From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1031997Ab2I1UW4 (ORCPT ); Fri, 28 Sep 2012 16:22:56 -0400 Received: from mail-pa0-f46.google.com ([209.85.220.46]:61998 "EHLO mail-pa0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1031981Ab2I1UWs (ORCPT ); Fri, 28 Sep 2012 16:22:48 -0400 From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org, stable@vger.kernel.org Cc: Greg Kroah-Hartman , alan@lxorguk.ukuu.org.uk, Yuchung Cheng , Nandita Dukkipati , Neal Cardwell , "David S. Miller" Subject: [ 025/218] tcp: fix cwnd reduction for non-sack recovery Date: Fri, 28 Sep 2012 13:14:02 -0700 Message-Id: <20120928201504.258719298@linuxfoundation.org> X-Mailer: git-send-email 1.7.12.1.428.g652398a In-Reply-To: <20120928201501.208384923@linuxfoundation.org> References: <20120928201501.208384923@linuxfoundation.org> User-Agent: quilt/0.60-2.1.2 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org 3.4-stable review patch. If anyone has any objections, please let me know. ------------------ From: Yuchung Cheng [ Upstream commit 7c4a56fec379ac0d7754e0d4da6a7361f1a4fe64 ] The cwnd reduction in fast recovery is based on the number of packets newly delivered per ACK. For non-sack connections every DUPACK signifies a packet has been delivered, but the sender mistakenly skips counting them for cwnd reduction. The fix is to compute newly_acked_sacked after DUPACKs are accounted in sacked_out for non-sack connections. Signed-off-by: Yuchung Cheng Acked-by: Nandita Dukkipati Acked-by: Neal Cardwell Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -3037,13 +3037,14 @@ static void tcp_update_cwnd_in_recovery( * tcp_xmit_retransmit_queue(). */ static void tcp_fastretrans_alert(struct sock *sk, int pkts_acked, - int newly_acked_sacked, bool is_dupack, + int prior_sacked, bool is_dupack, int flag) { struct inet_connection_sock *icsk = inet_csk(sk); struct tcp_sock *tp = tcp_sk(sk); int do_lost = is_dupack || ((flag & FLAG_DATA_SACKED) && (tcp_fackets_out(tp) > tp->reordering)); + int newly_acked_sacked = 0; int fast_rexmit = 0, mib_idx; if (WARN_ON(!tp->packets_out && tp->sacked_out)) @@ -3103,6 +3104,7 @@ static void tcp_fastretrans_alert(struct tcp_add_reno_sack(sk); } else do_lost = tcp_try_undo_partial(sk, pkts_acked); + newly_acked_sacked = pkts_acked + tp->sacked_out - prior_sacked; break; case TCP_CA_Loss: if (flag & FLAG_DATA_ACKED) @@ -3124,6 +3126,7 @@ static void tcp_fastretrans_alert(struct if (is_dupack) tcp_add_reno_sack(sk); } + newly_acked_sacked = pkts_acked + tp->sacked_out - prior_sacked; if (icsk->icsk_ca_state <= TCP_CA_Disorder) tcp_try_undo_dsack(sk); @@ -3695,7 +3698,6 @@ static int tcp_ack(struct sock *sk, cons int prior_packets; int prior_sacked = tp->sacked_out; int pkts_acked = 0; - int newly_acked_sacked = 0; int frto_cwnd = 0; /* If the ack is older than previous acks @@ -3768,8 +3770,6 @@ static int tcp_ack(struct sock *sk, cons flag |= tcp_clean_rtx_queue(sk, prior_fackets, prior_snd_una); pkts_acked = prior_packets - tp->packets_out; - newly_acked_sacked = (prior_packets - prior_sacked) - - (tp->packets_out - tp->sacked_out); if (tp->frto_counter) frto_cwnd = tcp_process_frto(sk, flag); @@ -3783,7 +3783,7 @@ static int tcp_ack(struct sock *sk, cons tcp_may_raise_cwnd(sk, flag)) tcp_cong_avoid(sk, ack, prior_in_flight); is_dupack = !(flag & (FLAG_SND_UNA_ADVANCED | FLAG_NOT_DUP)); - tcp_fastretrans_alert(sk, pkts_acked, newly_acked_sacked, + tcp_fastretrans_alert(sk, pkts_acked, prior_sacked, is_dupack, flag); } else { if ((flag & FLAG_DATA_ACKED) && !frto_cwnd) @@ -3798,7 +3798,7 @@ static int tcp_ack(struct sock *sk, cons no_queue: /* If data was DSACKed, see if we can undo a cwnd reduction. */ if (flag & FLAG_DSACKING_ACK) - tcp_fastretrans_alert(sk, pkts_acked, newly_acked_sacked, + tcp_fastretrans_alert(sk, pkts_acked, prior_sacked, is_dupack, flag); /* If this ack opens up a zero window, clear backoff. It was * being used to time the probes, and is probably far higher than @@ -3818,8 +3818,7 @@ old_ack: */ if (TCP_SKB_CB(skb)->sacked) { flag |= tcp_sacktag_write_queue(sk, skb, prior_snd_una); - newly_acked_sacked = tp->sacked_out - prior_sacked; - tcp_fastretrans_alert(sk, pkts_acked, newly_acked_sacked, + tcp_fastretrans_alert(sk, pkts_acked, prior_sacked, is_dupack, flag); }