[linux-2.6-block.git] / net / sched / sch_cbs.c

/*
 * net/sched/sch_cbs.c	Credit Based Shaper
 *
 *		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:	Vinicius Costa Gomes <vinicius.gomes@intel.com>
 *
 */

/* Credit Based Shaper (CBS)
 * =========================
 *
 * This is a simple rate-limiting shaper aimed at TSN applications on
 * systems with known traffic workloads.
 *
 * Its algorithm is defined by the IEEE 802.1Q-2014 Specification,
 * Section 8.6.8.2, and explained in more detail in the Annex L of the
 * same specification.
 *
 * There are four tunables to be considered:
 *
 *	'idleslope': Idleslope is the rate of credits that is
 *	accumulated (in kilobits per second) when there is at least
 *	one packet waiting for transmission. Packets are transmitted
 *	when the current value of credits is equal or greater than
 *	zero. When there is no packet to be transmitted the amount of
 *	credits is set to zero. This is the main tunable of the CBS
 *	algorithm.
 *
 *	'sendslope':
 *	Sendslope is the rate of credits that is depleted (it should be a
 *	negative number of kilobits per second) when a transmission is
 *	ocurring. It can be calculated as follows, (IEEE 802.1Q-2014 Section
 *	8.6.8.2 item g):
 *
 *	sendslope = idleslope - port_transmit_rate
 *
 *	'hicredit': Hicredit defines the maximum amount of credits (in
 *	bytes) that can be accumulated. Hicredit depends on the
 *	characteristics of interfering traffic,
 *	'max_interference_size' is the maximum size of any burst of
 *	traffic that can delay the transmission of a frame that is
 *	available for transmission for this traffic class, (IEEE
 *	802.1Q-2014 Annex L, Equation L-3):
 *
 *	hicredit = max_interference_size * (idleslope / port_transmit_rate)
 *
 *	'locredit': Locredit is the minimum amount of credits that can
 *	be reached. It is a function of the traffic flowing through
 *	this qdisc (IEEE 802.1Q-2014 Annex L, Equation L-2):
 *
 *	locredit = max_frame_size * (sendslope / port_transmit_rate)
 */

#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/sch_generic.h>
#include <net/pkt_sched.h>

#define BYTES_PER_KBIT (1000LL / 8)

struct cbs_sched_data {
	bool offload;
	int queue;
	s64 port_rate; /* in bytes/s */
	s64 last; /* timestamp in ns */
	s64 credits; /* in bytes */
	s32 locredit; /* in bytes */
	s32 hicredit; /* in bytes */
	s64 sendslope; /* in bytes/s */
	s64 idleslope; /* in bytes/s */
	struct qdisc_watchdog watchdog;
	int (*enqueue)(struct sk_buff *skb, struct Qdisc *sch,
		       struct sk_buff **to_free);
	struct sk_buff *(*dequeue)(struct Qdisc *sch);
	struct Qdisc *qdisc;
};

static int cbs_child_enqueue(struct sk_buff *skb, struct Qdisc *sch,
			     struct Qdisc *child,
			     struct sk_buff **to_free)
{
	int err;

	err = child->ops->enqueue(skb, child, to_free);
	if (err != NET_XMIT_SUCCESS)
		return err;

	qdisc_qstats_backlog_inc(sch, skb);
	sch->q.qlen++;

	return NET_XMIT_SUCCESS;
}

static int cbs_enqueue_offload(struct sk_buff *skb, struct Qdisc *sch,
			       struct sk_buff **to_free)
{
	struct cbs_sched_data *q = qdisc_priv(sch);
	struct Qdisc *qdisc = q->qdisc;

	return cbs_child_enqueue(skb, sch, qdisc, to_free);
}

static int cbs_enqueue_soft(struct sk_buff *skb, struct Qdisc *sch,
			    struct sk_buff **to_free)
{
	struct cbs_sched_data *q = qdisc_priv(sch);
	struct Qdisc *qdisc = q->qdisc;

	if (sch->q.qlen == 0 && q->credits > 0) {
		/* We need to stop accumulating credits when there's
		 * no enqueued packets and q->credits is positive.
		 */
		q->credits = 0;
		q->last = ktime_get_ns();
	}

	return cbs_child_enqueue(skb, sch, qdisc, to_free);
}

static int cbs_enqueue(struct sk_buff *skb, struct Qdisc *sch,
		       struct sk_buff **to_free)
{
	struct cbs_sched_data *q = qdisc_priv(sch);

