[TCP]: Minimum congestion window consolidation.

Many of the TCP congestion methods all just use ssthresh
as the minimum congestion window on decrease.  Rather than
duplicating the code, just have that be the default if that
handle in the ops structure is not set.

Minor behaviour change to TCP compound.  It probably wants
to use this (ssthresh) as lower bound, rather than ssthresh/2
because the latter causes undershoot on loss.

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

diff --git a/include/net/tcp.h b/include/net/tcp.h
index f1f4727..de88c54 100644 (file)
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -632,7 +632,7 @@ struct tcp_congestion_ops {
        /* return slow start threshold (required) */
        u32 (*ssthresh)(struct sock *sk);
        /* lower bound for congestion window (optional) */
-       u32 (*min_cwnd)(struct sock *sk);
+       u32 (*min_cwnd)(const struct sock *sk);
        /* do new cwnd calculation (required) */
        void (*cong_avoid)(struct sock *sk, u32 ack,
                           u32 rtt, u32 in_flight, int good_ack);
@@ -667,7 +667,7 @@ extern struct tcp_congestion_ops tcp_init_congestion_ops;
 extern u32 tcp_reno_ssthresh(struct sock *sk);
 extern void tcp_reno_cong_avoid(struct sock *sk, u32 ack,
                                u32 rtt, u32 in_flight, int flag);
-extern u32 tcp_reno_min_cwnd(struct sock *sk);
+extern u32 tcp_reno_min_cwnd(const struct sock *sk);
 extern struct tcp_congestion_ops tcp_reno;
 
 static inline void tcp_set_ca_state(struct sock *sk, const u8 ca_state)

diff --git a/net/ipv4/tcp_bic.c b/net/ipv4/tcp_bic.c
index 035f209..b2d9021 100644 (file)
--- a/net/ipv4/tcp_bic.c
+++ b/net/ipv4/tcp_bic.c
@@ -198,12 +198,6 @@ static u32 bictcp_undo_cwnd(struct sock *sk)
        return max(tp->snd_cwnd, ca->last_max_cwnd);
 }
 
-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 sock *sk, u8 new_state)
 {
        if (new_state == TCP_CA_Loss)
@@ -231,7 +225,6 @@ static struct tcp_congestion_ops bictcp = {
        .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           = "bic",

diff --git a/net/ipv4/tcp_compound.c b/net/ipv4/tcp_compound.c
index ec68cb8..bc54f7e 100644 (file)
--- a/net/ipv4/tcp_compound.c
+++ b/net/ipv4/tcp_compound.c
@@ -419,7 +419,6 @@ static struct tcp_congestion_ops tcp_compound = {
        .init           = tcp_compound_init,
        .ssthresh       = tcp_reno_ssthresh,
        .cong_avoid     = tcp_compound_cong_avoid,
-       .min_cwnd       = tcp_reno_min_cwnd,
        .rtt_sample     = tcp_compound_rtt_calc,
        .set_state      = tcp_compound_state,
        .cwnd_event     = tcp_compound_cwnd_event,

diff --git a/net/ipv4/tcp_cong.c b/net/ipv4/tcp_cong.c
index 91c2f41..857eefc 100644 (file)
--- a/net/ipv4/tcp_cong.c
+++ b/net/ipv4/tcp_cong.c
@@ -38,7 +38,7 @@ 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) {
+       if (!ca->ssthresh || !ca->cong_avoid) {
                printk(KERN_ERR "TCP %s does not implement required ops\n",
                       ca->name);
                return -EINVAL;
@@ -251,8 +251,8 @@ u32 tcp_reno_ssthresh(struct sock *sk)
 }
 EXPORT_SYMBOL_GPL(tcp_reno_ssthresh);
 
-/* Lower bound on congestion window. */
-u32 tcp_reno_min_cwnd(struct sock *sk)
+/* 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;

diff --git a/net/ipv4/tcp_cubic.c b/net/ipv4/tcp_cubic.c
index 31a4986..78b7a6b 100644 (file)
--- a/net/ipv4/tcp_cubic.c
+++ b/net/ipv4/tcp_cubic.c
@@ -325,11 +325,6 @@ static u32 bictcp_undo_cwnd(struct sock *sk)
        return max(tcp_sk(sk)->snd_cwnd, ca->last_max_cwnd);
 }
 
-static u32 bictcp_min_cwnd(struct sock *sk)
-{
-       return tcp_sk(sk)->snd_ssthresh;
-}
-
 static void bictcp_state(struct sock *sk, u8 new_state)
 {
        if (new_state == TCP_CA_Loss)
@@ -357,7 +352,6 @@ static struct tcp_congestion_ops cubictcp = {
        .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",

diff --git a/net/ipv4/tcp_htcp.c b/net/ipv4/tcp_htcp.c
index 1b2ff53..3d92c18 100644 (file)
--- a/net/ipv4/tcp_htcp.c
+++ b/net/ipv4/tcp_htcp.c
@@ -246,14 +246,6 @@ static void htcp_cong_avoid(struct sock *sk, u32 ack, u32 rtt,
        }
 }
 
-/* Lower bound on congestion window. */
-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 sock *sk)
 {
        struct htcp *ca = inet_csk_ca(sk);
@@ -285,7 +277,6 @@ static void htcp_state(struct sock *sk, u8 new_state)
 static struct tcp_congestion_ops htcp = {
        .init           = htcp_init,
        .ssthresh       = htcp_recalc_ssthresh,
-       .min_cwnd       = htcp_min_cwnd,
        .cong_avoid     = htcp_cong_avoid,
        .set_state      = htcp_state,
        .undo_cwnd      = htcp_cwnd_undo,

