[mirror_ubuntu-hirsute-kernel.git] / net / sched / sch_sfb.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 * net/sched/sch_sfb.c	  Stochastic Fair Blue
 *
 * Copyright (c) 2008-2011 Juliusz Chroboczek <jch@pps.jussieu.fr>
 * Copyright (c) 2011 Eric Dumazet <eric.dumazet@gmail.com>
 *
 * W. Feng, D. Kandlur, D. Saha, K. Shin. Blue:
 * A New Class of Active Queue Management Algorithms.
 * U. Michigan CSE-TR-387-99, April 1999.
 *
 * http://www.thefengs.com/wuchang/blue/CSE-TR-387-99.pdf
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/random.h>
#include <linux/jhash.h>
#include <net/ip.h>
#include <net/pkt_sched.h>
#include <net/pkt_cls.h>
#include <net/inet_ecn.h>

/*
 * SFB uses two B[l][n] : L x N arrays of bins (L levels, N bins per level)
 * This implementation uses L = 8 and N = 16
 * This permits us to split one 32bit hash (provided per packet by rxhash or
 * external classifier) into 8 subhashes of 4 bits.
 */
#define SFB_BUCKET_SHIFT 4
#define SFB_NUMBUCKETS	(1 << SFB_BUCKET_SHIFT) /* N bins per Level */
#define SFB_BUCKET_MASK (SFB_NUMBUCKETS - 1)
#define SFB_LEVELS	(32 / SFB_BUCKET_SHIFT) /* L */

/* SFB algo uses a virtual queue, named "bin" */
struct sfb_bucket {
	u16		qlen; /* length of virtual queue */
	u16		p_mark; /* marking probability */
};

/* We use a double buffering right before hash change
 * (Section 4.4 of SFB reference : moving hash functions)
 */
struct sfb_bins {
	u32		  perturbation; /* jhash perturbation */
	struct sfb_bucket bins[SFB_LEVELS][SFB_NUMBUCKETS];
};

struct sfb_sched_data {
	struct Qdisc	*qdisc;
	struct tcf_proto __rcu *filter_list;
	struct tcf_block *block;
	unsigned long	rehash_interval;
	unsigned long	warmup_time;	/* double buffering warmup time in jiffies */
	u32		max;
	u32		bin_size;	/* maximum queue length per bin */
	u32		increment;	/* d1 */
	u32		decrement;	/* d2 */
	u32		limit;		/* HARD maximal queue length */
	u32		penalty_rate;
	u32		penalty_burst;
	u32		tokens_avail;
	unsigned long	rehash_time;
	unsigned long	token_time;

	u8		slot;		/* current active bins (0 or 1) */
	bool		double_buffering;
	struct sfb_bins bins[2];

	struct {
		u32	earlydrop;
		u32	penaltydrop;
		u32	bucketdrop;
		u32	queuedrop;
		u32	childdrop;	/* drops in child qdisc */
		u32	marked;		/* ECN mark */
	} stats;
};

/*
 * Each queued skb might be hashed on one or two bins
 * We store in skb_cb the two hash values.
 * (A zero value means double buffering was not used)
 */
struct sfb_skb_cb {
	u32 hashes[2];
};

static inline struct sfb_skb_cb *sfb_skb_cb(const struct sk_buff *skb)
{
	qdisc_cb_private_validate(skb, sizeof(struct sfb_skb_cb));
	return (struct sfb_skb_cb *)qdisc_skb_cb(skb)->data;
}

/*
 * If using 'internal' SFB flow classifier, hash comes from skb rxhash
 * If using external classifier, hash comes from the classid.
 */
static u32 sfb_hash(const struct sk_buff *skb, u32 slot)
{
	return sfb_skb_cb(skb)->hashes[slot];
}

/* Probabilities are coded as Q0.16 fixed-point values,
 * with 0xFFFF representing 65535/65536 (almost 1.0)
 * Addition and subtraction are saturating in [0, 65535]
 */
static u32 prob_plus(u32 p1, u32 p2)
{
	u32 res = p1 + p2;

	return min_t(u32, res, SFB_MAX_PROB);
}

static u32 prob_minus(u32 p1, u32 p2)
{
	return p1 > p2 ? p1 - p2 : 0;
}

static void increment_one_qlen(u32 sfbhash, u32 slot, struct sfb_sched_data *q)
{
	int i;
	struct sfb_bucket *b = &q->bins[slot].bins[0][0];

	for (i = 0; i < SFB_LEVELS; i++) {
		u32 hash = sfbhash & SFB_BUCKET_MASK;

		sfbhash >>= SFB_BUCKET_SHIFT;
		if (b[hash].qlen < 0xFFFF)
			b[hash].qlen++;
		b += SFB_NUMBUCKETS; /* next level */
	}
}

static void increment_qlen(const struct sk_buff *skb, struct sfb_sched_data *q)
{
	u32 sfbhash;

	sfbhash = sfb_hash(skb, 0);
	if (sfbhash)
		increment_one_qlen(sfbhash, 0, q);

	sfbhash = sfb_hash(skb, 1);
	if (sfbhash)
		increment_one_qlen(sfbhash, 1, q);
}

static void decrement_one_qlen(u32 sfbhash, u32 slot,
			       struct sfb_sched_data *q)
{
	int i;
	struct sfb_bucket *b = &q->bins[slot].bins[0][0];

	for (i = 0; i < SFB_LEVELS; i++) {
		u32 hash = sfbhash & SFB_BUCKET_MASK;

		sfbhash >>= SFB_BUCKET_SHIFT;
		if (b[hash].qlen > 0)
			b[hash].qlen--;
		b += SFB_NUMBUCKETS; /* next level */
	}
}

static void decrement_qlen(const struct sk_buff *skb, struct sfb_sched_data *q)
{
	u32 sfbhash;

	sfbhash = sfb_hash(skb, 0);
	if (sfbhash)
		decrement_one_qlen(sfbhash, 0, q);

	sfbhash = sfb_hash(skb, 1);
	if (sfbhash)
		decrement_one_qlen(sfbhash, 1, q);
}

static void decrement_prob(struct sfb_bucket *b, struct sfb_sched_data *q)
{
	b->p_mark = prob_minus(b->p_mark, q->decrement);
}

static void increment_prob(struct sfb_bucket *b, struct sfb_sched_data *q)
{
	b->p_mark = prob_plus(b->p_mark, q->increment);
}

static void sfb_zero_all_buckets(struct sfb_sched_data *q)
{
	memset(&q->bins, 0, sizeof(q->bins));
}

