1 files changed, 236 insertions, 155 deletions

diff --git a/net/ipv4/tcp_cubic.c b/net/ipv4/tcp_cubic.c
index 31a4986dfbf..a9bd8a4828a 100644
--- a/net/ipv4/tcp_cubic.c
+++ b/net/ipv4/tcp_cubic.c
@@ -1,49 +1,67 @@
 /*
- * TCP CUBIC: Binary Increase Congestion control for TCP v2.0
- *
+ * TCP CUBIC: Binary Increase Congestion control for TCP v2.3
+ * Home page:
+ *      http://netsrv.csc.ncsu.edu/twiki/bin/view/Main/BIC
  * This is from the implementation of CUBIC TCP in
- * Injong Rhee, Lisong Xu.
- *  "CUBIC: A New TCP-Friendly High-Speed TCP Variant
- *  in PFLDnet 2005
+ * Sangtae Ha, Injong Rhee and Lisong Xu,
+ *  "CUBIC: A New TCP-Friendly High-Speed TCP Variant"
+ *  in ACM SIGOPS Operating System Review, July 2008.
+ * Available from:
+ *  http://netsrv.csc.ncsu.edu/export/cubic_a_new_tcp_2008.pdf
+ *
+ * CUBIC integrates a new slow start algorithm, called HyStart.
+ * The details of HyStart are presented in
+ *  Sangtae Ha and Injong Rhee,
+ *  "Taming the Elephants: New TCP Slow Start", NCSU TechReport 2008.
  * Available from:
- *  http://www.csc.ncsu.edu/faculty/rhee/export/bitcp/cubic-paper.pdf
+ *  http://netsrv.csc.ncsu.edu/export/hystart_techreport_2008.pdf
+ *
+ * All testing results are available from:
+ * http://netsrv.csc.ncsu.edu/wiki/index.php/TCP_Testing
  *
  * Unless CUBIC is enabled and congestion window is large
  * this behaves the same as the original Reno.
  */
 
-#include <linux/config.h>
 #include <linux/mm.h>
 #include <linux/module.h>
+#include <linux/math64.h>
 #include <net/tcp.h>
-#include <asm/div64.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
-					  */
 #define	BICTCP_HZ		10	/* BIC HZ 2^10 = 1024 */
 
-static int fast_convergence = 1;
-static int max_increment = 16;
-static int beta = 819;		/* = 819/1024 (BICTCP_BETA_SCALE) */
-static int initial_ssthresh = 100;
-static int bic_scale = 41;
-static int tcp_friendliness = 1;
+/* Two methods of hybrid slow start */
+#define HYSTART_ACK_TRAIN	0x1
+#define HYSTART_DELAY		0x2
+
+/* Number of delay samples for detecting the increase of delay */
+#define HYSTART_MIN_SAMPLES	8
+#define HYSTART_DELAY_MIN	(4U<<3)
+#define HYSTART_DELAY_MAX	(16U<<3)
+#define HYSTART_DELAY_THRESH(x)	clamp(x, HYSTART_DELAY_MIN, HYSTART_DELAY_MAX)
+
+static int fast_convergence __read_mostly = 1;
+static int beta __read_mostly = 717;	/* = 717/1024 (BICTCP_BETA_SCALE) */
+static int initial_ssthresh __read_mostly;
+static int bic_scale __read_mostly = 41;
+static int tcp_friendliness __read_mostly = 1;
 
-static u32 cube_rtt_scale;
-static u32 beta_scale;
-static u64 cube_factor;
+static int hystart __read_mostly = 1;
+static int hystart_detect __read_mostly = HYSTART_ACK_TRAIN | HYSTART_DELAY;
+static int hystart_low_window __read_mostly = 16;
+static int hystart_ack_delta __read_mostly = 2;
+
+static u32 cube_rtt_scale __read_mostly;
+static u32 beta_scale __read_mostly;
+static u64 cube_factor __read_mostly;
 
