[<prev] [next>] [day] [month] [year] [list]
Message-Id: <1311600996-6712-6-git-send-email-gerrit@erg.abdn.ac.uk>
Date: Mon, 25 Jul 2011 07:36:34 -0600
From: Gerrit Renker <gerrit@....abdn.ac.uk>
To: davem@...emloft.net
Cc: dccp@...r.kernel.org, netdev@...r.kernel.org,
Samuel Jero <sj323707@...o.edu>
Subject: [PATCH 5/7] dccp ccid-2: prevent cwnd > Sequence Window
From: Samuel Jero <sj323707@...o.edu>
Add a check to prevent CCID-2 from increasing the cwnd greater than the
Sequence Window.
When the congestion window becomes bigger than the Sequence Window, CCID-2
will attempt to keep more data in the network than the DCCP Sequence Window
code considers possible. This results in the Sequence Window code issuing
a Sync, thereby inducing needless overhead. Further, if this occurs at the
sender, CCID-2 will never detect the problem because the Acks it receives
will indicate no losses. I have seen this cause a drop of 1/3rd in throughput
for a connection.
Also add code to adjust the Sequence Window to be about 5 times the number of
packets in the network (RFC 4340, 7.5.2) and to adjust the Ack Ratio so that
the remote Sequence Window will hold about 5 times the number of packets in
the network. This allows the congestion window to increase correctly without
being limited by the Sequence Window.
Signed-off-by: Samuel Jero <sj323707@...o.edu>
Acked-by: Gerrit Renker <gerrit@....abdn.ac.uk>
---
net/dccp/ccids/ccid2.c | 50 +++++++++++++++++++++++++++++++++--------------
net/dccp/ccids/ccid2.h | 6 +++++
2 files changed, 41 insertions(+), 15 deletions(-)
--- a/net/dccp/ccids/ccid2.h
+++ b/net/dccp/ccids/ccid2.h
@@ -43,6 +43,12 @@ struct ccid2_seq {
#define CCID2_SEQBUF_LEN 1024
#define CCID2_SEQBUF_MAX 128
+/*
+ * Multiple of congestion window to keep the sequence window at
+ * (RFC 4340 7.5.2)
+ */
+#define CCID2_WIN_CHANGE_FACTOR 5
+
/**
* struct ccid2_hc_tx_sock - CCID2 TX half connection
* @tx_{cwnd,ssthresh,pipe}: as per RFC 4341, section 5
--- a/net/dccp/ccids/ccid2.c
+++ b/net/dccp/ccids/ccid2.c
@@ -85,7 +85,6 @@ static int ccid2_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb)
static void ccid2_change_l_ack_ratio(struct sock *sk, u32 val)
{
- struct dccp_sock *dp = dccp_sk(sk);
u32 max_ratio = DIV_ROUND_UP(ccid2_hc_tx_sk(sk)->tx_cwnd, 2);
/*
@@ -98,14 +97,15 @@ static void ccid2_change_l_ack_ratio(struct sock *sk, u32 val)
DCCP_WARN("Limiting Ack Ratio (%u) to %u\n", val, max_ratio);
val = max_ratio;
}
- if (val > DCCPF_ACK_RATIO_MAX)
- val = DCCPF_ACK_RATIO_MAX;
-
- if (val == dp->dccps_l_ack_ratio)
- return;
+ dccp_feat_signal_nn_change(sk, DCCPF_ACK_RATIO,
+ min_t(u32, val, DCCPF_ACK_RATIO_MAX));
+}
- ccid2_pr_debug("changing local ack ratio to %u\n", val);
- dp->dccps_l_ack_ratio = val;
+static void ccid2_change_l_seq_window(struct sock *sk, u64 val)
+{
+ dccp_feat_signal_nn_change(sk, DCCPF_SEQUENCE_WINDOW,
+ clamp_val(val, DCCPF_SEQ_WMIN,
+ DCCPF_SEQ_WMAX));
}
static void ccid2_hc_tx_rto_expire(unsigned long data)
@@ -405,17 +405,37 @@ static void ccid2_new_ack(struct sock *sk, struct ccid2_seq *seqp,
unsigned int *maxincr)
{
struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk);
-
- if (hc->tx_cwnd < hc->tx_ssthresh) {
- if (*maxincr > 0 && ++hc->tx_packets_acked == 2) {
+ struct dccp_sock *dp = dccp_sk(sk);
+ int r_seq_used = hc->tx_cwnd / dp->dccps_l_ack_ratio;
+
+ if (hc->tx_cwnd < dp->dccps_l_seq_win &&
+ r_seq_used < dp->dccps_r_seq_win) {
+ if (hc->tx_cwnd < hc->tx_ssthresh) {
+ if (*maxincr > 0 && ++hc->tx_packets_acked == 2) {
+ hc->tx_cwnd += 1;
+ *maxincr -= 1;
+ hc->tx_packets_acked = 0;
+ }
+ } else if (++hc->tx_packets_acked >= hc->tx_cwnd) {
hc->tx_cwnd += 1;
- *maxincr -= 1;
hc->tx_packets_acked = 0;
}
- } else if (++hc->tx_packets_acked >= hc->tx_cwnd) {
- hc->tx_cwnd += 1;
- hc->tx_packets_acked = 0;
}
+
+ /*
+ * Adjust the local sequence window and the ack ratio to allow about
+ * 5 times the number of packets in the network (RFC 4340 7.5.2)
+ */
+ if (r_seq_used * CCID2_WIN_CHANGE_FACTOR >= dp->dccps_r_seq_win)
+ ccid2_change_l_ack_ratio(sk, dp->dccps_l_ack_ratio * 2);
+ else if (r_seq_used * CCID2_WIN_CHANGE_FACTOR < dp->dccps_r_seq_win/2)
+ ccid2_change_l_ack_ratio(sk, dp->dccps_l_ack_ratio / 2 ? : 1U);
+
+ if (hc->tx_cwnd * CCID2_WIN_CHANGE_FACTOR >= dp->dccps_l_seq_win)
+ ccid2_change_l_seq_window(sk, dp->dccps_l_seq_win * 2);
+ else if (hc->tx_cwnd * CCID2_WIN_CHANGE_FACTOR < dp->dccps_l_seq_win/2)
+ ccid2_change_l_seq_window(sk, dp->dccps_l_seq_win / 2);
+
/*
* FIXME: RTT is sampled several times per acknowledgment (for each
* entry in the Ack Vector), instead of once per Ack (as in TCP SACK).
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists