[TCP]: TCP Illinois congestion control (rev3)

This is an implementation of TCP Illinois invented by Shao Liu
at University of Illinois. It is a another variant of Reno which adapts
the alpha and beta parameters based on RTT. The basic idea is to increase
window less rapidly as delay approaches the maximum. See the papers
and talks to get a more complete description.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 284 insertions, 0 deletions

diff --git a/net/ipv4/tcp_illinois.c b/net/ipv4/tcp_illinois.c
new file mode 100644
index 0000000..91a2a34
--- /dev/null
+++ b/net/ipv4/tcp_illinois.c
@@ -0,0 +1,284 @@
+/*
+ * TCP Illinois congestion control.
+ * Home page:
+ *	http://www.ews.uiuc.edu/~shaoliu/tcpillinois/index.html
+ *
+ * The algorithm is described in:
+ * "TCP-Illinois: A Loss and Delay-Based Congestion Control Algorithm
+ *  for High-Speed Networks"
+ * http://www.ews.uiuc.edu/~shaoliu/papersandslides/liubassri06perf.pdf
+ *
+ * Implemented from description in paper and ns-2 simulation.
+ * Copyright (C) 2007 Stephen Hemminger <shemminger@linux-foundation.org>
+ */
+
+#include <linux/module.h>
+#include <linux/skbuff.h>
+#include <linux/inet_diag.h>
+#include <asm/div64.h>
+#include <net/tcp.h>
+
+#define ALPHA_SHIFT	7
+#define ALPHA_SCALE	(1u<<ALPHA_SHIFT)
+#define ALPHA_MIN	((3*ALPHA_SCALE)/10)	/* ~0.3 */
+#define ALPHA_MAX	(10*ALPHA_SCALE)	/* 10.0 */
+#define ALPHA_BASE	ALPHA_SCALE		/* 1.0 */
+
+#define BETA_SHIFT	6
+#define BETA_SCALE	(1u<<BETA_SHIFT)
+#define BETA_MIN	(BETA_SCALE/8)		/* 0.8 */
+#define BETA_MAX	(BETA_SCALE/2)
+#define BETA_BASE	BETA_MAX		/* 0.5 */
+
+#define THETA		5
+
+static int win_thresh __read_mostly = 15;
+module_param(win_thresh, int, 0644);
+MODULE_PARM_DESC(win_thresh, "Window threshold for starting adaptive sizing");
+
+#define MAX_RTT		0x7fffffff
+
+/* TCP Illinois Parameters */
+struct tcp_illinois {
+	u32	last_alpha;
+	u32	min_rtt;
+	u32	max_rtt;
+	u32	rtt_low;
+	u32	rtt_cnt;
+	u64	sum_rtt;
+};
+
+static void tcp_illinois_init(struct sock *sk)
+{
+	struct tcp_illinois *ca = inet_csk_ca(sk);
+
+	ca->last_alpha = ALPHA_BASE;
+	ca->min_rtt = 0x7fffffff;
+}
+
+/*
+ * Keep track of min, max and average RTT
+ */
+static void tcp_illinois_rtt_calc(struct sock *sk, u32 rtt)
+{
+	struct tcp_illinois *ca = inet_csk_ca(sk);
+
+	if (rtt < ca->min_rtt)
+		ca->min_rtt = rtt;
+	if (rtt > ca->max_rtt)
+		ca->max_rtt = rtt;
+
+	if (++ca->rtt_cnt == 1)
+		ca->sum_rtt = rtt;
+	else
+		ca->sum_rtt += rtt;
+}
+
+/* max queuing delay */
+static inline u32 max_delay(const struct tcp_illinois *ca)
+{
+	return ca->max_rtt - ca->min_rtt;
+}
+
+/* average queueing delay */
+static u32 avg_delay(struct tcp_illinois *ca)
+{
+	u64 avg_rtt = ca->sum_rtt;
+
+	do_div(avg_rtt, ca->rtt_cnt);
+
+	ca->sum_rtt = 0;
+	ca->rtt_cnt = 0;
+
+	return avg_rtt - ca->min_rtt;
+}
+
+/*
+ * Compute value of alpha used for additive increase.
+ * If small window then use 1.0, equivalent to Reno.
+ *
+ * For larger windows, adjust based on average delay.
+ * A. If average delay is at minimum (we are uncongested),
+ *    then use large alpha (10.0) to increase faster.
+ * B. If average delay is at maximum (getting congested)
+ *    then use small alpha (1.0)
+ *
+ * The result is a convex window growth curve.
+ */
+static u32 alpha(const struct sock *sk)
+{
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct tcp_illinois *ca = inet_csk_ca(sk);
+	u32 dm = max_delay(ca);
+	u32 da = avg_delay(ca);
+	u32 d1, a;
+
+	if (tp->snd_cwnd < win_thresh)
+		return ALPHA_BASE;	/* same as Reno (1.0) */
+
+	d1 = dm / 100;
+	if (da <= d1) {
+		/* Don't let noise force agressive response */
+		if (ca->rtt_low < THETA) {
+			++ca->rtt_low;
+			return ca->last_alpha;
+		} else
+			return ALPHA_MAX;
+	}
+
+	ca->rtt_low = 0;
+
+	/*
+	 * Based on:
+	 *
+	 *      (dm - d1) amin amax
+	 * k1 = -------------------
+	 *         amax - amin
+	 *
+	 *       (dm - d1) amin
+	 * k2 = ----------------  - d1
+	 *        amax - amin
+	 *
+	 *             k1
+	 * alpha = ----------
+	 *          k2 + da
+	 */
+
+	dm -= d1;
+	da -= d1;
+
+	a = (dm * ALPHA_MAX) / (dm - (da  * (ALPHA_MAX - ALPHA_MIN)) / ALPHA_MIN);
+	ca->last_alpha = a;
+	return a;
+}
+
+/*
+ * Increase window in response to successful acknowledgment.
