[linux-block.git] / net / sched / sch_red.c

/*
 * net/sched/sch_red.c	Random Early Detection 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>
 *
 * Changes:
 * J Hadi Salim 980914:	computation fixes
 * Alexey Makarenko <makar@phoenix.kharkov.ua> 990814: qave on idle link was calculated incorrectly.
 * J Hadi Salim 980816:  ECN support
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/skbuff.h>
#include <net/pkt_sched.h>
#include <net/pkt_cls.h>
#include <net/inet_ecn.h>
#include <net/red.h>


/*	Parameters, settable by user:
	-----------------------------

	limit		- bytes (must be > qth_max + burst)

	Hard limit on queue length, should be chosen >qth_max
	to allow packet bursts. This parameter does not
	affect the algorithms behaviour and can be chosen
	arbitrarily high (well, less than ram size)
	Really, this limit will never be reached
	if RED works correctly.
 */

struct red_sched_data {
	u32			limit;		/* HARD maximal queue length */
	unsigned char		flags;
	struct timer_list	adapt_timer;
	struct Qdisc		*sch;
	struct red_parms	parms;
	struct red_vars		vars;
	struct red_stats	stats;
	struct Qdisc		*qdisc;
};

static inline int red_use_ecn(struct red_sched_data *q)
{
	return q->flags & TC_RED_ECN;
}

static inline int red_use_harddrop(struct red_sched_data *q)
{
	return q->flags & TC_RED_HARDDROP;
}

static int red_enqueue(struct sk_buff *skb, struct Qdisc *sch,
		       struct sk_buff **to_free)
{
	struct red_sched_data *q = qdisc_priv(sch);
	struct Qdisc *child = q->qdisc;
	int ret;

	q->vars.qavg = red_calc_qavg(&q->parms,
				     &q->vars,
				     child->qstats.backlog);

	if (red_is_idling(&q->vars))
		red_end_of_idle_period(&q->vars);

	switch (red_action(&q->parms, &q->vars, q->vars.qavg)) {
	case RED_DONT_MARK:
		break;

	case RED_PROB_MARK:
		qdisc_qstats_overlimit(sch);
		if (!red_use_ecn(q) || !INET_ECN_set_ce(skb)) {
			q->stats.prob_drop++;
			goto congestion_drop;
		}

		q->stats.prob_mark++;
		break;

	case RED_HARD_MARK:
		qdisc_qstats_overlimit(sch);
		if (red_use_harddrop(q) || !red_use_ecn(q) ||
		    !INET_ECN_set_ce(skb)) {
			q->stats.forced_drop++;
			goto congestion_drop;
		}

		q->stats.forced_mark++;
		break;
	}

	ret = qdisc_enqueue(skb, child, to_free);
	if (likely(ret == NET_XMIT_SUCCESS)) {
		qdisc_qstats_backlog_inc(sch, skb);
		sch->q.qlen++;
	} else if (net_xmit_drop_count(ret)) {
		q->stats.pdrop++;
		qdisc_qstats_drop(sch);
	}
	return ret;

congestion_drop:
	qdisc_drop(skb, sch, to_free);
	return NET_XMIT_CN;
}

static struct sk_buff *red_dequeue(struct Qdisc *sch)
{
	struct sk_buff *skb;
	struct red_sched_data *q = qdisc_priv(sch);
	struct Qdisc *child = q->qdisc;

	skb = child->dequeue(child);
	if (skb) {
		qdisc_bstats_update(sch, skb);
		qdisc_qstats_backlog_dec(sch, skb);
		sch->q.qlen--;
	} else {
		if (!red_is_idling(&q->vars))
			red_start_of_idle_period(&q->vars);
	}
	return skb;
}

static struct sk_buff *red_peek(struct Qdisc *sch)
{
	struct red_sched_data *q = qdisc_priv(sch);
	struct Qdisc *child = q->qdisc;

	return child->ops->peek(child);
}

static void red_reset(struct Qdisc *sch)
{
	struct red_sched_data *q = qdisc_priv(sch);

	qdisc_reset(q->qdisc);
	sch->qstats.backlog = 0;
	sch->q.qlen = 0;
	red_restart(&q->vars);
}

static int red_offload(struct Qdisc *sch, bool enable)
{
	struct red_sched_data *q = qdisc_priv(sch);
	struct net_device *dev = qdisc_dev(sch);
	struct tc_red_qopt_offload opt = {
		.handle = sch->handle,
		.parent = sch->parent,
	};

