[TCP]: Add pluggable congestion control algorithm infrastructure.

Allow TCP to have multiple pluggable congestion control algorithms.
Algorithms are defined by a set of operations and can be built in
or modules.  The legacy "new RENO" algorithm is used as a starting
point and fallback.

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

13 files changed, 399 insertions, 999 deletions

diff --git a/include/linux/sysctl.h b/include/linux/sysctl.h
index 614e939c78a..72965bfe6cf 100644
--- a/include/linux/sysctl.h
+++ b/include/linux/sysctl.h
@@ -333,21 +333,14 @@ enum
 	NET_TCP_FRTO=92,
 	NET_TCP_LOW_LATENCY=93,
 	NET_IPV4_IPFRAG_SECRET_INTERVAL=94,
-	NET_TCP_WESTWOOD=95,
 	NET_IPV4_IGMP_MAX_MSF=96,
 	NET_TCP_NO_METRICS_SAVE=97,
-	NET_TCP_VEGAS=98,
-	NET_TCP_VEGAS_ALPHA=99,
-	NET_TCP_VEGAS_BETA=100,
-	NET_TCP_VEGAS_GAMMA=101,
- 	NET_TCP_BIC=102,
- 	NET_TCP_BIC_FAST_CONVERGENCE=103,
-	NET_TCP_BIC_LOW_WINDOW=104,
 	NET_TCP_DEFAULT_WIN_SCALE=105,
 	NET_TCP_MODERATE_RCVBUF=106,
 	NET_TCP_TSO_WIN_DIVISOR=107,
 	NET_TCP_BIC_BETA=108,
 	NET_IPV4_ICMP_ERRORS_USE_INBOUND_IFADDR=109,
+	NET_TCP_CONG_CONTROL=110,
 };
 
 enum {
diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 97a7c9e03df..3ea75dd6640 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -203,13 +203,6 @@ struct tcp_sack_block {
 	__u32	end_seq;
 };
 
-enum tcp_congestion_algo {
-	TCP_RENO=0,
-	TCP_VEGAS,
-	TCP_WESTWOOD,
-	TCP_BIC,
-};
-
 struct tcp_options_received {
 /*	PAWS/RTTM data	*/
 	long	ts_recent_stamp;/* Time we stored ts_recent (for aging) */
@@ -305,7 +298,7 @@ struct tcp_sock {
 	__u8	reordering;	/* Packet reordering metric.		*/
 	__u8	frto_counter;	/* Number of new acks after RTO */
 
-	__u8	adv_cong;	/* Using Vegas, Westwood, or BIC */
+	__u8	unused;
 	__u8	defer_accept;	/* User waits for some data after accept() */
 
 /* RTT measurement */
@@ -401,37 +394,10 @@ struct tcp_sock {
 		__u32	time;
 	} rcvq_space;
 
-/* TCP Westwood structure */
-        struct {
-                __u32    bw_ns_est;        /* first bandwidth estimation..not too smoothed 8) */
-                __u32    bw_est;           /* bandwidth estimate */
-                __u32    rtt_win_sx;       /* here starts a new evaluation... */
-                __u32    bk;
-                __u32    snd_una;          /* used for evaluating the number of acked bytes */
-                __u32    cumul_ack;
-                __u32    accounted;
-                __u32    rtt;
-                __u32    rtt_min;          /* minimum observed RTT */
-        } westwood;
-
-/* Vegas variables */
-	struct {
-		__u32	beg_snd_nxt;	/* right edge during last RTT */
-		__u32	beg_snd_una;	/* left edge  during last RTT */
-		__u32	beg_snd_cwnd;	/* saves the size of the cwnd */
-		__u8	doing_vegas_now;/* if true, do vegas for this RTT */
-		__u16	cntRTT;		/* # of RTTs measured within last RTT */
-		__u32	minRTT;		/* min of RTTs measured within last RTT (in usec) */
-		__u32	baseRTT;	/* the min of all Vegas RTT measurements seen (in usec) */
-	} vegas;
-
-	/* BI TCP Parameters */
-	struct {
-		__u32	cnt;		/* increase cwnd by 1 after this number of ACKs */
-		__u32 	last_max_cwnd;	/* last maximium snd_cwnd */
-		__u32	last_cwnd;	/* the last snd_cwnd */
-		__u32   last_stamp;     /* time when updated last_cwnd */
-	} bictcp;
+	/* Pluggable TCP congestion control hook */
+	struct tcp_congestion_ops *ca_ops;
+	u32	ca_priv[16];
+#define TCP_CA_PRIV_SIZE	(16*sizeof(u32))
 };
 
 static inline struct tcp_sock *tcp_sk(const struct sock *sk)
@@ -439,6 +405,11 @@ static inline struct tcp_sock *tcp_sk(const struct sock *sk)
 	return (struct tcp_sock *)sk;
 }
 
+static inline void *tcp_ca(const struct tcp_sock *tp)
+{
+	return (void *) tp->ca_priv;
+}
+
 #endif
 