/*
 * compute max qlen, max p_mark, and avg p_mark
 */
static u32 sfb_compute_qlen(u32 *prob_r, u32 *avgpm_r, const struct sfb_sched_data *q)
{
	int i;
	u32 qlen = 0, prob = 0, totalpm = 0;
	const struct sfb_bucket *b = &q->bins[q->slot].bins[0][0];

	for (i = 0; i < SFB_LEVELS * SFB_NUMBUCKETS; i++) {
		if (qlen < b->qlen)
			qlen = b->qlen;
		totalpm += b->p_mark;
		if (prob < b->p_mark)
			prob = b->p_mark;
		b++;
	}
	*prob_r = prob;
	*avgpm_r = totalpm / (SFB_LEVELS * SFB_NUMBUCKETS);
	return qlen;
}


static void sfb_init_perturbation(u32 slot, struct sfb_sched_data *q)
{
	q->bins[slot].perturbation = prandom_u32();
}

static void sfb_swap_slot(struct sfb_sched_data *q)
{
	sfb_init_perturbation(q->slot, q);
	q->slot ^= 1;
	q->double_buffering = false;
}

/* Non elastic flows are allowed to use part of the bandwidth, expressed
 * in "penalty_rate" packets per second, with "penalty_burst" burst
 */
static bool sfb_rate_limit(struct sk_buff *skb, struct sfb_sched_data *q)
{
	if (q->penalty_rate == 0 || q->penalty_burst == 0)
		return true;

	if (q->tokens_avail < 1) {
		unsigned long age = min(10UL * HZ, jiffies - q->token_time);

		q->tokens_avail = (age * q->penalty_rate) / HZ;
		if (q->tokens_avail > q->penalty_burst)
			q->tokens_avail = q->penalty_burst;
		q->token_time = jiffies;
		if (q->tokens_avail < 1)
			return true;
	}

	q->tokens_avail--;
	return false;
}

static bool sfb_classify(struct sk_buff *skb, struct tcf_proto *fl,
			 int *qerr, u32 *salt)
{
	struct tcf_result res;
	int result;

	result = tcf_classify(skb, fl, &res, false);
	if (result >= 0) {
#ifdef CONFIG_NET_CLS_ACT
		switch (result) {
		case TC_ACT_STOLEN:
		case TC_ACT_QUEUED:
		case TC_ACT_TRAP:
			*qerr = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN;
			/* fall through */
		case TC_ACT_SHOT:
			return false;
		}
#endif
		*salt = TC_H_MIN(res.classid);
		return true;
	}
	return false;
}

static int sfb_enqueue(struct sk_buff *skb, struct Qdisc *sch,
		       struct sk_buff **to_free)
{

	struct sfb_sched_data *q = qdisc_priv(sch);
	struct Qdisc *child = q->qdisc;
	struct tcf_proto *fl;
	int i;
	u32 p_min = ~0;
	u32 minqlen = ~0;
	u32 r, sfbhash;
	u32 slot = q->slot;
	int ret = NET_XMIT_SUCCESS | __NET_XMIT_BYPASS;

	if (unlikely(sch->q.qlen >= q->limit)) {
		qdisc_qstats_overlimit(sch);
		q->stats.queuedrop++;
		goto drop;
	}

	if (q->rehash_interval > 0) {
		unsigned long limit = q->rehash_time + q->rehash_interval;

		if (unlikely(time_after(jiffies, limit))) {
			sfb_swap_slot(q);
			q->rehash_time = jiffies;
		} else if (unlikely(!q->double_buffering && q->warmup_time > 0 &&
				    time_after(jiffies, limit - q->warmup_time))) {
			q->double_buffering = true;
		}
	}

	fl = rcu_dereference_bh(q->filter_list);
	if (fl) {
		u32 salt;

		/* If using external classifiers, get result and record it. */
		if (!sfb_classify(skb, fl, &ret, &salt))
			goto other_drop;
		sfbhash = jhash_1word(salt, q->bins[slot].perturbation);
	} else {
		sfbhash = skb_get_hash_perturb(skb, q->bins[slot].perturbation);
	}


	if (!sfbhash)
		sfbhash = 1;
	sfb_skb_cb(skb)->hashes[slot] = sfbhash;

	for (i = 0; i < SFB_LEVELS; i++) {
		u32 hash = sfbhash & SFB_BUCKET_MASK;
		struct sfb_bucket *b = &q->bins[slot].bins[i][hash];

		sfbhash >>= SFB_BUCKET_SHIFT;
		if (b->qlen == 0)
			decrement_prob(b, q);
		else if (b->qlen >= q->bin_size)
			increment_prob(b, q);
		if (minqlen > b->qlen)
			minqlen = b->qlen;
		if (p_min > b->p_mark)
			p_min = b->p_mark;
	}

	slot ^= 1;
	sfb_skb_cb(skb)->hashes[slot] = 0;

	if (unlikely(minqlen >= q->max)) {
		qdisc_qstats_overlimit(sch);
		q->stats.bucketdrop++;
		goto drop;
	}

	if (unlikely(p_min >= SFB_MAX_PROB)) {
		/* Inelastic flow */
		if (q->double_buffering) {
			sfbhash = skb_get_hash_perturb(skb,
			    q->bins[slot].perturbation);
			if (!sfbhash)
				sfbhash = 1;
			sfb_skb_cb(skb)->hashes[slot] = sfbhash;

			for (i = 0; i < SFB_LEVELS; i++) {
				u32 hash = sfbhash & SFB_BUCKET_MASK;
				struct sfb_bucket *b = &q->bins[slot].bins[i][hash];

				sfbhash >>= SFB_BUCKET_SHIFT;
				if (b->qlen == 0)
					decrement_prob(b, q);
				else if (b->qlen >= q->bin_size)
					increment_prob(b, q);
			}
		}
		if (sfb_rate_limit(skb, q)) {
			qdisc_qstats_overlimit(sch);
			q->stats.penaltydrop++;
			goto drop;
		}
		goto enqueue;
	}

	r = prandom_u32() & SFB_MAX_PROB;

	if (unlikely(r < p_min)) {
		if (unlikely(p_min > SFB_MAX_PROB / 2)) {
			/* If we're marking that many packets, then either
			 * this flow is unresponsive, or we're badly congested.
			 * In either case, we want to start dropping packets.
			 */
			if (r < (p_min - SFB_MAX_PROB / 2) * 2) {
				q->stats.earlydrop++;
				goto drop;
			}
		}
		if (INET_ECN_set_ce(skb)) {
			q->stats.marked++;
		} else {
			q->stats.earlydrop++;
			goto drop;
		}
	}