 /* Note parameters that are used for precomputing scale factors are read-only */
 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(beta, int, 0444);
+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");
@@ -51,8 +69,15 @@ module_param(bic_scale, int, 0444);
 MODULE_PARM_DESC(bic_scale, "scale (scaled by 1024) value for bic function (bic_scale/1024)");
 module_param(tcp_friendliness, int, 0644);
 MODULE_PARM_DESC(tcp_friendliness, "turn on/off tcp friendliness");
-
-#include <asm/div64.h>
+module_param(hystart, int, 0644);
+MODULE_PARM_DESC(hystart, "turn on/off hybrid slow start algorithm");
+module_param(hystart_detect, int, 0644);
+MODULE_PARM_DESC(hystart_detect, "hyrbrid slow start detection mechanisms"
+		 " 1: packet-train 2: delay 3: both packet-train and delay");
+module_param(hystart_low_window, int, 0644);
+MODULE_PARM_DESC(hystart_low_window, "lower bound cwnd for hybrid slow start");
+module_param(hystart_ack_delta, int, 0644);
+MODULE_PARM_DESC(hystart_ack_delta, "spacing between ack's indicating train (msecs)");
 
 /* BIC TCP Parameters */
 struct bictcp {
@@ -63,19 +88,25 @@ struct bictcp {
 	u32	last_time;	/* time when updated last_cwnd */
 	u32	bic_origin_point;/* origin point of bic function */
 	u32	bic_K;		/* time to origin point from the beginning of the current epoch */
-	u32	delay_min;	/* min delay */
+	u32	delay_min;	/* min delay (msec << 3) */
 	u32	epoch_start;	/* beginning of an epoch */
 	u32	ack_cnt;	/* number of acks */
 	u32	tcp_cwnd;	/* estimated tcp cwnd */
 #define ACK_RATIO_SHIFT	4
-	u32	delayed_ack;	/* estimate the ratio of Packets/ACKs << 4 */
+#define ACK_RATIO_LIMIT (32u << ACK_RATIO_SHIFT)
+	u16	delayed_ack;	/* estimate the ratio of Packets/ACKs << 4 */
+	u8	sample_cnt;	/* number of samples to decide curr_rtt */
+	u8	found;		/* the exit point is found? */
+	u32	round_start;	/* beginning of each round */
+	u32	end_seq;	/* end_seq of the round */
+	u32	last_ack;	/* last time when the ACK spacing is close */
+	u32	curr_rtt;	/* the minimum rtt of current round */
 };
 
 static inline void bictcp_reset(struct bictcp *ca)
 {
 	ca->cnt = 0;
 	ca->last_max_cwnd = 0;
-	ca->loss_cwnd = 0;
 	ca->last_cwnd = 0;
 	ca->last_time = 0;
 	ca->bic_origin_point = 0;
@@ -85,59 +116,88 @@ static inline void bictcp_reset(struct bictcp *ca)
 	ca->delayed_ack = 2 << ACK_RATIO_SHIFT;
 	ca->ack_cnt = 0;
 	ca->tcp_cwnd = 0;
+	ca->found = 0;
 }
 
-static void bictcp_init(struct sock *sk)
+static inline u32 bictcp_clock(void)
 {
-	bictcp_reset(inet_csk_ca(sk));
-	if (initial_ssthresh)
-		tcp_sk(sk)->snd_ssthresh = initial_ssthresh;
+#if HZ < 1000
+	return ktime_to_ms(ktime_get_real());
+#else
+	return jiffies_to_msecs(jiffies);
+#endif
 }
 
-/* 64bit divisor, dividend and result. dynamic precision */
-static inline u_int64_t div64_64(u_int64_t dividend, u_int64_t divisor)
+static inline void bictcp_hystart_reset(struct sock *sk)
 {
-	u_int32_t d = divisor;
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct bictcp *ca = inet_csk_ca(sk);
 
-	if (divisor > 0xffffffffULL) {
-		unsigned int shift = fls(divisor >> 32);
+	ca->round_start = ca->last_ack = bictcp_clock();
+	ca->end_seq = tp->snd_nxt;
+	ca->curr_rtt = 0;
+	ca->sample_cnt = 0;
+}
 
