[ICSK]: Move TCP congestion avoidance members to icsk

This changeset basically moves tcp_sk()->{ca_ops,ca_state,etc} to inet_csk(),
minimal renaming/moving done in this changeset to ease review.

Most of it is just changes of struct tcp_sock * to struct sock * parameters.

With this we move to a state closer to two interesting goals:

1. Generalisation of net/ipv4/tcp_diag.c, becoming inet_diag.c, being used
   for any INET transport protocol that has struct inet_hashinfo and are
   derived from struct inet_connection_sock. Keeps the userspace API, that will
   just not display DCCP sockets, while newer versions of tools can support
   DCCP.

2. INET generic transport pluggable Congestion Avoidance infrastructure, using
   the current TCP CA infrastructure with DCCP.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

17 files changed, 352 insertions, 284 deletions

diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index 026630a15ea..fe3c6d3d0c9 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -508,7 +508,8 @@ struct sock *inet_csk_clone(struct sock *sk, const struct request_sock *req,
 		newsk->sk_write_space = sk_stream_write_space;
 
 		newicsk->icsk_retransmits = 0;
-		newicsk->icsk_backoff = 0;
+		newicsk->icsk_backoff	  = 0;
+		newicsk->icsk_probes_out  = 0;
 
 		/* Deinitialize accept_queue to trap illegal accesses. */
 		memset(&newicsk->icsk_accept_queue, 0, sizeof(newicsk->icsk_accept_queue));
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 0eed64a1991..02848e72e9c 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -1671,11 +1671,11 @@ int tcp_disconnect(struct sock *sk, int flags)
 		tp->write_seq = 1;
 	icsk->icsk_backoff = 0;
 	tp->snd_cwnd = 2;
-	tp->probes_out = 0;
+	icsk->icsk_probes_out = 0;
 	tp->packets_out = 0;
 	tp->snd_ssthresh = 0x7fffffff;
 	tp->snd_cwnd_cnt = 0;
-	tcp_set_ca_state(tp, TCP_CA_Open);
+	tcp_set_ca_state(sk, TCP_CA_Open);
 	tcp_clear_retrans(tp);
 	inet_csk_delack_init(sk);
 	sk->sk_send_head = NULL;
@@ -1718,7 +1718,7 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
 		name[val] = 0;
 
 		lock_sock(sk);
-		err = tcp_set_congestion_control(tp, name);
+		err = tcp_set_congestion_control(sk, name);
 		release_sock(sk);
 		return err;
 	}
@@ -1886,9 +1886,9 @@ void tcp_get_info(struct sock *sk, struct tcp_info *info)
 	memset(info, 0, sizeof(*info));
 
 	info->tcpi_state = sk->sk_state;
-	info->tcpi_ca_state = tp->ca_state;
+	info->tcpi_ca_state = icsk->icsk_ca_state;
 	info->tcpi_retransmits = icsk->icsk_retransmits;
-	info->tcpi_probes = tp->probes_out;
+	info->tcpi_probes = icsk->icsk_probes_out;
 	info->tcpi_backoff = icsk->icsk_backoff;
 
 	if (tp->rx_opt.tstamp_ok)
@@ -2016,7 +2016,7 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
 		len = min_t(unsigned int, len, TCP_CA_NAME_MAX);
 		if (put_user(len, optlen))
 			return -EFAULT;
-		if (copy_to_user(optval, tp->ca_ops->name, len))
+		if (copy_to_user(optval, icsk->icsk_ca_ops->name, len))
 			return -EFAULT;
 		return 0;
 	default:
diff --git a/net/ipv4/tcp_bic.c b/net/ipv4/tcp_bic.c
index ec38d45d664..b940346de4e 100644
--- a/net/ipv4/tcp_bic.c
+++ b/net/ipv4/tcp_bic.c
@@ -86,11 +86,11 @@ static inline void bictcp_reset(struct bictcp *ca)
 	ca->delayed_ack = 2 << ACK_RATIO_SHIFT;
 }
 
-static void bictcp_init(struct tcp_sock *tp)
+static void bictcp_init(struct sock *sk)
 {
-	bictcp_reset(tcp_ca(tp));
+	bictcp_reset(inet_csk_ca(sk));
 	if (initial_ssthresh)
-		tp->snd_ssthresh = initial_ssthresh;
+		tcp_sk(sk)->snd_ssthresh = initial_ssthresh;
 }
 
 /*
@@ -156,9 +156,10 @@ static inline void bictcp_update(struct bictcp *ca, u32 cwnd)
 
 
 /* Detect low utilization in congestion avoidance */
-static inline void bictcp_low_utilization(struct tcp_sock *tp, int flag)
+static inline void bictcp_low_utilization(struct sock *sk, int flag)
 {
-	struct bictcp *ca = tcp_ca(tp);
+	const struct tcp_sock *tp = tcp_sk(sk);
+	struct bictcp *ca = inet_csk_ca(sk);
 	u32 dist, delay;
 
 	/* No time stamp */
@@ -208,12 +209,13 @@ static inline void bictcp_low_utilization(struct tcp_sock *tp, int flag)
 
