	156
	157	return ret;
	158	}
	159
	160	static unsigned int tbf_drop(struct Qdisc* sch)
	161	{
	162	struct tbf_sched_data *q = qdisc_priv(sch);
6d037a26	163	unsigned int len = 0;
1da177e4	164
6d037a26	165	if (q->qdisc->ops->drop && (len = q->qdisc->ops->drop(q->qdisc)) != 0) {
1da177e4 LT	166	sch->q.qlen--;
	167	sch->qstats.drops++;
	168	}
	169	return len;
	170	}
	171
1da177e4 LT	172	static struct sk_buff tbf_dequeue(struct Qdisc sch)
	173	{
	174	struct tbf_sched_data *q = qdisc_priv(sch);
	175	struct sk_buff *skb;
	176
	177	skb = q->qdisc->dequeue(q->qdisc);
	178
	179	if (skb) {
	180	psched_time_t now;
f7f593e3	181	long toks;
1da177e4 LT	182	long ptoks = 0;
	183	unsigned int len = skb->len;
	184
3bebcda2	185	now = psched_get_time();
03cc45c0	186	toks = psched_tdiff_bounded(now, q->t_c, q->buffer);
1da177e4 LT	187
	188	if (q->P_tab) {
	189	ptoks = toks + q->ptokens;
	190	if (ptoks > (long)q->mtu)
	191	ptoks = q->mtu;
	192	ptoks -= L2T_P(q, len);
	193	}
	194	toks += q->tokens;
	195	if (toks > (long)q->buffer)
	196	toks = q->buffer;
	197	toks -= L2T(q, len);
	198
	199	if ((toks\|ptoks) >= 0) {
	200	q->t_c = now;
	201	q->tokens = toks;
	202	q->ptokens = ptoks;
	203	sch->q.qlen--;
	204	sch->flags &= ~TCQ_F_THROTTLED;
	205	return skb;
	206	}
	207
f7f593e3 PM	208	qdisc_watchdog_schedule(&q->watchdog,
f7f593e3 PM	209	now + max_t(long, -toks, -ptoks));
1da177e4 LT	210
	211	/* Maybe we have a shorter packet in the queue,
	212	which can be sent now. It sounds cool,
	213	but, however, this is wrong in principle.
	214	We MUST NOT reorder packets under these circumstances.
	215
	216	Really, if we split the flow into independent
	217	subflows, it would be a very good solution.
	218	This is the main idea of all FQ algorithms
	219	(cf. CSZ, HPFQ, HFSC)
	220	*/
	221
	222	if (q->qdisc->ops->requeue(skb, q->qdisc) != NET_XMIT_SUCCESS) {
	223	/* When requeue fails skb is dropped */
e488eafc	224	qdisc_tree_decrease_qlen(q->qdisc, 1);
1da177e4 LT	225	sch->qstats.drops++;
	226	}
	227
1da177e4 LT	228	sch->qstats.overlimits++;
	229	}
	230	return NULL;
	231	}
	232
	233	static void tbf_reset(struct Qdisc* sch)
	234	{
	235	struct tbf_sched_data *q = qdisc_priv(sch);
	236
	237	qdisc_reset(q->qdisc);
	238	sch->q.qlen = 0;
3bebcda2	239	q->t_c = psched_get_time();
1da177e4 LT	240	q->tokens = q->buffer;
1da177e4 LT	241	q->ptokens = q->mtu;
f7f593e3	242	qdisc_watchdog_cancel(&q->watchdog);
1da177e4 LT	243	}
1da177e4 LT	244
9f9afec4	245	static struct Qdisc tbf_create_dflt_qdisc(struct Qdisc sch, u32 limit)
1da177e4	246	{
9f9afec4	247	struct Qdisc *q;
1e90474c	248	struct nlattr *nla;
1da177e4 LT	249	int ret;
1da177e4 LT	250
9f9afec4 PM	251	q = qdisc_create_dflt(sch->dev, &bfifo_qdisc_ops,
9f9afec4 PM	252	TC_H_MAKE(sch->handle, 1));
1da177e4	253	if (q) {
1e90474c PM	254	nla = kmalloc(nla_attr_size(sizeof(struct tc_fifo_qopt)),
	255	GFP_KERNEL);
	256	if (nla) {
	257	nla->nla_type = RTM_NEWQDISC;
	258	nla->nla_len = nla_attr_size(sizeof(struct tc_fifo_qopt));
	259	((struct tc_fifo_qopt *)nla_data(nla))->limit = limit;
1da177e4	260
1e90474c PM	261	ret = q->ops->change(q, nla);
1e90474c PM	262	kfree(nla);
1da177e4 LT	263
	264	if (ret == 0)
	265	return q;
	266	}
	267	qdisc_destroy(q);
	268	}
	269
	270	return NULL;
	271	}
	272
1e90474c	273	static int tbf_change(struct Qdisc* sch, struct nlattr *opt)