-		d = divisor >> shift;
-		dividend >>= shift;
-	}
+static void bictcp_init(struct sock *sk)
+{
+	struct bictcp *ca = inet_csk_ca(sk);
 
-	/* avoid 64 bit division if possible */
-	if (dividend >> 32)
-		do_div(dividend, d);
-	else
-		dividend = (uint32_t) dividend / d;
+	bictcp_reset(ca);
+	ca->loss_cwnd = 0;
 
-	return dividend;
+	if (hystart)
+		bictcp_hystart_reset(sk);
+
+	if (!hystart && initial_ssthresh)
+		tcp_sk(sk)->snd_ssthresh = initial_ssthresh;
 }
 
-/*
- * calculate the cubic root of x using Newton-Raphson
+/* calculate the cubic root of x using a table lookup followed by one
+ * Newton-Raphson iteration.
+ * Avg err ~= 0.195%
  */
 static u32 cubic_root(u64 a)
 {
-	u32 x, x1;
-
-	/* Initial estimate is based on:
-	 * cbrt(x) = exp(log(x) / 3)
+	u32 x, b, shift;
+	/*
+	 * cbrt(x) MSB values for x MSB values in [0..63].
+	 * Precomputed then refined by hand - Willy Tarreau
+	 *
+	 * For x in [0..63],
+	 *   v = cbrt(x << 18) - 1
+	 *   cbrt(x) = (v[x] + 10) >> 6
 	 */
-	x = 1u << (fls64(a)/3);
+	static const u8 v[] = {
+		/* 0x00 */    0,   54,   54,   54,  118,  118,  118,  118,
+		/* 0x08 */  123,  129,  134,  138,  143,  147,  151,  156,
+		/* 0x10 */  157,  161,  164,  168,  170,  173,  176,  179,
+		/* 0x18 */  181,  185,  187,  190,  192,  194,  197,  199,
+		/* 0x20 */  200,  202,  204,  206,  209,  211,  213,  215,
+		/* 0x28 */  217,  219,  221,  222,  224,  225,  227,  229,
+		/* 0x30 */  231,  232,  234,  236,  237,  239,  240,  242,
+		/* 0x38 */  244,  245,  246,  248,  250,  251,  252,  254,
+	};
+
+	b = fls64(a);
+	if (b < 7) {
+		/* a in [0..63] */
+		return ((u32)v[(u32)a] + 35) >> 6;
+	}
+
+	b = ((b * 84) >> 8) - 1;
+	shift = (a >> (b * 3));
+
+	x = ((u32)(((u32)v[shift] + 10) << b)) >> 6;
 
 	/*
-	 * Iteration based on:
+	 * Newton-Raphson iteration
 	 *                         2
 	 * x    = ( 2 * x  +  a / x  ) / 3
 	 *  k+1          k         k
 	 */
-	do {
-		x1 = x;
-		x = (2 * x + (uint32_t) div64_64(a, x*x)) / 3;
-	} while (abs(x1 - x) > 1);
-
+	x = (2 * x + (u32)div64_u64(a, (u64)x * (u64)(x - 1)));
+	x = ((x * 341) >> 10);
 	return x;
 }
 
@@ -146,8 +206,8 @@ static u32 cubic_root(u64 a)
  */
 static inline void bictcp_update(struct bictcp *ca, u32 cwnd)
 {
-	u64 offs;
-	u32 delta, t, bic_target, min_cnt, max_cnt;
+	u32 delta, bic_target, max_cnt;
+	u64 offs, t;
 
 	ca->ack_cnt++;	/* count the number of ACKs */
 
@@ -176,61 +236,59 @@ static inline void bictcp_update(struct bictcp *ca, u32 cwnd)
 		}
 	}
 
