From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rick Jones Subject: Re: [PATCH] make _minimum_ TCP retransmission timeout configurable take 2 Date: Fri, 31 Aug 2007 15:20:52 -0700 Message-ID: <46D89444.3080508@hp.com> References: <46D859D9.6030407@hp.com> <20070831.115706.97292718.davem@davemloft.net> <46D88146.9090401@hp.com> <20070831.143855.88699905.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org To: David Miller Return-path: Received: from palrel12.hp.com ([156.153.255.237]:38436 "EHLO palrel12.hp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1759425AbXHaWVU (ORCPT ); Fri, 31 Aug 2007 18:21:20 -0400 In-Reply-To: <20070831.143855.88699905.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org David Miller wrote: > From: Rick Jones > Date: Fri, 31 Aug 2007 13:59:50 -0700 > > >>ip is at tcp_rto_min+0x20/0x40 > > > diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c > index 1ee7212..bbad2cd 100644 > --- a/net/ipv4/tcp_input.c > +++ b/net/ipv4/tcp_input.c > @@ -560,7 +560,7 @@ static u32 tcp_rto_min(struct sock *sk) > struct dst_entry *dst = __sk_dst_get(sk); > u32 rto_min = TCP_RTO_MIN; > > - if (dst_metric_locked(dst, RTAX_RTO_MIN)) > + if (dst && dst_metric_locked(dst, RTAX_RTO_MIN)) > rto_min = dst->metrics[RTAX_RTO_MIN-1]; > return rto_min; > } Applied and beating on it with a while loop doing a bunch of ip route del add change stuff while netperf TCP_CRR tests are running. Thusfar things seem OK wrt the system staying alive, but since I only saw the failure once I'm not sure how much that is really saying. I'm going to go ahead and take a look at input vs output units and differences between those with rto_min vs rtt. rick jones