	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
		return -EOPNOTSUPP;

	if (enable) {
		opt.command = TC_RED_REPLACE;
		opt.set.min = q->parms.qth_min >> q->parms.Wlog;
		opt.set.max = q->parms.qth_max >> q->parms.Wlog;
		opt.set.probability = q->parms.max_P;
		opt.set.is_ecn = red_use_ecn(q);
	} else {
		opt.command = TC_RED_DESTROY;
	}

	return dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED, &opt);
}

static void red_destroy(struct Qdisc *sch)
{
	struct red_sched_data *q = qdisc_priv(sch);

	del_timer_sync(&q->adapt_timer);
	red_offload(sch, false);
	qdisc_destroy(q->qdisc);
}

static const struct nla_policy red_policy[TCA_RED_MAX + 1] = {
	[TCA_RED_PARMS]	= { .len = sizeof(struct tc_red_qopt) },
	[TCA_RED_STAB]	= { .len = RED_STAB_SIZE },
	[TCA_RED_MAX_P] = { .type = NLA_U32 },
};

static int red_change(struct Qdisc *sch, struct nlattr *opt,
		      struct netlink_ext_ack *extack)
{
	struct red_sched_data *q = qdisc_priv(sch);
	struct nlattr *tb[TCA_RED_MAX + 1];
	struct tc_red_qopt *ctl;
	struct Qdisc *child = NULL;
	int err;
	u32 max_P;

	if (opt == NULL)
		return -EINVAL;

	err = nla_parse_nested(tb, TCA_RED_MAX, opt, red_policy, NULL);
	if (err < 0)
		return err;

	if (tb[TCA_RED_PARMS] == NULL ||
	    tb[TCA_RED_STAB] == NULL)
		return -EINVAL;

	max_P = tb[TCA_RED_MAX_P] ? nla_get_u32(tb[TCA_RED_MAX_P]) : 0;

	ctl = nla_data(tb[TCA_RED_PARMS]);
	if (!red_check_params(ctl->qth_min, ctl->qth_max, ctl->Wlog))
		return -EINVAL;

	if (ctl->limit > 0) {
		child = fifo_create_dflt(sch, &bfifo_qdisc_ops, ctl->limit,
					 extack);
		if (IS_ERR(child))
			return PTR_ERR(child);
	}

	if (child != &noop_qdisc)
		qdisc_hash_add(child, true);
	sch_tree_lock(sch);
	q->flags = ctl->flags;
	q->limit = ctl->limit;
	if (child) {
		qdisc_tree_reduce_backlog(q->qdisc, q->qdisc->q.qlen,
					  q->qdisc->qstats.backlog);
		qdisc_destroy(q->qdisc);
		q->qdisc = child;
	}

	red_set_parms(&q->parms,
		      ctl->qth_min, ctl->qth_max, ctl->Wlog,
		      ctl->Plog, ctl->Scell_log,
		      nla_data(tb[TCA_RED_STAB]),
		      max_P);
	red_set_vars(&q->vars);

	del_timer(&q->adapt_timer);
	if (ctl->flags & TC_RED_ADAPTATIVE)
		mod_timer(&q->adapt_timer, jiffies + HZ/2);

	if (!q->qdisc->q.qlen)
		red_start_of_idle_period(&q->vars);

	sch_tree_unlock(sch);
	red_offload(sch, true);
	return 0;
}

static inline void red_adaptative_timer(struct timer_list *t)
{
	struct red_sched_data *q = from_timer(q, t, adapt_timer);
	struct Qdisc *sch = q->sch;
	spinlock_t *root_lock = qdisc_lock(qdisc_root_sleeping(sch));

	spin_lock(root_lock);
	red_adaptative_algo(&q->parms, &q->vars);
	mod_timer(&q->adapt_timer, jiffies + HZ/2);
	spin_unlock(root_lock);
}

static int red_init(struct Qdisc *sch, struct nlattr *opt,
		    struct netlink_ext_ack *extack)
{
	struct red_sched_data *q = qdisc_priv(sch);

	q->qdisc = &noop_qdisc;
	q->sch = sch;
	timer_setup(&q->adapt_timer, red_adaptative_timer, 0);
	return red_change(sch, opt, extack);
}

static int red_dump_offload_stats(struct Qdisc *sch, struct tc_red_qopt *opt)
{
	struct net_device *dev = qdisc_dev(sch);
	struct tc_red_qopt_offload hw_stats = {
		.command = TC_RED_STATS,
		.handle = sch->handle,
		.parent = sch->parent,
		{
			.stats.bstats = &sch->bstats,
			.stats.qstats = &sch->qstats,
		},
	};
	int err;

	sch->flags &= ~TCQ_F_OFFLOADED;