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

drop:
	qdisc_drop(skb, sch, to_free);
	return NET_XMIT_CN;
other_drop:
	if (ret & __NET_XMIT_BYPASS)
		qdisc_qstats_drop(sch);
	kfree_skb(skb);
	return ret;
}

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

	skb = child->dequeue(q->qdisc);

	if (skb) {
		qdisc_bstats_update(sch, skb);
		qdisc_qstats_backlog_dec(sch, skb);
		sch->q.qlen--;
		decrement_qlen(skb, q);
	}

	return skb;
}

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

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

/* No sfb_drop -- impossible since the child doesn't return the dropped skb. */

static void sfb_reset(struct Qdisc *sch)
{
	struct sfb_sched_data *q = qdisc_priv(sch);

	qdisc_reset(q->qdisc);
	sch->qstats.backlog = 0;
	sch->q.qlen = 0;
	q->slot = 0;
	q->double_buffering = false;
	sfb_zero_all_buckets(q);
	sfb_init_perturbation(0, q);
}

static void sfb_destroy(struct Qdisc *sch)
{
	struct sfb_sched_data *q = qdisc_priv(sch);

	tcf_block_put(q->block);
	qdisc_put(q->qdisc);
}

static const struct nla_policy sfb_policy[TCA_SFB_MAX + 1] = {
	[TCA_SFB_PARMS]	= { .len = sizeof(struct tc_sfb_qopt) },
};

static const struct tc_sfb_qopt sfb_default_ops = {
	.rehash_interval = 600 * MSEC_PER_SEC,
	.warmup_time = 60 * MSEC_PER_SEC,
	.limit = 0,
	.max = 25,
	.bin_size = 20,
	.increment = (SFB_MAX_PROB + 500) / 1000, /* 0.1 % */
	.decrement = (SFB_MAX_PROB + 3000) / 6000,
	.penalty_rate = 10,
	.penalty_burst = 20,
};

static int sfb_change(struct Qdisc *sch, struct nlattr *opt,
		      struct netlink_ext_ack *extack)
{
	struct sfb_sched_data *q = qdisc_priv(sch);
	struct Qdisc *child;
	struct nlattr *tb[TCA_SFB_MAX + 1];
	const struct tc_sfb_qopt *ctl = &sfb_default_ops;
	u32 limit;
	int err;

	if (opt) {
		err = nla_parse_nested_deprecated(tb, TCA_SFB_MAX, opt,
						  sfb_policy, NULL);
		if (err < 0)
			return -EINVAL;

		if (tb[TCA_SFB_PARMS] == NULL)
			return -EINVAL;

		ctl = nla_data(tb[TCA_SFB_PARMS]);
	}

	limit = ctl->limit;
	if (limit == 0)
		limit = qdisc_dev(sch)->tx_queue_len;

	child = fifo_create_dflt(sch, &pfifo_qdisc_ops, limit, extack);
	if (IS_ERR(child))
		return PTR_ERR(child);

	if (child != &noop_qdisc)
		qdisc_hash_add(child, true);
	sch_tree_lock(sch);

	qdisc_tree_flush_backlog(q->qdisc);
	qdisc_put(q->qdisc);
	q->qdisc = child;

	q->rehash_interval = msecs_to_jiffies(ctl->rehash_interval);
	q->warmup_time = msecs_to_jiffies(ctl->warmup_time);
	q->rehash_time = jiffies;
	q->limit = limit;
	q->increment = ctl->increment;
	q->decrement = ctl->decrement;
	q->max = ctl->max;
	q->bin_size = ctl->bin_size;
	q->penalty_rate = ctl->penalty_rate;
	q->penalty_burst = ctl->penalty_burst;
	q->tokens_avail = ctl->penalty_burst;
	q->token_time = jiffies;

	q->slot = 0;
	q->double_buffering = false;
	sfb_zero_all_buckets(q);
	sfb_init_perturbation(0, q);
	sfb_init_perturbation(1, q);

	sch_tree_unlock(sch);

	return 0;
}

static int sfb_init(struct Qdisc *sch, struct nlattr *opt,
		    struct netlink_ext_ack *extack)
{
	struct sfb_sched_data *q = qdisc_priv(sch);
	int err;

	err = tcf_block_get(&q->block, &q->filter_list, sch, extack);
	if (err)
		return err;

	q->qdisc = &noop_qdisc;
	return sfb_change(sch, opt, extack);
}

static int sfb_dump(struct Qdisc *sch, struct sk_buff *skb)
{
	struct sfb_sched_data *q = qdisc_priv(sch);
	struct nlattr *opts;
	struct tc_sfb_qopt opt = {
		.rehash_interval = jiffies_to_msecs(q->rehash_interval),
		.warmup_time = jiffies_to_msecs(q->warmup_time),
		.limit = q->limit,
		.max = q->max,
		.bin_size = q->bin_size,
		.increment = q->increment,
		.decrement = q->decrement,
		.penalty_rate = q->penalty_rate,
		.penalty_burst = q->penalty_burst,
	};

	sch->qstats.backlog = q->qdisc->qstats.backlog;
	opts = nla_nest_start_noflag(skb, TCA_OPTIONS);
	if (opts == NULL)
		goto nla_put_failure;
	if (nla_put(skb, TCA_SFB_PARMS, sizeof(opt), &opt))
		goto nla_put_failure;
	return nla_nest_end(skb, opts);

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

static int sfb_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
{
	struct sfb_sched_data *q = qdisc_priv(sch);
	struct tc_sfb_xstats st = {
		.earlydrop = q->stats.earlydrop,
		.penaltydrop = q->stats.penaltydrop,
		.bucketdrop = q->stats.bucketdrop,
		.queuedrop = q->stats.queuedrop,
		.childdrop = q->stats.childdrop,
		.marked = q->stats.marked,
	};

	st.maxqlen = sfb_compute_qlen(&st.maxprob, &st.avgprob, q);

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

static int sfb_dump_class(struct Qdisc *sch, unsigned long cl,
			  struct sk_buff *skb, struct tcmsg *tcm)
{
	return -ENOSYS;
}

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

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

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

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

	return q->qdisc;
}

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

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

static int sfb_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
			    struct nlattr **tca, unsigned long *arg,
			    struct netlink_ext_ack *extack)
{
	return -ENOSYS;
}