 }
 
-static void bictcp_cong_avoid(struct tcp_sock *tp, u32 ack,
+static void bictcp_cong_avoid(struct sock *sk, u32 ack,
 			      u32 seq_rtt, u32 in_flight, int data_acked)
 {
-	struct bictcp *ca = tcp_ca(tp);
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct bictcp *ca = inet_csk_ca(sk);
 
-	bictcp_low_utilization(tp, data_acked);
+	bictcp_low_utilization(sk, data_acked);
 
 	if (in_flight < tp->snd_cwnd)
 		return;
@@ -242,9 +244,10 @@ static void bictcp_cong_avoid(struct tcp_sock *tp, u32 ack,
  *	behave like Reno until low_window is reached,
  *	then increase congestion window slowly
  */
-static u32 bictcp_recalc_ssthresh(struct tcp_sock *tp)
+static u32 bictcp_recalc_ssthresh(struct sock *sk)
 {
-	struct bictcp *ca = tcp_ca(tp);
+	const struct tcp_sock *tp = tcp_sk(sk);
+	struct bictcp *ca = inet_csk_ca(sk);
 
 	ca->epoch_start = 0;	/* end of epoch */
 
@@ -269,31 +272,34 @@ static u32 bictcp_recalc_ssthresh(struct tcp_sock *tp)
 		return max((tp->snd_cwnd * beta) / BICTCP_BETA_SCALE, 2U);
 }
 
-static u32 bictcp_undo_cwnd(struct tcp_sock *tp)
+static u32 bictcp_undo_cwnd(struct sock *sk)
 {
-	struct bictcp *ca = tcp_ca(tp);
-
+	const struct tcp_sock *tp = tcp_sk(sk);
+	const struct bictcp *ca = inet_csk_ca(sk);
 	return max(tp->snd_cwnd, ca->last_max_cwnd);
 }
 
-static u32 bictcp_min_cwnd(struct tcp_sock *tp)
+static u32 bictcp_min_cwnd(struct sock *sk)
 {
+	const struct tcp_sock *tp = tcp_sk(sk);
 	return tp->snd_ssthresh;
 }
 
-static void bictcp_state(struct tcp_sock *tp, u8 new_state)
+static void bictcp_state(struct sock *sk, u8 new_state)
 {
 	if (new_state == TCP_CA_Loss)
-		bictcp_reset(tcp_ca(tp));
+		bictcp_reset(inet_csk_ca(sk));
 }
 
 /* Track delayed acknowledgement ratio using sliding window
  * ratio = (15*ratio + sample) / 16
  */
-static void bictcp_acked(struct tcp_sock *tp, u32 cnt)
+static void bictcp_acked(struct sock *sk, u32 cnt)
 {
-	if (cnt > 0 && 	tp->ca_state == TCP_CA_Open) {
-		struct bictcp *ca = tcp_ca(tp);
+	const struct inet_connection_sock *icsk = inet_csk(sk);
+
+	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;
 	}
@@ -314,7 +320,7 @@ static struct tcp_congestion_ops bictcp = {
 
 static int __init bictcp_register(void)
 {
-	BUG_ON(sizeof(struct bictcp) > TCP_CA_PRIV_SIZE);
+	BUG_ON(sizeof(struct bictcp) > ICSK_CA_PRIV_SIZE);
 	return tcp_register_congestion_control(&bictcp);
 }
 
diff --git a/net/ipv4/tcp_cong.c b/net/ipv4/tcp_cong.c
index 4970d10a778..bbf2d6624e8 100644
--- a/net/ipv4/tcp_cong.c
+++ b/net/ipv4/tcp_cong.c
@@ -73,33 +73,36 @@ void tcp_unregister_congestion_control(struct tcp_congestion_ops *ca)
 EXPORT_SYMBOL_GPL(tcp_unregister_congestion_control);
 
 /* Assign choice of congestion control. */
-void tcp_init_congestion_control(struct tcp_sock *tp)
+void tcp_init_congestion_control(struct sock *sk)
 {
+	struct inet_connection_sock *icsk = inet_csk(sk);
 	struct tcp_congestion_ops *ca;
 
-	if (tp->ca_ops != &tcp_init_congestion_ops)
+	if (icsk->icsk_ca_ops != &tcp_init_congestion_ops)
 		return;
 
 	rcu_read_lock();
 	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
 		if (try_module_get(ca->owner)) {
-			tp->ca_ops = ca;
+			icsk->icsk_ca_ops = ca;
 			break;
 		}
 
 	}
 	rcu_read_unlock();
 
-	if (tp->ca_ops->init)
-		tp->ca_ops->init(tp);
+	if (icsk->icsk_ca_ops->init)
+		icsk->icsk_ca_ops->init(sk);
 }
 
 /* Manage refcounts on socket close. */
-void tcp_cleanup_congestion_control(struct tcp_sock *tp)
+void tcp_cleanup_congestion_control(struct sock *sk)
 {
-	if (tp->ca_ops->release)
-		tp->ca_ops->release(tp);
-	module_put(tp->ca_ops->owner);
+	struct inet_connection_sock *icsk = inet_csk(sk);
+
+	if (icsk->icsk_ca_ops->release)
+		icsk->icsk_ca_ops->release(sk);
+	module_put(icsk->icsk_ca_ops->owner);
 }
 
 /* Used by sysctl to change default congestion control */
@@ -143,14 +146,15 @@ void tcp_get_default_congestion_control(char *name)
 }
 
