From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stefan Wahren Subject: Re: Packet loss when txqueuelen is zero Date: Mon, 30 Jun 2014 19:08:16 +0200 Message-ID: <53B19980.7050606@lategoodbye.de> References: <53AEA00F.6060500@lategoodbye.de> <1403966418.15139.27.camel@edumazet-glaptop2.roam.corp.google.com> <53B019E5.4060905@lategoodbye.de> <1404107363.15139.29.camel@edumazet-glaptop2.roam.corp.google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: netdev@vger.kernel.org, arnd@arndb.de To: Eric Dumazet Return-path: Received: from www.augenpunkt.de ([213.239.207.9]:49230 "EHLO www.augenpunkt.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752488AbaF3Qvx (ORCPT ); Mon, 30 Jun 2014 12:51:53 -0400 In-Reply-To: <1404107363.15139.29.camel@edumazet-glaptop2.roam.corp.google.com> Sender: netdev-owner@vger.kernel.org List-ID: Hi Eric, Am 30.06.2014 07:49, schrieb Eric Dumazet: > On Sun, 2014-06-29 at 15:51 +0200, Stefan Wahren wrote: > >> But fq_codel requires BQL in the QCA7000 network driver. Right? > > Not at all. 10 skb in TX ring is very small, BQL wont really help in > this case. > > BQL is good for NIC with TSO support, mostly. > Dave T=C3=A4ht made the suggestion to reduce the skbs to 2. But i'm not= sure=20 if it's a good solution, because the SPI overhead becomes bigger. I have another idea. The QCA has an internal tx buffer something above=20 3000 bytes. So it would be helpful if the tx ring stores the estimated=20 byte size of all skbs. In the case this sum is so big that skb with a=20 MTU (1500 byte) won't fit into the QCA buffer, the tx queue can be stop= ped. I commited my implementation [1] in the morning. [1] -=20 https://github.com/I2SE/qca7000/commit/4539475a4ce1d0f5ab6ed111df9d2a9a= b5f9bcc3