 #endif	/* _LINUX_TCP_H */
diff --git a/include/net/tcp.h b/include/net/tcp.h
index f730935b824..e427cf35915 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -505,25 +505,6 @@ static __inline__ int tcp_sk_listen_hashfn(struct sock *sk)
 #else
 # define TCP_TW_RECYCLE_TICK (12+2-TCP_TW_RECYCLE_SLOTS_LOG)
 #endif
-
-#define BICTCP_BETA_SCALE    1024	/* Scale factor beta calculation
-					 * max_cwnd = snd_cwnd * beta
-					 */
-#define BICTCP_MAX_INCREMENT 32		/*
-					 * Limit on the amount of
-					 * increment allowed during
-					 * binary search.
-					 */
-#define BICTCP_FUNC_OF_MIN_INCR 11	/*
-					 * log(B/Smin)/log(B/(B-1))+1,
-					 * Smin:min increment
-					 * B:log factor
-					 */
-#define BICTCP_B		4	 /*
-					  * In binary search,
-					  * go to point (max+min)/N
-					  */
-
 /*
  *	TCP option
  */
@@ -596,16 +577,7 @@ extern int sysctl_tcp_adv_win_scale;
 extern int sysctl_tcp_tw_reuse;
 extern int sysctl_tcp_frto;
 extern int sysctl_tcp_low_latency;
-extern int sysctl_tcp_westwood;
-extern int sysctl_tcp_vegas_cong_avoid;
-extern int sysctl_tcp_vegas_alpha;
-extern int sysctl_tcp_vegas_beta;
-extern int sysctl_tcp_vegas_gamma;
 extern int sysctl_tcp_nometrics_save;
-extern int sysctl_tcp_bic;
-extern int sysctl_tcp_bic_fast_convergence;
-extern int sysctl_tcp_bic_low_window;
-extern int sysctl_tcp_bic_beta;
 extern int sysctl_tcp_moderate_rcvbuf;
 extern int sysctl_tcp_tso_win_divisor;
 
@@ -1136,6 +1108,80 @@ static inline void tcp_packets_out_dec(struct tcp_sock *tp,
 	tp->packets_out -= tcp_skb_pcount(skb);
 }
 
+/* Events passed to congestion control interface */
+enum tcp_ca_event {
+	CA_EVENT_TX_START,	/* first transmit when no packets in flight */
+	CA_EVENT_CWND_RESTART,	/* congestion window restart */
+	CA_EVENT_COMPLETE_CWR,	/* end of congestion recovery */
+	CA_EVENT_FRTO,		/* fast recovery timeout */
+	CA_EVENT_LOSS,		/* loss timeout */
+	CA_EVENT_FAST_ACK,	/* in sequence ack */
+	CA_EVENT_SLOW_ACK,	/* other ack */
+};
+
+/*
+ * Interface for adding new TCP congestion control handlers
+ */
+#define TCP_CA_NAME_MAX	16
+struct tcp_congestion_ops {
+	struct list_head	list;
+
+	/* initialize private data (optional) */
+	void (*init)(struct tcp_sock *tp);
+	/* cleanup private data  (optional) */
+	void (*release)(struct tcp_sock *tp);
+
+	/* return slow start threshold (required) */
+	u32 (*ssthresh)(struct tcp_sock *tp);
+	/* lower bound for congestion window (optional) */
+	u32 (*min_cwnd)(struct tcp_sock *tp);
+	/* do new cwnd calculation (required) */
+	void (*cong_avoid)(struct tcp_sock *tp, u32 ack,
+			   u32 rtt, u32 in_flight, int good_ack);
+	/* round trip time sample per acked packet (optional) */
+	void (*rtt_sample)(struct tcp_sock *tp, u32 usrtt);
+	/* call before changing ca_state (optional) */
+	void (*set_state)(struct tcp_sock *tp, u8 new_state);
+	/* call when cwnd event occurs (optional) */
+	void (*cwnd_event)(struct tcp_sock *tp, enum tcp_ca_event ev);
+	/* new value of cwnd after loss (optional) */
+	u32  (*undo_cwnd)(struct tcp_sock *tp);
+	/* hook for packet ack accounting (optional) */
+	void (*pkts_acked)(struct tcp_sock *tp, u32 num_acked);
+	/* get info for tcp_diag (optional) */
+	void (*get_info)(struct tcp_sock *tp, u32 ext, struct sk_buff *skb);
+
+	char 		name[TCP_CA_NAME_MAX];
+	struct module 	*owner;
+};
+
+extern int tcp_register_congestion_control(struct tcp_congestion_ops *type);
+extern void tcp_unregister_congestion_control(struct tcp_congestion_ops *type);
+
+extern void tcp_init_congestion_control(struct tcp_sock *tp);
+extern void tcp_cleanup_congestion_control(struct tcp_sock *tp);
+extern int tcp_set_default_congestion_control(const char *name);
+extern void tcp_get_default_congestion_control(char *name);
+
+extern struct tcp_congestion_ops tcp_reno;
+extern u32 tcp_reno_ssthresh(struct tcp_sock *tp);
+extern void tcp_reno_cong_avoid(struct tcp_sock *tp, u32 ack,
+				u32 rtt, u32 in_flight, int flag);
+extern u32 tcp_reno_min_cwnd(struct tcp_sock *tp);
+
+static inline void tcp_set_ca_state(struct tcp_sock *tp, u8 ca_state)
+{
+	if (tp->ca_ops->set_state)
+		tp->ca_ops->set_state(tp, ca_state);
+	tp->ca_state = ca_state;
+}
+
+static inline void tcp_ca_event(struct tcp_sock *tp, enum tcp_ca_event event)
+{
+	if (tp->ca_ops->cwnd_event)
+		tp->ca_ops->cwnd_event(tp, event);
+}
+
 /* This determines how many packets are "in the network" to the best
  * of our knowledge.  In many cases it is conservative, but where
  * detailed information is available from the receiver (via SACK
@@ -1155,91 +1201,6 @@ static __inline__ unsigned int tcp_packets_in_flight(const struct tcp_sock *tp)
 	return (tp->packets_out - tp->left_out + tp->retrans_out);
 }
 
