[linux.git] / include / net / red.h

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef __NET_SCHED_RED_H
#define __NET_SCHED_RED_H

#include <linux/types.h>
#include <linux/bug.h>
#include <net/pkt_sched.h>
#include <net/inet_ecn.h>
#include <net/dsfield.h>
#include <linux/reciprocal_div.h>

/*	Random Early Detection (RED) algorithm.
	=======================================

	Source: Sally Floyd and Van Jacobson, "Random Early Detection Gateways
	for Congestion Avoidance", 1993, IEEE/ACM Transactions on Networking.

	This file codes a "divisionless" version of RED algorithm
	as written down in Fig.17 of the paper.

	Short description.
	------------------

	When a new packet arrives we calculate the average queue length:

	avg = (1-W)*avg + W*current_queue_len,

	W is the filter time constant (chosen as 2^(-Wlog)), it controls
	the inertia of the algorithm. To allow larger bursts, W should be
	decreased.

	if (avg > th_max) -> packet marked (dropped).
	if (avg < th_min) -> packet passes.
	if (th_min < avg < th_max) we calculate probability:

	Pb = max_P * (avg - th_min)/(th_max-th_min)

	and mark (drop) packet with this probability.
	Pb changes from 0 (at avg==th_min) to max_P (avg==th_max).
	max_P should be small (not 1), usually 0.01..0.02 is good value.

	max_P is chosen as a number, so that max_P/(th_max-th_min)
	is a negative power of two in order arithmetic to contain
	only shifts.


	Parameters, settable by user:
	-----------------------------

	qth_min		- bytes (should be < qth_max/2)
	qth_max		- bytes (should be at least 2*qth_min and less limit)
	Wlog	       	- bits (<32) log(1/W).
	Plog	       	- bits (<32)

	Plog is related to max_P by formula:

	max_P = (qth_max-qth_min)/2^Plog;

	F.e. if qth_max=128K and qth_min=32K, then Plog=22
	corresponds to max_P=0.02

	Scell_log
	Stab

	Lookup table for log((1-W)^(t/t_ave).


	NOTES:

	Upper bound on W.
	-----------------

	If you want to allow bursts of L packets of size S,
	you should choose W:

	L + 1 - th_min/S < (1-(1-W)^L)/W

	th_min/S = 32         th_min/S = 4

	log(W)	L
	-1	33
	-2	35
	-3	39
	-4	46
	-5	57
	-6	75
	-7	101
	-8	135
	-9	190
	etc.
 */

/*
 * Adaptative RED : An Algorithm for Increasing the Robustness of RED's AQM
 * (Sally FLoyd, Ramakrishna Gummadi, and Scott Shenker) August 2001
 *
 * Every 500 ms:
 *  if (avg > target and max_p <= 0.5)
 *   increase max_p : max_p += alpha;
 *  else if (avg < target and max_p >= 0.01)
 *   decrease max_p : max_p *= beta;
 *
 * target :[qth_min + 0.4*(qth_min - qth_max),
 *          qth_min + 0.6*(qth_min - qth_max)].
 * alpha : min(0.01, max_p / 4)
 * beta : 0.9
 * max_P is a Q0.32 fixed point number (with 32 bits mantissa)
 * max_P between 0.01 and 0.5 (1% - 50%) [ Its no longer a negative power of two ]
 */
#define RED_ONE_PERCENT ((u32)DIV_ROUND_CLOSEST(1ULL<<32, 100))

#define MAX_P_MIN (1 * RED_ONE_PERCENT)
#define MAX_P_MAX (50 * RED_ONE_PERCENT)
#define MAX_P_ALPHA(val) min(MAX_P_MIN, val / 4)

#define RED_STAB_SIZE	256
#define RED_STAB_MASK	(RED_STAB_SIZE - 1)

struct red_stats {
	u32		prob_drop;	/* Early probability drops */
	u32		prob_mark;	/* Early probability marks */
	u32		forced_drop;	/* Forced drops, qavg > max_thresh */
	u32		forced_mark;	/* Forced marks, qavg > max_thresh */
	u32		pdrop;          /* Drops due to queue limits */
};