-        /* cubic function - calc*/
-        /* calculate c * time^3 / rtt,
-         *  while considering overflow in calculation of time^3
+	/* cubic function - calc*/
+	/* calculate c * time^3 / rtt,
+	 *  while considering overflow in calculation of time^3
 	 * (so time^3 is done by using 64 bit)
 	 * and without the support of division of 64bit numbers
 	 * (so all divisions are done by using 32 bit)
-         *  also NOTE the unit of those veriables
-         *	  time  = (t - K) / 2^bictcp_HZ
-         *	  c = bic_scale >> 10
+	 *  also NOTE the unit of those veriables
+	 *	  time  = (t - K) / 2^bictcp_HZ
+	 *	  c = bic_scale >> 10
 	 * rtt  = (srtt >> 3) / HZ
 	 * !!! The following code does not have overflow problems,
 	 * if the cwnd < 1 million packets !!!
-         */
+	 */
 
+	t = (s32)(tcp_time_stamp - ca->epoch_start);
+	t += msecs_to_jiffies(ca->delay_min >> 3);
 	/* change the unit from HZ to bictcp_HZ */
-        t = ((tcp_time_stamp + ca->delay_min - ca->epoch_start)
-	     << BICTCP_HZ) / HZ;
+	t <<= BICTCP_HZ;
+	do_div(t, HZ);
 
-        if (t < ca->bic_K)		/* t - K */
+	if (t < ca->bic_K)		/* t - K */
 		offs = ca->bic_K - t;
-        else
-                offs = t - ca->bic_K;
+	else
+		offs = t - ca->bic_K;
 
 	/* c/rtt * (t-K)^3 */
 	delta = (cube_rtt_scale * offs * offs * offs) >> (10+3*BICTCP_HZ);
-        if (t < ca->bic_K)                                	/* below origin*/
-                bic_target = ca->bic_origin_point - delta;
-        else                                                	/* above origin*/
-                bic_target = ca->bic_origin_point + delta;
+	if (t < ca->bic_K)                                	/* below origin*/
+		bic_target = ca->bic_origin_point - delta;
+	else                                                	/* above origin*/
+		bic_target = ca->bic_origin_point + delta;
 
-        /* cubic function - calc bictcp_cnt*/
-        if (bic_target > cwnd) {
+	/* cubic function - calc bictcp_cnt*/
+	if (bic_target > cwnd) {
 		ca->cnt = cwnd / (bic_target - cwnd);
-        } else {
-                ca->cnt = 100 * cwnd;              /* very small increment*/
-        }
-
-	if (ca->delay_min > 0) {
-		/* max increment = Smax * rtt / 0.1  */
-		min_cnt = (cwnd * HZ * 8)/(10 * max_increment * ca->delay_min);
-		if (ca->cnt < min_cnt)
-			ca->cnt = min_cnt;
+	} else {
+		ca->cnt = 100 * cwnd;              /* very small increment*/
 	}
 
-        /* slow start and low utilization  */
-	if (ca->loss_cwnd == 0)		/* could be aggressive in slow start */
-		ca->cnt = 50;
+	/*
+	 * The initial growth of cubic function may be too conservative
+	 * when the available bandwidth is still unknown.
+	 */
+	if (ca->last_max_cwnd == 0 && ca->cnt > 20)
+		ca->cnt = 20;	/* increase cwnd 5% per RTT */
 
 	/* TCP Friendly */
 	if (tcp_friendliness) {
 		u32 scale = beta_scale;
 		delta = (cwnd * scale) >> 3;
-	        while (ca->ack_cnt > delta) {		/* update tcp cwnd */
-	                ca->ack_cnt -= delta;
-        	        ca->tcp_cwnd++;
+		while (ca->ack_cnt > delta) {		/* update tcp cwnd */
+			ca->ack_cnt -= delta;
+			ca->tcp_cwnd++;
 		}
 
 		if (ca->tcp_cwnd > cwnd){	/* if bic is slower than tcp */
@@ -239,62 +297,28 @@ static inline void bictcp_update(struct bictcp *ca, u32 cwnd)
 			if (ca->cnt > max_cnt)
 				ca->cnt = max_cnt;
 		}
-        }
+	}
 
 	ca->cnt = (ca->cnt << ACK_RATIO_SHIFT) / ca->delayed_ack;
 	if (ca->cnt == 0)			/* cannot be zero */
 		ca->cnt = 1;
 }
 
