1 files changed, 52 insertions, 83 deletions

diff --git a/net/ipv4/tcp_cong.c b/net/ipv4/tcp_cong.c
index ccd88407e0c..7b09d8b49fa 100644
--- a/net/ipv4/tcp_cong.c
+++ b/net/ipv4/tcp_cong.c
@@ -1,19 +1,20 @@
 /*
  * Plugable TCP congestion control support and newReno
  * congestion control.
- * Based on ideas from I/O scheduler suport and Web100.
+ * Based on ideas from I/O scheduler support and Web100.
  *
  * Copyright (C) 2005 Stephen Hemminger <shemminger@osdl.org>
  */
 
+#define pr_fmt(fmt) "TCP: " fmt
+
 #include <linux/module.h>
 #include <linux/mm.h>
 #include <linux/types.h>
 #include <linux/list.h>
+#include <linux/gfp.h>
 #include <net/tcp.h>
 
-int sysctl_tcp_max_ssthresh = 0;
-
 static DEFINE_SPINLOCK(tcp_cong_list_lock);
 static LIST_HEAD(tcp_cong_list);
 
@@ -40,18 +41,17 @@ int tcp_register_congestion_control(struct tcp_congestion_ops *ca)
 
 	/* all algorithms must implement ssthresh and cong_avoid ops */
 	if (!ca->ssthresh || !ca->cong_avoid) {
-		printk(KERN_ERR "TCP %s does not implement required ops\n",
-		       ca->name);
+		pr_err("%s does not implement required ops\n", ca->name);
 		return -EINVAL;
 	}
 
 	spin_lock(&tcp_cong_list_lock);
 	if (tcp_ca_find(ca->name)) {
-		printk(KERN_NOTICE "TCP %s already registered\n", ca->name);
+		pr_notice("%s already registered\n", ca->name);
 		ret = -EEXIST;
 	} else {
 		list_add_tail_rcu(&ca->list, &tcp_cong_list);
-		printk(KERN_INFO "TCP %s registered\n", ca->name);
+		pr_info("%s registered\n", ca->name);
 	}
 	spin_unlock(&tcp_cong_list_lock);
 
@@ -115,8 +115,8 @@ int tcp_set_default_congestion_control(const char *name)
 
 	spin_lock(&tcp_cong_list_lock);
 	ca = tcp_ca_find(name);
-#ifdef CONFIG_KMOD
-	if (!ca && capable(CAP_SYS_MODULE)) {
+#ifdef CONFIG_MODULES
+	if (!ca && capable(CAP_NET_ADMIN)) {
 		spin_unlock(&tcp_cong_list_lock);
 
 		request_module("tcp_%s", name);
@@ -126,7 +126,7 @@ int tcp_set_default_congestion_control(const char *name)
 #endif
 
 	if (ca) {
-		ca->non_restricted = 1;	/* default is always allowed */
+		ca->flags |= TCP_CONG_NON_RESTRICTED;	/* default is always allowed */
 		list_move(&ca->list, &tcp_cong_list);
 		ret = 0;
 	}
@@ -181,7 +181,7 @@ void tcp_get_allowed_congestion_control(char *buf, size_t maxlen)
 	*buf = '\0';
 	rcu_read_lock();
 	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
-		if (!ca->non_restricted)
+		if (!(ca->flags & TCP_CONG_NON_RESTRICTED))
 			continue;
 		offs += snprintf(buf + offs, maxlen - offs,
 				 "%s%s",
@@ -195,10 +195,10 @@ void tcp_get_allowed_congestion_control(char *buf, size_t maxlen)
 int tcp_set_allowed_congestion_control(char *val)
 {
 	struct tcp_congestion_ops *ca;
-	char *clone, *name;
+	char *saved_clone, *clone, *name;
 	int ret = 0;
 
-	clone = kstrdup(val, GFP_USER);
+	saved_clone = clone = kstrdup(val, GFP_USER);
 	if (!clone)
 		return -ENOMEM;
 