	return q->enqueue(skb, sch, to_free);
}

/* timediff is in ns, slope is in bytes/s */
static s64 timediff_to_credits(s64 timediff, s64 slope)
{
	return div64_s64(timediff * slope, NSEC_PER_SEC);
}

static s64 delay_from_credits(s64 credits, s64 slope)
{
	if (unlikely(slope == 0))
		return S64_MAX;

	return div64_s64(-credits * NSEC_PER_SEC, slope);
}

static s64 credits_from_len(unsigned int len, s64 slope, s64 port_rate)
{
	if (unlikely(port_rate == 0))
		return S64_MAX;

	return div64_s64(len * slope, port_rate);
}

static struct sk_buff *cbs_child_dequeue(struct Qdisc *sch, struct Qdisc *child)
{
	struct sk_buff *skb;

	skb = child->ops->dequeue(child);
	if (!skb)
		return NULL;

	qdisc_qstats_backlog_dec(sch, skb);
	qdisc_bstats_update(sch, skb);
	sch->q.qlen--;

	return skb;
}

static struct sk_buff *cbs_dequeue_soft(struct Qdisc *sch)
{
	struct cbs_sched_data *q = qdisc_priv(sch);
	struct Qdisc *qdisc = q->qdisc;
	s64 now = ktime_get_ns();
	struct sk_buff *skb;
	s64 credits;
	int len;

	if (q->credits < 0) {
		credits = timediff_to_credits(now - q->last, q->idleslope);

		credits = q->credits + credits;
		q->credits = min_t(s64, credits, q->hicredit);

		if (q->credits < 0) {
			s64 delay;

			delay = delay_from_credits(q->credits, q->idleslope);
			qdisc_watchdog_schedule_ns(&q->watchdog, now + delay);

			q->last = now;

			return NULL;
		}
	}
	skb = cbs_child_dequeue(sch, qdisc);
	if (!skb)
		return NULL;

	len = qdisc_pkt_len(skb);

	/* As sendslope is a negative number, this will decrease the
	 * amount of q->credits.
	 */
	credits = credits_from_len(len, q->sendslope, q->port_rate);
	credits += q->credits;

	q->credits = max_t(s64, credits, q->locredit);
	q->last = now;

	return skb;
}

static struct sk_buff *cbs_dequeue_offload(struct Qdisc *sch)
{
	struct cbs_sched_data *q = qdisc_priv(sch);
	struct Qdisc *qdisc = q->qdisc;

	return cbs_child_dequeue(sch, qdisc);
}

static struct sk_buff *cbs_dequeue(struct Qdisc *sch)
{
	struct cbs_sched_data *q = qdisc_priv(sch);

	return q->dequeue(sch);
}

static const struct nla_policy cbs_policy[TCA_CBS_MAX + 1] = {
	[TCA_CBS_PARMS]	= { .len = sizeof(struct tc_cbs_qopt) },
};

static void cbs_disable_offload(struct net_device *dev,
				struct cbs_sched_data *q)
{
	struct tc_cbs_qopt_offload cbs = { };
	const struct net_device_ops *ops;
	int err;

	if (!q->offload)
		return;

	q->enqueue = cbs_enqueue_soft;
	q->dequeue = cbs_dequeue_soft;

	ops = dev->netdev_ops;
	if (!ops->ndo_setup_tc)
		return;

	cbs.queue = q->queue;
	cbs.enable = 0;

	err = ops->ndo_setup_tc(dev, TC_SETUP_QDISC_CBS, &cbs);
	if (err < 0)
		pr_warn("Couldn't disable CBS offload for queue %d\n",
			cbs.queue);
}

static int cbs_enable_offload(struct net_device *dev, struct cbs_sched_data *q,
			      const struct tc_cbs_qopt *opt,
			      struct netlink_ext_ack *extack)
{
	const struct net_device_ops *ops = dev->netdev_ops;
	struct tc_cbs_qopt_offload cbs = { };
	int err;

	if (!ops->ndo_setup_tc) {
		NL_SET_ERR_MSG(extack, "Specified device does not support cbs offload");
		return -EOPNOTSUPP;
	}

	cbs.queue = q->queue;

	cbs.enable = 1;
	cbs.hicredit = opt->hicredit;
	cbs.locredit = opt->locredit;
	cbs.idleslope = opt->idleslope;
	cbs.sendslope = opt->sendslope;