	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
		return 0;

	err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED,
					    &hw_stats);
	if (err == -EOPNOTSUPP)
		return 0;

	if (!err)
		sch->flags |= TCQ_F_OFFLOADED;

	return err;
}

static int red_dump(struct Qdisc *sch, struct sk_buff *skb)
{
	struct red_sched_data *q = qdisc_priv(sch);
	struct nlattr *opts = NULL;
	struct tc_red_qopt opt = {
		.limit		= q->limit,
		.flags		= q->flags,
		.qth_min	= q->parms.qth_min >> q->parms.Wlog,
		.qth_max	= q->parms.qth_max >> q->parms.Wlog,
		.Wlog		= q->parms.Wlog,
		.Plog		= q->parms.Plog,
		.Scell_log	= q->parms.Scell_log,
	};
	int err;

	sch->qstats.backlog = q->qdisc->qstats.backlog;
	err = red_dump_offload_stats(sch, &opt);
	if (err)
		goto nla_put_failure;

	opts = nla_nest_start(skb, TCA_OPTIONS);
	if (opts == NULL)
		goto nla_put_failure;
	if (nla_put(skb, TCA_RED_PARMS, sizeof(opt), &opt) ||
	    nla_put_u32(skb, TCA_RED_MAX_P, q->parms.max_P))
		goto nla_put_failure;
	return nla_nest_end(skb, opts);

nla_put_failure:
	nla_nest_cancel(skb, opts);
	return -EMSGSIZE;
}

static int red_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
{
	struct red_sched_data *q = qdisc_priv(sch);
	struct net_device *dev = qdisc_dev(sch);
	struct tc_red_xstats st = {
		.early	= q->stats.prob_drop + q->stats.forced_drop,
		.pdrop	= q->stats.pdrop,
		.other	= q->stats.other,
		.marked	= q->stats.prob_mark + q->stats.forced_mark,
	};

	if (sch->flags & TCQ_F_OFFLOADED) {
		struct red_stats hw_stats = {0};
		struct tc_red_qopt_offload hw_stats_request = {
			.command = TC_RED_XSTATS,
			.handle = sch->handle,
			.parent = sch->parent,
			{
				.xstats = &hw_stats,
			},
		};
		if (!dev->netdev_ops->ndo_setup_tc(dev,
						   TC_SETUP_QDISC_RED,
						   &hw_stats_request)) {
			st.early += hw_stats.prob_drop + hw_stats.forced_drop;
			st.pdrop += hw_stats.pdrop;
			st.other += hw_stats.other;
			st.marked += hw_stats.prob_mark + hw_stats.forced_mark;
		}
	}

	return gnet_stats_copy_app(d, &st, sizeof(st));
}

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

	tcm->tcm_handle |= TC_H_MIN(1);
	tcm->tcm_info = q->qdisc->handle;
	return 0;
}

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

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

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

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

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

static void red_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 red_class_ops = {
	.graft		=	red_graft,
	.leaf		=	red_leaf,
	.find		=	red_find,
	.walk		=	red_walk,
	.dump		=	red_dump_class,
};

static struct Qdisc_ops red_qdisc_ops __read_mostly = {
	.id		=	"red",
	.priv_size	=	sizeof(struct red_sched_data),
	.cl_ops		=	&red_class_ops,
	.enqueue	=	red_enqueue,
	.dequeue	=	red_dequeue,
	.peek		=	red_peek,
	.init		=	red_init,
	.reset		=	red_reset,
	.destroy	=	red_destroy,
	.change		=	red_change,
	.dump		=	red_dump,
	.dump_stats	=	red_dump_stats,
	.owner		=	THIS_MODULE,
};

static int __init red_module_init(void)
{
	return register_qdisc(&red_qdisc_ops);
}

static void __exit red_module_exit(void)
{
	unregister_qdisc(&red_qdisc_ops);
}

module_init(red_module_init)
module_exit(red_module_exit)

