[linux-2.6-block.git] / net / ipv4 / tcp_bic.c

/*
 * Binary Increase Congestion control for TCP
 * Home page:
 *      http://netsrv.csc.ncsu.edu/twiki/bin/view/Main/BIC
 * This is from the implementation of BICTCP in
 * Lison-Xu, Kahaled Harfoush, and Injong Rhee.
 *  "Binary Increase Congestion Control for Fast, Long Distance
 *  Networks" in InfoComm 2004
 * Available from:
 *  http://netsrv.csc.ncsu.edu/export/bitcp.pdf
 *
 * Unless BIC is enabled and congestion window is large
 * this behaves the same as the original Reno.
 */

#include <linux/mm.h>
#include <linux/module.h>
#include <net/tcp.h>

#define BICTCP_BETA_SCALE    1024	/* Scale factor beta calculation
					 * max_cwnd = snd_cwnd * beta
					 */
#define BICTCP_B		4	 /*
					  * In binary search,
					  * go to point (max+min)/N
					  */

static int fast_convergence = 1;
static int max_increment = 16;
static int low_window = 14;
static int beta = 819;		/* = 819/1024 (BICTCP_BETA_SCALE) */
static int initial_ssthresh;
static int smooth_part = 20;

module_param(fast_convergence, int, 0644);
MODULE_PARM_DESC(fast_convergence, "turn on/off fast convergence");
module_param(max_increment, int, 0644);
MODULE_PARM_DESC(max_increment, "Limit on increment allowed during binary search");
module_param(low_window, int, 0644);
MODULE_PARM_DESC(low_window, "lower bound on congestion window (for TCP friendliness)");
module_param(beta, int, 0644);
MODULE_PARM_DESC(beta, "beta for multiplicative increase");
module_param(initial_ssthresh, int, 0644);
MODULE_PARM_DESC(initial_ssthresh, "initial value of slow start threshold");
module_param(smooth_part, int, 0644);
MODULE_PARM_DESC(smooth_part, "log(B/(B*Smin))/log(B/(B-1))+B, # of RTT from Wmax-B to Wmax");

/* BIC TCP Parameters */
struct bictcp {
	u32	cnt;		/* increase cwnd by 1 after ACKs */
	u32	last_max_cwnd;	/* last maximum snd_cwnd */
	u32	last_cwnd;	/* the last snd_cwnd */
	u32	last_time;	/* time when updated last_cwnd */
	u32	epoch_start;	/* beginning of an epoch */
#define ACK_RATIO_SHIFT	4
	u32	delayed_ack;	/* estimate the ratio of Packets/ACKs << 4 */
};

static inline void bictcp_reset(struct bictcp *ca)
{
	ca->cnt = 0;
	ca->last_max_cwnd = 0;
	ca->last_cwnd = 0;
	ca->last_time = 0;
	ca->epoch_start = 0;
	ca->delayed_ack = 2 << ACK_RATIO_SHIFT;
}

static void bictcp_init(struct sock *sk)
{
	struct bictcp *ca = inet_csk_ca(sk);

	bictcp_reset(ca);

	if (initial_ssthresh)
		tcp_sk(sk)->snd_ssthresh = initial_ssthresh;
}

/*
 * Compute congestion window to use.
 */
static inline void bictcp_update(struct bictcp *ca, u32 cwnd)
{
	if (ca->last_cwnd == cwnd &&
	    (s32)(tcp_jiffies32 - ca->last_time) <= HZ / 32)
		return;

	ca->last_cwnd = cwnd;
	ca->last_time = tcp_jiffies32;

	if (ca->epoch_start == 0) /* record the beginning of an epoch */
		ca->epoch_start = tcp_jiffies32;

	/* start off normal */
	if (cwnd <= low_window) {
		ca->cnt = cwnd;
		return;
	}

	/* binary increase */
	if (cwnd < ca->last_max_cwnd) {
		__u32	dist = (ca->last_max_cwnd - cwnd)
			/ BICTCP_B;

		if (dist > max_increment)
			/* linear increase */
			ca->cnt = cwnd / max_increment;
		else if (dist <= 1U)
			/* binary search increase */
			ca->cnt = (cwnd * smooth_part) / BICTCP_B;
		else
			/* binary search increase */
			ca->cnt = cwnd / dist;
	} else {
		/* slow start AMD linear increase */
		if (cwnd < ca->last_max_cwnd + BICTCP_B)
			/* slow start */
			ca->cnt = (cwnd * smooth_part) / BICTCP_B;
		else if (cwnd < ca->last_max_cwnd + max_increment*(BICTCP_B-1))
			/* slow start */
			ca->cnt = (cwnd * (BICTCP_B-1))
				/ (cwnd - ca->last_max_cwnd);
		else
			/* linear increase */
			ca->cnt = cwnd / max_increment;
	}