1da177e4	274	{
cee63723	275	int err;
1da177e4	276	struct tbf_sched_data *q = qdisc_priv(sch);
1e90474c	277	struct nlattr *tb[TCA_TBF_PTAB + 1];
1da177e4 LT	278	struct tc_tbf_qopt *qopt;
	279	struct qdisc_rate_table *rtab = NULL;
	280	struct qdisc_rate_table *ptab = NULL;
	281	struct Qdisc *child = NULL;
	282	int max_size,n;
	283
cee63723 PM	284	err = nla_parse_nested(tb, TCA_TBF_PTAB, opt, NULL);
	285	if (err < 0)
	286	return err;
	287
	288	err = -EINVAL;
	289	if (tb[TCA_TBF_PARMS] == NULL \|\|
1e90474c	290	nla_len(tb[TCA_TBF_PARMS]) < sizeof(*qopt))
1da177e4 LT	291	goto done;
1da177e4 LT	292
1e90474c PM	293	qopt = nla_data(tb[TCA_TBF_PARMS]);
1e90474c PM	294	rtab = qdisc_get_rtab(&qopt->rate, tb[TCA_TBF_RTAB]);
1da177e4 LT	295	if (rtab == NULL)
	296	goto done;
	297
	298	if (qopt->peakrate.rate) {
	299	if (qopt->peakrate.rate > qopt->rate.rate)
1e90474c	300	ptab = qdisc_get_rtab(&qopt->peakrate, tb[TCA_TBF_PTAB]);
1da177e4 LT	301	if (ptab == NULL)
	302	goto done;
	303	}
	304
	305	for (n = 0; n < 256; n++)
	306	if (rtab->data[n] > qopt->buffer) break;
	307	max_size = (n << qopt->rate.cell_log)-1;
	308	if (ptab) {
	309	int size;
	310
	311	for (n = 0; n < 256; n++)
	312	if (ptab->data[n] > qopt->mtu) break;
	313	size = (n << qopt->peakrate.cell_log)-1;
	314	if (size < max_size) max_size = size;
	315	}
	316	if (max_size < 0)
	317	goto done;
	318
053cfed7	319	if (qopt->limit > 0) {
9f9afec4	320	if ((child = tbf_create_dflt_qdisc(sch, qopt->limit)) == NULL)
1da177e4 LT	321	goto done;
	322	}
	323
	324	sch_tree_lock(sch);
5e50da01 PM	325	if (child) {
5e50da01 PM	326	qdisc_tree_decrease_qlen(q->qdisc, q->qdisc->q.qlen);
053cfed7	327	qdisc_destroy(xchg(&q->qdisc, child));
5e50da01	328	}
1da177e4 LT	329	q->limit = qopt->limit;
	330	q->mtu = qopt->mtu;
	331	q->max_size = max_size;
	332	q->buffer = qopt->buffer;
	333	q->tokens = q->buffer;
	334	q->ptokens = q->mtu;
	335	rtab = xchg(&q->R_tab, rtab);
	336	ptab = xchg(&q->P_tab, ptab);
	337	sch_tree_unlock(sch);
	338	err = 0;
	339	done:
	340	if (rtab)
	341	qdisc_put_rtab(rtab);
	342	if (ptab)
	343	qdisc_put_rtab(ptab);
	344	return err;
	345	}
	346
1e90474c	347	static int tbf_init(struct Qdisc* sch, struct nlattr *opt)
1da177e4 LT	348	{
	349	struct tbf_sched_data *q = qdisc_priv(sch);
	350
	351	if (opt == NULL)
	352	return -EINVAL;
	353
3bebcda2	354	q->t_c = psched_get_time();
f7f593e3	355	qdisc_watchdog_init(&q->watchdog, sch);
1da177e4 LT	356	q->qdisc = &noop_qdisc;
	357
	358	return tbf_change(sch, opt);
	359	}
	360
	361	static void tbf_destroy(struct Qdisc *sch)
	362	{
	363	struct tbf_sched_data *q = qdisc_priv(sch);
	364
f7f593e3	365	qdisc_watchdog_cancel(&q->watchdog);
1da177e4 LT	366
	367	if (q->P_tab)
	368	qdisc_put_rtab(q->P_tab);
	369	if (q->R_tab)
	370	qdisc_put_rtab(q->R_tab);
	371
	372	qdisc_destroy(q->qdisc);
	373	}
	374
	375	static int tbf_dump(struct Qdisc sch, struct sk_buff skb)
	376	{
	377	struct tbf_sched_data *q = qdisc_priv(sch);
27a884dc	378	unsigned char *b = skb_tail_pointer(skb);
1e90474c	379	struct nlattr *nla;
1da177e4 LT	380	struct tc_tbf_qopt opt;
1da177e4 LT	381
1e90474c PM	382	nla = (struct nlattr*)b;
1e90474c PM	383	NLA_PUT(skb, TCA_OPTIONS, 0, NULL);
1da177e4 LT	384
	385	opt.limit = q->limit;
	386	opt.rate = q->R_tab->rate;
	387	if (q->P_tab)
	388	opt.peakrate = q->P_tab->rate;