	err = ops->ndo_setup_tc(dev, TC_SETUP_QDISC_CBS, &cbs);
	if (err < 0) {
		NL_SET_ERR_MSG(extack, "Specified device failed to setup cbs hardware offload");
		return err;
	}

	q->enqueue = cbs_enqueue_offload;
	q->dequeue = cbs_dequeue_offload;

	return 0;
}

static int cbs_change(struct Qdisc *sch, struct nlattr *opt,
		      struct netlink_ext_ack *extack)
{
	struct cbs_sched_data *q = qdisc_priv(sch);
	struct net_device *dev = qdisc_dev(sch);
	struct nlattr *tb[TCA_CBS_MAX + 1];
	struct tc_cbs_qopt *qopt;
	int err;

	err = nla_parse_nested(tb, TCA_CBS_MAX, opt, cbs_policy, extack);
	if (err < 0)
		return err;

	if (!tb[TCA_CBS_PARMS]) {
		NL_SET_ERR_MSG(extack, "Missing CBS parameter which are mandatory");
		return -EINVAL;
	}

	qopt = nla_data(tb[TCA_CBS_PARMS]);

	if (!qopt->offload) {
		struct ethtool_link_ksettings ecmd;
		s64 link_speed;

		if (!__ethtool_get_link_ksettings(dev, &ecmd))
			link_speed = ecmd.base.speed;
		else
			link_speed = SPEED_1000;

		q->port_rate = link_speed * 1000 * BYTES_PER_KBIT;

		cbs_disable_offload(dev, q);
	} else {
		err = cbs_enable_offload(dev, q, qopt, extack);
		if (err < 0)
			return err;
	}

	/* Everything went OK, save the parameters used. */
	q->hicredit = qopt->hicredit;
	q->locredit = qopt->locredit;
	q->idleslope = qopt->idleslope * BYTES_PER_KBIT;
	q->sendslope = qopt->sendslope * BYTES_PER_KBIT;
	q->offload = qopt->offload;

	return 0;
}

static int cbs_init(struct Qdisc *sch, struct nlattr *opt,
		    struct netlink_ext_ack *extack)
{
	struct cbs_sched_data *q = qdisc_priv(sch);
	struct net_device *dev = qdisc_dev(sch);

	if (!opt) {
		NL_SET_ERR_MSG(extack, "Missing CBS qdisc options  which are mandatory");
		return -EINVAL;
	}

	q->qdisc = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops,
				     sch->handle, extack);
	if (!q->qdisc)
		return -ENOMEM;

	qdisc_hash_add(q->qdisc, false);

	q->queue = sch->dev_queue - netdev_get_tx_queue(dev, 0);

	q->enqueue = cbs_enqueue_soft;
	q->dequeue = cbs_dequeue_soft;

	qdisc_watchdog_init(&q->watchdog, sch);

	return cbs_change(sch, opt, extack);
}

static void cbs_destroy(struct Qdisc *sch)
{
	struct cbs_sched_data *q = qdisc_priv(sch);
	struct net_device *dev = qdisc_dev(sch);

	qdisc_watchdog_cancel(&q->watchdog);

	cbs_disable_offload(dev, q);

	if (q->qdisc)
		qdisc_put(q->qdisc);
}

static int cbs_dump(struct Qdisc *sch, struct sk_buff *skb)
{
	struct cbs_sched_data *q = qdisc_priv(sch);
	struct tc_cbs_qopt opt = { };
	struct nlattr *nest;

	nest = nla_nest_start(skb, TCA_OPTIONS);
	if (!nest)
		goto nla_put_failure;

	opt.hicredit = q->hicredit;
	opt.locredit = q->locredit;
	opt.sendslope = div64_s64(q->sendslope, BYTES_PER_KBIT);
	opt.idleslope = div64_s64(q->idleslope, BYTES_PER_KBIT);
	opt.offload = q->offload;

	if (nla_put(skb, TCA_CBS_PARMS, sizeof(opt), &opt))
		goto nla_put_failure;

	return nla_nest_end(skb, nest);

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

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

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

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

	return 0;
}

static int cbs_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new,
		     struct Qdisc **old, struct netlink_ext_ack *extack)
{
	struct cbs_sched_data *q = qdisc_priv(sch);

	if (!new) {
		new = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops,
					sch->handle, NULL);
		if (!new)
			new = &noop_qdisc;
	}

