Patchwork dccp test-tree [RFC] [Patch 2/2] dccp: CCID2 check ack ratio when reducing cwnd

login
register
mail settings
Submitter Samuel Jero
Date March 15, 2011, 4:53 a.m.
Message ID <1300164797.20638.65.camel@jero-laptop>
Download mbox | patch
Permalink /patch/86853/
State Awaiting Upstream
Delegated to: David Miller
Headers show

Comments

Samuel Jero - March 15, 2011, 4:53 a.m.
This patch causes CCID2 to check the ack ratio after reducing the
congestion window. If the ack ratio is greater than the congestion
window, it is reduced. This prevents timeouts caused by an ack ratio
larger than the congestion window. 

In this situation, we choose to set the ack ratio to half the congestion
window (or one if that's zero) so that if we loose one ack we don't
trigger a timeout.

--- 
Signed-off-by: Samuel Jero
Gerrit Renker - March 18, 2011, 11:33 a.m.
| This patch causes CCID2 to check the ack ratio after reducing the
| congestion window.
Thanks a lot, it has been integrated with the other changes:
  http://eden-feed.erg.abdn.ac.uk/cgi-bin/gitweb.cgi?p=dccp_exp.git;a=commitdiff;h=f7bcfc7ddf5c1f61c861c3d2b475df7874833ed3
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/net/dccp/ccids/ccid2.c b/net/dccp/ccids/ccid2.c
index 1475ba3..b7f7dbd 100644
--- a/net/dccp/ccids/ccid2.c
+++ b/net/dccp/ccids/ccid2.c
@@ -118,4 +118,22 @@  static void ccid2_change_l_seq_window(struct sock *sk, u64 val)
 		dccp_feat_signal_nn_change(sk, DCCPF_SEQUENCE_WINDOW, val);
 }
+
+static void ccid2_check_l_ack_ratio(struct sock *sk)
+{
+	struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk);
+
+	/*
+	* After a loss, idle period, application limited period, or RTO we
+	* need tocheck that the ack ratio is still less than the congestion
+	* window. Otherwise, we will send an entire congestion window of
+	* packets and got no response because we haven't sent ack ratio
+	* packets yet.
+	* If the ack ratio does need to be reduced, we reduce it to half of
+	* the congestion window (or 1 if that's zero) instead of to the
+	* congestion window. This prevents problems if one ack is lost.
+	*/
+	if (dccp_feat_get_nn_next_val(sk, DCCPF_ACK_RATIO) > hc->tx_cwnd)
+		ccid2_change_l_ack_ratio(sk, hc->tx_cwnd/2 ? : 1U);
+}
 
 static void ccid2_hc_tx_rto_expire(unsigned long data)
@@ -203,6 +214,8 @@  static void ccid2_cwnd_application_limited(struct sock *sk, const u32 now)
 	}
 	hc->tx_cwnd_used  = 0;
 	hc->tx_cwnd_stamp = now;
+
+	ccid2_check_l_ack_ratio(sk);
 }
 
 /* This borrows the code of tcp_cwnd_restart() */
@@ -221,6 +234,8 @@  static void ccid2_cwnd_restart(struct sock *sk, const u32 now)
 
 	hc->tx_cwnd_stamp = now;
 	hc->tx_cwnd_used  = 0;
+
+	ccid2_check_l_ack_ratio(sk);
 }
 
 static void ccid2_hc_tx_packet_sent(struct sock *sk, unsigned int len)
@@ -490,9 +505,7 @@  static void ccid2_congestion_event(struct sock *sk, struct ccid2_seq *seqp)
 	hc->tx_cwnd      = hc->tx_cwnd / 2 ? : 1U;
 	hc->tx_ssthresh  = max(hc->tx_cwnd, 2U);
 
-	/* Avoid spurious timeouts resulting from Ack Ratio > cwnd */
-	if (dccp_sk(sk)->dccps_l_ack_ratio > hc->tx_cwnd)
-		ccid2_change_l_ack_ratio(sk, hc->tx_cwnd);
+	ccid2_check_l_ack_ratio(sk);
 }
 
 static int ccid2_hc_tx_parse_options(struct sock *sk, u8 packet_type,