-
-/* Keep track of minimum rtt */
-static inline void measure_delay(struct sock *sk)
-{
-	const struct tcp_sock *tp = tcp_sk(sk);
-	struct bictcp *ca = inet_csk_ca(sk);
-	u32 delay;
-
-	/* No time stamp */
-	if (!(tp->rx_opt.saw_tstamp && tp->rx_opt.rcv_tsecr) ||
-	     /* Discard delay samples right after fast recovery */
-	    (s32)(tcp_time_stamp - ca->epoch_start) < HZ)
-		return;
-
-	delay = tcp_time_stamp - tp->rx_opt.rcv_tsecr;
-	if (delay == 0)
-		delay = 1;
-
-	/* first time call or link delay decreases */
-	if (ca->delay_min == 0 || ca->delay_min > delay)
-		ca->delay_min = delay;
-}
-
-static void bictcp_cong_avoid(struct sock *sk, u32 ack,
-			      u32 seq_rtt, u32 in_flight, int data_acked)
+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 (data_acked)
-		measure_delay(sk);
-
-	if (!tcp_is_cwnd_limited(sk, in_flight))
+	if (!tcp_is_cwnd_limited(sk))
 		return;
 
-	if (tp->snd_cwnd <= tp->snd_ssthresh)
-		tcp_slow_start(tp);
-	else {
+	if (tp->snd_cwnd <= tp->snd_ssthresh) {
+		if (hystart && after(ack, ca->end_seq))
+			bictcp_hystart_reset(sk);
+		tcp_slow_start(tp, acked);
+	} else {
 		bictcp_update(ca, tp->snd_cwnd);
-
-		/* In dangerous area, increase slowly.
-		 * In theory this is tp->snd_cwnd += 1 / tp->snd_cwnd
-		 */
-		if (tp->snd_cwnd_cnt >= ca->cnt) {
-			if (tp->snd_cwnd < tp->snd_cwnd_clamp)
-				tp->snd_cwnd++;
-			tp->snd_cwnd_cnt = 0;
-		} else
-			tp->snd_cwnd_cnt++;
+		tcp_cong_avoid_ai(tp, ca->cnt);
 	}
 
 }
@@ -322,42 +346,99 @@ static u32 bictcp_undo_cwnd(struct sock *sk)
 {
 	struct bictcp *ca = inet_csk_ca(sk);
 
-	return max(tcp_sk(sk)->snd_cwnd, ca->last_max_cwnd);
+	return max(tcp_sk(sk)->snd_cwnd, ca->loss_cwnd);
 }
 
-static u32 bictcp_min_cwnd(struct sock *sk)
+static void bictcp_state(struct sock *sk, u8 new_state)
 {
-	return tcp_sk(sk)->snd_ssthresh;
+	if (new_state == TCP_CA_Loss) {
+		bictcp_reset(inet_csk_ca(sk));
+		bictcp_hystart_reset(sk);
+	}
 }
 
-static void bictcp_state(struct sock *sk, u8 new_state)
+static void hystart_update(struct sock *sk, u32 delay)
 {
-	if (new_state == TCP_CA_Loss)
-		bictcp_reset(inet_csk_ca(sk));
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct bictcp *ca = inet_csk_ca(sk);
+
+	if (!(ca->found & hystart_detect)) {
+		u32 now = bictcp_clock();
+
+		/* first detection parameter - ack-train detection */
+		if ((s32)(now - ca->last_ack) <= hystart_ack_delta) {
+			ca->last_ack = now;
+			if ((s32)(now - ca->round_start) > ca->delay_min >> 4)
+				ca->found |= HYSTART_ACK_TRAIN;
+		}
+
+		/* obtain the minimum delay of more than sampling packets */
+		if (ca->sample_cnt < HYSTART_MIN_SAMPLES) {
+			if (ca->curr_rtt == 0 || ca->curr_rtt > delay)
+				ca->curr_rtt = delay;
+
+			ca->sample_cnt++;
+		} else {
+			if (ca->curr_rtt > ca->delay_min +
+			    HYSTART_DELAY_THRESH(ca->delay_min>>4))
+				ca->found |= HYSTART_DELAY;
+		}
+		/*
+		 * Either one of two conditions are met,
+		 * we exit from slow start immediately.
+		 */
+		if (ca->found & hystart_detect)
+			tp->snd_ssthresh = tp->snd_cwnd;
+	}
 }
 
