From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jesper Dangaard Brouer Subject: Re: [bpf-next V1 PATCH 0/8] bpf/xdp: add flags argument to ndo_xdp_xmit and flag flush operation Date: Thu, 31 May 2018 09:39:16 +0200 Message-ID: <20180531093916.78404201@redhat.com> References: <152770312703.20510.5854417568847239931.stgit@firesoul> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: Networking , Daniel Borkmann , Alexei Starovoitov , John Fastabend , makita.toshiaki@lab.ntt.co.jp, brouer@redhat.com To: Song Liu Return-path: Received: from mx3-rdu2.redhat.com ([66.187.233.73]:38472 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1753740AbeEaHjV (ORCPT ); Thu, 31 May 2018 03:39:21 -0400 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On Wed, 30 May 2018 15:18:02 -0700 Song Liu wrote: > Overall, this set looks good to me. The only suggestion I have is to add more > documentation on the expected behavior of XDP_XMIT_FLUSH in netdevice.h > (as part of 01/08). I do see your point, as the behavior of XDP_XMIT_FLUSH is actually more a "doorbell" functionality. I still choose to call it "flush", because it is replacing a function called ndo_xdp_flush, and providing the exact same code-function as ndo_xdp_flush. (IMHO it should have been called ndo_xdo_doorbell). Any opinions about renaming XDP_XMIT_FLUSH to XDP_XMIT_DOORBELL? If you look at virtio_net and tun usage, the effect is a wakeup (virtqueue_kick and sk_data_ready). Still I like the name "doorbell" better than "wakeup", as it also maps to NIC usage which often call this "doorbell" or "tail" pointer update. -- Best regards, Jesper Dangaard Brouer MSc.CS, Principal Kernel Engineer at Red Hat LinkedIn: http://www.linkedin.com/in/brouer