MODULE_LICENSE("GPL");
Commit	Line	Data
1da177e4 LT	1	/*
	2	* net/sched/sch_red.c Random Early Detection 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	*
	11	* Changes:
dba051f3	12	* J Hadi Salim 980914: computation fixes
1da177e4	13	* Alexey Makarenko <makar@phoenix.kharkov.ua> 990814: qave on idle link was calculated incorrectly.
dba051f3	14	* J Hadi Salim 980816: ECN support
1da177e4 LT	15	*/
1da177e4 LT	16
1da177e4	17	#include <linux/module.h>
1da177e4 LT	18	#include <linux/types.h>
1da177e4 LT	19	#include <linux/kernel.h>
1da177e4	20	#include <linux/skbuff.h>
1da177e4	21	#include <net/pkt_sched.h>
602f3baf	22	#include <net/pkt_cls.h>
1da177e4	23	#include <net/inet_ecn.h>
6b31b28a	24	#include <net/red.h>
1da177e4 LT	25
1da177e4 LT	26
6b31b28a	27	/* Parameters, settable by user:
1da177e4 LT	28	-----------------------------
	29
	30	limit - bytes (must be > qth_max + burst)
	31
	32	Hard limit on queue length, should be chosen >qth_max
	33	to allow packet bursts. This parameter does not
	34	affect the algorithms behaviour and can be chosen
	35	arbitrarily high (well, less than ram size)
	36	Really, this limit will never be reached
	37	if RED works correctly.
1da177e4 LT	38	*/
1da177e4 LT	39
cc7ec456	40	struct red_sched_data {
6b31b28a TG	41	u32 limit; /* HARD maximal queue length */
6b31b28a TG	42	unsigned char flags;
8af2a218	43	struct timer_list adapt_timer;
cdeabbb8	44	struct Qdisc *sch;
6b31b28a	45	struct red_parms parms;
eeca6688	46	struct red_vars vars;
6b31b28a	47	struct red_stats stats;
f38c39d6	48	struct Qdisc *qdisc;
1da177e4 LT	49	};
1da177e4 LT	50
6b31b28a	51	static inline int red_use_ecn(struct red_sched_data *q)
1da177e4	52	{
6b31b28a	53	return q->flags & TC_RED_ECN;
1da177e4 LT	54	}
1da177e4 LT	55
bdc450a0 TG	56	static inline int red_use_harddrop(struct red_sched_data *q)
	57	{
	58	return q->flags & TC_RED_HARDDROP;
	59	}
	60
520ac30f ED	61	static int red_enqueue(struct sk_buff skb, struct Qdisc sch,
520ac30f ED	62	struct sk_buff **to_free)
1da177e4 LT	63	{
1da177e4 LT	64	struct red_sched_data *q = qdisc_priv(sch);
f38c39d6 PM	65	struct Qdisc *child = q->qdisc;
f38c39d6 PM	66	int ret;
1da177e4	67
eeca6688 ED	68	q->vars.qavg = red_calc_qavg(&q->parms,
	69	&q->vars,
	70	child->qstats.backlog);
1da177e4	71
eeca6688 ED	72	if (red_is_idling(&q->vars))
eeca6688 ED	73	red_end_of_idle_period(&q->vars);
1da177e4	74
eeca6688	75	switch (red_action(&q->parms, &q->vars, q->vars.qavg)) {
cc7ec456 ED	76	case RED_DONT_MARK:
	77	break;
	78
	79	case RED_PROB_MARK:
25331d6c	80	qdisc_qstats_overlimit(sch);
cc7ec456 ED	81	if (!red_use_ecn(q) \|\| !INET_ECN_set_ce(skb)) {
	82	q->stats.prob_drop++;
	83	goto congestion_drop;
	84	}
	85
	86	q->stats.prob_mark++;
	87	break;
	88
	89	case RED_HARD_MARK:
25331d6c	90	qdisc_qstats_overlimit(sch);
cc7ec456 ED	91	if (red_use_harddrop(q) \|\| !red_use_ecn(q) \|\|
	92	!INET_ECN_set_ce(skb)) {
	93	q->stats.forced_drop++;
	94	goto congestion_drop;
	95	}