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

static void sfb_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_block *sfb_tcf_block(struct Qdisc *sch, unsigned long cl,
				       struct netlink_ext_ack *extack)
{
	struct sfb_sched_data *q = qdisc_priv(sch);

	if (cl)
		return NULL;
	return q->block;
}

static unsigned long sfb_bind(struct Qdisc *sch, unsigned long parent,
			      u32 classid)
{
	return 0;
}


static const struct Qdisc_class_ops sfb_class_ops = {
	.graft		=	sfb_graft,
	.leaf		=	sfb_leaf,
	.find		=	sfb_find,
	.change		=	sfb_change_class,
	.delete		=	sfb_delete,
	.walk		=	sfb_walk,
	.tcf_block	=	sfb_tcf_block,
	.bind_tcf	=	sfb_bind,
	.unbind_tcf	=	sfb_unbind,
	.dump		=	sfb_dump_class,
};

static struct Qdisc_ops sfb_qdisc_ops __read_mostly = {
	.id		=	"sfb",
	.priv_size	=	sizeof(struct sfb_sched_data),
	.cl_ops		=	&sfb_class_ops,
	.enqueue	=	sfb_enqueue,
	.dequeue	=	sfb_dequeue,
	.peek		=	sfb_peek,
	.init		=	sfb_init,
	.reset		=	sfb_reset,
	.destroy	=	sfb_destroy,
	.change		=	sfb_change,
	.dump		=	sfb_dump,
	.dump_stats	=	sfb_dump_stats,
	.owner		=	THIS_MODULE,
};

static int __init sfb_module_init(void)
{
	return register_qdisc(&sfb_qdisc_ops);
}

static void __exit sfb_module_exit(void)
{
	unregister_qdisc(&sfb_qdisc_ops);
}

module_init(sfb_module_init)
module_exit(sfb_module_exit)