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 6d16788..e08245b 100644 (file)
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -1689,17 +1689,26 @@ static inline void tcp_moderate_cwnd(struct tcp_sock *tp)
        tp->snd_cwnd_stamp = tcp_time_stamp;
 }
 
+/* Lower bound on congestion window is slow start threshold
+ * unless congestion avoidance choice decides to overide it.
+ */
+static inline u32 tcp_cwnd_min(const struct sock *sk)
+{
+       const struct tcp_congestion_ops *ca_ops = inet_csk(sk)->icsk_ca_ops;
+
+       return ca_ops->min_cwnd ? ca_ops->min_cwnd(sk) : tcp_sk(sk)->snd_ssthresh;
+}
+
 /* Decrease cwnd each second ack. */
 static void tcp_cwnd_down(struct sock *sk)
 {
-       const struct inet_connection_sock *icsk = inet_csk(sk);
        struct tcp_sock *tp = tcp_sk(sk);
        int decr = tp->snd_cwnd_cnt + 1;
 
        tp->snd_cwnd_cnt = decr&1;
        decr >>= 1;
 
-       if (decr && tp->snd_cwnd > icsk->icsk_ca_ops->min_cwnd(sk))
+       if (decr && tp->snd_cwnd > tcp_cwnd_min(sk))
                tp->snd_cwnd -= decr;
 
        tp->snd_cwnd = min(tp->snd_cwnd, tcp_packets_in_flight(tp)+1);

diff --git a/net/ipv4/tcp_veno.c b/net/ipv4/tcp_veno.c
index 1091671..11b42a7 100644 (file)
--- a/net/ipv4/tcp_veno.c
+++ b/net/ipv4/tcp_veno.c
@@ -199,17 +199,10 @@ static u32 tcp_veno_ssthresh(struct sock *sk)
                return max(tp->snd_cwnd >> 1U, 2U);
 }
 
-static u32 tcp_veno_min_cwnd(struct sock * sk)
-{
-       const struct tcp_sock *tp = tcp_sk(sk);
-       return tp->snd_ssthresh;
-}
-
 static struct tcp_congestion_ops tcp_veno = {
        .init           = tcp_veno_init,
        .ssthresh       = tcp_veno_ssthresh,
        .cong_avoid     = tcp_veno_cong_avoid,
-       .min_cwnd       = tcp_veno_min_cwnd,
        .rtt_sample     = tcp_veno_rtt_calc,
        .set_state      = tcp_veno_state,
        .cwnd_event     = tcp_veno_cwnd_event,

diff --git a/net/ipv4/tcp_westwood.c b/net/ipv4/tcp_westwood.c
index 0c340c3..29eb258 100644 (file)
--- a/net/ipv4/tcp_westwood.c
+++ b/net/ipv4/tcp_westwood.c
@@ -162,12 +162,6 @@ static inline u32 westwood_acked_count(struct sock *sk)
        return w->cumul_ack;
 }
 
-static inline u32 westwood_bw_rttmin(const struct sock *sk)
-{
-       const struct tcp_sock *tp = tcp_sk(sk);
-       const struct westwood *w = inet_csk_ca(sk);
-       return max_t(u32, (w->bw_est * w->rtt_min) / tp->mss_cache, 2);
-}
 
 /*
  * TCP Westwood
@@ -175,9 +169,11 @@ static inline u32 westwood_bw_rttmin(const struct sock *sk)
  * in packets we use mss_cache). Rttmin is guaranteed to be >= 2
  * so avoids ever returning 0.
  */
-static u32 tcp_westwood_cwnd_min(struct sock *sk)
+static u32 tcp_westwood_bw_rttmin(const struct sock *sk)
 {
-       return westwood_bw_rttmin(sk);
+       const struct tcp_sock *tp = tcp_sk(sk);
+       const struct westwood *w = inet_csk_ca(sk);
+       return max_t(u32, (w->bw_est * w->rtt_min) / tp->mss_cache, 2);
 }
 
 static void tcp_westwood_event(struct sock *sk, enum tcp_ca_event event)
@@ -191,11 +187,11 @@ static void tcp_westwood_event(struct sock *sk, enum tcp_ca_event event)
                break;
 
        case CA_EVENT_COMPLETE_CWR:
-               tp->snd_cwnd = tp->snd_ssthresh = westwood_bw_rttmin(sk);
+               tp->snd_cwnd = tp->snd_ssthresh = tcp_westwood_bw_rttmin(sk);
                break;
 
        case CA_EVENT_FRTO:
-               tp->snd_ssthresh = westwood_bw_rttmin(sk);
+               tp->snd_ssthresh = tcp_westwood_bw_rttmin(sk);
                break;
 
        case CA_EVENT_SLOW_ACK:
@@ -235,7 +231,7 @@ static struct tcp_congestion_ops tcp_westwood = {
        .init           = tcp_westwood_init,
        .ssthresh       = tcp_reno_ssthresh,
        .cong_avoid     = tcp_reno_cong_avoid,
-       .min_cwnd       = tcp_westwood_cwnd_min,
+       .min_cwnd       = tcp_westwood_bw_rttmin,
        .cwnd_event     = tcp_westwood_event,
        .get_info       = tcp_westwood_info,
        .pkts_acked     = tcp_westwood_pkts_acked,