	96
	97	q->stats.forced_mark++;
	98	break;
1da177e4 LT	99	}
1da177e4 LT	100
520ac30f	101	ret = qdisc_enqueue(skb, child, to_free);
f38c39d6	102	if (likely(ret == NET_XMIT_SUCCESS)) {
d7f4f332	103	qdisc_qstats_backlog_inc(sch, skb);
f38c39d6	104	sch->q.qlen++;
378a2f09	105	} else if (net_xmit_drop_count(ret)) {
f38c39d6	106	q->stats.pdrop++;
25331d6c	107	qdisc_qstats_drop(sch);
f38c39d6 PM	108	}
f38c39d6 PM	109	return ret;
6b31b28a TG	110
6b31b28a TG	111	congestion_drop:
520ac30f	112	qdisc_drop(skb, sch, to_free);
1da177e4 LT	113	return NET_XMIT_CN;
	114	}
	115
cc7ec456	116	static struct sk_buff red_dequeue(struct Qdisc sch)
1da177e4 LT	117	{
	118	struct sk_buff *skb;
	119	struct red_sched_data *q = qdisc_priv(sch);
f38c39d6	120	struct Qdisc *child = q->qdisc;
1da177e4	121
f38c39d6	122	skb = child->dequeue(child);
9190b3b3 ED	123	if (skb) {
9190b3b3 ED	124	qdisc_bstats_update(sch, skb);
d7f4f332	125	qdisc_qstats_backlog_dec(sch, skb);
f38c39d6	126	sch->q.qlen--;
9190b3b3	127	} else {
eeca6688 ED	128	if (!red_is_idling(&q->vars))
eeca6688 ED	129	red_start_of_idle_period(&q->vars);
9190b3b3	130	}
9e178ff2	131	return skb;
1da177e4 LT	132	}
1da177e4 LT	133
cc7ec456	134	static struct sk_buff red_peek(struct Qdisc sch)
8e3af978 JP	135	{
	136	struct red_sched_data *q = qdisc_priv(sch);
	137	struct Qdisc *child = q->qdisc;
	138
	139	return child->ops->peek(child);
	140	}
	141
cc7ec456	142	static void red_reset(struct Qdisc *sch)
1da177e4 LT	143	{
	144	struct red_sched_data *q = qdisc_priv(sch);
	145
f38c39d6	146	qdisc_reset(q->qdisc);
d7f4f332	147	sch->qstats.backlog = 0;
f38c39d6	148	sch->q.qlen = 0;
eeca6688	149	red_restart(&q->vars);
1da177e4 LT	150	}
1da177e4 LT	151
602f3baf NF	152	static int red_offload(struct Qdisc *sch, bool enable)
	153	{
	154	struct red_sched_data *q = qdisc_priv(sch);
	155	struct net_device *dev = qdisc_dev(sch);
	156	struct tc_red_qopt_offload opt = {
	157	.handle = sch->handle,
	158	.parent = sch->parent,
	159	};
	160
	161	if (!tc_can_offload(dev) \|\| !dev->netdev_ops->ndo_setup_tc)
	162	return -EOPNOTSUPP;
	163
	164	if (enable) {
	165	opt.command = TC_RED_REPLACE;
	166	opt.set.min = q->parms.qth_min >> q->parms.Wlog;
	167	opt.set.max = q->parms.qth_max >> q->parms.Wlog;
	168	opt.set.probability = q->parms.max_P;
	169	opt.set.is_ecn = red_use_ecn(q);
	170	} else {
	171	opt.command = TC_RED_DESTROY;
	172	}
	173
8234af2d	174	return dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED, &opt);
602f3baf NF	175	}
602f3baf NF	176
f38c39d6 PM	177	static void red_destroy(struct Qdisc *sch)
	178	{
	179	struct red_sched_data *q = qdisc_priv(sch);
8af2a218 ED	180
8af2a218 ED	181	del_timer_sync(&q->adapt_timer);
602f3baf	182	red_offload(sch, false);
f38c39d6 PM	183	qdisc_destroy(q->qdisc);
	184	}
	185
27a3421e PM	186	static const struct nla_policy red_policy[TCA_RED_MAX + 1] = {
	187	[TCA_RED_PARMS] = { .len = sizeof(struct tc_red_qopt) },
	188	[TCA_RED_STAB] = { .len = RED_STAB_SIZE },