MODULE_DESCRIPTION("Stochastic Fair Blue queue discipline");
MODULE_AUTHOR("Juliusz Chroboczek");
MODULE_AUTHOR("Eric Dumazet");
MODULE_LICENSE("GPL");
Commit	Line	Data
d2912cb1	1	// SPDX-License-Identifier: GPL-2.0-only
e13e02a3 ED	2	/*
	3	* net/sched/sch_sfb.c Stochastic Fair Blue
	4	*
	5	* Copyright (c) 2008-2011 Juliusz Chroboczek <jch@pps.jussieu.fr>
	6	* Copyright (c) 2011 Eric Dumazet <eric.dumazet@gmail.com>
	7	*
e13e02a3 ED	8	* W. Feng, D. Kandlur, D. Saha, K. Shin. Blue:
	9	* A New Class of Active Queue Management Algorithms.
	10	* U. Michigan CSE-TR-387-99, April 1999.
	11	*
	12	* http://www.thefengs.com/wuchang/blue/CSE-TR-387-99.pdf
e13e02a3 ED	13	*/
	14
	15	#include <linux/module.h>
	16	#include <linux/types.h>
	17	#include <linux/kernel.h>
	18	#include <linux/errno.h>
	19	#include <linux/skbuff.h>
	20	#include <linux/random.h>
	21	#include <linux/jhash.h>
	22	#include <net/ip.h>
	23	#include <net/pkt_sched.h>
cf1facda	24	#include <net/pkt_cls.h>
e13e02a3 ED	25	#include <net/inet_ecn.h>
	26
	27	/*
	28	* SFB uses two B[l][n] : L x N arrays of bins (L levels, N bins per level)
	29	* This implementation uses L = 8 and N = 16
	30	* This permits us to split one 32bit hash (provided per packet by rxhash or
	31	* external classifier) into 8 subhashes of 4 bits.
	32	*/
	33	#define SFB_BUCKET_SHIFT 4
	34	#define SFB_NUMBUCKETS (1 << SFB_BUCKET_SHIFT) /* N bins per Level */
	35	#define SFB_BUCKET_MASK (SFB_NUMBUCKETS - 1)
	36	#define SFB_LEVELS (32 / SFB_BUCKET_SHIFT) /* L */
	37
	38	/* SFB algo uses a virtual queue, named "bin" */
	39	struct sfb_bucket {
	40	u16 qlen; /* length of virtual queue */
	41	u16 p_mark; /* marking probability */
	42	};
	43
	44	/* We use a double buffering right before hash change
	45	* (Section 4.4 of SFB reference : moving hash functions)
	46	*/
	47	struct sfb_bins {
	48	u32 perturbation; /* jhash perturbation */
	49	struct sfb_bucket bins[SFB_LEVELS][SFB_NUMBUCKETS];
	50	};
	51
	52	struct sfb_sched_data {
	53	struct Qdisc *qdisc;
25d8c0d5	54	struct tcf_proto __rcu *filter_list;
6529eaba	55	struct tcf_block *block;
e13e02a3 ED	56	unsigned long rehash_interval;
	57	unsigned long warmup_time; /* double buffering warmup time in jiffies */
	58	u32 max;
	59	u32 bin_size; /* maximum queue length per bin */
	60	u32 increment; /* d1 */
	61	u32 decrement; /* d2 */
	62	u32 limit; /* HARD maximal queue length */
	63	u32 penalty_rate;
	64	u32 penalty_burst;
	65	u32 tokens_avail;
	66	unsigned long rehash_time;
	67	unsigned long token_time;
	68
	69	u8 slot; /* current active bins (0 or 1) */
	70	bool double_buffering;
	71	struct sfb_bins bins[2];
	72
	73	struct {
	74	u32 earlydrop;
	75	u32 penaltydrop;
	76	u32 bucketdrop;
	77	u32 queuedrop;
	78	u32 childdrop; /* drops in child qdisc */
	79	u32 marked; /* ECN mark */
	80	} stats;
	81	};
	82
	83	/*
	84	* Each queued skb might be hashed on one or two bins
	85	* We store in skb_cb the two hash values.
	86	* (A zero value means double buffering was not used)
	87	*/
	88	struct sfb_skb_cb {
	89	u32 hashes[2];
	90	};
	91
	92	static inline struct sfb_skb_cb sfb_skb_cb(const struct sk_buff skb)
	93	{
16bda13d	94	qdisc_cb_private_validate(skb, sizeof(struct sfb_skb_cb));
e13e02a3 ED	95	return (struct sfb_skb_cb *)qdisc_skb_cb(skb)->data;
	96	}
	97
	98	/*
	99	* If using 'internal' SFB flow classifier, hash comes from skb rxhash
	100	* If using external classifier, hash comes from the classid.
	101	*/
	102	static u32 sfb_hash(const struct sk_buff *skb, u32 slot)
	103	{
	104	return sfb_skb_cb(skb)->hashes[slot];
	105	}
	106
	107	/* Probabilities are coded as Q0.16 fixed-point values,
	108	* with 0xFFFF representing 65535/65536 (almost 1.0)
	109	* Addition and subtraction are saturating in [0, 65535]
	110	*/
	111	static u32 prob_plus(u32 p1, u32 p2)
	112	{
	113	u32 res = p1 + p2;
	114
	115	return min_t(u32, res, SFB_MAX_PROB);
	116	}
	117
	118	static u32 prob_minus(u32 p1, u32 p2)
	119	{
	120	return p1 > p2 ? p1 - p2 : 0;
	121	}
	122
	123	static void increment_one_qlen(u32 sfbhash, u32 slot, struct sfb_sched_data *q)
	124	{
	125	int i;
	126	struct sfb_bucket *b = &q->bins[slot].bins[0][0];
	127
	128	for (i = 0; i < SFB_LEVELS; i++) {
	129	u32 hash = sfbhash & SFB_BUCKET_MASK;
	130
	131	sfbhash >>= SFB_BUCKET_SHIFT;
	132	if (b[hash].qlen < 0xFFFF)
	133	b[hash].qlen++;
	134	b += SFB_NUMBUCKETS; /* next level */
	135	}
	136	}
	137
	138	static void increment_qlen(const struct sk_buff skb, struct sfb_sched_data q)
	139	{
	140	u32 sfbhash;
	141
	142	sfbhash = sfb_hash(skb, 0);
	143	if (sfbhash)
	144	increment_one_qlen(sfbhash, 0, q);
	145
	146	sfbhash = sfb_hash(skb, 1);
	147	if (sfbhash)
	148	increment_one_qlen(sfbhash, 1, q);
	149	}
	150
	151	static void decrement_one_qlen(u32 sfbhash, u32 slot,
	152	struct sfb_sched_data *q)
	153	{
	154	int i;
	155	struct sfb_bucket *b = &q->bins[slot].bins[0][0];
	156
	157	for (i = 0; i < SFB_LEVELS; i++) {
	158	u32 hash = sfbhash & SFB_BUCKET_MASK;
159
160	sfbhash >>= SFB_BUCKET_SHIFT;
161	if (b[hash].qlen > 0)
162	b[hash].qlen--;
163	b += SFB_NUMBUCKETS; /* next level */
164	}
165	}
166
167	static void decrement_qlen(const struct sk_buff skb, struct sfb_sched_data q)
168	{
169	u32 sfbhash;
170
171	sfbhash = sfb_hash(skb, 0);
172	if (sfbhash)
173	decrement_one_qlen(sfbhash, 0, q);
174