-/*
- * Which congestion algorithim is in use on the connection.
- */
-#define tcp_is_vegas(__tp)	((__tp)->adv_cong == TCP_VEGAS)
-#define tcp_is_westwood(__tp)	((__tp)->adv_cong == TCP_WESTWOOD)
-#define tcp_is_bic(__tp)	((__tp)->adv_cong == TCP_BIC)
-
-/* Recalculate snd_ssthresh, we want to set it to:
- *
- * Reno:
- * 	one half the current congestion window, but no
- *	less than two segments
- *
- * BIC:
- *	behave like Reno until low_window is reached,
- *	then increase congestion window slowly
- */
-static inline __u32 tcp_recalc_ssthresh(struct tcp_sock *tp)
-{
-	if (tcp_is_bic(tp)) {
-		if (sysctl_tcp_bic_fast_convergence &&
-		    tp->snd_cwnd < tp->bictcp.last_max_cwnd)
-			tp->bictcp.last_max_cwnd = (tp->snd_cwnd * 
-						    (BICTCP_BETA_SCALE
-						     + sysctl_tcp_bic_beta))
-				/ (2 * BICTCP_BETA_SCALE);
-		else
-			tp->bictcp.last_max_cwnd = tp->snd_cwnd;
-
-		if (tp->snd_cwnd > sysctl_tcp_bic_low_window)
-			return max((tp->snd_cwnd * sysctl_tcp_bic_beta)
-				   / BICTCP_BETA_SCALE, 2U);
-	}
-
-	return max(tp->snd_cwnd >> 1U, 2U);
-}
-
-/* Stop taking Vegas samples for now. */
-#define tcp_vegas_disable(__tp)	((__tp)->vegas.doing_vegas_now = 0)
-    
-static inline void tcp_vegas_enable(struct tcp_sock *tp)
-{
-	/* There are several situations when we must "re-start" Vegas:
-	 *
-	 *  o when a connection is established
-	 *  o after an RTO
-	 *  o after fast recovery
-	 *  o when we send a packet and there is no outstanding
-	 *    unacknowledged data (restarting an idle connection)
-	 *
-	 * In these circumstances we cannot do a Vegas calculation at the
-	 * end of the first RTT, because any calculation we do is using
-	 * stale info -- both the saved cwnd and congestion feedback are
-	 * stale.
-	 *
-	 * Instead we must wait until the completion of an RTT during
-	 * which we actually receive ACKs.
-	 */
-    
-	/* Begin taking Vegas samples next time we send something. */
-	tp->vegas.doing_vegas_now = 1;
-     
-	/* Set the beginning of the next send window. */
-	tp->vegas.beg_snd_nxt = tp->snd_nxt;
-
-	tp->vegas.cntRTT = 0;
-	tp->vegas.minRTT = 0x7fffffff;
-}
-
-/* Should we be taking Vegas samples right now? */
-#define tcp_vegas_enabled(__tp)	((__tp)->vegas.doing_vegas_now)
-
-extern void tcp_ca_init(struct tcp_sock *tp);
-
-static inline void tcp_set_ca_state(struct tcp_sock *tp, u8 ca_state)
-{
-	if (tcp_is_vegas(tp)) {
-		if (ca_state == TCP_CA_Open) 
-			tcp_vegas_enable(tp);
-		else
-			tcp_vegas_disable(tp);
-	}
-	tp->ca_state = ca_state;
-}
-
 /* If cwnd > ssthresh, we may raise ssthresh to be half-way to cwnd.
  * The exception is rate halving phase, when cwnd is decreasing towards
  * ssthresh.
@@ -1288,7 +1249,7 @@ static inline void tcp_cwnd_validate(struct sock *sk, struct tcp_sock *tp)
 static inline void __tcp_enter_cwr(struct tcp_sock *tp)
 {
 	tp->undo_marker = 0;
-	tp->snd_ssthresh = tcp_recalc_ssthresh(tp);
+	tp->snd_ssthresh = tp->ca_ops->ssthresh(tp);
 	tp->snd_cwnd = min(tp->snd_cwnd,
 			   tcp_packets_in_flight(tp) + 1U);
 	tp->snd_cwnd_cnt = 0;
@@ -1876,52 +1837,4 @@ struct tcp_iter_state {
 extern int tcp_proc_register(struct tcp_seq_afinfo *afinfo);
 extern void tcp_proc_unregister(struct tcp_seq_afinfo *afinfo);
 