a73ed26b	189	[TCA_RED_MAX_P] = { .type = NLA_U32 },
27a3421e PM	190	};
27a3421e PM	191
2030721c AA	192	static int red_change(struct Qdisc sch, struct nlattr opt,
2030721c AA	193	struct netlink_ext_ack *extack)
1da177e4 LT	194	{
1da177e4 LT	195	struct red_sched_data *q = qdisc_priv(sch);
1e90474c	196	struct nlattr *tb[TCA_RED_MAX + 1];
1da177e4	197	struct tc_red_qopt *ctl;
f38c39d6	198	struct Qdisc *child = NULL;
cee63723	199	int err;
a73ed26b	200	u32 max_P;
1da177e4	201
cee63723	202	if (opt == NULL)
dba051f3 TG	203	return -EINVAL;
dba051f3 TG	204
fceb6435	205	err = nla_parse_nested(tb, TCA_RED_MAX, opt, red_policy, NULL);
cee63723 PM	206	if (err < 0)
	207	return err;
	208
1e90474c	209	if (tb[TCA_RED_PARMS] == NULL \|\|
27a3421e	210	tb[TCA_RED_STAB] == NULL)
1da177e4 LT	211	return -EINVAL;
1da177e4 LT	212
a73ed26b ED	213	max_P = tb[TCA_RED_MAX_P] ? nla_get_u32(tb[TCA_RED_MAX_P]) : 0;
a73ed26b ED	214
1e90474c	215	ctl = nla_data(tb[TCA_RED_PARMS]);
8afa10cb NF	216	if (!red_check_params(ctl->qth_min, ctl->qth_max, ctl->Wlog))
8afa10cb NF	217	return -EINVAL;
1da177e4	218
f38c39d6	219	if (ctl->limit > 0) {
a38a9882 AA	220	child = fifo_create_dflt(sch, &bfifo_qdisc_ops, ctl->limit,
a38a9882 AA	221	extack);
fb0305ce PM	222	if (IS_ERR(child))
fb0305ce PM	223	return PTR_ERR(child);
f38c39d6 PM	224	}
f38c39d6 PM	225
49b49971 JK	226	if (child != &noop_qdisc)
49b49971 JK	227	qdisc_hash_add(child, true);
1da177e4 LT	228	sch_tree_lock(sch);
1da177e4 LT	229	q->flags = ctl->flags;
1da177e4	230	q->limit = ctl->limit;
5e50da01	231	if (child) {
2ccccf5f WC	232	qdisc_tree_reduce_backlog(q->qdisc, q->qdisc->q.qlen,
2ccccf5f WC	233	q->qdisc->qstats.backlog);
b94c8afc PM	234	qdisc_destroy(q->qdisc);
b94c8afc PM	235	q->qdisc = child;
5e50da01	236	}
1da177e4	237
eeca6688 ED	238	red_set_parms(&q->parms,
eeca6688 ED	239	ctl->qth_min, ctl->qth_max, ctl->Wlog,
a73ed26b ED	240	ctl->Plog, ctl->Scell_log,
	241	nla_data(tb[TCA_RED_STAB]),
	242	max_P);
eeca6688	243	red_set_vars(&q->vars);
6b31b28a	244
8af2a218 ED	245	del_timer(&q->adapt_timer);
	246	if (ctl->flags & TC_RED_ADAPTATIVE)
	247	mod_timer(&q->adapt_timer, jiffies + HZ/2);
	248
1ee5fa1e	249	if (!q->qdisc->q.qlen)
eeca6688	250	red_start_of_idle_period(&q->vars);
dba051f3	251
1da177e4	252	sch_tree_unlock(sch);
602f3baf	253	red_offload(sch, true);
1da177e4 LT	254	return 0;
	255	}
	256
cdeabbb8	257	static inline void red_adaptative_timer(struct timer_list *t)
8af2a218	258	{
cdeabbb8 KC	259	struct red_sched_data *q = from_timer(q, t, adapt_timer);
cdeabbb8 KC	260	struct Qdisc *sch = q->sch;
8af2a218 ED	261	spinlock_t *root_lock = qdisc_lock(qdisc_root_sleeping(sch));
	262
	263	spin_lock(root_lock);
eeca6688	264	red_adaptative_algo(&q->parms, &q->vars);
8af2a218 ED	265	mod_timer(&q->adapt_timer, jiffies + HZ/2);
	266	spin_unlock(root_lock);
	267	}
	268
e63d7dfd AA	269	static int red_init(struct Qdisc sch, struct nlattr opt,
e63d7dfd AA	270	struct netlink_ext_ack *extack)