	*old = qdisc_replace(sch, new, &q->qdisc);
	return 0;
}

static struct Qdisc *cbs_leaf(struct Qdisc *sch, unsigned long arg)
{
	struct cbs_sched_data *q = qdisc_priv(sch);

	return q->qdisc;
}

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

static void cbs_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 cbs_class_ops = {
	.graft		=	cbs_graft,
	.leaf		=	cbs_leaf,
	.find		=	cbs_find,
	.walk		=	cbs_walk,
	.dump		=	cbs_dump_class,
};

static struct Qdisc_ops cbs_qdisc_ops __read_mostly = {
	.id		=	"cbs",
	.cl_ops		=	&cbs_class_ops,
	.priv_size	=	sizeof(struct cbs_sched_data),
	.enqueue	=	cbs_enqueue,
	.dequeue	=	cbs_dequeue,
	.peek		=	qdisc_peek_dequeued,
	.init		=	cbs_init,
	.reset		=	qdisc_reset_queue,
	.destroy	=	cbs_destroy,
	.change		=	cbs_change,
	.dump		=	cbs_dump,
	.owner		=	THIS_MODULE,
};

static int __init cbs_module_init(void)
{
	return register_qdisc(&cbs_qdisc_ops);
}

static void __exit cbs_module_exit(void)
{
	unregister_qdisc(&cbs_qdisc_ops);
}
module_init(cbs_module_init)
module_exit(cbs_module_exit)
MODULE_LICENSE("GPL");
Commit	Line	Data
585d763a VCG	1	/*
	2	* net/sched/sch_cbs.c Credit Based Shaper
	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: Vinicius Costa Gomes <vinicius.gomes@intel.com>
	10	*
	11	*/
	12
	13	/* Credit Based Shaper (CBS)
	14	* =========================
	15	*
	16	* This is a simple rate-limiting shaper aimed at TSN applications on
	17	* systems with known traffic workloads.
	18	*
	19	* Its algorithm is defined by the IEEE 802.1Q-2014 Specification,
	20	* Section 8.6.8.2, and explained in more detail in the Annex L of the
	21	* same specification.
	22	*
	23	* There are four tunables to be considered:
	24	*
	25	* 'idleslope': Idleslope is the rate of credits that is
	26	* accumulated (in kilobits per second) when there is at least
	27	* one packet waiting for transmission. Packets are transmitted
	28	* when the current value of credits is equal or greater than
	29	* zero. When there is no packet to be transmitted the amount of
	30	* credits is set to zero. This is the main tunable of the CBS
	31	* algorithm.
	32	*
	33	* 'sendslope':
	34	* Sendslope is the rate of credits that is depleted (it should be a
	35	* negative number of kilobits per second) when a transmission is
	36	* ocurring. It can be calculated as follows, (IEEE 802.1Q-2014 Section
	37	* 8.6.8.2 item g):
	38	*
	39	* sendslope = idleslope - port_transmit_rate
	40	*
	41	* 'hicredit': Hicredit defines the maximum amount of credits (in
	42	* bytes) that can be accumulated. Hicredit depends on the
	43	* characteristics of interfering traffic,
	44	* 'max_interference_size' is the maximum size of any burst of
	45	* traffic that can delay the transmission of a frame that is
	46	* available for transmission for this traffic class, (IEEE
	47	* 802.1Q-2014 Annex L, Equation L-3):
	48	*
	49	* hicredit = max_interference_size * (idleslope / port_transmit_rate)
	50	*
	51	* 'locredit': Locredit is the minimum amount of credits that can
	52	* be reached. It is a function of the traffic flowing through
	53	* this qdisc (IEEE 802.1Q-2014 Annex L, Equation L-2):
	54	*
	55	* locredit = max_frame_size * (sendslope / port_transmit_rate)
	56	*/
	57