-/* TCP Westwood functions and constants */
-
-#define TCP_WESTWOOD_INIT_RTT  (20*HZ)           /* maybe too conservative?! */
-#define TCP_WESTWOOD_RTT_MIN   (HZ/20)           /* 50ms */
-
-static inline void tcp_westwood_update_rtt(struct tcp_sock *tp, __u32 rtt_seq)
-{
-        if (tcp_is_westwood(tp))
-                tp->westwood.rtt = rtt_seq;
-}
-
-static inline __u32 __tcp_westwood_bw_rttmin(const struct tcp_sock *tp)
-{
-        return max((tp->westwood.bw_est) * (tp->westwood.rtt_min) /
-		   (__u32) (tp->mss_cache_std),
-		   2U);
-}
-
-static inline __u32 tcp_westwood_bw_rttmin(const struct tcp_sock *tp)
-{
-	return tcp_is_westwood(tp) ? __tcp_westwood_bw_rttmin(tp) : 0;
-}
-
-static inline int tcp_westwood_ssthresh(struct tcp_sock *tp)
-{
-	__u32 ssthresh = 0;
-
-	if (tcp_is_westwood(tp)) {
-		ssthresh = __tcp_westwood_bw_rttmin(tp);
-		if (ssthresh)
-			tp->snd_ssthresh = ssthresh;  
-	}
-
-	return (ssthresh != 0);
-}
-
-static inline int tcp_westwood_cwnd(struct tcp_sock *tp)
-{
-	__u32 cwnd = 0;
-
-	if (tcp_is_westwood(tp)) {
-		cwnd = __tcp_westwood_bw_rttmin(tp);
-		if (cwnd)
-			tp->snd_cwnd = cwnd;
-	}
-
-	return (cwnd != 0);
-}
 #endif	/* _TCP_H */
diff --git a/net/ipv4/Makefile b/net/ipv4/Makefile
index 65d57d8e1ad..89c0b4cb470 100644
--- a/net/ipv4/Makefile
+++ b/net/ipv4/Makefile
@@ -5,7 +5,8 @@
 obj-y     := utils.o route.o inetpeer.o protocol.o \
 	     ip_input.o ip_fragment.o ip_forward.o ip_options.o \
 	     ip_output.o ip_sockglue.o \
-	     tcp.o tcp_input.o tcp_output.o tcp_timer.o tcp_ipv4.o tcp_minisocks.o \
+	     tcp.o tcp_input.o tcp_output.o tcp_timer.o tcp_ipv4.o \
+	     tcp_minisocks.o tcp_cong.o \
 	     datagram.o raw.o udp.o arp.o icmp.o devinet.o af_inet.o igmp.o \
 	     sysctl_net_ipv4.o fib_frontend.o fib_semantics.o
 
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 23068bddbf0..e3289453241 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -118,6 +118,45 @@ static int ipv4_sysctl_forward_strategy(ctl_table *table,
 	return 1;
 }
 
+static int proc_tcp_congestion_control(ctl_table *ctl, int write, struct file * filp,
+				       void __user *buffer, size_t *lenp, loff_t *ppos)
+{
+	char val[TCP_CA_NAME_MAX];
+	ctl_table tbl = {
+		.data = val,
+		.maxlen = TCP_CA_NAME_MAX,
+	};
+	int ret;
+
+	tcp_get_default_congestion_control(val);
+
+	ret = proc_dostring(&tbl, write, filp, buffer, lenp, ppos);
+	if (write && ret == 0)
+		ret = tcp_set_default_congestion_control(val);
+	return ret;
+}
+
+int sysctl_tcp_congestion_control(ctl_table *table, int __user *name, int nlen,
+				  void __user *oldval, size_t __user *oldlenp,
+				  void __user *newval, size_t newlen,
+				  void **context)
+{
+	char val[TCP_CA_NAME_MAX];
+	ctl_table tbl = {
+		.data = val,
+		.maxlen = TCP_CA_NAME_MAX,
+	};
+	int ret;
+
+	tcp_get_default_congestion_control(val);
+	ret = sysctl_string(&tbl, name, nlen, oldval, oldlenp, newval, newlen,
+			    context);
+	if (ret == 0 && newval && newlen)
+		ret = tcp_set_default_congestion_control(val);
+	return ret;
+}
+
+
 ctl_table ipv4_table[] = {
         {
 		.ctl_name	= NET_IPV4_TCP_TIMESTAMPS,
@@ -612,70 +651,6 @@ ctl_table ipv4_table[] = {
 		.proc_handler	= &proc_dointvec,
 	},
 	{
-		.ctl_name	= NET_TCP_WESTWOOD, 
-		.procname	= "tcp_westwood",
-		.data		= &sysctl_tcp_westwood,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
-		.ctl_name	= NET_TCP_VEGAS,
-		.procname	= "tcp_vegas_cong_avoid",
-		.data		= &sysctl_tcp_vegas_cong_avoid,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
-		.ctl_name	= NET_TCP_VEGAS_ALPHA,
-		.procname	= "tcp_vegas_alpha",
-		.data		= &sysctl_tcp_vegas_alpha,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
-		.ctl_name	= NET_TCP_VEGAS_BETA,
-		.procname	= "tcp_vegas_beta",
-		.data		= &sysctl_tcp_vegas_beta,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
-		.ctl_name	= NET_TCP_VEGAS_GAMMA,
-		.procname	= "tcp_vegas_gamma",
-		.data		= &sysctl_tcp_vegas_gamma,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
-		.ctl_name	= NET_TCP_BIC,
-		.procname	= "tcp_bic",
-		.data		= &sysctl_tcp_bic,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
-		.ctl_name	= NET_TCP_BIC_FAST_CONVERGENCE,
-		.procname	= "tcp_bic_fast_convergence",
-		.data		= &sysctl_tcp_bic_fast_convergence,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
-		.ctl_name	= NET_TCP_BIC_LOW_WINDOW,
-		.procname	= "tcp_bic_low_window",
-		.data		= &sysctl_tcp_bic_low_window,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
 		.ctl_name	= NET_TCP_MODERATE_RCVBUF,
 		.procname	= "tcp_moderate_rcvbuf",
 		.data		= &sysctl_tcp_moderate_rcvbuf,
@@ -692,13 +667,14 @@ ctl_table ipv4_table[] = {
 		.proc_handler	= &proc_dointvec,
 	},
 	{
-		.ctl_name	= NET_TCP_BIC_BETA,
-		.procname	= "tcp_bic_beta",
-		.data		= &sysctl_tcp_bic_beta,
-		.maxlen		= sizeof(int),
+		.ctl_name	= NET_TCP_CONG_CONTROL,
+		.procname	= "tcp_congestion_control",
 		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
+		.maxlen		= TCP_CA_NAME_MAX,
+		.proc_handler	= &proc_tcp_congestion_control,
+		.strategy	= &sysctl_tcp_congestion_control,
 	},
+
 	{ .ctl_name = 0 }
 };
 
