From mboxrd@z Thu Jan 1 00:00:00 1970 From: Luka Perkov Subject: Re: [PATCH v2] net: mvneta: add basic XDP_DROP support Date: Mon, 24 Dec 2018 11:16:27 +0100 Message-ID: <20181224101627.GA8770@archlinux> References: <20181221132223.14012-1-domagoj.pintaric@sartura.hr> <20181222115254.2001201b@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Domagoj Pintaric , netdev@vger.kernel.org, Thomas Petazzoni , "David S . Miller" , Ilias Apalodimas , Marcin Wojtas , damir.samardzic@sartura.hr To: Jesper Dangaard Brouer Return-path: Received: from mail-wr1-f67.google.com ([209.85.221.67]:42687 "EHLO mail-wr1-f67.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1725268AbeLXKQc (ORCPT ); Mon, 24 Dec 2018 05:16:32 -0500 Received: by mail-wr1-f67.google.com with SMTP id q18so11127443wrx.9 for ; Mon, 24 Dec 2018 02:16:31 -0800 (PST) Content-Disposition: inline In-Reply-To: <20181222115254.2001201b@redhat.com> Sender: netdev-owner@vger.kernel.org List-ID: Hi Jesper, On Sat, Dec 22, 2018 at 11:52:54AM +0100, Jesper Dangaard Brouer wrote: > On Fri, 21 Dec 2018 14:22:23 +0100 > Domagoj Pintaric wrote: > > > Add initial mvneta XDP support for hardware buffer management enabled > > devices only. > > Hi Domagoj, > > I would really appreciate if we could coordinate our work on the mvneta > driver. Ilias (Cc'ed) and I are also working on adding XDP support for > this driver, although this is the software-buffer side of the driver we > have functioning now. > > You can directly follow our progress here: [1][2] > [1] https://github.com/xdp-project/xdp-project/blob/master/areas/arm64/board_espressobin04_bench_xdp.org > [2] https://github.com/apalos/bpf-next/commits/mvneta_04_page_pool_recycle_xdp > > You XDP-setup function is actually more correct that ours[3], as you > handle BPF per queue (which we were planning to fix before upstreaming). > > That said, adding XDP_DROP is easy, but I want to see more of the XDP > features/actions added, as those require a lot more work. I always > worry that a driver will stop at just XDP_DROP, so what are your plans > for adding the harder features? Thank you for point out the you already made so much progress as we were not aware of it. Generally speaking, in Sartura our goal was to leverage all the development done in eBPF and XDP mainly for datacenter use-cases and use the technology in embedded systems where our primary focus is. That said, in OpenWrt summit two months ago Damir gave a talk about these technologies [1]. You are correct that we have sent a patch only for the easy part. But that is only what we have implemented so far :) As this was low-priority project on our side we didn't have concrete timelines for other features. Looking at what you have done and what we have I'd be happy to see if we can somehow divide the efforts in most optimal way. Except the mvneta driver (and we have a lot of boards utilizing that driver) we would also like to collaborate in mainline XDP support for mvpp2 which I have seen you have in plans as well. Thanks, Luka [1] https://openwrtsummit.files.wordpress.com/2018/11/20181029-state-of-fast-path-networking-in-linux.pdf > Even-thought XDP_DROP looks easy in this patch, then you are actually > doing some wrong, as XDP can also modify frames before doing XDP_PASS, > and (1) you have non-standard-head room (MVNETA_MH_SIZE + NET_SKB_PAD), > (2) and you don't handle if XDP changed the xdp.data header pointer, > and (3) you backing memory either comes from page-fragments or kmalloc > which also goes against the XDP memory requirements. > > [3] https://github.com/apalos/bpf-next/commit/4b567e74552d3cdf55 > -- > Best regards, > Jesper Dangaard Brouer > MSc.CS, Principal Kernel Engineer at Red Hat > LinkedIn: http://www.linkedin.com/in/brouer