	/* if in slow start or link utilization is very low */
	if (ca->last_max_cwnd == 0) {
		if (ca->cnt > 20) /* increase cwnd 5% per RTT */
			ca->cnt = 20;
	}

	ca->cnt = (ca->cnt << ACK_RATIO_SHIFT) / ca->delayed_ack;
	if (ca->cnt == 0)			/* cannot be zero */
		ca->cnt = 1;
}

static void bictcp_cong_avoid(struct sock *sk, u32 ack, u32 acked)
{
	struct tcp_sock *tp = tcp_sk(sk);
	struct bictcp *ca = inet_csk_ca(sk);

	if (!tcp_is_cwnd_limited(sk))
		return;

	if (tcp_in_slow_start(tp))
		tcp_slow_start(tp, acked);
	else {
		bictcp_update(ca, tp->snd_cwnd);
		tcp_cong_avoid_ai(tp, ca->cnt, 1);
	}
}

/*
 *	behave like Reno until low_window is reached,
 *	then increase congestion window slowly
 */
static u32 bictcp_recalc_ssthresh(struct sock *sk)
{
	const struct tcp_sock *tp = tcp_sk(sk);
	struct bictcp *ca = inet_csk_ca(sk);

	ca->epoch_start = 0;	/* end of epoch */

	/* Wmax and fast convergence */
	if (tp->snd_cwnd < ca->last_max_cwnd && fast_convergence)
		ca->last_max_cwnd = (tp->snd_cwnd * (BICTCP_BETA_SCALE + beta))
			/ (2 * BICTCP_BETA_SCALE);
	else
		ca->last_max_cwnd = tp->snd_cwnd;

	if (tp->snd_cwnd <= low_window)
		return max(tp->snd_cwnd >> 1U, 2U);
	else
		return max((tp->snd_cwnd * beta) / BICTCP_BETA_SCALE, 2U);
}

static void bictcp_state(struct sock *sk, u8 new_state)
{
	if (new_state == TCP_CA_Loss)
		bictcp_reset(inet_csk_ca(sk));
}

/* Track delayed acknowledgment ratio using sliding window
 * ratio = (15*ratio + sample) / 16
 */
static void bictcp_acked(struct sock *sk, const struct ack_sample *sample)
{
	const struct inet_connection_sock *icsk = inet_csk(sk);

	if (icsk->icsk_ca_state == TCP_CA_Open) {
		struct bictcp *ca = inet_csk_ca(sk);

		ca->delayed_ack += sample->pkts_acked -
			(ca->delayed_ack >> ACK_RATIO_SHIFT);
	}
}

static struct tcp_congestion_ops bictcp __read_mostly = {
	.init		= bictcp_init,
	.ssthresh	= bictcp_recalc_ssthresh,
	.cong_avoid	= bictcp_cong_avoid,
	.set_state	= bictcp_state,
	.undo_cwnd	= tcp_reno_undo_cwnd,
	.pkts_acked     = bictcp_acked,
	.owner		= THIS_MODULE,
	.name		= "bic",
};

static int __init bictcp_register(void)
{
	BUILD_BUG_ON(sizeof(struct bictcp) > ICSK_CA_PRIV_SIZE);
	return tcp_register_congestion_control(&bictcp);
}

static void __exit bictcp_unregister(void)
{
	tcp_unregister_congestion_control(&bictcp);
}

module_init(bictcp_register);
module_exit(bictcp_unregister);