 /* Change congestion control for socket */
-int tcp_set_congestion_control(struct tcp_sock *tp, const char *name)
+int tcp_set_congestion_control(struct sock *sk, const char *name)
 {
+	struct inet_connection_sock *icsk = inet_csk(sk);
 	struct tcp_congestion_ops *ca;
 	int err = 0;
 
 	rcu_read_lock();
 	ca = tcp_ca_find(name);
-	if (ca == tp->ca_ops)
+	if (ca == icsk->icsk_ca_ops)
 		goto out;
 
 	if (!ca)
@@ -160,10 +164,10 @@ int tcp_set_congestion_control(struct tcp_sock *tp, const char *name)
 		err = -EBUSY;
 
 	else {
-		tcp_cleanup_congestion_control(tp);
-		tp->ca_ops = ca;
-		if (tp->ca_ops->init)
-			tp->ca_ops->init(tp);
+		tcp_cleanup_congestion_control(sk);
+		icsk->icsk_ca_ops = ca;
+		if (icsk->icsk_ca_ops->init)
+			icsk->icsk_ca_ops->init(sk);
 	}
  out:
 	rcu_read_unlock();
@@ -177,9 +181,11 @@ int tcp_set_congestion_control(struct tcp_sock *tp, const char *name)
 /* 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,
+void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 rtt, u32 in_flight,
 			 int flag)
 {
+	struct tcp_sock *tp = tcp_sk(sk);
+
 	if (in_flight < tp->snd_cwnd)
 		return;
 
@@ -202,15 +208,17 @@ void tcp_reno_cong_avoid(struct tcp_sock *tp, u32 ack, u32 rtt, u32 in_flight,
 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)
+u32 tcp_reno_ssthresh(struct sock *sk)
 {
+	const struct tcp_sock *tp = tcp_sk(sk);
 	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)
+u32 tcp_reno_min_cwnd(struct sock *sk)
 {
+	const struct tcp_sock *tp = tcp_sk(sk);
 	return tp->snd_ssthresh/2;
 }
 EXPORT_SYMBOL_GPL(tcp_reno_min_cwnd);
diff --git a/net/ipv4/tcp_diag.c b/net/ipv4/tcp_diag.c
index 5f4c74f45e8..4288ecfec9a 100644
--- a/net/ipv4/tcp_diag.c
+++ b/net/ipv4/tcp_diag.c
@@ -66,10 +66,10 @@ 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 (ext & (1<<(TCPDIAG_CONG-1))) {
-			size_t len = strlen(tp->ca_ops->name);
+		if ((ext & (1 << (TCPDIAG_CONG - 1))) && icsk->icsk_ca_ops) {
+			size_t len = strlen(icsk->icsk_ca_ops->name);
 			strcpy(TCPDIAG_PUT(skb, TCPDIAG_CONG, len+1),
-			       tp->ca_ops->name);
+			       icsk->icsk_ca_ops->name);
 		}
 	}
 	r->tcpdiag_family = sk->sk_family;
@@ -136,18 +136,17 @@ static int tcpdiag_fill(struct sk_buff *skb, struct sock *sk,
 		r->tcpdiag_expires = EXPIRES_IN_MS(icsk->icsk_timeout);
 	} else if (icsk->icsk_pending == ICSK_TIME_PROBE0) {
 		r->tcpdiag_timer = 4;
-		r->tcpdiag_retrans = tp->probes_out;
+		r->tcpdiag_retrans = icsk->icsk_probes_out;
 		r->tcpdiag_expires = EXPIRES_IN_MS(icsk->icsk_timeout);
 	} else if (timer_pending(&sk->sk_timer)) {
 		r->tcpdiag_timer = 2;
-		r->tcpdiag_retrans = tp->probes_out;
+		r->tcpdiag_retrans = icsk->icsk_probes_out;
 		r->tcpdiag_expires = EXPIRES_IN_MS(sk->sk_timer.expires);
 	} else {
 		r->tcpdiag_timer = 0;
 		r->tcpdiag_expires = 0;
 	}
 #undef EXPIRES_IN_MS
-
 	r->tcpdiag_rqueue = tp->rcv_nxt - tp->copied_seq;
 	r->tcpdiag_wqueue = tp->write_seq - tp->snd_una;
 	r->tcpdiag_uid = sock_i_uid(sk);
@@ -163,8 +162,9 @@ static int tcpdiag_fill(struct sk_buff *skb, struct sock *sk,
 	if (info) 
 		tcp_get_info(sk, info);
 
-	if (sk->sk_state < TCP_TIME_WAIT && tp->ca_ops->get_info)
-		tp->ca_ops->get_info(tp, ext, skb);
+	if (sk->sk_state < TCP_TIME_WAIT &&
+	    icsk->icsk_ca_ops && icsk->icsk_ca_ops->get_info)
+		icsk->icsk_ca_ops->get_info(sk, ext, skb);
 
 	nlh->nlmsg_len = skb->tail - b;
 	return skb->len;
diff --git a/net/ipv4/tcp_highspeed.c b/net/ipv4/tcp_highspeed.c
index 36c51f8136b..6acc04bde08 100644
--- a/net/ipv4/tcp_highspeed.c
+++ b/net/ipv4/tcp_highspeed.c
@@ -98,9 +98,10 @@ struct hstcp {
 	u32	ai;
 };
 