struct red_parms {
	/* Parameters */
	u32		qth_min;	/* Min avg length threshold: Wlog scaled */
	u32		qth_max;	/* Max avg length threshold: Wlog scaled */
	u32		Scell_max;
	u32		max_P;		/* probability, [0 .. 1.0] 32 scaled */
	/* reciprocal_value(max_P / qth_delta) */
	struct reciprocal_value	max_P_reciprocal;
	u32		qth_delta;	/* max_th - min_th */
	u32		target_min;	/* min_th + 0.4*(max_th - min_th) */
	u32		target_max;	/* min_th + 0.6*(max_th - min_th) */
	u8		Scell_log;
	u8		Wlog;		/* log(W)		*/
	u8		Plog;		/* random number bits	*/
	u8		Stab[RED_STAB_SIZE];
};

struct red_vars {
	/* Variables */
	int		qcount;		/* Number of packets since last random
					   number generation */
	u32		qR;		/* Cached random number */

	unsigned long	qavg;		/* Average queue length: Wlog scaled */
	ktime_t		qidlestart;	/* Start of current idle period */
};

static inline u32 red_maxp(u8 Plog)
{
	return Plog < 32 ? (~0U >> Plog) : ~0U;
}

static inline void red_set_vars(struct red_vars *v)
{
	/* Reset average queue length, the value is strictly bound
	 * to the parameters below, resetting hurts a bit but leaving
	 * it might result in an unreasonable qavg for a while. --TGR
	 */
	v->qavg		= 0;

	v->qcount	= -1;
}

static inline bool red_check_params(u32 qth_min, u32 qth_max, u8 Wlog,
				    u8 Scell_log, u8 *stab)
{
	if (fls(qth_min) + Wlog >= 32)
		return false;
	if (fls(qth_max) + Wlog >= 32)
		return false;
	if (Scell_log >= 32)
		return false;
	if (qth_max < qth_min)
		return false;
	if (stab) {
		int i;

		for (i = 0; i < RED_STAB_SIZE; i++)
			if (stab[i] >= 32)
				return false;
	}
	return true;
}

static inline int red_get_flags(unsigned char qopt_flags,
				unsigned char historic_mask,
				struct nlattr *flags_attr,
				unsigned char supported_mask,
				struct nla_bitfield32 *p_flags,
				unsigned char *p_userbits,
				struct netlink_ext_ack *extack)
{
	struct nla_bitfield32 flags;

	if (qopt_flags && flags_attr) {
		NL_SET_ERR_MSG_MOD(extack, "flags should be passed either through qopt, or through a dedicated attribute");
		return -EINVAL;
	}

	if (flags_attr) {
		flags = nla_get_bitfield32(flags_attr);
	} else {
		flags.selector = historic_mask;
		flags.value = qopt_flags & historic_mask;
	}

	*p_flags = flags;
	*p_userbits = qopt_flags & ~historic_mask;
	return 0;
}

static inline int red_validate_flags(unsigned char flags,
				     struct netlink_ext_ack *extack)
{
	if ((flags & TC_RED_NODROP) && !(flags & TC_RED_ECN)) {
		NL_SET_ERR_MSG_MOD(extack, "nodrop mode is only meaningful with ECN");
		return -EINVAL;
	}

	return 0;
}

static inline void red_set_parms(struct red_parms *p,
				 u32 qth_min, u32 qth_max, u8 Wlog, u8 Plog,
				 u8 Scell_log, u8 *stab, u32 max_P)
{
	int delta = qth_max - qth_min;
	u32 max_p_delta;

	WRITE_ONCE(p->qth_min, qth_min << Wlog);
	WRITE_ONCE(p->qth_max, qth_max << Wlog);
	WRITE_ONCE(p->Wlog, Wlog);
	WRITE_ONCE(p->Plog, Plog);
	if (delta <= 0)
		delta = 1;
	p->qth_delta	= delta;
	if (!max_P) {
		max_P = red_maxp(Plog);
		max_P *= delta; /* max_P = (qth_max - qth_min)/2^Plog */
	}
	WRITE_ONCE(p->max_P, max_P);
	max_p_delta = max_P / delta;
	max_p_delta = max(max_p_delta, 1U);
	p->max_P_reciprocal  = reciprocal_value(max_p_delta);

	/* RED Adaptative target :
	 * [min_th + 0.4*(min_th - max_th),
	 *  min_th + 0.6*(min_th - max_th)].
	 */
	delta /= 5;
	p->target_min = qth_min + 2*delta;
	p->target_max = qth_min + 3*delta;

