From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Re: [PATCH] tbf scheduler: TSO support Date: Fri, 11 May 2007 03:04:21 +0200 Message-ID: <4643C115.3020802@trash.net> References: <20070510.210556.05145104.taka@valinux.co.jp> <46431687.7060600@trash.net> <20070510.141344.127196214.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Cc: taka@valinux.co.jp, linux-net@vger.kernel.org, netdev@vger.kernel.org To: David Miller Return-path: In-Reply-To: <20070510.141344.127196214.davem@davemloft.net> Sender: linux-net-owner@vger.kernel.org List-Id: netdev.vger.kernel.org David Miller wrote: > From: Patrick McHardy > Date: Thu, 10 May 2007 14:56:39 +0200 > >>I don't see why this is needed, the correct way to use TBF with TSO >>is to specify a larger MTU value, in which case it won't drop TSO >>packets. > > > Why should a user have to know anything in the world about TSO in > order to configure TBF properly? I don't think they should have > to at all. The user shouldn't necessarily, but userspace should. The way I see it the MTU is a fundamental parameter for TBF (the peakrate bucket size) and just because userspace picks a bad default (2000) this is no reason to change the implementation to something that is not really TBF anymore and even affects non-TSO packets _and_ TSO packets even when the MTU is chosen large enough (granted, the first point is an implementation detail). The much better solution would be to let userspace pick an appropriate default value and still let the user override it.