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 674bbd8cfd3..f3dbc8dc126 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -2333,6 +2333,8 @@ void __init tcp_init(void)
 	printk(KERN_INFO "TCP: Hash tables configured "
 	       "(established %d bind %d)\n",
 	       tcp_ehash_size << 1, tcp_bhash_size);
+
+	tcp_register_congestion_control(&tcp_reno);
 }
 
 EXPORT_SYMBOL(tcp_accept);
diff --git a/net/ipv4/tcp_cong.c b/net/ipv4/tcp_cong.c
new file mode 100644
index 00000000000..665394a63ae
--- /dev/null
+++ b/net/ipv4/tcp_cong.c
@@ -0,0 +1,195 @@
+/*
+ * Plugable TCP congestion control support and newReno
+ * congestion control.
+ * Based on ideas from I/O scheduler suport and Web100.
+ *
+ * Copyright (C) 2005 Stephen Hemminger <shemminger@osdl.org>
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/types.h>
+#include <linux/list.h>
+#include <net/tcp.h>
+
+static DEFINE_SPINLOCK(tcp_cong_list_lock);
+static LIST_HEAD(tcp_cong_list);
+
+/* Simple linear search, don't expect many entries! */
+static struct tcp_congestion_ops *tcp_ca_find(const char *name)
+{
+	struct tcp_congestion_ops *e;
+
+	list_for_each_entry(e, &tcp_cong_list, list) {
+		if (strcmp(e->name, name) == 0)
+			return e;
+	}
+
+	return NULL;
+}
+
+/*
+ * Attach new congestion control algorthim to the list
+ * of available options.
+ */
+int tcp_register_congestion_control(struct tcp_congestion_ops *ca)
+{
+	int ret = 0;
+
+	/* all algorithms must implement ssthresh and cong_avoid ops */
+	if (!ca->ssthresh || !ca->cong_avoid || !ca->min_cwnd) {
+		printk(KERN_ERR "TCP %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);
+		ret = -EEXIST;
+	} else {
+		list_add_rcu(&ca->list, &tcp_cong_list);
+		printk(KERN_INFO "TCP %s registered\n", ca->name);
+	}
+	spin_unlock(&tcp_cong_list_lock);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(tcp_register_congestion_control);
+
+/*
+ * Remove congestion control algorithm, called from
+ * the module's remove function.  Module ref counts are used
+ * to ensure that this can't be done till all sockets using
+ * that method are closed.
+ */
+void tcp_unregister_congestion_control(struct tcp_congestion_ops *ca)
+{
+	spin_lock(&tcp_cong_list_lock);
+	list_del_rcu(&ca->list);
+	spin_unlock(&tcp_cong_list_lock);
+}
+EXPORT_SYMBOL_GPL(tcp_unregister_congestion_control);
+
+/* Assign choice of congestion control. */
+void tcp_init_congestion_control(struct tcp_sock *tp)
+{
+	struct tcp_congestion_ops *ca;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
+		if (try_module_get(ca->owner)) {
+			tp->ca_ops = ca;
+			break;
+		}
+
+	}
+	rcu_read_unlock();
+
+	if (tp->ca_ops->init)
+		tp->ca_ops->init(tp);
+}
+
+/* Manage refcounts on socket close. */
+void tcp_cleanup_congestion_control(struct tcp_sock *tp)
+{
+	if (tp->ca_ops->release)
+		tp->ca_ops->release(tp);
+	module_put(tp->ca_ops->owner);
+}
+
+/* Used by sysctl to change default congestion control */
+int tcp_set_default_congestion_control(const char *name)
+{
+	struct tcp_congestion_ops *ca;
+	int ret = -ENOENT;
+
+	spin_lock(&tcp_cong_list_lock);
+	ca = tcp_ca_find(name);
+#ifdef CONFIG_KMOD