-static void hstcp_init(struct tcp_sock *tp)
+static void hstcp_init(struct sock *sk)
 {
-	struct hstcp *ca = tcp_ca(tp);
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct hstcp *ca = inet_csk_ca(sk);
 
 	ca->ai = 0;
 
@@ -109,10 +110,11 @@ static void hstcp_init(struct tcp_sock *tp)
 	tp->snd_cwnd_clamp = min_t(u32, tp->snd_cwnd_clamp, 0xffffffff/128);
 }
 
-static void hstcp_cong_avoid(struct tcp_sock *tp, u32 adk, u32 rtt,
+static void hstcp_cong_avoid(struct sock *sk, u32 adk, u32 rtt,
 			     u32 in_flight, int good)
 {
-	struct hstcp *ca = tcp_ca(tp);
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct hstcp *ca = inet_csk_ca(sk);
 
 	if (in_flight < tp->snd_cwnd)
 		return;
@@ -143,9 +145,10 @@ static void hstcp_cong_avoid(struct tcp_sock *tp, u32 adk, u32 rtt,
 	}
 }
 
-static u32 hstcp_ssthresh(struct tcp_sock *tp)
+static u32 hstcp_ssthresh(struct sock *sk)
 {
-	struct hstcp *ca = tcp_ca(tp);
+	const struct tcp_sock *tp = tcp_sk(sk);
+	const struct hstcp *ca = inet_csk_ca(sk);
 
 	/* Do multiplicative decrease */
 	return max(tp->snd_cwnd - ((tp->snd_cwnd * hstcp_aimd_vals[ca->ai].md) >> 8), 2U);
@@ -164,7 +167,7 @@ static struct tcp_congestion_ops tcp_highspeed = {
 
 static int __init hstcp_register(void)
 {
-	BUG_ON(sizeof(struct hstcp) > TCP_CA_PRIV_SIZE);
+	BUG_ON(sizeof(struct hstcp) > ICSK_CA_PRIV_SIZE);
 	return tcp_register_congestion_control(&tcp_highspeed);
 }
 
diff --git a/net/ipv4/tcp_htcp.c b/net/ipv4/tcp_htcp.c
index 40168275acf..e47b37984e9 100644
--- a/net/ipv4/tcp_htcp.c
+++ b/net/ipv4/tcp_htcp.c
@@ -55,18 +55,21 @@ static inline void htcp_reset(struct htcp *ca)
 	ca->snd_cwnd_cnt2 = 0;
 }
 
-static u32 htcp_cwnd_undo(struct tcp_sock *tp)
+static u32 htcp_cwnd_undo(struct sock *sk)
 {
-	struct htcp *ca = tcp_ca(tp);
+	const struct tcp_sock *tp = tcp_sk(sk);
+	struct htcp *ca = inet_csk_ca(sk);
 	ca->ccount = ca->undo_ccount;
 	ca->maxRTT = ca->undo_maxRTT;
 	ca->old_maxB = ca->undo_old_maxB;
 	return max(tp->snd_cwnd, (tp->snd_ssthresh<<7)/ca->beta);
 }
 
-static inline void measure_rtt(struct tcp_sock *tp)
+static inline void measure_rtt(struct sock *sk)
 {
-	struct htcp *ca = tcp_ca(tp);
+	const struct inet_connection_sock *icsk = inet_csk(sk);
+	const struct tcp_sock *tp = tcp_sk(sk);
+	struct htcp *ca = inet_csk_ca(sk);
 	u32 srtt = tp->srtt>>3;
 
 	/* keep track of minimum RTT seen so far, minRTT is zero at first */
@@ -74,7 +77,7 @@ static inline void measure_rtt(struct tcp_sock *tp)
 		ca->minRTT = srtt;
 
 	/* max RTT */
-	if (tp->ca_state == TCP_CA_Open && tp->snd_ssthresh < 0xFFFF && ca->ccount > 3) {
+	if (icsk->icsk_ca_state == TCP_CA_Open && tp->snd_ssthresh < 0xFFFF && ca->ccount > 3) {
 		if (ca->maxRTT < ca->minRTT)
 			ca->maxRTT = ca->minRTT;
 		if (ca->maxRTT < srtt && srtt <= ca->maxRTT+HZ/50)
@@ -82,13 +85,16 @@ static inline void measure_rtt(struct tcp_sock *tp)
 	}
 }
 
-static void measure_achieved_throughput(struct tcp_sock *tp, u32 pkts_acked)
+static void measure_achieved_throughput(struct sock *sk, u32 pkts_acked)
 {
-	struct htcp *ca = tcp_ca(tp);
+	const struct inet_connection_sock *icsk = inet_csk(sk);
+	const struct tcp_sock *tp = tcp_sk(sk);
+	struct htcp *ca = inet_csk_ca(sk);
 	u32 now = tcp_time_stamp;
 
