From mboxrd@z Thu Jan  1 00:00:00 1970
From: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Subject: Re: [net PATCH] i40e/i40evf: Limit TSO to 7 descriptors for payload
 instead of 8 per packet
Date: Wed, 30 Mar 2016 16:40:23 -0400
Message-ID: <20160330204023.GK27540@oracle.com>
References: <20160330064213.12927.46852.stgit@localhost.localdomain>
 <20160330170011.GB27540@oracle.com>
 <CAKgT0UdTgaeamyLetGPK0mqMNS-80BTc8t-WdSGJuuLDfz4JOg@mail.gmail.com>
 <20160330172001.GD27540@oracle.com>
 <CAKgT0Ufb-_X-DD2jYkVJQwLZ-JBfWD2Biuhi6YaxWBeNy9K2fQ@mail.gmail.com>
 <20160330124135.000054c6@unknown>
 <20160330200959.GI27540@oracle.com>
 <1459368904.6473.207.camel@edumazet-glaptop3.roam.corp.google.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>,
	Alexander Duyck <alexander.duyck@gmail.com>,
	Alexander Duyck <aduyck@mirantis.com>,
	Netdev <netdev@vger.kernel.org>,
	intel-wired-lan <intel-wired-lan@lists.osuosl.org>,
	Jeff Kirsher <jeffrey.t.kirsher@intel.com>
To: Eric Dumazet <eric.dumazet@gmail.com>
Return-path: <netdev-owner@vger.kernel.org>
Received: from aserp1040.oracle.com ([141.146.126.69]:38306 "EHLO
	aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1752085AbcC3Ukc (ORCPT
	<rfc822;netdev@vger.kernel.org>); Wed, 30 Mar 2016 16:40:32 -0400
Content-Disposition: inline
In-Reply-To: <1459368904.6473.207.camel@edumazet-glaptop3.roam.corp.google.com>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

On (03/30/16 13:15), Eric Dumazet wrote:
> You might try netperf -t TCP_SENDFILE -- -m 150
> 
> to let netperf use sendfile() on small frags.

that still did not reproduce it but let me try beating on
that approach with more permutations.

BTW, another data-point that may help debug this: even with i40e,
if you use the "-o" option to the rds-stress invocation, there are
no problems: the "-o" option  enforces uni-directional data
transfer, so one side is pure-Tx, other side is pure-Rx. 

It is only when both sides simultaneously do both Tx and Rx
on the tcp socket that you see the issue. I dont know if
that provides any clues.

--Sowmini