	WRITE_ONCE(p->Scell_log, Scell_log);
	p->Scell_max	= (255 << Scell_log);

	if (stab)
		memcpy(p->Stab, stab, sizeof(p->Stab));
}

static inline int red_is_idling(const struct red_vars *v)
{
	return v->qidlestart != 0;
}

static inline void red_start_of_idle_period(struct red_vars *v)
{
	v->qidlestart = ktime_get();
}

static inline void red_end_of_idle_period(struct red_vars *v)
{
	v->qidlestart = 0;
}

static inline void red_restart(struct red_vars *v)
{
	red_end_of_idle_period(v);
	v->qavg = 0;
	v->qcount = -1;
}

static inline unsigned long red_calc_qavg_from_idle_time(const struct red_parms *p,
							 const struct red_vars *v)
{
	s64 delta = ktime_us_delta(ktime_get(), v->qidlestart);
	long us_idle = min_t(s64, delta, p->Scell_max);
	int  shift;

	/*
	 * The problem: ideally, average length queue recalculation should
	 * be done over constant clock intervals. This is too expensive, so
	 * that the calculation is driven by outgoing packets.
	 * When the queue is idle we have to model this clock by hand.
	 *
	 * SF+VJ proposed to "generate":
	 *
	 *	m = idletime / (average_pkt_size / bandwidth)
	 *
	 * dummy packets as a burst after idle time, i.e.
	 *
	 * 	v->qavg *= (1-W)^m
	 *
	 * This is an apparently overcomplicated solution (f.e. we have to
	 * precompute a table to make this calculation in reasonable time)
	 * I believe that a simpler model may be used here,
	 * but it is field for experiments.
	 */

	shift = p->Stab[(us_idle >> p->Scell_log) & RED_STAB_MASK];

	if (shift)
		return v->qavg >> shift;
	else {
		/* Approximate initial part of exponent with linear function:
		 *
		 * 	(1-W)^m ~= 1-mW + ...
		 *
		 * Seems, it is the best solution to
		 * problem of too coarse exponent tabulation.
		 */
		us_idle = (v->qavg * (u64)us_idle) >> p->Scell_log;

		if (us_idle < (v->qavg >> 1))
			return v->qavg - us_idle;
		else
			return v->qavg >> 1;
	}
}

static inline unsigned long red_calc_qavg_no_idle_time(const struct red_parms *p,
						       const struct red_vars *v,
						       unsigned int backlog)
{
	/*
	 * NOTE: v->qavg is fixed point number with point at Wlog.
	 * The formula below is equivalent to floating point
	 * version:
	 *
	 * 	qavg = qavg*(1-W) + backlog*W;
	 *
	 * --ANK (980924)
	 */
	return v->qavg + (backlog - (v->qavg >> p->Wlog));
}

static inline unsigned long red_calc_qavg(const struct red_parms *p,
					  const struct red_vars *v,
					  unsigned int backlog)
{
	if (!red_is_idling(v))
		return red_calc_qavg_no_idle_time(p, v, backlog);
	else
		return red_calc_qavg_from_idle_time(p, v);
}


static inline u32 red_random(const struct red_parms *p)
{
	return reciprocal_divide(get_random_u32(), p->max_P_reciprocal);
}

static inline int red_mark_probability(const struct red_parms *p,
				       const struct red_vars *v,
				       unsigned long qavg)
{
	/* The formula used below causes questions.

	   OK. qR is random number in the interval
		(0..1/max_P)*(qth_max-qth_min)
	   i.e. 0..(2^Plog). If we used floating point
	   arithmetic, it would be: (2^Plog)*rnd_num,
	   where rnd_num is less 1.

	   Taking into account, that qavg have fixed
	   point at Wlog, two lines
	   below have the following floating point equivalent:

	   max_P*(qavg - qth_min)/(qth_max-qth_min) < rnd/qcount

	   Any questions? --ANK (980924)
	 */
	return !(((qavg - p->qth_min) >> p->Wlog) * v->qcount < v->qR);
}

enum {
	RED_BELOW_MIN_THRESH,
	RED_BETWEEN_TRESH,
	RED_ABOVE_MAX_TRESH,
};

static inline int red_cmp_thresh(const struct red_parms *p, unsigned long qavg)
{
	if (qavg < p->qth_min)
		return RED_BELOW_MIN_THRESH;
	else if (qavg >= p->qth_max)
		return RED_ABOVE_MAX_TRESH;
	else
		return RED_BETWEEN_TRESH;
}

enum {
	RED_DONT_MARK,
	RED_PROB_MARK,
	RED_HARD_MARK,
};