 	/* achieved throughput calculations */
-	if (tp->ca_state != TCP_CA_Open && tp->ca_state != TCP_CA_Disorder) {
+	if (icsk->icsk_ca_state != TCP_CA_Open &&
+	    icsk->icsk_ca_state != TCP_CA_Disorder) {
 		ca->packetcount = 0;
 		ca->lasttime = now;
 		return;
@@ -173,9 +179,9 @@ static inline void htcp_alpha_update(struct htcp *ca)
  * that point do we really have a real sense of maxRTT (the queues en route
  * were getting just too full now).
  */
-static void htcp_param_update(struct tcp_sock *tp)
+static void htcp_param_update(struct sock *sk)
 {
-	struct htcp *ca = tcp_ca(tp);
+	struct htcp *ca = inet_csk_ca(sk);
 	u32 minRTT = ca->minRTT;
 	u32 maxRTT = ca->maxRTT;
 
@@ -187,17 +193,19 @@ static void htcp_param_update(struct tcp_sock *tp)
 		ca->maxRTT = minRTT + ((maxRTT-minRTT)*95)/100;
 }
 
-static u32 htcp_recalc_ssthresh(struct tcp_sock *tp)
+static u32 htcp_recalc_ssthresh(struct sock *sk)
 {
-	struct htcp *ca = tcp_ca(tp);
-	htcp_param_update(tp);
+	const struct tcp_sock *tp = tcp_sk(sk);
+	const struct htcp *ca = inet_csk_ca(sk);
+	htcp_param_update(sk);
 	return max((tp->snd_cwnd * ca->beta) >> 7, 2U);
 }
 
-static void htcp_cong_avoid(struct tcp_sock *tp, u32 ack, u32 rtt,
+static void htcp_cong_avoid(struct sock *sk, u32 ack, u32 rtt,
 			    u32 in_flight, int data_acked)
 {
-	struct htcp *ca = tcp_ca(tp);
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct htcp *ca = inet_csk_ca(sk);
 
 	if (in_flight < tp->snd_cwnd)
 		return;
@@ -207,7 +215,7 @@ static void htcp_cong_avoid(struct tcp_sock *tp, u32 ack, u32 rtt,
 		if (tp->snd_cwnd < tp->snd_cwnd_clamp)
 			tp->snd_cwnd++;
 	} else {
-		measure_rtt(tp);
+		measure_rtt(sk);
 
 		/* keep track of number of round-trip times since last backoff event */
 		if (ca->snd_cwnd_cnt2++ > tp->snd_cwnd) {
@@ -229,28 +237,29 @@ static void htcp_cong_avoid(struct tcp_sock *tp, u32 ack, u32 rtt,
 }
 
 /* Lower bound on congestion window. */
-static u32 htcp_min_cwnd(struct tcp_sock *tp)
+static u32 htcp_min_cwnd(struct sock *sk)
 {
+	const struct tcp_sock *tp = tcp_sk(sk);
 	return tp->snd_ssthresh;
 }
 
 
-static void htcp_init(struct tcp_sock *tp)
+static void htcp_init(struct sock *sk)
 {
-	struct htcp *ca = tcp_ca(tp);
+	struct htcp *ca = inet_csk_ca(sk);
 
 	memset(ca, 0, sizeof(struct htcp));
 	ca->alpha = ALPHA_BASE;
 	ca->beta = BETA_MIN;
 }
 
-static void htcp_state(struct tcp_sock *tp, u8 new_state)
+static void htcp_state(struct sock *sk, u8 new_state)
 {
 	switch (new_state) {
 	case TCP_CA_CWR:
 	case TCP_CA_Recovery:
 	case TCP_CA_Loss:
-		htcp_reset(tcp_ca(tp));
+		htcp_reset(inet_csk_ca(sk));
 		break;
 	}
 }
@@ -269,7 +278,7 @@ static struct tcp_congestion_ops htcp = {
 
 static int __init htcp_register(void)
 {
-	BUG_ON(sizeof(struct htcp) > TCP_CA_PRIV_SIZE);
+	BUG_ON(sizeof(struct htcp) > ICSK_CA_PRIV_SIZE);
 	BUILD_BUG_ON(BETA_MIN >= BETA_MAX);
 	if (!use_bandwidth_switch)
 		htcp.pkts_acked = NULL;
diff --git a/net/ipv4/tcp_hybla.c b/net/ipv4/tcp_hybla.c
index 13a66342c30..77add63623d 100644
--- a/net/ipv4/tcp_hybla.c
+++ b/net/ipv4/tcp_hybla.c
@@ -33,19 +33,20 @@ MODULE_PARM_DESC(rtt0, "reference rout trip time (ms)");
 
 
 /* This is called to refresh values for hybla parameters */
-static inline void hybla_recalc_param (struct tcp_sock *tp)
+static inline void hybla_recalc_param (struct sock *sk)
 {
-	struct hybla *ca = tcp_ca(tp);
+	struct hybla *ca = inet_csk_ca(sk);
 
-	ca->rho_3ls = max_t(u32, tp->srtt / msecs_to_jiffies(rtt0), 8);
+	ca->rho_3ls = max_t(u32, tcp_sk(sk)->srtt / msecs_to_jiffies(rtt0), 8);
 	ca->rho = ca->rho_3ls >> 3;
 	ca->rho2_7ls = (ca->rho_3ls * ca->rho_3ls) << 1;
 	ca->rho2 = ca->rho2_7ls >>7;
 }
 