1da177e4	271	{
f38c39d6 PM	272	struct red_sched_data *q = qdisc_priv(sch);
	273
	274	q->qdisc = &noop_qdisc;
cdeabbb8 KC	275	q->sch = sch;
cdeabbb8 KC	276	timer_setup(&q->adapt_timer, red_adaptative_timer, 0);
2030721c	277	return red_change(sch, opt, extack);
1da177e4 LT	278	}
1da177e4 LT	279
428a68af	280	static int red_dump_offload_stats(struct Qdisc sch, struct tc_red_qopt opt)
602f3baf NF	281	{
	282	struct net_device *dev = qdisc_dev(sch);
	283	struct tc_red_qopt_offload hw_stats = {
ee9d3429	284	.command = TC_RED_STATS,
602f3baf NF	285	.handle = sch->handle,
602f3baf NF	286	.parent = sch->parent,
ee9d3429 AM	287	{
	288	.stats.bstats = &sch->bstats,
	289	.stats.qstats = &sch->qstats,
	290	},
602f3baf	291	};
8234af2d NF	292	int err;
	293
	294	sch->flags &= ~TCQ_F_OFFLOADED;
602f3baf	295
8234af2d NF	296	if (!tc_can_offload(dev) \|\| !dev->netdev_ops->ndo_setup_tc)
	297	return 0;
	298
	299	err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED,
	300	&hw_stats);
	301	if (err == -EOPNOTSUPP)
602f3baf NF	302	return 0;
602f3baf NF	303
8234af2d NF	304	if (!err)
	305	sch->flags \|= TCQ_F_OFFLOADED;
	306
	307	return err;
602f3baf NF	308	}
602f3baf NF	309
1da177e4 LT	310	static int red_dump(struct Qdisc sch, struct sk_buff skb)
	311	{
	312	struct red_sched_data *q = qdisc_priv(sch);
1e90474c	313	struct nlattr *opts = NULL;
6b31b28a TG	314	struct tc_red_qopt opt = {
	315	.limit = q->limit,
	316	.flags = q->flags,
	317	.qth_min = q->parms.qth_min >> q->parms.Wlog,
	318	.qth_max = q->parms.qth_max >> q->parms.Wlog,
	319	.Wlog = q->parms.Wlog,
	320	.Plog = q->parms.Plog,
	321	.Scell_log = q->parms.Scell_log,
	322	};
602f3baf	323	int err;
1da177e4	324
0dfb33a0	325	sch->qstats.backlog = q->qdisc->qstats.backlog;
428a68af	326	err = red_dump_offload_stats(sch, &opt);
602f3baf NF	327	if (err)
	328	goto nla_put_failure;
	329
1e90474c PM	330	opts = nla_nest_start(skb, TCA_OPTIONS);
	331	if (opts == NULL)
	332	goto nla_put_failure;
1b34ec43 DM	333	if (nla_put(skb, TCA_RED_PARMS, sizeof(opt), &opt) \|\|
	334	nla_put_u32(skb, TCA_RED_MAX_P, q->parms.max_P))
	335	goto nla_put_failure;
1e90474c	336	return nla_nest_end(skb, opts);
1da177e4	337
1e90474c	338	nla_put_failure:
bc3ed28c TG	339	nla_nest_cancel(skb, opts);
bc3ed28c TG	340	return -EMSGSIZE;
1da177e4 LT	341	}
	342
	343	static int red_dump_stats(struct Qdisc sch, struct gnet_dump d)
	344	{
	345	struct red_sched_data *q = qdisc_priv(sch);
602f3baf	346	struct net_device *dev = qdisc_dev(sch);
6b31b28a TG	347	struct tc_red_xstats st = {
	348	.early = q->stats.prob_drop + q->stats.forced_drop,
	349	.pdrop = q->stats.pdrop,
	350	.other = q->stats.other,
	351	.marked = q->stats.prob_mark + q->stats.forced_mark,
	352	};
	353
428a68af	354	if (sch->flags & TCQ_F_OFFLOADED) {
602f3baf NF	355	struct red_stats hw_stats = {0};
602f3baf NF	356	struct tc_red_qopt_offload hw_stats_request = {
ee9d3429	357	.command = TC_RED_XSTATS,
602f3baf NF	358	.handle = sch->handle,