static inline int red_action(const struct red_parms *p,
			     struct red_vars *v,
			     unsigned long qavg)
{
	switch (red_cmp_thresh(p, qavg)) {
		case RED_BELOW_MIN_THRESH:
			v->qcount = -1;
			return RED_DONT_MARK;

		case RED_BETWEEN_TRESH:
			if (++v->qcount) {
				if (red_mark_probability(p, v, qavg)) {
					v->qcount = 0;
					v->qR = red_random(p);
					return RED_PROB_MARK;
				}
			} else
				v->qR = red_random(p);

			return RED_DONT_MARK;

		case RED_ABOVE_MAX_TRESH:
			v->qcount = -1;
			return RED_HARD_MARK;
	}

	BUG();
	return RED_DONT_MARK;
}

static inline void red_adaptative_algo(struct red_parms *p, struct red_vars *v)
{
	unsigned long qavg;
	u32 max_p_delta;

	qavg = v->qavg;
	if (red_is_idling(v))
		qavg = red_calc_qavg_from_idle_time(p, v);

	/* v->qavg is fixed point number with point at Wlog */
	qavg >>= p->Wlog;

	if (qavg > p->target_max && p->max_P <= MAX_P_MAX)
		p->max_P += MAX_P_ALPHA(p->max_P); /* maxp = maxp + alpha */
	else if (qavg < p->target_min && p->max_P >= MAX_P_MIN)
		p->max_P = (p->max_P/10)*9; /* maxp = maxp * Beta */