+	if (!ca) {
+		spin_unlock(&tcp_cong_list_lock);
+
+		request_module("tcp_%s", name);
+		spin_lock(&tcp_cong_list_lock);
+		ca = tcp_ca_find(name);
+	}
+#endif
+
+	if (ca) {
+		list_move(&ca->list, &tcp_cong_list);
+		ret = 0;
+	}
+	spin_unlock(&tcp_cong_list_lock);
+
+	return ret;
+}
+
+/* Get current default congestion control */
+void tcp_get_default_congestion_control(char *name)
+{
+	struct tcp_congestion_ops *ca;
+	/* We will always have reno... */
+	BUG_ON(list_empty(&tcp_cong_list));
+
+	rcu_read_lock();
+	ca = list_entry(tcp_cong_list.next, struct tcp_congestion_ops, list);
+	strncpy(name, ca->name, TCP_CA_NAME_MAX);
+	rcu_read_unlock();
+}
+
+/*
+ * TCP Reno congestion control
+ * This is special case used for fallback as well.
+ */
+/* This is Jacobson's slow start and congestion avoidance.
+ * SIGCOMM '88, p. 328.
+ */
+void tcp_reno_cong_avoid(struct tcp_sock *tp, u32 ack, u32 rtt, u32 in_flight,
+			 int flag)
+{
+	if (in_flight < tp->snd_cwnd)
+		return;
+
+        if (tp->snd_cwnd <= tp->snd_ssthresh) {
+                /* In "safe" area, increase. */
+		if (tp->snd_cwnd < tp->snd_cwnd_clamp)
+			tp->snd_cwnd++;
+	} else {
+                /* In dangerous area, increase slowly.
+		 * 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++;
+	}
+}
+EXPORT_SYMBOL_GPL(tcp_reno_cong_avoid);
+
+/* Slow start threshold is half the congestion window (min 2) */
+u32 tcp_reno_ssthresh(struct tcp_sock *tp)
+{
+	return max(tp->snd_cwnd >> 1U, 2U);
+}
+EXPORT_SYMBOL_GPL(tcp_reno_ssthresh);
+
+/* Lower bound on congestion window. */
+u32 tcp_reno_min_cwnd(struct tcp_sock *tp)
+{
+	return tp->snd_ssthresh/2;
+}
+EXPORT_SYMBOL_GPL(tcp_reno_min_cwnd);
+
+struct tcp_congestion_ops tcp_reno = {
+	.name		= "reno",
+	.owner		= THIS_MODULE,
+	.ssthresh	= tcp_reno_ssthresh,
+	.cong_avoid	= tcp_reno_cong_avoid,
+	.min_cwnd	= tcp_reno_min_cwnd,
+};
+
+EXPORT_SYMBOL_GPL(tcp_reno);
diff --git a/net/ipv4/tcp_diag.c b/net/ipv4/tcp_diag.c
index 634befc0792..867acc0f79d 100644
--- a/net/ipv4/tcp_diag.c
+++ b/net/ipv4/tcp_diag.c
@@ -42,7 +42,6 @@ struct tcpdiag_entry
 
 static struct sock *tcpnl;
 
-
 #define TCPDIAG_PUT(skb, attrtype, attrlen) \
 ({ int rtalen = RTA_LENGTH(attrlen);        \
    struct rtattr *rta;                      \
@@ -61,7 +60,6 @@ static int tcpdiag_fill(struct sk_buff *skb, struct sock *sk,
 	struct nlmsghdr  *nlh;
 	struct tcp_info  *info = NULL;
 	struct tcpdiag_meminfo  *minfo = NULL;
-	struct tcpvegas_info *vinfo = NULL;
 	unsigned char	 *b = skb->tail;
 
 	nlh = NLMSG_PUT(skb, pid, seq, TCPDIAG_GETSOCK, sizeof(*r));
@@ -73,9 +71,6 @@ static int tcpdiag_fill(struct sk_buff *skb, struct sock *sk,
 		if (ext & (1<<(TCPDIAG_INFO-1)))
 			info = TCPDIAG_PUT(skb, TCPDIAG_INFO, sizeof(*info));
 		
-		if ((tcp_is_westwood(tp) || tcp_is_vegas(tp))
-		    && (ext & (1<<(TCPDIAG_VEGASINFO-1))))
-			vinfo = TCPDIAG_PUT(skb, TCPDIAG_VEGASINFO, sizeof(*vinfo));
 	}
 	r->tcpdiag_family = sk->sk_family;
 	r->tcpdiag_state = sk->sk_state;
@@ -166,19 +161,8 @@ static int tcpdiag_fill(struct sk_buff *skb, struct sock *sk,
 	if (info) 
 		tcp_get_info(sk, info);
 