@@ -212,19 +212,20 @@ int tcp_set_allowed_congestion_control(char *val)
 		}
 	}
 
-	/* pass 2 clear */
+	/* pass 2 clear old values */
 	list_for_each_entry_rcu(ca, &tcp_cong_list, list)
-		ca->non_restricted = 0;
+		ca->flags &= ~TCP_CONG_NON_RESTRICTED;
 
 	/* pass 3 mark as allowed */
 	while ((name = strsep(&val, " ")) && *name) {
 		ca = tcp_ca_find(name);
 		WARN_ON(!ca);
 		if (ca)
-			ca->non_restricted = 1;
+			ca->flags |= TCP_CONG_NON_RESTRICTED;
 	}
 out:
 	spin_unlock(&tcp_cong_list_lock);
+	kfree(saved_clone);
 
 	return ret;
 }
@@ -244,9 +245,9 @@ int tcp_set_congestion_control(struct sock *sk, const char *name)
 	if (ca == icsk->icsk_ca_ops)
 		goto out;
 
-#ifdef CONFIG_KMOD
+#ifdef CONFIG_MODULES
 	/* not found attempt to autoload module */
-	if (!ca && capable(CAP_SYS_MODULE)) {
+	if (!ca && capable(CAP_NET_ADMIN)) {
 		rcu_read_unlock();
 		request_module("tcp_%s", name);
 		rcu_read_lock();
@@ -256,7 +257,8 @@ int tcp_set_congestion_control(struct sock *sk, const char *name)
 	if (!ca)
 		err = -ENOENT;
 
-	else if (!(ca->non_restricted || capable(CAP_NET_ADMIN)))
+	else if (!((ca->flags & TCP_CONG_NON_RESTRICTED) ||
+		   ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)))
 		err = -EPERM;
 
 	else if (!try_module_get(ca->owner))
@@ -274,44 +276,39 @@ int tcp_set_congestion_control(struct sock *sk, const char *name)
 	return err;
 }
 
-
-/*
- * Slow start (exponential increase) with
- * RFC3742 Limited Slow Start (fast linear increase) support.
+/* Slow start is used when congestion window is no greater than the slow start
+ * threshold. We base on RFC2581 and also handle stretch ACKs properly.
+ * We do not implement RFC3465 Appropriate Byte Counting (ABC) per se but
+ * something better;) a packet is only considered (s)acked in its entirety to
+ * defend the ACK attacks described in the RFC. Slow start processes a stretch
+ * ACK of degree N as if N acks of degree 1 are received back to back except
+ * ABC caps N to 2. Slow start exits when cwnd grows over ssthresh and
+ * returns the leftover acks to adjust cwnd in congestion avoidance mode.
  */
-void tcp_slow_start(struct tcp_sock *tp)
+int tcp_slow_start(struct tcp_sock *tp, u32 acked)
 {
-	int cnt = 0;
-
-	if (sysctl_tcp_abc) {
-		/* RFC3465: Slow Start
-		 * TCP sender SHOULD increase cwnd by the number of
-		 * previously unacknowledged bytes ACKed by each incoming
-		 * acknowledgment, provided the increase is not more than L
-		 */
-		if (tp->bytes_acked < tp->mss_cache)
-			return;
-	}
+	u32 cwnd = tp->snd_cwnd + acked;
 
-	if (sysctl_tcp_max_ssthresh > 0 &&
-	    tp->snd_cwnd > sysctl_tcp_max_ssthresh)
-		cnt += sysctl_tcp_max_ssthresh>>1;
-	else
-		cnt += tp->snd_cwnd;
-
-	/* RFC3465: We MAY increase by 2 if discovered delayed ack */
-	if (sysctl_tcp_abc > 1 && tp->bytes_acked >= 2*tp->mss_cache)
-		cnt <<= 1;
-	tp->bytes_acked = 0;
+	if (cwnd > tp->snd_ssthresh)
+		cwnd = tp->snd_ssthresh + 1;
+	acked -= cwnd - tp->snd_cwnd;
+	tp->snd_cwnd = min(cwnd, tp->snd_cwnd_clamp);
+	return acked;
+}
+EXPORT_SYMBOL_GPL(tcp_slow_start);
 