	max_p_delta = DIV_ROUND_CLOSEST(p->max_P, p->qth_delta);
	max_p_delta = max(max_p_delta, 1U);
	p->max_P_reciprocal = reciprocal_value(max_p_delta);
}
#endif
Commit	Line	Data
b2441318	1	/* SPDX-License-Identifier: GPL-2.0 */
a7834745 TG	2	#ifndef __NET_SCHED_RED_H
	3	#define __NET_SCHED_RED_H
	4
a7834745	5	#include <linux/types.h>
187f1882	6	#include <linux/bug.h>
a7834745 TG	7	#include <net/pkt_sched.h>
	8	#include <net/inet_ecn.h>
	9	#include <net/dsfield.h>
8af2a218	10	#include <linux/reciprocal_div.h>
a7834745 TG	11
	12	/* Random Early Detection (RED) algorithm.
	13	=======================================
	14
	15	Source: Sally Floyd and Van Jacobson, "Random Early Detection Gateways
	16	for Congestion Avoidance", 1993, IEEE/ACM Transactions on Networking.
	17
	18	This file codes a "divisionless" version of RED algorithm
	19	as written down in Fig.17 of the paper.
	20
	21	Short description.
	22	------------------
	23
	24	When a new packet arrives we calculate the average queue length:
	25
	26	avg = (1-W)avg + Wcurrent_queue_len,
	27
	28	W is the filter time constant (chosen as 2^(-Wlog)), it controls
	29	the inertia of the algorithm. To allow larger bursts, W should be
	30	decreased.
	31
	32	if (avg > th_max) -> packet marked (dropped).
	33	if (avg < th_min) -> packet passes.
	34	if (th_min < avg < th_max) we calculate probability:
	35
	36	Pb = max_P * (avg - th_min)/(th_max-th_min)
	37
	38	and mark (drop) packet with this probability.
	39	Pb changes from 0 (at avg==th_min) to max_P (avg==th_max).
	40	max_P should be small (not 1), usually 0.01..0.02 is good value.
	41
	42	max_P is chosen as a number, so that max_P/(th_max-th_min)
a7a45f02	43	is a negative power of two in order arithmetic to contain
a7834745 TG	44	only shifts.
	45
	46
	47	Parameters, settable by user:
	48	-----------------------------
	49
	50	qth_min - bytes (should be < qth_max/2)
	51	qth_max - bytes (should be at least 2*qth_min and less limit)
	52	Wlog - bits (<32) log(1/W).
	53	Plog - bits (<32)
	54
	55	Plog is related to max_P by formula:
	56
	57	max_P = (qth_max-qth_min)/2^Plog;
	58
	59	F.e. if qth_max=128K and qth_min=32K, then Plog=22
	60	corresponds to max_P=0.02
	61
	62	Scell_log
	63	Stab
	64
	65	Lookup table for log((1-W)^(t/t_ave).
	66
	67
	68	NOTES:
	69
	70	Upper bound on W.
	71	-----------------
	72
	73	If you want to allow bursts of L packets of size S,
	74	you should choose W:
	75
	76	L + 1 - th_min/S < (1-(1-W)^L)/W
	77
	78	th_min/S = 32 th_min/S = 4
	79
	80	log(W) L
	81	-1 33
	82	-2 35
	83	-3 39
	84	-4 46
	85	-5 57
	86	-6 75
	87	-7 101
	88	-8 135
	89	-9 190
	90	etc.
	91	*/
	92
8af2a218 ED	93	/*
	94	* Adaptative RED : An Algorithm for Increasing the Robustness of RED's AQM
	95	* (Sally FLoyd, Ramakrishna Gummadi, and Scott Shenker) August 2001
	96	*
	97	* Every 500 ms:
	98	* if (avg > target and max_p <= 0.5)
	99	* increase max_p : max_p += alpha;
	100	* else if (avg < target and max_p >= 0.01)
	101	* decrease max_p : max_p *= beta;
	102	*
	103	* target :[qth_min + 0.4*(qth_min - qth_max),
	104	* qth_min + 0.6*(qth_min - qth_max)].
	105	* alpha : min(0.01, max_p / 4)
	106	* beta : 0.9
	107	* max_P is a Q0.32 fixed point number (with 32 bits mantissa)
	108	* max_P between 0.01 and 0.5 (1% - 50%) [ Its no longer a negative power of two ]
	109	*/
	110	#define RED_ONE_PERCENT ((u32)DIV_ROUND_CLOSEST(1ULL<<32, 100))
	111
	112	#define MAX_P_MIN (1 * RED_ONE_PERCENT)
	113	#define MAX_P_MAX (50 * RED_ONE_PERCENT)
	114	#define MAX_P_ALPHA(val) min(MAX_P_MIN, val / 4)
	115
a7834745 TG	116	#define RED_STAB_SIZE 256
	117	#define RED_STAB_MASK (RED_STAB_SIZE - 1)
	118
fd2c3ef7	119	struct red_stats {
a7834745 TG	120	u32 prob_drop; /* Early probability drops */
	121	u32 prob_mark; /* Early probability marks */