	58	#include <linux/module.h>
	59	#include <linux/types.h>
	60	#include <linux/kernel.h>
	61	#include <linux/string.h>
	62	#include <linux/errno.h>
	63	#include <linux/skbuff.h>
	64	#include <net/netlink.h>
65	#include <net/sch_generic.h>
66	#include <net/pkt_sched.h>
67
68	#define BYTES_PER_KBIT (1000LL / 8)
69
70	struct cbs_sched_data {
3d0bd028 VCG	71	bool offload;
3d0bd028 VCG	72	int queue;
585d763a VCG	73	s64 port_rate; /* in bytes/s */
	74	s64 last; /* timestamp in ns */
	75	s64 credits; /* in bytes */
	76	s32 locredit; /* in bytes */
	77	s32 hicredit; /* in bytes */
	78	s64 sendslope; /* in bytes/s */
	79	s64 idleslope; /* in bytes/s */
	80	struct qdisc_watchdog watchdog;
990e35ec VCG	81	int (enqueue)(struct sk_buff skb, struct Qdisc *sch,
990e35ec VCG	82	struct sk_buff **to_free);
585d763a	83	struct sk_buff (dequeue)(struct Qdisc *sch);
990e35ec	84	struct Qdisc *qdisc;
585d763a VCG	85	};
585d763a VCG	86
990e35ec VCG	87	static int cbs_child_enqueue(struct sk_buff skb, struct Qdisc sch,
	88	struct Qdisc *child,
	89	struct sk_buff **to_free)
3d0bd028	90	{
990e35ec VCG	91	int err;
	92
	93	err = child->ops->enqueue(skb, child, to_free);
	94	if (err != NET_XMIT_SUCCESS)
	95	return err;
	96
	97	qdisc_qstats_backlog_inc(sch, skb);
	98	sch->q.qlen++;
	99
	100	return NET_XMIT_SUCCESS;
3d0bd028 VCG	101	}
3d0bd028 VCG	102
990e35ec VCG	103	static int cbs_enqueue_offload(struct sk_buff skb, struct Qdisc sch,
990e35ec VCG	104	struct sk_buff **to_free)
585d763a VCG	105	{
585d763a VCG	106	struct cbs_sched_data *q = qdisc_priv(sch);
990e35ec VCG	107	struct Qdisc *qdisc = q->qdisc;
	108
	109	return cbs_child_enqueue(skb, sch, qdisc, to_free);
	110	}
	111
	112	static int cbs_enqueue_soft(struct sk_buff skb, struct Qdisc sch,
	113	struct sk_buff **to_free)
	114	{
	115	struct cbs_sched_data *q = qdisc_priv(sch);
	116	struct Qdisc *qdisc = q->qdisc;
585d763a VCG	117
	118	if (sch->q.qlen == 0 && q->credits > 0) {
	119	/* We need to stop accumulating credits when there's
	120	* no enqueued packets and q->credits is positive.
	121	*/
	122	q->credits = 0;
	123	q->last = ktime_get_ns();
	124	}
	125
990e35ec	126	return cbs_child_enqueue(skb, sch, qdisc, to_free);
585d763a VCG	127	}
	128
	129	static int cbs_enqueue(struct sk_buff skb, struct Qdisc sch,
	130	struct sk_buff **to_free)
	131	{
	132	struct cbs_sched_data *q = qdisc_priv(sch);
	133
990e35ec	134	return q->enqueue(skb, sch, to_free);
585d763a VCG	135	}
	136
	137	/* timediff is in ns, slope is in bytes/s */
	138	static s64 timediff_to_credits(s64 timediff, s64 slope)
	139	{
	140	return div64_s64(timediff * slope, NSEC_PER_SEC);
	141	}
	142
	143	static s64 delay_from_credits(s64 credits, s64 slope)
	144	{
	145	if (unlikely(slope == 0))
	146	return S64_MAX;
	147
	148	return div64_s64(-credits * NSEC_PER_SEC, slope);
	149	}
	150
	151	static s64 credits_from_len(unsigned int len, s64 slope, s64 port_rate)
	152	{
	153	if (unlikely(port_rate == 0))
	154	return S64_MAX;
	155
	156	return div64_s64(len * slope, port_rate);
	157	}
	158
990e35ec VCG	159	static struct sk_buff cbs_child_dequeue(struct Qdisc sch, struct Qdisc *child)
	160	{
	161	struct sk_buff *skb;
	162
	163	skb = child->ops->dequeue(child);
	164	if (!skb)