+ */
+static void tcp_illinois_cong_avoid(struct sock *sk, u32 ack, u32 rtt,
+				    u32 in_flight, int flag)
+{
+	struct tcp_sock *tp = tcp_sk(sk);
+
+	/* RFC2861 only increase cwnd if fully utilized */
+	if (!tcp_is_cwnd_limited(sk, in_flight))
+		return;
+
+	/* In slow start */
+	if (tp->snd_cwnd <= tp->snd_ssthresh)
+		tcp_slow_start(tp);
+
+	else {
+		/* additive increase  cwnd += alpha / cwnd */
+		if ((tp->snd_cwnd_cnt * alpha(sk)) >> ALPHA_SHIFT >= tp->snd_cwnd) {
+			if (tp->snd_cwnd < tp->snd_cwnd_clamp)
+				tp->snd_cwnd++;
+			tp->snd_cwnd_cnt = 0;
+		} else
+			tp->snd_cwnd_cnt++;
+	}
+}
+
+/*
+ * Beta used for multiplicative decrease.
+ * For small window sizes returns same value as Reno (0.5)
+ *
+ * If delay is small (10% of max) then beta = 1/8
+ * If delay is up to 80% of max then beta = 1/2
+ * In between is a linear function
+ */
+static inline u32 beta(struct sock *sk)
+{
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct tcp_illinois *ca = inet_csk_ca(sk);
+	u32 dm = max_delay(ca);
+	u32 da = avg_delay(ca);
+	u32 d2, d3;
+
+	if (tp->snd_cwnd < win_thresh)
+		return BETA_BASE;
+
+	d2 = dm / 10;
+	if (da <= d2)
+		return BETA_MIN;
+	d3 = (8 * dm) / 10;
+	if (da >= d3 || d3 <= d2)
+		return BETA_MAX;
+
+	/*
+	 * Based on:
+	 *
+	 *       bmin d3 - bmax d2
+	 * k3 = -------------------
+	 *         d3 - d2
+	 *
+	 *       bmax - bmin
+	 * k4 = -------------
+	 *         d3 - d2
+	 *
+	 * b = k3 + k4 da
+	 */
+	return (BETA_MIN * d3 - BETA_MAX * d2 + (BETA_MAX - BETA_MIN) * da)
+		/ (d3 - d2);
+}
+
+static u32 tcp_illinois_ssthresh(struct sock *sk)
+{
+	struct tcp_sock *tp = tcp_sk(sk);
+
+	/* Multiplicative decrease */
+	return max((tp->snd_cwnd * beta(sk)) >> BETA_SHIFT, 2U);
+}
+
+/* Extract info for TCP socket info provided via netlink.
+ * We aren't really doing Vegas, but we can provide RTT info
+ */
+static void tcp_illinois_get_info(struct sock *sk, u32 ext,
+			       struct sk_buff *skb)
+{
+	const struct tcp_illinois *ca = inet_csk_ca(sk);
+
+	if (ext & (1 << (INET_DIAG_VEGASINFO - 1))) {
+		struct tcpvegas_info info = {
+			.tcpv_enabled = 1,
+			.tcpv_rttcnt = ca->rtt_cnt,
+			.tcpv_minrtt = ca->min_rtt,
+		};
+		u64 avg_rtt = ca->sum_rtt;
+		do_div(avg_rtt, ca->rtt_cnt);
+		info.tcpv_rtt = avg_rtt;
+
+		nla_put(skb, INET_DIAG_VEGASINFO, sizeof(info), &info);
+	}
+}
+
+static struct tcp_congestion_ops tcp_illinois = {
+	.init		= tcp_illinois_init,
+	.ssthresh	= tcp_illinois_ssthresh,
+	.min_cwnd	= tcp_reno_min_cwnd,
+	.cong_avoid	= tcp_illinois_cong_avoid,
+	.rtt_sample	= tcp_illinois_rtt_calc,
+	.get_info	= tcp_illinois_get_info,
+
+	.owner		= THIS_MODULE,
+	.name		= "illinois",
+};
+
+static int __init tcp_illinois_register(void)
+{
+	BUILD_BUG_ON(sizeof(struct tcp_illinois) > ICSK_CA_PRIV_SIZE);
+	return tcp_register_congestion_control(&tcp_illinois);
+}
+
+static void __exit tcp_illinois_unregister(void)
+{
+	tcp_unregister_congestion_control(&tcp_illinois);
+}
+
+module_init(tcp_illinois_register);
+module_exit(tcp_illinois_unregister);
+
+MODULE_AUTHOR("Stephen Hemminger, Shao Liu");
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("TCP Illinois");
+MODULE_VERSION("0.3");