	122	u32 forced_drop; /* Forced drops, qavg > max_thresh */
	123	u32 forced_mark; /* Forced marks, qavg > max_thresh */
	124	u32 pdrop; /* Drops due to queue limits */
a7834745 TG	125	};
a7834745 TG	126
fd2c3ef7	127	struct red_parms {
a7834745	128	/* Parameters */
8af2a218 ED	129	u32 qth_min; /* Min avg length threshold: Wlog scaled */
8af2a218 ED	130	u32 qth_max; /* Max avg length threshold: Wlog scaled */
a7834745	131	u32 Scell_max;
8af2a218	132	u32 max_P; /* probability, [0 .. 1.0] 32 scaled */
809fa972 HFS	133	/* reciprocal_value(max_P / qth_delta) */
809fa972 HFS	134	struct reciprocal_value max_P_reciprocal;
8af2a218 ED	135	u32 qth_delta; /* max_th - min_th */
	136	u32 target_min; /* min_th + 0.4(max_th - min_th) /
	137	u32 target_max; /* min_th + 0.6(max_th - min_th) /
a7834745 TG	138	u8 Scell_log;
	139	u8 Wlog; /* log(W) */
	140	u8 Plog; /* random number bits */
	141	u8 Stab[RED_STAB_SIZE];
eeca6688	142	};
a7834745	143
eeca6688	144	struct red_vars {
a7834745 TG	145	/* Variables */
	146	int qcount; /* Number of packets since last random
	147	number generation */
	148	u32 qR; /* Cached random number */
	149
8af2a218	150	unsigned long qavg; /* Average queue length: Wlog scaled */
ea6a5d3b	151	ktime_t qidlestart; /* Start of current idle period */
a7834745 TG	152	};
a7834745 TG	153
8af2a218	154	static inline u32 red_maxp(u8 Plog)
a7834745	155	{
8af2a218	156	return Plog < 32 ? (~0U >> Plog) : ~0U;
a7834745 TG	157	}
a7834745 TG	158
eeca6688 ED	159	static inline void red_set_vars(struct red_vars *v)
	160	{
	161	/* Reset average queue length, the value is strictly bound
a7a45f02	162	* to the parameters below, resetting hurts a bit but leaving
eeca6688 ED	163	* it might result in an unreasonable qavg for a while. --TGR
	164	*/
	165	v->qavg = 0;
	166
	167	v->qcount = -1;
	168	}
8af2a218	169
e323d865 ED	170	static inline bool red_check_params(u32 qth_min, u32 qth_max, u8 Wlog,
e323d865 ED	171	u8 Scell_log, u8 *stab)
8afa10cb	172	{
3a87571f	173	if (fls(qth_min) + Wlog >= 32)
8afa10cb	174	return false;
3a87571f	175	if (fls(qth_max) + Wlog >= 32)
8afa10cb	176	return false;
bd1248f1 RD	177	if (Scell_log >= 32)
bd1248f1 RD	178	return false;
8afa10cb NF	179	if (qth_max < qth_min)
8afa10cb NF	180	return false;
e323d865 ED	181	if (stab) {
	182	int i;
	183
	184	for (i = 0; i < RED_STAB_SIZE; i++)
	185	if (stab[i] >= 32)
	186	return false;
	187	}
8afa10cb NF	188	return true;
	189	}
	190
14bc175d PM	191	static inline int red_get_flags(unsigned char qopt_flags,
	192	unsigned char historic_mask,
	193	struct nlattr *flags_attr,
	194	unsigned char supported_mask,
	195	struct nla_bitfield32 *p_flags,
	196	unsigned char *p_userbits,
	197	struct netlink_ext_ack *extack)
	198	{
	199	struct nla_bitfield32 flags;
	200
	201	if (qopt_flags && flags_attr) {
	202	NL_SET_ERR_MSG_MOD(extack, "flags should be passed either through qopt, or through a dedicated attribute");
	203	return -EINVAL;
	204	}
	205
	206	if (flags_attr) {
	207	flags = nla_get_bitfield32(flags_attr);
	208	} else {
	209	flags.selector = historic_mask;
	210	flags.value = qopt_flags & historic_mask;
	211	}
	212
	213	*p_flags = flags;
	214	*p_userbits = qopt_flags & ~historic_mask;
	215	return 0;
	216	}
	217
	218	static inline int red_validate_flags(unsigned char flags,
	219	struct netlink_ext_ack *extack)
	220	{
0a7fad23 PM	221	if ((flags & TC_RED_NODROP) && !(flags & TC_RED_ECN)) {
	222	NL_SET_ERR_MSG_MOD(extack, "nodrop mode is only meaningful with ECN");
	223	return -EINVAL;
	224	}
	225
14bc175d PM	226	return 0;
	227	}
	228
a7834745 TG	229	static inline void red_set_parms(struct red_parms *p,
a7834745 TG	230	u32 qth_min, u32 qth_max, u8 Wlog, u8 Plog,
a73ed26b	231	u8 Scell_log, u8 *stab, u32 max_P)
a7834745	232	{
8af2a218	233	int delta = qth_max - qth_min;