175	sfbhash = sfb_hash(skb, 1);
176	if (sfbhash)
177	decrement_one_qlen(sfbhash, 1, q);
178	}
179
180	static void decrement_prob(struct sfb_bucket b, struct sfb_sched_data q)
181	{
182	b->p_mark = prob_minus(b->p_mark, q->decrement);
183	}
184
185	static void increment_prob(struct sfb_bucket b, struct sfb_sched_data q)
186	{
187	b->p_mark = prob_plus(b->p_mark, q->increment);
188	}
189
190	static void sfb_zero_all_buckets(struct sfb_sched_data *q)
191	{
192	memset(&q->bins, 0, sizeof(q->bins));
193	}
194
195	/*
196	* compute max qlen, max p_mark, and avg p_mark
197	*/
198	static u32 sfb_compute_qlen(u32 prob_r, u32 avgpm_r, const struct sfb_sched_data *q)
199	{
200	int i;
201	u32 qlen = 0, prob = 0, totalpm = 0;
202	const struct sfb_bucket *b = &q->bins[q->slot].bins[0][0];
203
204	for (i = 0; i < SFB_LEVELS * SFB_NUMBUCKETS; i++) {
205	if (qlen < b->qlen)
206	qlen = b->qlen;
207	totalpm += b->p_mark;
208	if (prob < b->p_mark)
209	prob = b->p_mark;
210	b++;
211	}
212	*prob_r = prob;
213	avgpm_r = totalpm / (SFB_LEVELS SFB_NUMBUCKETS);
214	return qlen;
215	}
216
217
218	static void sfb_init_perturbation(u32 slot, struct sfb_sched_data *q)
219	{
63862b5b	220	q->bins[slot].perturbation = prandom_u32();
e13e02a3 ED	221	}
	222
	223	static void sfb_swap_slot(struct sfb_sched_data *q)
	224	{
	225	sfb_init_perturbation(q->slot, q);
	226	q->slot ^= 1;
	227	q->double_buffering = false;
	228	}
	229
	230	/* Non elastic flows are allowed to use part of the bandwidth, expressed
	231	* in "penalty_rate" packets per second, with "penalty_burst" burst
	232	*/
	233	static bool sfb_rate_limit(struct sk_buff skb, struct sfb_sched_data q)
	234	{
	235	if (q->penalty_rate == 0 \|\| q->penalty_burst == 0)
	236	return true;
	237
	238	if (q->tokens_avail < 1) {
	239	unsigned long age = min(10UL * HZ, jiffies - q->token_time);
	240
	241	q->tokens_avail = (age * q->penalty_rate) / HZ;
	242	if (q->tokens_avail > q->penalty_burst)
	243	q->tokens_avail = q->penalty_burst;
	244	q->token_time = jiffies;
	245	if (q->tokens_avail < 1)
	246	return true;
	247	}
	248
	249	q->tokens_avail--;
	250	return false;
	251	}
	252
25d8c0d5	253	static bool sfb_classify(struct sk_buff skb, struct tcf_proto fl,
e13e02a3 ED	254	int qerr, u32 salt)
	255	{
	256	struct tcf_result res;
	257	int result;
	258
87d83093	259	result = tcf_classify(skb, fl, &res, false);
e13e02a3 ED	260	if (result >= 0) {
	261	#ifdef CONFIG_NET_CLS_ACT
	262	switch (result) {
	263	case TC_ACT_STOLEN:
	264	case TC_ACT_QUEUED:
e25ea21f	265	case TC_ACT_TRAP:
e13e02a3	266	*qerr = NET_XMIT_SUCCESS \| __NET_XMIT_STOLEN;
f3ae608e	267	/* fall through */
e13e02a3 ED	268	case TC_ACT_SHOT:
	269	return false;
	270	}
	271	#endif
	272	*salt = TC_H_MIN(res.classid);
	273	return true;
	274	}
	275	return false;
	276	}
	277
520ac30f ED	278	static int sfb_enqueue(struct sk_buff skb, struct Qdisc sch,
520ac30f ED	279	struct sk_buff **to_free)
e13e02a3 ED	280	{
	281
	282	struct sfb_sched_data *q = qdisc_priv(sch);
	283	struct Qdisc *child = q->qdisc;
25d8c0d5	284	struct tcf_proto *fl;
e13e02a3 ED	285	int i;
	286	u32 p_min = ~0;
	287	u32 minqlen = ~0;
63c0ad4d TH	288	u32 r, sfbhash;
63c0ad4d TH	289	u32 slot = q->slot;
e13e02a3 ED	290	int ret = NET_XMIT_SUCCESS \| __NET_XMIT_BYPASS;
e13e02a3 ED	291
363437f4	292	if (unlikely(sch->q.qlen >= q->limit)) {
25331d6c	293	qdisc_qstats_overlimit(sch);
363437f4 ED	294	q->stats.queuedrop++;
	295	goto drop;
	296	}
	297
e13e02a3 ED	298	if (q->rehash_interval > 0) {
	299	unsigned long limit = q->rehash_time + q->rehash_interval;
	300
	301	if (unlikely(time_after(jiffies, limit))) {
	302	sfb_swap_slot(q);
	303	q->rehash_time = jiffies;
	304	} else if (unlikely(!q->double_buffering && q->warmup_time > 0 &&
	305	time_after(jiffies, limit - q->warmup_time))) {
	306	q->double_buffering = true;
	307	}
	308	}
	309
25d8c0d5 JF	310	fl = rcu_dereference_bh(q->filter_list);
25d8c0d5 JF	311	if (fl) {
63c0ad4d TH	312	u32 salt;
63c0ad4d TH	313
e13e02a3	314	/* If using external classifiers, get result and record it. */
25d8c0d5	315	if (!sfb_classify(skb, fl, &ret, &salt))
e13e02a3	316	goto other_drop;
63c0ad4d	317	sfbhash = jhash_1word(salt, q->bins[slot].perturbation);
e13e02a3	318	} else {
63c0ad4d	319	sfbhash = skb_get_hash_perturb(skb, q->bins[slot].perturbation);
e13e02a3 ED	320	}
e13e02a3 ED	321
e13e02a3	322
e13e02a3 ED	323	if (!sfbhash)
	324	sfbhash = 1;
	325	sfb_skb_cb(skb)->hashes[slot] = sfbhash;
	326
	327	for (i = 0; i < SFB_LEVELS; i++) {
	328	u32 hash = sfbhash & SFB_BUCKET_MASK;
	329	struct sfb_bucket *b = &q->bins[slot].bins[i][hash];
	330
	331	sfbhash >>= SFB_BUCKET_SHIFT;
	332	if (b->qlen == 0)
	333	decrement_prob(b, q);
	334	else if (b->qlen >= q->bin_size)
	335	increment_prob(b, q);
	336	if (minqlen > b->qlen)
	337	minqlen = b->qlen;
	338	if (p_min > b->p_mark)
	339	p_min = b->p_mark;
	340	}
	341
	342	slot ^= 1;
	343	sfb_skb_cb(skb)->hashes[slot] = 0;
	344
363437f4	345	if (unlikely(minqlen >= q->max)) {
25331d6c	346	qdisc_qstats_overlimit(sch);
363437f4	347	q->stats.bucketdrop++;
e13e02a3 ED	348	goto drop;
	349	}
	350
	351	if (unlikely(p_min >= SFB_MAX_PROB)) {
	352	/* Inelastic flow */
	353	if (q->double_buffering) {
63c0ad4d TH	354	sfbhash = skb_get_hash_perturb(skb,
63c0ad4d TH	355	q->bins[slot].perturbation);
e13e02a3 ED	356	if (!sfbhash)