	389	else
	390	memset(&opt.peakrate, 0, sizeof(opt.peakrate));
	391	opt.mtu = q->mtu;
	392	opt.buffer = q->buffer;
1e90474c PM	393	NLA_PUT(skb, TCA_TBF_PARMS, sizeof(opt), &opt);
1e90474c PM	394	nla->nla_len = skb_tail_pointer(skb) - b;
1da177e4 LT	395
	396	return skb->len;
	397
1e90474c	398	nla_put_failure:
dc5fc579	399	nlmsg_trim(skb, b);
1da177e4 LT	400	return -1;
	401	}
	402
	403	static int tbf_dump_class(struct Qdisc *sch, unsigned long cl,
	404	struct sk_buff skb, struct tcmsg tcm)
	405	{
	406	struct tbf_sched_data *q = qdisc_priv(sch);
	407
	408	if (cl != 1) /* only one class */
	409	return -ENOENT;
	410
	411	tcm->tcm_handle \|= TC_H_MIN(1);
	412	tcm->tcm_info = q->qdisc->handle;
	413
	414	return 0;
	415	}
	416
	417	static int tbf_graft(struct Qdisc sch, unsigned long arg, struct Qdisc new,
	418	struct Qdisc **old)
	419	{
	420	struct tbf_sched_data *q = qdisc_priv(sch);
	421
	422	if (new == NULL)
	423	new = &noop_qdisc;
	424
	425	sch_tree_lock(sch);
	426	*old = xchg(&q->qdisc, new);
5e50da01	427	qdisc_tree_decrease_qlen(old, (old)->q.qlen);
1da177e4	428	qdisc_reset(*old);
1da177e4 LT	429	sch_tree_unlock(sch);
	430
	431	return 0;
	432	}
	433
	434	static struct Qdisc tbf_leaf(struct Qdisc sch, unsigned long arg)
	435	{
	436	struct tbf_sched_data *q = qdisc_priv(sch);
	437	return q->qdisc;
	438	}
	439
	440	static unsigned long tbf_get(struct Qdisc *sch, u32 classid)
	441	{
	442	return 1;
	443	}
	444
	445	static void tbf_put(struct Qdisc *sch, unsigned long arg)
	446	{
	447	}
	448
10297b99	449	static int tbf_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
1e90474c	450	struct nlattr *tca, unsigned long arg)
1da177e4 LT	451	{
	452	return -ENOSYS;
	453	}
	454
	455	static int tbf_delete(struct Qdisc *sch, unsigned long arg)
	456	{
	457	return -ENOSYS;
	458	}
	459
	460	static void tbf_walk(struct Qdisc sch, struct qdisc_walker walker)
	461	{
	462	if (!walker->stop) {
	463	if (walker->count >= walker->skip)
	464	if (walker->fn(sch, 1, walker) < 0) {
	465	walker->stop = 1;
	466	return;
	467	}
	468	walker->count++;
	469	}
	470	}
	471
	472	static struct tcf_proto *tbf_find_tcf(struct Qdisc sch, unsigned long cl)
	473	{
	474	return NULL;
	475	}
	476
20fea08b	477	static const struct Qdisc_class_ops tbf_class_ops =
1da177e4 LT	478	{
	479	.graft = tbf_graft,
	480	.leaf = tbf_leaf,
	481	.get = tbf_get,
	482	.put = tbf_put,
	483	.change = tbf_change_class,
	484	.delete = tbf_delete,
	485	.walk = tbf_walk,
	486	.tcf_chain = tbf_find_tcf,
	487	.dump = tbf_dump_class,
	488	};
	489
20fea08b	490	static struct Qdisc_ops tbf_qdisc_ops __read_mostly = {
1da177e4 LT	491	.next = NULL,
	492	.cl_ops = &tbf_class_ops,
	493	.id = "tbf",
	494	.priv_size = sizeof(struct tbf_sched_data),
	495	.enqueue = tbf_enqueue,
	496	.dequeue = tbf_dequeue,
	497	.requeue = tbf_requeue,
	498	.drop = tbf_drop,
	499	.init = tbf_init,
	500	.reset = tbf_reset,
	501	.destroy = tbf_destroy,
	502	.change = tbf_change,
	503	.dump = tbf_dump,
	504	.owner = THIS_MODULE,
	505	};
	506
	507	static int __init tbf_module_init(void)
	508	{
	509	return register_qdisc(&tbf_qdisc_ops);
	510	}
	511
	512	static void __exit tbf_module_exit(void)
	513	{
	514	unregister_qdisc(&tbf_qdisc_ops);
	515	}
	516	module_init(tbf_module_init)
	517	module_exit(tbf_module_exit)
	518	MODULE_LICENSE("GPL");