602f3baf NF	359	.parent = sch->parent,
ee9d3429 AM	360	{
	361	.xstats = &hw_stats,
	362	},
602f3baf NF	363	};
	364	if (!dev->netdev_ops->ndo_setup_tc(dev,
	365	TC_SETUP_QDISC_RED,
	366	&hw_stats_request)) {
	367	st.early += hw_stats.prob_drop + hw_stats.forced_drop;
	368	st.pdrop += hw_stats.pdrop;
	369	st.other += hw_stats.other;
	370	st.marked += hw_stats.prob_mark + hw_stats.forced_mark;
	371	}
	372	}
	373
6b31b28a	374	return gnet_stats_copy_app(d, &st, sizeof(st));
1da177e4 LT	375	}
1da177e4 LT	376
f38c39d6 PM	377	static int red_dump_class(struct Qdisc *sch, unsigned long cl,
	378	struct sk_buff skb, struct tcmsg tcm)
	379	{
	380	struct red_sched_data *q = qdisc_priv(sch);
	381
f38c39d6 PM	382	tcm->tcm_handle \|= TC_H_MIN(1);
	383	tcm->tcm_info = q->qdisc->handle;
	384	return 0;
	385	}
	386
	387	static int red_graft(struct Qdisc sch, unsigned long arg, struct Qdisc new,
653d6fd6	388	struct Qdisc *old, struct netlink_ext_ack extack)
f38c39d6 PM	389	{
	390	struct red_sched_data *q = qdisc_priv(sch);
	391
	392	if (new == NULL)
	393	new = &noop_qdisc;
	394
86a7996c	395	*old = qdisc_replace(sch, new, &q->qdisc);
f38c39d6 PM	396	return 0;
	397	}
	398
	399	static struct Qdisc red_leaf(struct Qdisc sch, unsigned long arg)
	400	{
	401	struct red_sched_data *q = qdisc_priv(sch);
	402	return q->qdisc;
	403	}
	404
143976ce	405	static unsigned long red_find(struct Qdisc *sch, u32 classid)
f38c39d6 PM	406	{
	407	return 1;
	408	}
	409
f38c39d6 PM	410	static void red_walk(struct Qdisc sch, struct qdisc_walker walker)
	411	{
	412	if (!walker->stop) {
	413	if (walker->count >= walker->skip)
	414	if (walker->fn(sch, 1, walker) < 0) {
	415	walker->stop = 1;
	416	return;
	417	}
	418	walker->count++;
	419	}
	420	}
	421
20fea08b	422	static const struct Qdisc_class_ops red_class_ops = {
f38c39d6 PM	423	.graft = red_graft,
f38c39d6 PM	424	.leaf = red_leaf,
143976ce	425	.find = red_find,
f38c39d6	426	.walk = red_walk,
f38c39d6 PM	427	.dump = red_dump_class,
	428	};
	429
20fea08b	430	static struct Qdisc_ops red_qdisc_ops __read_mostly = {
1da177e4 LT	431	.id = "red",
1da177e4 LT	432	.priv_size = sizeof(struct red_sched_data),
f38c39d6	433	.cl_ops = &red_class_ops,
1da177e4 LT	434	.enqueue = red_enqueue,
1da177e4 LT	435	.dequeue = red_dequeue,
8e3af978	436	.peek = red_peek,
1da177e4 LT	437	.init = red_init,
1da177e4 LT	438	.reset = red_reset,
f38c39d6	439	.destroy = red_destroy,
1da177e4 LT	440	.change = red_change,
	441	.dump = red_dump,
	442	.dump_stats = red_dump_stats,
	443	.owner = THIS_MODULE,
	444	};
	445
	446	static int __init red_module_init(void)
	447	{
	448	return register_qdisc(&red_qdisc_ops);
	449	}
dba051f3 TG	450
dba051f3 TG	451	static void __exit red_module_exit(void)
1da177e4 LT	452	{
	453	unregister_qdisc(&red_qdisc_ops);
	454	}
dba051f3	455
1da177e4 LT	456	module_init(red_module_init)
1da177e4 LT	457	module_exit(red_module_exit)
dba051f3	458
1da177e4	459	MODULE_LICENSE("GPL");