-static void hybla_init(struct tcp_sock *tp)
+static void hybla_init(struct sock *sk)
 {
-	struct hybla *ca = tcp_ca(tp);
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct hybla *ca = inet_csk_ca(sk);
 
 	ca->rho = 0;
 	ca->rho2 = 0;
@@ -57,17 +58,16 @@ static void hybla_init(struct tcp_sock *tp)
 	tp->snd_cwnd_clamp = 65535;
 
 	/* 1st Rho measurement based on initial srtt */
-	hybla_recalc_param(tp);
+	hybla_recalc_param(sk);
 
 	/* set minimum rtt as this is the 1st ever seen */
 	ca->minrtt = tp->srtt;
 	tp->snd_cwnd = ca->rho;
 }
 
-static void hybla_state(struct tcp_sock *tp, u8 ca_state)
+static void hybla_state(struct sock *sk, u8 ca_state)
 {
-	struct hybla *ca = tcp_ca(tp);
-
+	struct hybla *ca = inet_csk_ca(sk);
 	ca->hybla_en = (ca_state == TCP_CA_Open);
 }
 
@@ -86,27 +86,28 @@ static inline u32 hybla_fraction(u32 odds)
  *     o Give cwnd a new value based on the model proposed
  *     o remember increments <1
  */
-static void hybla_cong_avoid(struct tcp_sock *tp, u32 ack, u32 rtt,
+static void hybla_cong_avoid(struct sock *sk, u32 ack, u32 rtt,
 			    u32 in_flight, int flag)
 {
-	struct hybla *ca = tcp_ca(tp);
+	struct tcp_sock *tp = tcp_sk(sk);
+	struct hybla *ca = inet_csk_ca(sk);
 	u32 increment, odd, rho_fractions;
 	int is_slowstart = 0;
 
 	/*  Recalculate rho only if this srtt is the lowest */
 	if (tp->srtt < ca->minrtt){
-		hybla_recalc_param(tp);
+		hybla_recalc_param(sk);
 		ca->minrtt = tp->srtt;
 	}
 
 	if (!ca->hybla_en)
-		return tcp_reno_cong_avoid(tp, ack, rtt, in_flight, flag);
+		return tcp_reno_cong_avoid(sk, ack, rtt, in_flight, flag);
 
 	if (in_flight < tp->snd_cwnd)
 		return;
 
 	if (ca->rho == 0)
-		hybla_recalc_param(tp);
+		hybla_recalc_param(sk);
 
 	rho_fractions = ca->rho_3ls - (ca->rho << 3);
 
@@ -170,7 +171,7 @@ static struct tcp_congestion_ops tcp_hybla = {
 
 static int __init hybla_register(void)
 {
-	BUG_ON(sizeof(struct hybla) > TCP_CA_PRIV_SIZE);
+	BUG_ON(sizeof(struct hybla) > ICSK_CA_PRIV_SIZE);
 	return tcp_register_congestion_control(&tcp_hybla);
 }
 
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 71d456148de..fdd9547fb78 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -325,11 +325,12 @@ static void tcp_init_buffer_space(struct sock *sk)
 /* 5. Recalculate window clamp after socket hit its memory bounds. */
 static void tcp_clamp_window(struct sock *sk, struct tcp_sock *tp)
 {
+	struct inet_connection_sock *icsk = inet_csk(sk);
 	struct sk_buff *skb;
 	unsigned int app_win = tp->rcv_nxt - tp->copied_seq;
 	int ofo_win = 0;
 
-	inet_csk(sk)->icsk_ack.quick = 0;
+	icsk->icsk_ack.quick = 0;
 
 	skb_queue_walk(&tp->out_of_order_queue, skb) {
 		ofo_win += skb->len;
@@ -350,8 +351,8 @@ static void tcp_clamp_window(struct sock *sk, struct tcp_sock *tp)
 		app_win += ofo_win;
 		if (atomic_read(&sk->sk_rmem_alloc) >= 2 * sk->sk_rcvbuf)
 			app_win >>= 1;
-		if (app_win > inet_csk(sk)->icsk_ack.rcv_mss)
-			app_win -= inet_csk(sk)->icsk_ack.rcv_mss;
+		if (app_win > icsk->icsk_ack.rcv_mss)
+			app_win -= icsk->icsk_ack.rcv_mss;
 		app_win = max(app_win, 2U*tp->advmss);
 
 		if (!ofo_win)
@@ -549,8 +550,10 @@ static void tcp_event_data_recv(struct sock *sk, struct tcp_sock *tp, struct sk_
  * 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, u32 *usrtt)
+static void tcp_rtt_estimator(struct sock *sk, const __u32 mrtt, u32 *usrtt)
 {
+	struct tcp_sock *tp = tcp_sk(sk);
+	const struct inet_connection_sock *icsk = inet_csk(sk);
 	long m = mrtt; /* RTT */
 
 	/*	The following amusing code comes from Jacobson's
@@ -610,8 +613,8 @@ static void tcp_rtt_estimator(struct tcp_sock *tp, __u32 mrtt, u32 *usrtt)
 		tp->rtt_seq = tp->snd_nxt;
 	}
 
-	if (tp->ca_ops->rtt_sample)
-		tp->ca_ops->rtt_sample(tp, *usrtt);
+	if (icsk->icsk_ca_ops->rtt_sample)
+		icsk->icsk_ca_ops->rtt_sample(sk, *usrtt);
 }
 