MODULE_AUTHOR("Stephen Hemminger");
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("BIC TCP");
Commit	Line	Data
83803034 SH	1	/*
83803034 SH	2	* Binary Increase Congestion control for TCP
0bc8c7bf SH	3	* Home page:
0bc8c7bf SH	4	* http://netsrv.csc.ncsu.edu/twiki/bin/view/Main/BIC
83803034 SH	5	* This is from the implementation of BICTCP in
	6	* Lison-Xu, Kahaled Harfoush, and Injong Rhee.
	7	* "Binary Increase Congestion Control for Fast, Long Distance
	8	* Networks" in InfoComm 2004
	9	* Available from:
0bc8c7bf	10	* http://netsrv.csc.ncsu.edu/export/bitcp.pdf
83803034 SH	11	*
	12	* Unless BIC is enabled and congestion window is large
	13	* this behaves the same as the original Reno.
	14	*/
	15
83803034 SH	16	#include <linux/mm.h>
	17	#include <linux/module.h>
	18	#include <net/tcp.h>
	19
83803034 SH	20	#define BICTCP_BETA_SCALE 1024 /* Scale factor beta calculation
	21	* max_cwnd = snd_cwnd * beta
	22	*/
	23	#define BICTCP_B 4 /*
	24	* In binary search,
	25	* go to point (max+min)/N
	26	*/
	27
	28	static int fast_convergence = 1;
450b5b18	29	static int max_increment = 16;
83803034 SH	30	static int low_window = 14;
83803034 SH	31	static int beta = 819; /* = 819/1024 (BICTCP_BETA_SCALE) */
66e1e3b2	32	static int initial_ssthresh;
83803034 SH	33	static int smooth_part = 20;
	34
	35	module_param(fast_convergence, int, 0644);
	36	MODULE_PARM_DESC(fast_convergence, "turn on/off fast convergence");
	37	module_param(max_increment, int, 0644);
	38	MODULE_PARM_DESC(max_increment, "Limit on increment allowed during binary search");
	39	module_param(low_window, int, 0644);
	40	MODULE_PARM_DESC(low_window, "lower bound on congestion window (for TCP friendliness)");
	41	module_param(beta, int, 0644);
	42	MODULE_PARM_DESC(beta, "beta for multiplicative increase");
83803034 SH	43	module_param(initial_ssthresh, int, 0644);
	44	MODULE_PARM_DESC(initial_ssthresh, "initial value of slow start threshold");
	45	module_param(smooth_part, int, 0644);
	46	MODULE_PARM_DESC(smooth_part, "log(B/(B*Smin))/log(B/(B-1))+B, # of RTT from Wmax-B to Wmax");
	47
83803034 SH	48	/* BIC TCP Parameters */
	49	struct bictcp {
	50	u32 cnt; /* increase cwnd by 1 after ACKs */
688d1945	51	u32 last_max_cwnd; /* last maximum snd_cwnd */
83803034 SH	52	u32 last_cwnd; /* the last snd_cwnd */
83803034 SH	53	u32 last_time; /* time when updated last_cwnd */
83803034 SH	54	u32 epoch_start; /* beginning of an epoch */
	55	#define ACK_RATIO_SHIFT 4
	56	u32 delayed_ack; /* estimate the ratio of Packets/ACKs << 4 */
	57	};
	58
	59	static inline void bictcp_reset(struct bictcp *ca)
	60	{
	61	ca->cnt = 0;
	62	ca->last_max_cwnd = 0;
83803034 SH	63	ca->last_cwnd = 0;
83803034 SH	64	ca->last_time = 0;
83803034 SH	65	ca->epoch_start = 0;
	66	ca->delayed_ack = 2 << ACK_RATIO_SHIFT;
	67	}
	68
6687e988	69	static void bictcp_init(struct sock *sk)
83803034	70	{
fc16dcd8 NC	71	struct bictcp *ca = inet_csk_ca(sk);
	72
	73	bictcp_reset(ca);
fc16dcd8	74
83803034	75	if (initial_ssthresh)
6687e988	76	tcp_sk(sk)->snd_ssthresh = initial_ssthresh;
83803034 SH	77	}
	78
	79	/*
	80	* Compute congestion window to use.
	81	*/
	82	static inline void bictcp_update(struct bictcp *ca, u32 cwnd)
	83	{
	84	if (ca->last_cwnd == cwnd &&
ac35f562	85	(s32)(tcp_jiffies32 - ca->last_time) <= HZ / 32)
83803034 SH	86	return;
	87
	88	ca->last_cwnd = cwnd;
ac35f562	89	ca->last_time = tcp_jiffies32;
83803034 SH	90
83803034 SH	91	if (ca->epoch_start == 0) /* record the beginning of an epoch */
ac35f562	92	ca->epoch_start = tcp_jiffies32;
83803034 SH	93
	94	/* start off normal */
	95	if (cwnd <= low_window) {
	96	ca->cnt = cwnd;
	97	return;
	98	}
	99
	100	/* binary increase */
	101	if (cwnd < ca->last_max_cwnd) {
688d1945	102	__u32 dist = (ca->last_max_cwnd - cwnd)
83803034 SH	103	/ BICTCP_B;
	104
	105	if (dist > max_increment)
	106	/* linear increase */
	107	ca->cnt = cwnd / max_increment;
	108	else if (dist <= 1U)
	109	/* binary search increase */