 /* Track delayed acknowledgment ratio using sliding window
  * ratio = (15*ratio + sample) / 16
  */
-static void bictcp_acked(struct sock *sk, u32 cnt)
+static void bictcp_acked(struct sock *sk, u32 cnt, s32 rtt_us)
 {
 	const struct inet_connection_sock *icsk = inet_csk(sk);
+	const struct tcp_sock *tp = tcp_sk(sk);
+	struct bictcp *ca = inet_csk_ca(sk);
+	u32 delay;
+
+	if (icsk->icsk_ca_state == TCP_CA_Open) {
+		u32 ratio = ca->delayed_ack;
+
+		ratio -= ca->delayed_ack >> ACK_RATIO_SHIFT;
+		ratio += cnt;
 
-	if (cnt > 0 && icsk->icsk_ca_state == TCP_CA_Open) {
-		struct bictcp *ca = inet_csk_ca(sk);
-		cnt -= ca->delayed_ack >> ACK_RATIO_SHIFT;
-		ca->delayed_ack += cnt;
+		ca->delayed_ack = clamp(ratio, 1U, ACK_RATIO_LIMIT);
 	}
-}
 
+	/* Some calls are for duplicates without timetamps */
+	if (rtt_us < 0)
+		return;
+
+	/* Discard delay samples right after fast recovery */
+	if (ca->epoch_start && (s32)(tcp_time_stamp - ca->epoch_start) < HZ)
+		return;
+
+	delay = (rtt_us << 3) / USEC_PER_MSEC;
+	if (delay == 0)
+		delay = 1;
+
+	/* first time call or link delay decreases */
+	if (ca->delay_min == 0 || ca->delay_min > delay)
+		ca->delay_min = delay;
+
+	/* hystart triggers when cwnd is larger than some threshold */
+	if (hystart && tp->snd_cwnd <= tp->snd_ssthresh &&
+	    tp->snd_cwnd >= hystart_low_window)
+		hystart_update(sk, delay);
+}
 
-static struct tcp_congestion_ops cubictcp = {
+static struct tcp_congestion_ops cubictcp __read_mostly = {
 	.init		= bictcp_init,
 	.ssthresh	= bictcp_recalc_ssthresh,
 	.cong_avoid	= bictcp_cong_avoid,
 	.set_state	= bictcp_state,
 	.undo_cwnd	= bictcp_undo_cwnd,
-	.min_cwnd	= bictcp_min_cwnd,
 	.pkts_acked     = bictcp_acked,
 	.owner		= THIS_MODULE,
 	.name		= "cubic",
@@ -365,7 +446,7 @@ static struct tcp_congestion_ops cubictcp = {
 
 static int __init cubictcp_register(void)
 {
-	BUG_ON(sizeof(struct bictcp) > ICSK_CA_PRIV_SIZE);
+	BUILD_BUG_ON(sizeof(struct bictcp) > ICSK_CA_PRIV_SIZE);
 
 	/* Precompute a bunch of the scaling factors that are used per-packet
 	 * based on SRTT of 100ms
@@ -373,7 +454,7 @@ static int __init cubictcp_register(void)
 
 	beta_scale = 8*(BICTCP_BETA_SCALE+beta)/ 3 / (BICTCP_BETA_SCALE - beta);
 
-	cube_rtt_scale = (bic_scale << 3) / 10;	/* 1024*c/rtt */
+	cube_rtt_scale = (bic_scale * 10);	/* 1024*c/rtt */
 
 	/* calculate the "K" for (wmax-cwnd) = c/rtt * K^3
 	 *  so K = cubic_root( (wmax-cwnd)*rtt/c )
@@ -408,4 +489,4 @@ module_exit(cubictcp_unregister);
 MODULE_AUTHOR("Sangtae Ha, Stephen Hemminger");
 MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("CUBIC TCP");
-MODULE_VERSION("2.0");
+MODULE_VERSION("2.3");