 /* Calculate rto without backoff.  This is the second half of Van Jacobson's
@@ -663,9 +666,10 @@ void tcp_update_metrics(struct sock *sk)
 	dst_confirm(dst);
 
 	if (dst && (dst->flags&DST_HOST)) {
+		const struct inet_connection_sock *icsk = inet_csk(sk);
 		int m;
 
-		if (inet_csk(sk)->icsk_backoff || !tp->srtt) {
+		if (icsk->icsk_backoff || !tp->srtt) {
 			/* This session failed to estimate rtt. Why?
 			 * Probably, no packets returned in time.
 			 * Reset our results.
@@ -714,7 +718,7 @@ void tcp_update_metrics(struct sock *sk)
 			    tp->snd_cwnd > dst_metric(dst, RTAX_CWND))
 				dst->metrics[RTAX_CWND-1] = tp->snd_cwnd;
 		} else if (tp->snd_cwnd > tp->snd_ssthresh &&
-			   tp->ca_state == TCP_CA_Open) {
+			   icsk->icsk_ca_state == TCP_CA_Open) {
 			/* Cong. avoidance phase, cwnd is reliable. */
 			if (!dst_metric_locked(dst, RTAX_SSTHRESH))
 				dst->metrics[RTAX_SSTHRESH-1] =
@@ -828,8 +832,10 @@ reset:
 	}
 }
 
-static void tcp_update_reordering(struct tcp_sock *tp, int metric, int ts)
+static void tcp_update_reordering(struct sock *sk, const int metric,
+				  const int ts)
 {
+	struct tcp_sock *tp = tcp_sk(sk);
 	if (metric > tp->reordering) {
 		tp->reordering = min(TCP_MAX_REORDERING, metric);
 
@@ -844,7 +850,7 @@ static void tcp_update_reordering(struct tcp_sock *tp, int metric, int ts)
 			NET_INC_STATS_BH(LINUX_MIB_TCPSACKREORDER);
 #if FASTRETRANS_DEBUG > 1
 		printk(KERN_DEBUG "Disorder%d %d %u f%u s%u rr%d\n",
-		       tp->rx_opt.sack_ok, tp->ca_state,
+		       tp->rx_opt.sack_ok, inet_csk(sk)->icsk_ca_state,
 		       tp->reordering,
 		       tp->fackets_out,
 		       tp->sacked_out,
@@ -906,6 +912,7 @@ static void tcp_update_reordering(struct tcp_sock *tp, int metric, int ts)
 static int
 tcp_sacktag_write_queue(struct sock *sk, struct sk_buff *ack_skb, u32 prior_snd_una)
 {
+	const struct inet_connection_sock *icsk = inet_csk(sk);
 	struct tcp_sock *tp = tcp_sk(sk);
 	unsigned char *ptr = ack_skb->h.raw + TCP_SKB_CB(ack_skb)->sacked;
 	struct tcp_sack_block *sp = (struct tcp_sack_block *)(ptr+2);
@@ -1071,7 +1078,7 @@ tcp_sacktag_write_queue(struct sock *sk, struct sk_buff *ack_skb, u32 prior_snd_
 	 * we have to account for reordering! Ugly,
 	 * but should help.
 	 */
-	if (lost_retrans && tp->ca_state == TCP_CA_Recovery) {
+	if (lost_retrans && icsk->icsk_ca_state == TCP_CA_Recovery) {
 		struct sk_buff *skb;
 
 		sk_stream_for_retrans_queue(skb, sk) {
@@ -1100,8 +1107,8 @@ tcp_sacktag_write_queue(struct sock *sk, struct sk_buff *ack_skb, u32 prior_snd_
 
 	tp->left_out = tp->sacked_out + tp->lost_out;
 
-	if ((reord < tp->fackets_out) && tp->ca_state != TCP_CA_Loss)
-		tcp_update_reordering(tp, ((tp->fackets_out + 1) - reord), 0);
+	if ((reord < tp->fackets_out) && icsk->icsk_ca_state != TCP_CA_Loss)
+		tcp_update_reordering(sk, ((tp->fackets_out + 1) - reord), 0);
 
 #if FASTRETRANS_DEBUG > 0
 	BUG_TRAP((int)tp->sacked_out >= 0);
@@ -1118,17 +1125,18 @@ tcp_sacktag_write_queue(struct sock *sk, struct sk_buff *ack_skb, u32 prior_snd_
  */
 void tcp_enter_frto(struct sock *sk)
 {
+	const struct inet_connection_sock *icsk = inet_csk(sk);
 	struct tcp_sock *tp = tcp_sk(sk);
 	struct sk_buff *skb;
 
 	tp->frto_counter = 1;
 
-	if (tp->ca_state <= TCP_CA_Disorder ||
+	if (icsk->icsk_ca_state <= TCP_CA_Disorder ||
             tp->snd_una == tp->high_seq ||
-            (tp->ca_state == TCP_CA_Loss && !inet_csk(sk)->icsk_retransmits)) {
-		tp->prior_ssthresh = tcp_current_ssthresh(tp);
-		tp->snd_ssthresh = tp->ca_ops->ssthresh(tp);
-		tcp_ca_event(tp, CA_EVENT_FRTO);
+            (icsk->icsk_ca_state == TCP_CA_Loss && !icsk->icsk_retransmits)) {
+		tp->prior_ssthresh = tcp_current_ssthresh(sk);
+		tp->snd_ssthresh = icsk->icsk_ca_ops->ssthresh(sk);
+		tcp_ca_event(sk, CA_EVENT_FRTO);
 	}
 