a73ed26b	234	u32 max_p_delta;
8af2a218	235
7253c1d1 ED	236	WRITE_ONCE(p->qth_min, qth_min << Wlog);
	237	WRITE_ONCE(p->qth_max, qth_max << Wlog);
	238	WRITE_ONCE(p->Wlog, Wlog);
	239	WRITE_ONCE(p->Plog, Plog);
5c472203	240	if (delta <= 0)
8af2a218 ED	241	delta = 1;
8af2a218 ED	242	p->qth_delta = delta;
a73ed26b ED	243	if (!max_P) {
	244	max_P = red_maxp(Plog);
	245	max_P = delta; / max_P = (qth_max - qth_min)/2^Plog */
	246	}
7253c1d1	247	WRITE_ONCE(p->max_P, max_P);
a73ed26b ED	248	max_p_delta = max_P / delta;
	249	max_p_delta = max(max_p_delta, 1U);
	250	p->max_P_reciprocal = reciprocal_value(max_p_delta);
8af2a218 ED	251
	252	/* RED Adaptative target :
	253	* [min_th + 0.4*(min_th - max_th),
	254	* min_th + 0.6*(min_th - max_th)].
	255	*/
	256	delta /= 5;
	257	p->target_min = qth_min + 2*delta;
	258	p->target_max = qth_min + 3*delta;
	259
7253c1d1	260	WRITE_ONCE(p->Scell_log, Scell_log);
a7834745 TG	261	p->Scell_max = (255 << Scell_log);
a7834745 TG	262
ddecf0f4 ED	263	if (stab)
ddecf0f4 ED	264	memcpy(p->Stab, stab, sizeof(p->Stab));
a7834745 TG	265	}
a7834745 TG	266
eeca6688	267	static inline int red_is_idling(const struct red_vars *v)
a7834745	268	{
2456e855	269	return v->qidlestart != 0;
a7834745 TG	270	}
a7834745 TG	271
eeca6688	272	static inline void red_start_of_idle_period(struct red_vars *v)
a7834745	273	{
eeca6688	274	v->qidlestart = ktime_get();
a7834745 TG	275	}
a7834745 TG	276
eeca6688	277	static inline void red_end_of_idle_period(struct red_vars *v)
a7834745	278	{
2456e855	279	v->qidlestart = 0;
a7834745 TG	280	}
a7834745 TG	281
eeca6688	282	static inline void red_restart(struct red_vars *v)
a7834745	283	{
eeca6688 ED	284	red_end_of_idle_period(v);
	285	v->qavg = 0;
	286	v->qcount = -1;
a7834745 TG	287	}
a7834745 TG	288
eeca6688 ED	289	static inline unsigned long red_calc_qavg_from_idle_time(const struct red_parms *p,
eeca6688 ED	290	const struct red_vars *v)
a7834745	291	{
eeca6688	292	s64 delta = ktime_us_delta(ktime_get(), v->qidlestart);
ea6a5d3b	293	long us_idle = min_t(s64, delta, p->Scell_max);
a7834745 TG	294	int shift;
a7834745 TG	295
a7834745	296	/*
8a2dc6af	297	* The problem: ideally, average length queue recalculation should
a7834745 TG	298	* be done over constant clock intervals. This is too expensive, so
	299	* that the calculation is driven by outgoing packets.
	300	* When the queue is idle we have to model this clock by hand.
	301	*
	302	* SF+VJ proposed to "generate":
	303	*
	304	* m = idletime / (average_pkt_size / bandwidth)
	305	*
	306	* dummy packets as a burst after idle time, i.e.
	307	*
4362aaf6	308	* v->qavg *= (1-W)^m
a7834745 TG	309	*
	310	* This is an apparently overcomplicated solution (f.e. we have to
	311	* precompute a table to make this calculation in reasonable time)
	312	* I believe that a simpler model may be used here,
	313	* but it is field for experiments.
	314	*/
	315
	316	shift = p->Stab[(us_idle >> p->Scell_log) & RED_STAB_MASK];
	317
	318	if (shift)
eeca6688	319	return v->qavg >> shift;
a7834745 TG	320	else {
	321	/* Approximate initial part of exponent with linear function:
	322	*
	323	* (1-W)^m ~= 1-mW + ...
	324	*
	325	* Seems, it is the best solution to
	326	* problem of too coarse exponent tabulation.
	327	*/
eeca6688	328	us_idle = (v->qavg * (u64)us_idle) >> p->Scell_log;
a7834745	329
eeca6688 ED	330	if (us_idle < (v->qavg >> 1))
eeca6688 ED	331	return v->qavg - us_idle;
a7834745	332	else
eeca6688	333	return v->qavg >> 1;
a7834745 TG	334	}
	335	}
	336
8af2a218	337	static inline unsigned long red_calc_qavg_no_idle_time(const struct red_parms *p,
eeca6688	338	const struct red_vars *v,
a7834745 TG	339	unsigned int backlog)
	340	{
	341	/*
4362aaf6	342	* NOTE: v->qavg is fixed point number with point at Wlog.