-	if (vinfo) {
-		if (tcp_is_vegas(tp)) {
-			vinfo->tcpv_enabled = tp->vegas.doing_vegas_now;
-			vinfo->tcpv_rttcnt = tp->vegas.cntRTT;
-			vinfo->tcpv_rtt = jiffies_to_usecs(tp->vegas.baseRTT);
-			vinfo->tcpv_minrtt = jiffies_to_usecs(tp->vegas.minRTT);
-		} else {
-			vinfo->tcpv_enabled = 0;
-			vinfo->tcpv_rttcnt = 0;
-			vinfo->tcpv_rtt = jiffies_to_usecs(tp->westwood.rtt);
-			vinfo->tcpv_minrtt = jiffies_to_usecs(tp->westwood.rtt_min);
-		}
-	}
+	if (sk->sk_state < TCP_TIME_WAIT && tp->ca_ops->get_info)
+		tp->ca_ops->get_info(tp, ext, skb);
 
 	nlh->nlmsg_len = skb->tail - b;
 	return skb->len;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 5bad504630a..7bbbbc33eb4 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -61,7 +61,6 @@
  *		Panu Kuhlberg:		Experimental audit of TCP (re)transmission
  *					engine. Lots of bugs are found.
  *		Pasi Sarolahti:		F-RTO for dealing with spurious RTOs
- *		Angelo Dell'Aera:	TCP Westwood+ support
  */
 
 #include <linux/config.h>
@@ -88,23 +87,9 @@ int sysctl_tcp_rfc1337;
 int sysctl_tcp_max_orphans = NR_FILE;
 int sysctl_tcp_frto;
 int sysctl_tcp_nometrics_save;
-int sysctl_tcp_westwood;
-int sysctl_tcp_vegas_cong_avoid;
 
 int sysctl_tcp_moderate_rcvbuf = 1;
 
-/* Default values of the Vegas variables, in fixed-point representation
- * with V_PARAM_SHIFT bits to the right of the binary point.
- */
-#define V_PARAM_SHIFT 1
-int sysctl_tcp_vegas_alpha = 1<<V_PARAM_SHIFT;
-int sysctl_tcp_vegas_beta  = 3<<V_PARAM_SHIFT;
-int sysctl_tcp_vegas_gamma = 1<<V_PARAM_SHIFT;
-int sysctl_tcp_bic = 1;
-int sysctl_tcp_bic_fast_convergence = 1;
-int sysctl_tcp_bic_low_window = 14;
-int sysctl_tcp_bic_beta = 819;		/* = 819/1024 (BICTCP_BETA_SCALE) */
-
 #define FLAG_DATA		0x01 /* Incoming frame contained data.		*/
 #define FLAG_WIN_UPDATE		0x02 /* Incoming ACK was a window update.	*/
 #define FLAG_DATA_ACKED		0x04 /* This ACK acknowledged new data.		*/
@@ -333,15 +318,6 @@ static void tcp_init_buffer_space(struct sock *sk)
 	tp->snd_cwnd_stamp = tcp_time_stamp;
 }
 
-static void init_bictcp(struct tcp_sock *tp)
-{
-	tp->bictcp.cnt = 0;
-
-	tp->bictcp.last_max_cwnd = 0;
-	tp->bictcp.last_cwnd = 0;
-	tp->bictcp.last_stamp = 0;
-}
-
 /* 5. Recalculate window clamp after socket hit its memory bounds. */
 static void tcp_clamp_window(struct sock *sk, struct tcp_sock *tp)
 {
@@ -558,45 +534,6 @@ static void tcp_event_data_recv(struct sock *sk, struct tcp_sock *tp, struct sk_
 		tcp_grow_window(sk, tp, skb);
 }
 
-/* When starting a new connection, pin down the current choice of 
- * congestion algorithm.
- */
-void tcp_ca_init(struct tcp_sock *tp)
-{
-	if (sysctl_tcp_westwood) 
-		tp->adv_cong = TCP_WESTWOOD;
-	else if (sysctl_tcp_bic)
-		tp->adv_cong = TCP_BIC;
-	else if (sysctl_tcp_vegas_cong_avoid) {
-		tp->adv_cong = TCP_VEGAS;
-		tp->vegas.baseRTT = 0x7fffffff;
-		tcp_vegas_enable(tp);
-	} 
-}
-
-/* Do RTT sampling needed for Vegas.
- * Basically we:
- *   o min-filter RTT samples from within an RTT to get the current
- *     propagation delay + queuing delay (we are min-filtering to try to
- *     avoid the effects of delayed ACKs)
- *   o min-filter RTT samples from a much longer window (forever for now)
- *     to find the propagation delay (baseRTT)
- */
-static inline void vegas_rtt_calc(struct tcp_sock *tp, __u32 rtt)
-{
-	__u32 vrtt = rtt + 1; /* Never allow zero rtt or baseRTT */
-
-	/* Filter to find propagation delay: */
-	if (vrtt < tp->vegas.baseRTT) 
-		tp->vegas.baseRTT = vrtt;
-
-	/* Find the min RTT during the last RTT to find
-	 * the current prop. delay + queuing delay:
-	 */
-	tp->vegas.minRTT = min(tp->vegas.minRTT, vrtt);
-	tp->vegas.cntRTT++;
-}
-
 /* Called to compute a smoothed rtt estimate. The data fed to this
  * routine either comes from timestamps, or from segments that were
  * known _not_ to have been retransmitted [see Karn/Partridge
@@ -606,13 +543,10 @@ static inline void vegas_rtt_calc(struct tcp_sock *tp, __u32 rtt)
  * To save cycles in the RFC 1323 implementation it was better to break
  * it up into three procedures. -- erics
  */
-static void tcp_rtt_estimator(struct tcp_sock *tp, __u32 mrtt)
+static void tcp_rtt_estimator(struct tcp_sock *tp, __u32 mrtt, u32 *usrtt)
 {
 	long m = mrtt; /* RTT */
 