	165	return NULL;
	166
	167	qdisc_qstats_backlog_dec(sch, skb);
	168	qdisc_bstats_update(sch, skb);
	169	sch->q.qlen--;
	170
	171	return skb;
	172	}
	173
585d763a VCG	174	static struct sk_buff cbs_dequeue_soft(struct Qdisc sch)
	175	{
	176	struct cbs_sched_data *q = qdisc_priv(sch);
990e35ec	177	struct Qdisc *qdisc = q->qdisc;
585d763a VCG	178	s64 now = ktime_get_ns();
	179	struct sk_buff *skb;
	180	s64 credits;
	181	int len;
	182
	183	if (q->credits < 0) {
	184	credits = timediff_to_credits(now - q->last, q->idleslope);
	185
	186	credits = q->credits + credits;
	187	q->credits = min_t(s64, credits, q->hicredit);
	188
	189	if (q->credits < 0) {
	190	s64 delay;
	191
	192	delay = delay_from_credits(q->credits, q->idleslope);
	193	qdisc_watchdog_schedule_ns(&q->watchdog, now + delay);
	194
	195	q->last = now;
	196
	197	return NULL;
	198	}
	199	}
990e35ec	200	skb = cbs_child_dequeue(sch, qdisc);
585d763a VCG	201	if (!skb)
	202	return NULL;
	203
	204	len = qdisc_pkt_len(skb);
	205
	206	/* As sendslope is a negative number, this will decrease the
	207	* amount of q->credits.
	208	*/
	209	credits = credits_from_len(len, q->sendslope, q->port_rate);
	210	credits += q->credits;
	211
	212	q->credits = max_t(s64, credits, q->locredit);
	213	q->last = now;
	214
	215	return skb;
	216	}
	217
3d0bd028 VCG	218	static struct sk_buff cbs_dequeue_offload(struct Qdisc sch)
3d0bd028 VCG	219	{
990e35ec VCG	220	struct cbs_sched_data *q = qdisc_priv(sch);
	221	struct Qdisc *qdisc = q->qdisc;
	222
	223	return cbs_child_dequeue(sch, qdisc);
3d0bd028 VCG	224	}
3d0bd028 VCG	225
585d763a VCG	226	static struct sk_buff cbs_dequeue(struct Qdisc sch)
	227	{
	228	struct cbs_sched_data *q = qdisc_priv(sch);
	229
	230	return q->dequeue(sch);
	231	}
	232
	233	static const struct nla_policy cbs_policy[TCA_CBS_MAX + 1] = {
	234	[TCA_CBS_PARMS] = { .len = sizeof(struct tc_cbs_qopt) },
	235	};
	236
3d0bd028 VCG	237	static void cbs_disable_offload(struct net_device *dev,
	238	struct cbs_sched_data *q)
	239	{
	240	struct tc_cbs_qopt_offload cbs = { };
	241	const struct net_device_ops *ops;
	242	int err;
	243
	244	if (!q->offload)
	245	return;
	246
	247	q->enqueue = cbs_enqueue_soft;
	248	q->dequeue = cbs_dequeue_soft;
	249
	250	ops = dev->netdev_ops;
	251	if (!ops->ndo_setup_tc)
	252	return;
	253
	254	cbs.queue = q->queue;
	255	cbs.enable = 0;
	256
8521db4c	257	err = ops->ndo_setup_tc(dev, TC_SETUP_QDISC_CBS, &cbs);
3d0bd028 VCG	258	if (err < 0)
	259	pr_warn("Couldn't disable CBS offload for queue %d\n",
	260	cbs.queue);
	261	}
	262
	263	static int cbs_enable_offload(struct net_device dev, struct cbs_sched_data q,
710fb396 AA	264	const struct tc_cbs_qopt *opt,
710fb396 AA	265	struct netlink_ext_ack *extack)
3d0bd028 VCG	266	{
	267	const struct net_device_ops *ops = dev->netdev_ops;
	268	struct tc_cbs_qopt_offload cbs = { };
	269	int err;
	270
710fb396 AA	271	if (!ops->ndo_setup_tc) {
710fb396 AA	272	NL_SET_ERR_MSG(extack, "Specified device does not support cbs offload");
3d0bd028	273	return -EOPNOTSUPP;
710fb396	274	}
3d0bd028 VCG	275
	276	cbs.queue = q->queue;
	277
	278	cbs.enable = 1;
	279	cbs.hicredit = opt->hicredit;
	280	cbs.locredit = opt->locredit;
	281	cbs.idleslope = opt->idleslope;