a7a45f02	343	* The formula below is equivalent to floating point
a7834745 TG	344	* version:
	345	*
	346	* qavg = qavg(1-W) + backlogW;
	347	*
	348	* --ANK (980924)
	349	*/
eeca6688	350	return v->qavg + (backlog - (v->qavg >> p->Wlog));
a7834745 TG	351	}
a7834745 TG	352
8af2a218	353	static inline unsigned long red_calc_qavg(const struct red_parms *p,
eeca6688	354	const struct red_vars *v,
a7834745 TG	355	unsigned int backlog)
a7834745 TG	356	{
eeca6688 ED	357	if (!red_is_idling(v))
eeca6688 ED	358	return red_calc_qavg_no_idle_time(p, v, backlog);
a7834745	359	else
eeca6688	360	return red_calc_qavg_from_idle_time(p, v);
a7834745 TG	361	}
a7834745 TG	362
8af2a218 ED	363
8af2a218 ED	364	static inline u32 red_random(const struct red_parms *p)
a7834745	365	{
a251c17a	366	return reciprocal_divide(get_random_u32(), p->max_P_reciprocal);
a7834745 TG	367	}
a7834745 TG	368
eeca6688 ED	369	static inline int red_mark_probability(const struct red_parms *p,
	370	const struct red_vars *v,
	371	unsigned long qavg)
a7834745 TG	372	{
	373	/* The formula used below causes questions.
	374
8af2a218 ED	375	OK. qR is random number in the interval
8af2a218 ED	376	(0..1/max_P)*(qth_max-qth_min)
a7834745	377	i.e. 0..(2^Plog). If we used floating point
a7a45f02	378	arithmetic, it would be: (2^Plog)*rnd_num,
a7834745 TG	379	where rnd_num is less 1.
	380
	381	Taking into account, that qavg have fixed
8af2a218	382	point at Wlog, two lines
a7834745 TG	383	below have the following floating point equivalent:
	384
	385	max_P*(qavg - qth_min)/(qth_max-qth_min) < rnd/qcount
	386
	387	Any questions? --ANK (980924)
	388	*/
eeca6688	389	return !(((qavg - p->qth_min) >> p->Wlog) * v->qcount < v->qR);
a7834745 TG	390	}
	391
	392	enum {
	393	RED_BELOW_MIN_THRESH,
	394	RED_BETWEEN_TRESH,
	395	RED_ABOVE_MAX_TRESH,
	396	};
	397
eeca6688	398	static inline int red_cmp_thresh(const struct red_parms *p, unsigned long qavg)
a7834745 TG	399	{
	400	if (qavg < p->qth_min)
	401	return RED_BELOW_MIN_THRESH;
	402	else if (qavg >= p->qth_max)
	403	return RED_ABOVE_MAX_TRESH;
	404	else
	405	return RED_BETWEEN_TRESH;
	406	}
	407
	408	enum {
	409	RED_DONT_MARK,
	410	RED_PROB_MARK,
	411	RED_HARD_MARK,
	412	};
	413
eeca6688 ED	414	static inline int red_action(const struct red_parms *p,
	415	struct red_vars *v,
	416	unsigned long qavg)
a7834745 TG	417	{
	418	switch (red_cmp_thresh(p, qavg)) {
	419	case RED_BELOW_MIN_THRESH:
eeca6688	420	v->qcount = -1;
a7834745 TG	421	return RED_DONT_MARK;
	422
	423	case RED_BETWEEN_TRESH:
eeca6688 ED	424	if (++v->qcount) {
	425	if (red_mark_probability(p, v, qavg)) {
	426	v->qcount = 0;
	427	v->qR = red_random(p);
a7834745 TG	428	return RED_PROB_MARK;
	429	}
	430	} else
eeca6688	431	v->qR = red_random(p);
a7834745 TG	432
	433	return RED_DONT_MARK;
	434
	435	case RED_ABOVE_MAX_TRESH:
eeca6688	436	v->qcount = -1;
a7834745 TG	437	return RED_HARD_MARK;
	438	}
	439
	440	BUG();
	441	return RED_DONT_MARK;
	442	}
	443
eeca6688	444	static inline void red_adaptative_algo(struct red_parms p, struct red_vars v)
8af2a218 ED	445	{
	446	unsigned long qavg;
	447	u32 max_p_delta;
	448
eeca6688 ED	449	qavg = v->qavg;
	450	if (red_is_idling(v))
	451	qavg = red_calc_qavg_from_idle_time(p, v);
8af2a218	452
4362aaf6	453	/* v->qavg is fixed point number with point at Wlog */
8af2a218 ED	454	qavg >>= p->Wlog;
	455
	456	if (qavg > p->target_max && p->max_P <= MAX_P_MAX)
	457	p->max_P += MAX_P_ALPHA(p->max_P); /* maxp = maxp + alpha */
	458	else if (qavg < p->target_min && p->max_P >= MAX_P_MIN)
	459	p->max_P = (p->max_P/10)9; / maxp = maxp * Beta */
	460
	461	max_p_delta = DIV_ROUND_CLOSEST(p->max_P, p->qth_delta);
a73ed26b	462	max_p_delta = max(max_p_delta, 1U);
8af2a218 ED	463	p->max_P_reciprocal = reciprocal_value(max_p_delta);
8af2a218 ED	464	}
a7834745	465	#endif