	110	ca->cnt = (cwnd * smooth_part) / BICTCP_B;
	111	else
	112	/* binary search increase */
	113	ca->cnt = cwnd / dist;
	114	} else {
	115	/* slow start AMD linear increase */
	116	if (cwnd < ca->last_max_cwnd + BICTCP_B)
	117	/* slow start */
	118	ca->cnt = (cwnd * smooth_part) / BICTCP_B;
	119	else if (cwnd < ca->last_max_cwnd + max_increment*(BICTCP_B-1))
	120	/* slow start */
	121	ca->cnt = (cwnd * (BICTCP_B-1))
42a39450	122	/ (cwnd - ca->last_max_cwnd);
83803034 SH	123	else
	124	/* linear increase */
	125	ca->cnt = cwnd / max_increment;
	126	}
	127
	128	/* if in slow start or link utilization is very low */
fc16dcd8	129	if (ca->last_max_cwnd == 0) {
83803034 SH	130	if (ca->cnt > 20) /* increase cwnd 5% per RTT */
	131	ca->cnt = 20;
	132	}
	133
	134	ca->cnt = (ca->cnt << ACK_RATIO_SHIFT) / ca->delayed_ack;
	135	if (ca->cnt == 0) /* cannot be zero */
	136	ca->cnt = 1;
	137	}
	138
24901551	139	static void bictcp_cong_avoid(struct sock *sk, u32 ack, u32 acked)
83803034	140	{
6687e988 ACM	141	struct tcp_sock *tp = tcp_sk(sk);
6687e988 ACM	142	struct bictcp *ca = inet_csk_ca(sk);
83803034	143
24901551	144	if (!tcp_is_cwnd_limited(sk))
83803034 SH	145	return;
83803034 SH	146
071d5080	147	if (tcp_in_slow_start(tp))
9f9843a7	148	tcp_slow_start(tp, acked);
7faffa1c	149	else {
83803034	150	bictcp_update(ca, tp->snd_cwnd);
e73ebb08	151	tcp_cong_avoid_ai(tp, ca->cnt, 1);
83803034	152	}
83803034 SH	153	}
	154
	155	/*
	156	* behave like Reno until low_window is reached,
	157	* then increase congestion window slowly
	158	*/
6687e988	159	static u32 bictcp_recalc_ssthresh(struct sock *sk)
83803034	160	{
6687e988 ACM	161	const struct tcp_sock *tp = tcp_sk(sk);
6687e988 ACM	162	struct bictcp *ca = inet_csk_ca(sk);
83803034 SH	163
	164	ca->epoch_start = 0; /* end of epoch */
	165
83803034 SH	166	/* Wmax and fast convergence */
	167	if (tp->snd_cwnd < ca->last_max_cwnd && fast_convergence)
	168	ca->last_max_cwnd = (tp->snd_cwnd * (BICTCP_BETA_SCALE + beta))
	169	/ (2 * BICTCP_BETA_SCALE);
	170	else
	171	ca->last_max_cwnd = tp->snd_cwnd;
	172
83803034 SH	173	if (tp->snd_cwnd <= low_window)
	174	return max(tp->snd_cwnd >> 1U, 2U);
	175	else
	176	return max((tp->snd_cwnd * beta) / BICTCP_BETA_SCALE, 2U);
	177	}
	178
6687e988	179	static void bictcp_state(struct sock *sk, u8 new_state)
83803034 SH	180	{
83803034 SH	181	if (new_state == TCP_CA_Loss)
6687e988	182	bictcp_reset(inet_csk_ca(sk));
83803034 SH	183	}
83803034 SH	184
05d05450	185	/* Track delayed acknowledgment ratio using sliding window
83803034 SH	186	* ratio = (15*ratio + sample) / 16
83803034 SH	187	*/
756ee172	188	static void bictcp_acked(struct sock sk, const struct ack_sample sample)
83803034	189	{
6687e988 ACM	190	const struct inet_connection_sock *icsk = inet_csk(sk);
6687e988 ACM	191
35e86941	192	if (icsk->icsk_ca_state == TCP_CA_Open) {
6687e988	193	struct bictcp *ca = inet_csk_ca(sk);
688d1945	194
756ee172 LB	195	ca->delayed_ack += sample->pkts_acked -
756ee172 LB	196	(ca->delayed_ack >> ACK_RATIO_SHIFT);
83803034 SH	197	}
	198	}
	199
a252bebe	200	static struct tcp_congestion_ops bictcp __read_mostly = {
83803034 SH	201	.init = bictcp_init,
	202	.ssthresh = bictcp_recalc_ssthresh,
	203	.cong_avoid = bictcp_cong_avoid,
	204	.set_state = bictcp_state,
f1722a1b	205	.undo_cwnd = tcp_reno_undo_cwnd,
83803034 SH	206	.pkts_acked = bictcp_acked,
	207	.owner = THIS_MODULE,
	208	.name = "bic",
	209	};
	210
	211	static int __init bictcp_register(void)
	212	{
65e3d726	213	BUILD_BUG_ON(sizeof(struct bictcp) > ICSK_CA_PRIV_SIZE);
83803034 SH	214	return tcp_register_congestion_control(&bictcp);
	215	}
	216
	217	static void __exit bictcp_unregister(void)
	218	{
	219	tcp_unregister_congestion_control(&bictcp);
	220	}
	221
	222	module_init(bictcp_register);
	223	module_exit(bictcp_unregister);
	224
	225	MODULE_AUTHOR("Stephen Hemminger");
	226	MODULE_LICENSE("GPL");
	227	MODULE_DESCRIPTION("BIC TCP");