	282	cbs.sendslope = opt->sendslope;
	283
8521db4c	284	err = ops->ndo_setup_tc(dev, TC_SETUP_QDISC_CBS, &cbs);
710fb396 AA	285	if (err < 0) {
710fb396 AA	286	NL_SET_ERR_MSG(extack, "Specified device failed to setup cbs hardware offload");
3d0bd028	287	return err;
710fb396	288	}
3d0bd028 VCG	289
	290	q->enqueue = cbs_enqueue_offload;
	291	q->dequeue = cbs_dequeue_offload;
	292
	293	return 0;
	294	}
	295
2030721c AA	296	static int cbs_change(struct Qdisc sch, struct nlattr opt,
2030721c AA	297	struct netlink_ext_ack *extack)
585d763a VCG	298	{
	299	struct cbs_sched_data *q = qdisc_priv(sch);
	300	struct net_device *dev = qdisc_dev(sch);
	301	struct nlattr *tb[TCA_CBS_MAX + 1];
585d763a	302	struct tc_cbs_qopt *qopt;
585d763a VCG	303	int err;
585d763a VCG	304
710fb396	305	err = nla_parse_nested(tb, TCA_CBS_MAX, opt, cbs_policy, extack);
585d763a VCG	306	if (err < 0)
	307	return err;
	308
710fb396 AA	309	if (!tb[TCA_CBS_PARMS]) {
710fb396 AA	310	NL_SET_ERR_MSG(extack, "Missing CBS parameter which are mandatory");
585d763a	311	return -EINVAL;
710fb396	312	}
585d763a VCG	313
	314	qopt = nla_data(tb[TCA_CBS_PARMS]);
	315
3d0bd028 VCG	316	if (!qopt->offload) {
	317	struct ethtool_link_ksettings ecmd;
	318	s64 link_speed;
585d763a	319
3d0bd028 VCG	320	if (!__ethtool_get_link_ksettings(dev, &ecmd))
	321	link_speed = ecmd.base.speed;
	322	else
	323	link_speed = SPEED_1000;
585d763a	324
3d0bd028	325	q->port_rate = link_speed * 1000 * BYTES_PER_KBIT;
585d763a	326
3d0bd028 VCG	327	cbs_disable_offload(dev, q);
3d0bd028 VCG	328	} else {
710fb396	329	err = cbs_enable_offload(dev, q, qopt, extack);
3d0bd028 VCG	330	if (err < 0)
	331	return err;
	332	}
585d763a	333
3d0bd028	334	/* Everything went OK, save the parameters used. */
585d763a VCG	335	q->hicredit = qopt->hicredit;
	336	q->locredit = qopt->locredit;
	337	q->idleslope = qopt->idleslope * BYTES_PER_KBIT;
	338	q->sendslope = qopt->sendslope * BYTES_PER_KBIT;
3d0bd028	339	q->offload = qopt->offload;
585d763a VCG	340
	341	return 0;
	342	}
	343
e63d7dfd AA	344	static int cbs_init(struct Qdisc sch, struct nlattr opt,
e63d7dfd AA	345	struct netlink_ext_ack *extack)
585d763a VCG	346	{
585d763a VCG	347	struct cbs_sched_data *q = qdisc_priv(sch);
3d0bd028	348	struct net_device *dev = qdisc_dev(sch);
585d763a	349
710fb396 AA	350	if (!opt) {
710fb396 AA	351	NL_SET_ERR_MSG(extack, "Missing CBS qdisc options which are mandatory");
585d763a	352	return -EINVAL;
710fb396	353	}
585d763a	354
990e35ec VCG	355	q->qdisc = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops,
	356	sch->handle, extack);
	357	if (!q->qdisc)
	358	return -ENOMEM;
	359
	360	qdisc_hash_add(q->qdisc, false);
	361
3d0bd028 VCG	362	q->queue = sch->dev_queue - netdev_get_tx_queue(dev, 0);
	363
	364	q->enqueue = cbs_enqueue_soft;
	365	q->dequeue = cbs_dequeue_soft;
	366
585d763a VCG	367	qdisc_watchdog_init(&q->watchdog, sch);
585d763a VCG	368
2030721c	369	return cbs_change(sch, opt, extack);
585d763a VCG	370	}
	371
	372	static void cbs_destroy(struct Qdisc *sch)
	373	{
	374	struct cbs_sched_data *q = qdisc_priv(sch);
3d0bd028	375	struct net_device *dev = qdisc_dev(sch);
585d763a VCG	376
585d763a VCG	377	qdisc_watchdog_cancel(&q->watchdog);
3d0bd028 VCG	378
