From mboxrd@z Thu Jan 1 00:00:00 1970 From: Vlad Yasevich Subject: Re: [PATCH net 1/2] macvlan: forbid L2 fowarding offload for macvtap Date: Fri, 10 Jan 2014 11:40:51 -0500 Message-ID: <52D02293.9060304@redhat.com> References: <52CA5CDA.1020503@gmail.com> <52CA612D.2000902@redhat.com> <20140106122628.GA24280@hmsreliant.think-freely.org> <52CB7009.2030903@redhat.com> <52CB8D54.9090506@gmail.com> <52CB9D1A.1050101@redhat.com> <52CBAC2A.90005@intel.com> <52CBC22D.3050002@redhat.com> <20140108125528.GC14741@redhat.com> <52CDA186.1080601@gmail.com> <20140109071721.GD19559@redhat.com> <52CE63EB.5000109@redhat.com> <20140109133908.354a4d73@nehalam.linuxnetplumber.net> <52CF9C0C.2020808@redhat.com> Reply-To: vyasevic@redhat.com Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: "Michael S. Tsirkin" , John Fastabend , John Fastabend , Neil Horman , davem@davemloft.net, netdev@vger.kernel.org, linux-kernel@vger.kernel.org To: Jason Wang , Stephen Hemminger Return-path: Received: from mx1.redhat.com ([209.132.183.28]:46876 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751880AbaAJQlE (ORCPT ); Fri, 10 Jan 2014 11:41:04 -0500 In-Reply-To: <52CF9C0C.2020808@redhat.com> Sender: netdev-owner@vger.kernel.org List-ID: On 01/10/2014 02:06 AM, Jason Wang wrote: > On 01/10/2014 05:39 AM, Stephen Hemminger wrote: >> On Thu, 09 Jan 2014 16:55:07 +0800 >> Jason Wang wrote: >> >>> What if use do want a qdisc and want to change the its queue length for >>> tun/macvlan? And the the name tx_queue_length is misleading. For tun it >>> may make sense since it was used in transmission path. For macvtap it >>> was not. So maybe what we need is just a new ioctl for both tun/macvtap >>> and a new feature flag. If user create the device with new feature flag, >>> the socket receive queue length could be changed by ioctl instead of >>> dev->tx_queue_length. If not, the old behaviour could be kept. >> The overloading of tx_queue_len in macvtap was the original design mistake. >> Can't this just be undone and expose rx_queue_len as sysfs attribute? > > That works. But we current allow user to change the socket sndbuf > through TUNSNDBUF. Maybe we need a similar one for receive. > That would make sense. Since the user interacts with tun fd almost as a socket and there is actually a socket hiding in the kernel, it almost begs for actual SO_SNDBUF/SO_RCVBUF support :) -vlad