-	tp->snd_cwnd_cnt += cnt;
-	while (tp->snd_cwnd_cnt >= tp->snd_cwnd) {
-		tp->snd_cwnd_cnt -= tp->snd_cwnd;
+/* In theory this is tp->snd_cwnd += 1 / tp->snd_cwnd (or alternative w) */
+void tcp_cong_avoid_ai(struct tcp_sock *tp, u32 w)
+{
+	if (tp->snd_cwnd_cnt >= w) {
 		if (tp->snd_cwnd < tp->snd_cwnd_clamp)
 			tp->snd_cwnd++;
+		tp->snd_cwnd_cnt = 0;
+	} else {
+		tp->snd_cwnd_cnt++;
 	}
 }
-EXPORT_SYMBOL_GPL(tcp_slow_start);
+EXPORT_SYMBOL_GPL(tcp_cong_avoid_ai);
 
 /*
  * TCP Reno congestion control
@@ -320,37 +317,19 @@ EXPORT_SYMBOL_GPL(tcp_slow_start);
 /* This is Jacobson's slow start and congestion avoidance.
  * SIGCOMM '88, p. 328.
  */
-void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 rtt, u32 in_flight,
-			 int flag)
+void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 acked)
 {
 	struct tcp_sock *tp = tcp_sk(sk);
 
-	if (!tcp_is_cwnd_limited(sk, in_flight))
+	if (!tcp_is_cwnd_limited(sk))
 		return;
 
 	/* In "safe" area, increase. */
 	if (tp->snd_cwnd <= tp->snd_ssthresh)
-		tcp_slow_start(tp);
-
+		tcp_slow_start(tp, acked);
 	/* In dangerous area, increase slowly. */
-	else if (sysctl_tcp_abc) {
-		/* RFC3465: Appropriate Byte Count
-		 * increase once for each full cwnd acked
-		 */
-		if (tp->bytes_acked >= tp->snd_cwnd*tp->mss_cache) {
-			tp->bytes_acked -= tp->snd_cwnd*tp->mss_cache;
-			if (tp->snd_cwnd < tp->snd_cwnd_clamp)
-				tp->snd_cwnd++;
-		}
-	} else {
-		/* In theory this is tp->snd_cwnd += 1 / tp->snd_cwnd */
-		if (tp->snd_cwnd_cnt >= tp->snd_cwnd) {
-			if (tp->snd_cwnd < tp->snd_cwnd_clamp)
-				tp->snd_cwnd++;
-			tp->snd_cwnd_cnt = 0;
-		} else
-			tp->snd_cwnd_cnt++;
-	}
+	else
+		tcp_cong_avoid_ai(tp, tp->snd_cwnd);
 }
 EXPORT_SYMBOL_GPL(tcp_reno_cong_avoid);
 
@@ -362,21 +341,12 @@ u32 tcp_reno_ssthresh(struct sock *sk)
 }
 EXPORT_SYMBOL_GPL(tcp_reno_ssthresh);
 
-/* Lower bound on congestion window with halving. */
-u32 tcp_reno_min_cwnd(const struct sock *sk)
-{
-	const struct tcp_sock *tp = tcp_sk(sk);
-	return tp->snd_ssthresh/2;
-}
-EXPORT_SYMBOL_GPL(tcp_reno_min_cwnd);
-
 struct tcp_congestion_ops tcp_reno = {
+	.flags		= TCP_CONG_NON_RESTRICTED,
 	.name		= "reno",
-	.non_restricted = 1,
 	.owner		= THIS_MODULE,
 	.ssthresh	= tcp_reno_ssthresh,
 	.cong_avoid	= tcp_reno_cong_avoid,
-	.min_cwnd	= tcp_reno_min_cwnd,
 };
 
 /* Initial congestion control used (until SYN)
@@ -388,6 +358,5 @@ struct tcp_congestion_ops tcp_init_congestion_ops  = {
 	.owner		= THIS_MODULE,
 	.ssthresh	= tcp_reno_ssthresh,
 	.cong_avoid	= tcp_reno_cong_avoid,
-	.min_cwnd	= tcp_reno_min_cwnd,
 };
 EXPORT_SYMBOL_GPL(tcp_init_congestion_ops);