3d0bd028 VCG	379	cbs_disable_offload(dev, q);
990e35ec VCG	380
990e35ec VCG	381	if (q->qdisc)
86bd446b	382	qdisc_put(q->qdisc);
585d763a VCG	383	}
	384
	385	static int cbs_dump(struct Qdisc sch, struct sk_buff skb)
	386	{
	387	struct cbs_sched_data *q = qdisc_priv(sch);
	388	struct tc_cbs_qopt opt = { };
	389	struct nlattr *nest;
	390
	391	nest = nla_nest_start(skb, TCA_OPTIONS);
	392	if (!nest)
	393	goto nla_put_failure;
	394
	395	opt.hicredit = q->hicredit;
	396	opt.locredit = q->locredit;
	397	opt.sendslope = div64_s64(q->sendslope, BYTES_PER_KBIT);
	398	opt.idleslope = div64_s64(q->idleslope, BYTES_PER_KBIT);
3d0bd028	399	opt.offload = q->offload;
585d763a VCG	400
	401	if (nla_put(skb, TCA_CBS_PARMS, sizeof(opt), &opt))
	402	goto nla_put_failure;
	403
	404	return nla_nest_end(skb, nest);
	405
	406	nla_put_failure:
	407	nla_nest_cancel(skb, nest);
	408	return -1;
	409	}
	410
990e35ec VCG	411	static int cbs_dump_class(struct Qdisc *sch, unsigned long cl,
	412	struct sk_buff skb, struct tcmsg tcm)
	413	{
	414	struct cbs_sched_data *q = qdisc_priv(sch);
	415
	416	if (cl != 1 \|\| !q->qdisc) /* only one class */
	417	return -ENOENT;
	418
	419	tcm->tcm_handle \|= TC_H_MIN(1);
	420	tcm->tcm_info = q->qdisc->handle;
	421
	422	return 0;
	423	}
	424
	425	static int cbs_graft(struct Qdisc sch, unsigned long arg, struct Qdisc new,
	426	struct Qdisc *old, struct netlink_ext_ack extack)
	427	{
	428	struct cbs_sched_data *q = qdisc_priv(sch);
	429
	430	if (!new) {
	431	new = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops,
	432	sch->handle, NULL);
	433	if (!new)
	434	new = &noop_qdisc;
	435	}
	436
	437	*old = qdisc_replace(sch, new, &q->qdisc);
	438	return 0;
	439	}
	440
	441	static struct Qdisc cbs_leaf(struct Qdisc sch, unsigned long arg)
	442	{
	443	struct cbs_sched_data *q = qdisc_priv(sch);
	444
	445	return q->qdisc;
	446	}
	447
	448	static unsigned long cbs_find(struct Qdisc *sch, u32 classid)
	449	{
	450	return 1;
	451	}
	452
	453	static void cbs_walk(struct Qdisc sch, struct qdisc_walker walker)
	454	{
	455	if (!walker->stop) {
	456	if (walker->count >= walker->skip) {
	457	if (walker->fn(sch, 1, walker) < 0) {
	458	walker->stop = 1;
	459	return;
	460	}
	461	}
	462	walker->count++;
	463	}
	464	}
	465
	466	static const struct Qdisc_class_ops cbs_class_ops = {
	467	.graft = cbs_graft,
	468	.leaf = cbs_leaf,
	469	.find = cbs_find,
	470	.walk = cbs_walk,
	471	.dump = cbs_dump_class,
	472	};
	473
585d763a VCG	474	static struct Qdisc_ops cbs_qdisc_ops __read_mostly = {
585d763a VCG	475	.id = "cbs",
990e35ec	476	.cl_ops = &cbs_class_ops,
585d763a VCG	477	.priv_size = sizeof(struct cbs_sched_data),
	478	.enqueue = cbs_enqueue,
	479	.dequeue = cbs_dequeue,
	480	.peek = qdisc_peek_dequeued,
	481	.init = cbs_init,
	482	.reset = qdisc_reset_queue,
	483	.destroy = cbs_destroy,
	484	.change = cbs_change,
	485	.dump = cbs_dump,
	486	.owner = THIS_MODULE,
	487	};
	488
	489	static int __init cbs_module_init(void)
	490	{
	491	return register_qdisc(&cbs_qdisc_ops);
	492	}
	493
	494	static void __exit cbs_module_exit(void)
	495	{
	496	unregister_qdisc(&cbs_qdisc_ops);
	497	}
	498	module_init(cbs_module_init)
	499	module_exit(cbs_module_exit)
	500	MODULE_LICENSE("GPL");