-	if (tcp_vegas_enabled(tp))
-		vegas_rtt_calc(tp, mrtt);
-
 	/*	The following amusing code comes from Jacobson's
 	 *	article in SIGCOMM '88.  Note that rtt and mdev
 	 *	are scaled versions of rtt and mean deviation.
@@ -670,7 +604,8 @@ static void tcp_rtt_estimator(struct tcp_sock *tp, __u32 mrtt)
 		tp->rtt_seq = tp->snd_nxt;
 	}
 
-	tcp_westwood_update_rtt(tp, tp->srtt >> 3);
+	if (tp->ca_ops->rtt_sample)
+		tp->ca_ops->rtt_sample(tp, *usrtt);
 }
 
 /* Calculate rto without backoff.  This is the second half of Van Jacobson's
@@ -1185,8 +1120,8 @@ void tcp_enter_frto(struct sock *sk)
             tp->snd_una == tp->high_seq ||
             (tp->ca_state == TCP_CA_Loss && !tp->retransmits)) {
 		tp->prior_ssthresh = tcp_current_ssthresh(tp);
-		if (!tcp_westwood_ssthresh(tp))
-			tp->snd_ssthresh = tcp_recalc_ssthresh(tp);
+		tp->snd_ssthresh = tp->ca_ops->ssthresh(tp);
+		tcp_ca_event(tp, CA_EVENT_FRTO);
 	}
 
 	/* Have to clear retransmission markers here to keep the bookkeeping
@@ -1252,8 +1187,6 @@ static void tcp_enter_frto_loss(struct sock *sk)
 	tcp_set_ca_state(tp, TCP_CA_Loss);
 	tp->high_seq = tp->frto_highmark;
 	TCP_ECN_queue_cwr(tp);
-
-	init_bictcp(tp);
 }
 
 void tcp_clear_retrans(struct tcp_sock *tp)
@@ -1283,7 +1216,8 @@ void tcp_enter_loss(struct sock *sk, int how)
 	if (tp->ca_state <= TCP_CA_Disorder || tp->snd_una == tp->high_seq ||
 	    (tp->ca_state == TCP_CA_Loss && !tp->retransmits)) {
 		tp->prior_ssthresh = tcp_current_ssthresh(tp);
-		tp->snd_ssthresh = tcp_recalc_ssthresh(tp);
+		tp->snd_ssthresh = tp->ca_ops->ssthresh(tp);
+		tcp_ca_event(tp, CA_EVENT_LOSS);
 	}
 	tp->snd_cwnd	   = 1;
 	tp->snd_cwnd_cnt   = 0;
@@ -1596,28 +1530,14 @@ static inline void tcp_moderate_cwnd(struct tcp_sock *tp)
 }
 
 /* Decrease cwnd each second ack. */
-
 static void tcp_cwnd_down(struct tcp_sock *tp)
 {
 	int decr = tp->snd_cwnd_cnt + 1;
-	__u32 limit;
-
-	/*
-	 * TCP Westwood
-	 * Here limit is evaluated as BWestimation*RTTmin (for obtaining it
-	 * in packets we use mss_cache). If sysctl_tcp_westwood is off
-	 * tcp_westwood_bw_rttmin() returns 0. In such case snd_ssthresh is
-	 * still used as usual. It prevents other strange cases in which
-	 * BWE*RTTmin could assume value 0. It should not happen but...
-	 */
-
-	if (!(limit = tcp_westwood_bw_rttmin(tp)))
-		limit = tp->snd_ssthresh/2;
 
 	tp->snd_cwnd_cnt = decr&1;
 	decr >>= 1;
 
-	if (decr && tp->snd_cwnd > limit)
+	if (decr && tp->snd_cwnd > tp->ca_ops->min_cwnd(tp))
 		tp->snd_cwnd -= decr;
 
 	tp->snd_cwnd = min(tp->snd_cwnd, tcp_packets_in_flight(tp)+1);
@@ -1654,8 +1574,8 @@ static void DBGUNDO(struct sock *sk, struct tcp_sock *tp, const char *msg)
 static void tcp_undo_cwr(struct tcp_s