	357	sfbhash = 1;
	358	sfb_skb_cb(skb)->hashes[slot] = sfbhash;
	359
	360	for (i = 0; i < SFB_LEVELS; i++) {
	361	u32 hash = sfbhash & SFB_BUCKET_MASK;
	362	struct sfb_bucket *b = &q->bins[slot].bins[i][hash];
	363
	364	sfbhash >>= SFB_BUCKET_SHIFT;
	365	if (b->qlen == 0)
	366	decrement_prob(b, q);
	367	else if (b->qlen >= q->bin_size)
	368	increment_prob(b, q);
	369	}
	370	}
	371	if (sfb_rate_limit(skb, q)) {
25331d6c	372	qdisc_qstats_overlimit(sch);
e13e02a3 ED	373	q->stats.penaltydrop++;
	374	goto drop;
	375	}
	376	goto enqueue;
	377	}
	378
63862b5b	379	r = prandom_u32() & SFB_MAX_PROB;
e13e02a3 ED	380
	381	if (unlikely(r < p_min)) {
	382	if (unlikely(p_min > SFB_MAX_PROB / 2)) {
	383	/* If we're marking that many packets, then either
	384	* this flow is unresponsive, or we're badly congested.
	385	* In either case, we want to start dropping packets.
	386	*/
	387	if (r < (p_min - SFB_MAX_PROB / 2) * 2) {
	388	q->stats.earlydrop++;
	389	goto drop;
	390	}
	391	}
	392	if (INET_ECN_set_ce(skb)) {
	393	q->stats.marked++;
	394	} else {
	395	q->stats.earlydrop++;
	396	goto drop;
	397	}
	398	}
	399
	400	enqueue:
520ac30f	401	ret = qdisc_enqueue(skb, child, to_free);
e13e02a3	402	if (likely(ret == NET_XMIT_SUCCESS)) {
3d4357fb	403	qdisc_qstats_backlog_inc(sch, skb);
e13e02a3 ED	404	sch->q.qlen++;
	405	increment_qlen(skb, q);
	406	} else if (net_xmit_drop_count(ret)) {
	407	q->stats.childdrop++;
25331d6c	408	qdisc_qstats_drop(sch);
e13e02a3 ED	409	}
	410	return ret;
	411
	412	drop:
520ac30f	413	qdisc_drop(skb, sch, to_free);
e13e02a3 ED	414	return NET_XMIT_CN;
	415	other_drop:
	416	if (ret & __NET_XMIT_BYPASS)
25331d6c	417	qdisc_qstats_drop(sch);
e13e02a3 ED	418	kfree_skb(skb);
	419	return ret;
	420	}
	421
	422	static struct sk_buff sfb_dequeue(struct Qdisc sch)
	423	{
	424	struct sfb_sched_data *q = qdisc_priv(sch);
	425	struct Qdisc *child = q->qdisc;
	426	struct sk_buff *skb;
	427
	428	skb = child->dequeue(q->qdisc);
	429
	430	if (skb) {
	431	qdisc_bstats_update(sch, skb);
3d4357fb	432	qdisc_qstats_backlog_dec(sch, skb);
e13e02a3 ED	433	sch->q.qlen--;
	434	decrement_qlen(skb, q);
	435	}
	436
	437	return skb;
	438	}
	439
	440	static struct sk_buff sfb_peek(struct Qdisc sch)
	441	{
	442	struct sfb_sched_data *q = qdisc_priv(sch);
	443	struct Qdisc *child = q->qdisc;
	444
	445	return child->ops->peek(child);
	446	}
	447
	448	/* No sfb_drop -- impossible since the child doesn't return the dropped skb. */
	449
	450	static void sfb_reset(struct Qdisc *sch)
	451	{
	452	struct sfb_sched_data *q = qdisc_priv(sch);
	453
	454	qdisc_reset(q->qdisc);
3d4357fb	455	sch->qstats.backlog = 0;
e13e02a3 ED	456	sch->q.qlen = 0;
	457	q->slot = 0;
	458	q->double_buffering = false;
	459	sfb_zero_all_buckets(q);
	460	sfb_init_perturbation(0, q);
	461	}
	462
	463	static void sfb_destroy(struct Qdisc *sch)
	464	{
	465	struct sfb_sched_data *q = qdisc_priv(sch);
	466
6529eaba	467	tcf_block_put(q->block);
86bd446b	468	qdisc_put(q->qdisc);
e13e02a3 ED	469	}
	470
	471	static const struct nla_policy sfb_policy[TCA_SFB_MAX + 1] = {
	472	[TCA_SFB_PARMS] = { .len = sizeof(struct tc_sfb_qopt) },
	473	};
	474
	475	static const struct tc_sfb_qopt sfb_default_ops = {
	476	.rehash_interval = 600 * MSEC_PER_SEC,
	477	.warmup_time = 60 * MSEC_PER_SEC,
	478	.limit = 0,
	479	.max = 25,
	480	.bin_size = 20,
	481	.increment = (SFB_MAX_PROB + 500) / 1000, /* 0.1 % */
	482	.decrement = (SFB_MAX_PROB + 3000) / 6000,
	483	.penalty_rate = 10,
	484	.penalty_burst = 20,
	485	};
	486
2030721c AA	487	static int sfb_change(struct Qdisc sch, struct nlattr opt,
2030721c AA	488	struct netlink_ext_ack *extack)
e13e02a3 ED	489	{
	490	struct sfb_sched_data *q = qdisc_priv(sch);
	491	struct Qdisc *child;
	492	struct nlattr *tb[TCA_SFB_MAX + 1];
	493	const struct tc_sfb_qopt *ctl = &sfb_default_ops;
	494	u32 limit;
	495	int err;
	496
	497	if (opt) {
8cb08174 JB	498	err = nla_parse_nested_deprecated(tb, TCA_SFB_MAX, opt,
8cb08174 JB	499	sfb_policy, NULL);
e13e02a3 ED	500	if (err < 0)
	501	return -EINVAL;
	502
	503	if (tb[TCA_SFB_PARMS] == NULL)
	504	return -EINVAL;
	505
	506	ctl = nla_data(tb[TCA_SFB_PARMS]);
	507	}
	508
	509	limit = ctl->limit;
	510	if (limit == 0)
348e3435	511	limit = qdisc_dev(sch)->tx_queue_len;
e13e02a3	512
a38a9882	513	child = fifo_create_dflt(sch, &pfifo_qdisc_ops, limit, extack);
e13e02a3 ED	514	if (IS_ERR(child))
	515	return PTR_ERR(child);
	516
49b49971 JK	517	if (child != &noop_qdisc)
49b49971 JK	518	qdisc_hash_add(child, true);
e13e02a3 ED	519	sch_tree_lock(sch);
e13e02a3 ED	520
e5f0e8f8	521	qdisc_tree_flush_backlog(q->qdisc);
86bd446b	522	qdisc_put(q->qdisc);
e13e02a3 ED	523	q->qdisc = child;
	524
	525	q->rehash_interval = msecs_to_jiffies(ctl->rehash_interval);
	526	q->warmup_time = msecs_to_jiffies(ctl->warmup_time);
	527	q->rehash_time = jiffies;
	528	q->limit = limit;
	529	q->increment = ctl->increment;
	530	q->decrement = ctl->decrement;
	531	q->max = ctl->max;
	532	q->bin_size = ctl->bin_size;
	533	q->penalty_rate = ctl->penalty_rate;
	534	q->penalty_burst = ctl->penalty_burst;
	535	q->tokens_avail = ctl->penalty_burst;
	536	q->token_time = jiffies;
	537
	538	q->slot = 0;
	539	q->double_buffering = false;
	540	sfb_zero_all_buckets(q);
	541	sfb_init_perturbation(0, q);
	542	sfb_init_perturbation(1, q);
	543