 	/* Have to clear retransmission markers here to keep the bookkeeping
@@ -1145,7 +1153,7 @@ void tcp_enter_frto(struct sock *sk)
 	}
 	tcp_sync_left_out(tp);
 
-	tcp_set_ca_state(tp, TCP_CA_Open);
+	tcp_set_ca_state(sk, TCP_CA_Open);
 	tp->frto_highmark = tp->snd_nxt;
 }
 
@@ -1191,7 +1199,7 @@ static void tcp_enter_frto_loss(struct sock *sk)
 
 	tp->reordering = min_t(unsigned int, tp->reordering,
 					     sysctl_tcp_reordering);
-	tcp_set_ca_state(tp, TCP_CA_Loss);
+	tcp_set_ca_state(sk, TCP_CA_Loss);
 	tp->high_seq = tp->frto_highmark;
 	TCP_ECN_queue_cwr(tp);
 }
@@ -1215,16 +1223,17 @@ void tcp_clear_retrans(struct tcp_sock *tp)
  */
 void tcp_enter_loss(struct sock *sk, int how)
 {
+	const struct inet_connection_sock *icsk = inet_csk(sk);
 	struct tcp_sock *tp = tcp_sk(sk);
 	struct sk_buff *skb;
 	int cnt = 0;
 
 	/* Reduce ssthresh if it has not yet been made inside this window. */
-	if (tp->ca_state <= TCP_CA_Disorder || tp->snd_una == tp->high_seq ||
-	    (tp->ca_state == TCP_CA_Loss && !inet_csk(sk)->icsk_retransmits)) {
-		tp->prior_ssthresh = tcp_current_ssthresh(tp);
-		tp->snd_ssthresh = tp->ca_ops->ssthresh(tp);
-		tcp_ca_event(tp, CA_EVENT_LOSS);
+	if (icsk->icsk_ca_state <= TCP_CA_Disorder || tp->snd_una == tp->high_seq ||
+	    (icsk->icsk_ca_state == TCP_CA_Loss && !icsk->icsk_retransmits)) {
+		tp->prior_ssthresh = tcp_current_ssthresh(sk);
+		tp->snd_ssthresh = icsk->icsk_ca_ops->ssthresh(sk);
+		tcp_ca_event(sk, CA_EVENT_LOSS);
 	}
 	tp->snd_cwnd	   = 1;
 	tp->snd_cwnd_cnt   = 0;
@@ -1255,7 +1264,7 @@ void tcp_enter_loss(struct sock *sk, int how)
 
 	tp->reordering = min_t(unsigned int, tp->reordering,
 					     sysctl_tcp_reordering);
-	tcp_set_ca_state(tp, TCP_CA_Loss);
+	tcp_set_ca_state(sk, TCP_CA_Loss);
 	tp->high_seq = tp->snd_nxt;
 	TCP_ECN_queue_cwr(tp);
 }
@@ -1272,13 +1281,14 @@ static int tcp_check_sack_reneging(struct sock *sk)
 	 */
 	if ((skb = skb_peek(&sk->sk_write_queue)) != NULL &&
 	    (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)) {
+		struct inet_connection_sock *icsk = inet_csk(sk);
 		NET_INC_STATS_BH(LINUX_MIB_TCPSACKRENEGING);
 
 		tcp_enter_loss(sk, 1);
-		inet_csk(sk)->icsk_retransmits++;
+		icsk->icsk_retransmits++;
 		tcp_retransmit_skb(sk, skb_peek(&sk->sk_write_queue));
 		inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
-					  inet_csk(sk)->icsk_rto, TCP_RTO_MAX);
+					  icsk->icsk_rto, TCP_RTO_MAX);
 		return 1;
 	}
 	return 0;
@@ -1431,8 +1441,9 @@ static int tcp_time_to_recover(struct sock *sk, struct tcp_sock *tp)
  * in assumption of absent reordering, interpret this as reordering.
  * The only another reason could be bug in receiver TCP.
  */
-static void tcp_check_reno_reordering(struct tcp_sock *tp, int addend)
+static void tcp_check_reno_reordering(struct sock *sk, const int addend)
 {
+	struct tcp_sock *tp = tcp_sk(sk);
 	u32 holes;
 
 	holes = max(tp->lost_out, 1U);
@@ -1440,16 +1451,17 @@ static void tcp_check_reno_reordering(struct tcp_sock *tp, int addend)
 
 	if ((tp->sacked_out + holes) > tp->packets_out) {
 		tp->sacked_out = tp->packets_out - holes;
-		tcp_update_reordering(tp, tp->packets_out+addend, 0);
+		tcp_update_reordering(sk, tp->packets_out + addend, 0);
 	}
 }
 
 /* Emulate SACKs for SACKless connection: account for a new dupack. */
 
-static void tcp_add_reno_sack(struct tcp_sock *tp)
+static void tcp_add_reno_sack(struct sock *sk)
 {
+	struct tcp_sock *tp = tcp_sk(sk);
 	tp->sacked_out++;
-	tcp_check_reno_reordering(t