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

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