	544	sch_tree_unlock(sch);
	545
	546	return 0;
	547	}
	548
e63d7dfd AA	549	static int sfb_init(struct Qdisc sch, struct nlattr opt,
e63d7dfd AA	550	struct netlink_ext_ack *extack)
e13e02a3 ED	551	{
e13e02a3 ED	552	struct sfb_sched_data *q = qdisc_priv(sch);
6529eaba JP	553	int err;
6529eaba JP	554
8d1a77f9	555	err = tcf_block_get(&q->block, &q->filter_list, sch, extack);
6529eaba JP	556	if (err)
6529eaba JP	557	return err;
e13e02a3 ED	558
e13e02a3 ED	559	q->qdisc = &noop_qdisc;
2030721c	560	return sfb_change(sch, opt, extack);
e13e02a3 ED	561	}
	562
	563	static int sfb_dump(struct Qdisc sch, struct sk_buff skb)
	564	{
	565	struct sfb_sched_data *q = qdisc_priv(sch);
	566	struct nlattr *opts;
	567	struct tc_sfb_qopt opt = {
	568	.rehash_interval = jiffies_to_msecs(q->rehash_interval),
	569	.warmup_time = jiffies_to_msecs(q->warmup_time),
	570	.limit = q->limit,
	571	.max = q->max,
	572	.bin_size = q->bin_size,
	573	.increment = q->increment,
	574	.decrement = q->decrement,
	575	.penalty_rate = q->penalty_rate,
	576	.penalty_burst = q->penalty_burst,
	577	};
	578
	579	sch->qstats.backlog = q->qdisc->qstats.backlog;
ae0be8de	580	opts = nla_nest_start_noflag(skb, TCA_OPTIONS);
7ac2908e AC	581	if (opts == NULL)
7ac2908e AC	582	goto nla_put_failure;
1b34ec43 DM	583	if (nla_put(skb, TCA_SFB_PARMS, sizeof(opt), &opt))
1b34ec43 DM	584	goto nla_put_failure;
e13e02a3 ED	585	return nla_nest_end(skb, opts);
	586
	587	nla_put_failure:
	588	nla_nest_cancel(skb, opts);
	589	return -EMSGSIZE;
	590	}
	591
	592	static int sfb_dump_stats(struct Qdisc sch, struct gnet_dump d)
	593	{
	594	struct sfb_sched_data *q = qdisc_priv(sch);
	595	struct tc_sfb_xstats st = {
	596	.earlydrop = q->stats.earlydrop,
	597	.penaltydrop = q->stats.penaltydrop,
	598	.bucketdrop = q->stats.bucketdrop,
	599	.queuedrop = q->stats.queuedrop,
	600	.childdrop = q->stats.childdrop,
	601	.marked = q->stats.marked,
	602	};
	603
	604	st.maxqlen = sfb_compute_qlen(&st.maxprob, &st.avgprob, q);
	605
	606	return gnet_stats_copy_app(d, &st, sizeof(st));
	607	}
	608
	609	static int sfb_dump_class(struct Qdisc *sch, unsigned long cl,
	610	struct sk_buff skb, struct tcmsg tcm)
	611	{
	612	return -ENOSYS;
	613	}
	614
	615	static int sfb_graft(struct Qdisc sch, unsigned long arg, struct Qdisc new,
653d6fd6	616	struct Qdisc *old, struct netlink_ext_ack extack)
e13e02a3 ED	617	{
	618	struct sfb_sched_data *q = qdisc_priv(sch);
	619
	620	if (new == NULL)
	621	new = &noop_qdisc;
	622
86a7996c	623	*old = qdisc_replace(sch, new, &q->qdisc);
e13e02a3 ED	624	return 0;
	625	}
	626
	627	static struct Qdisc sfb_leaf(struct Qdisc sch, unsigned long arg)
	628	{
	629	struct sfb_sched_data *q = qdisc_priv(sch);
	630
	631	return q->qdisc;
	632	}
	633
143976ce	634	static unsigned long sfb_find(struct Qdisc *sch, u32 classid)
e13e02a3 ED	635	{
	636	return 1;
	637	}
	638
143976ce	639	static void sfb_unbind(struct Qdisc *sch, unsigned long arg)
e13e02a3 ED	640	{
	641	}
	642
	643	static int sfb_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
793d81d6 AA	644	struct nlattr *tca, unsigned long arg,
793d81d6 AA	645	struct netlink_ext_ack *extack)
e13e02a3 ED	646	{
	647	return -ENOSYS;
	648	}
	649
	650	static int sfb_delete(struct Qdisc *sch, unsigned long cl)
	651	{
	652	return -ENOSYS;
	653	}
	654
	655	static void sfb_walk(struct Qdisc sch, struct qdisc_walker walker)
	656	{
	657	if (!walker->stop) {
	658	if (walker->count >= walker->skip)
	659	if (walker->fn(sch, 1, walker) < 0) {
	660	walker->stop = 1;
	661	return;
	662	}
	663	walker->count++;
	664	}
	665	}
	666
cbaacc4e AA	667	static struct tcf_block sfb_tcf_block(struct Qdisc sch, unsigned long cl,
cbaacc4e AA	668	struct netlink_ext_ack *extack)
e13e02a3 ED	669	{
	670	struct sfb_sched_data *q = qdisc_priv(sch);
	671
	672	if (cl)
	673	return NULL;
6529eaba	674	return q->block;
e13e02a3 ED	675	}
	676
	677	static unsigned long sfb_bind(struct Qdisc *sch, unsigned long parent,
	678	u32 classid)
	679	{
	680	return 0;
	681	}
	682
	683
	684	static const struct Qdisc_class_ops sfb_class_ops = {
	685	.graft = sfb_graft,
	686	.leaf = sfb_leaf,
143976ce	687	.find = sfb_find,
e13e02a3 ED	688	.change = sfb_change_class,
	689	.delete = sfb_delete,
	690	.walk = sfb_walk,
6529eaba	691	.tcf_block = sfb_tcf_block,
e13e02a3	692	.bind_tcf = sfb_bind,
143976ce	693	.unbind_tcf = sfb_unbind,
e13e02a3 ED	694	.dump = sfb_dump_class,
	695	};
	696
	697	static struct Qdisc_ops sfb_qdisc_ops __read_mostly = {
	698	.id = "sfb",
	699	.priv_size = sizeof(struct sfb_sched_data),
	700	.cl_ops = &sfb_class_ops,
	701	.enqueue = sfb_enqueue,
	702	.dequeue = sfb_dequeue,
	703	.peek = sfb_peek,
	704	.init = sfb_init,
	705	.reset = sfb_reset,
	706	.destroy = sfb_destroy,
	707	.change = sfb_change,
	708	.dump = sfb_dump,
	709	.dump_stats = sfb_dump_stats,
	710	.owner = THIS_MODULE,
	711	};
	712
	713	static int __init sfb_module_init(void)
	714	{
	715	return register_qdisc(&sfb_qdisc_ops);
	716	}
	717
	718	static void __exit sfb_module_exit(void)
	719	{
	720	unregister_qdisc(&sfb_qdisc_ops);
	721	}
	722
	723	module_init(sfb_module_init)
	724	module_exit(sfb_module_exit)
	725
	726	MODULE_DESCRIPTION("Stochastic Fair Blue queue discipline");
	727	MODULE_AUTHOR("Juliusz Chroboczek");
	728	MODULE_AUTHOR("Eric Dumazet");
	729	MODULE_LICENSE("GPL");