From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp.tuxdriver.com (charlotte.tuxdriver.com [70.61.120.58]) by dpdk.org (Postfix) with ESMTP id E5A8A959 for ; Mon, 14 Jul 2014 15:59:28 +0200 (CEST) Received: from uucp by smtp.tuxdriver.com with local-rmail (Exim 4.63) (envelope-from ) id 1X6gnZ-0002Da-TM; Mon, 14 Jul 2014 10:00:10 -0400 Received: from linville-x1.hq.tuxdriver.com (localhost.localdomain [127.0.0.1]) by linville-x1.hq.tuxdriver.com (8.14.8/8.14.6) with ESMTP id s6EDjB0T028090; Mon, 14 Jul 2014 09:45:11 -0400 Received: (from linville@localhost) by linville-x1.hq.tuxdriver.com (8.14.8/8.14.8/Submit) id s6EDjAwu028088; Mon, 14 Jul 2014 09:45:10 -0400 Date: Mon, 14 Jul 2014 09:45:10 -0400 From: "John W. Linville" To: "Zhou, Danny" Message-ID: <20140714134509.GB27848@tuxdriver.com> References: <1405024369-30058-1-git-send-email-linville@tuxdriver.com> <20140711174042.GE25478@tuxdriver.com> <20140711184648.GH25478@tuxdriver.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.23 (2014-03-12) Cc: "dev@dpdk.org" Subject: Re: [dpdk-dev] [PATCH] librte_pmd_packet: add PMD for AF_PACKET-based virtual devices X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 14 Jul 2014 13:59:29 -0000 On Sat, Jul 12, 2014 at 12:42:04AM +0000, Zhou, Danny wrote: > I just upgraded my kernel to 3.15.5 and hardcoded below captured from include/uapi/linux/if_packet.h to librte_pmd_packet.c to workaround it, now I can receive/transmit packet now. Commenting out PACKET_FANOUT_FLAG_ROLLOVER would cause no packet can be received. > > #define PACKET_QDISC_BYPASS 20 > #define PACKET_FANOUT_FLAG_ROLLOVER 0x1000 You shouldn't need PACKET_FANOUT_FLAG_ROLLOVER if all the queues are being used. Does the application you are running make use of all the queues? If not, you probably should use the qpairs option to limit the number of queues created by the eth_packet PMD. John > > > -----Original Message----- > > From: John W. Linville [mailto:linville@tuxdriver.com] > > Sent: Saturday, July 12, 2014 2:47 AM > > To: Zhou, Danny > > Cc: dev@dpdk.org > > Subject: Re: [dpdk-dev] [PATCH] librte_pmd_packet: add PMD for > > AF_PACKET-based virtual devices > > > > Not sure what the issue might be, PACKET_FANOUT_FLAG_ROLLOVER is defined > > in include/uapi/linux/if_packet.h in the v3.12 tree. > > > > On Fri, Jul 11, 2014 at 06:01:27PM +0000, Zhou, Danny wrote: > > > Tried on 3.12, both of them are undefined. Anyway, will comment them out and see > > what performance it could achieve. > > > > > > > -----Original Message----- > > > > From: John W. Linville [mailto:linville@tuxdriver.com] > > > > Sent: Saturday, July 12, 2014 1:41 AM > > > > To: Zhou, Danny > > > > Cc: dev@dpdk.org > > > > Subject: Re: [dpdk-dev] [PATCH] librte_pmd_packet: add PMD for > > > > AF_PACKET-based virtual devices > > > > > > > > On Fri, Jul 11, 2014 at 05:20:42PM +0000, Zhou, Danny wrote: > > > > > Looks like you used a pretty new kernel version with new socket > > > > > options that old > > > > kernel like my 3.12 does not support. When I tried this patch, it > > > > just cannot build, and compiler complains like below. Which Linux distribution > > does this patch work for? > > > > How to ensure it works for old kernels? > > > > > > > > > > /home/danny/dpdk.org/dpdk/lib/librte_pmd_packet/rte_eth_packet.c: > > > > > In function > > > > rte_pmd_init_internals: > > > > > /home/danny/dpdk.org/dpdk/lib/librte_pmd_packet/rte_eth_packet.c:5 > > > > > 24:1 > > > > > 7: error: PACKET_FANOUT_FLAG_ROLLOVER undeclared (first use in > > > > > this > > > > > function) > > > > > /home/danny/dpdk.org/dpdk/lib/librte_pmd_packet/rte_eth_packet.c:5 > > > > > 24:1 > > > > > 7: note: each undeclared identifier is reported only once for each > > > > > function it appears in > > > > > /home/danny/dpdk.org/dpdk/lib/librte_pmd_packet/rte_eth_packet.c:5 > > > > > 57:3 > > > > > 3: error: PACKET_QDISC_BYPASS undeclared (first use in this > > > > > function) > > > > > > > > Both of them are isolated, so for playing with it you could just comment those > > out. > > > > It looks like PACKET_FANOUT_FLAG_ROLLOVER should have been in 3.10, > > > > while PACKET_QDISC_BYPASS didn't show-up until 3.14... > > > > > > > > /home/linville/git/linux > > > > [linville-x1.hq.tuxdriver.com]:> git annotate > > > > include/uapi/linux/if_packet.h | grep PACKET_FANOUT_FLAG_ROLLOVER > > > > 77f65ebdca506 (Willem de Bruijn 2013-03-19 10:18:11 +0000 64)#define > > > > PACKET_FANOUT_FLAG_ROLLOVER 0x1000 > > > > > > > > /home/linville/git/linux > > > > [linville-x1.hq.tuxdriver.com]:> git show -s --format=short > > > > 77f65ebdca506 commit 77f65ebdca506870d99bfabe52bde222511022ec > > > > Author: Willem de Bruijn > > > > > > > > packet: packet fanout rollover during socket overload > > > > > > > > /home/linville/git/linux > > > > [linville-x1.hq.tuxdriver.com]:> git describe --contains > > > > 77f65ebdca506 > > > > v3.10-rc1~66^2~423 > > > > > > > > /home/linville/git/linux > > > > [linville-x1.hq.tuxdriver.com]:> git annotate > > > > include/uapi/linux/if_packet.h | grep PACKET_QDISC_BYPASS > > > > d346a3fae3ff1 (Daniel Borkmann 2013-12-06 11:36:17 +0100 56)#define > > > > PACKET_QDISC_BYPASS 20 > > > > > > > > /home/linville/git/linux > > > > [linville-x1.hq.tuxdriver.com]:> git show -s --format=short > > > > d346a3fae3ff1 commit > > > > d346a3fae3ff1d99f5d0c819bf86edf9094a26a1 > > > > Author: Daniel Borkmann > > > > > > > > packet: introduce PACKET_QDISC_BYPASS socket option > > > > > > > > /home/linville/git/linux > > > > [linville-x1.hq.tuxdriver.com]:> git describe --contains > > > > d346a3fae3ff1 > > > > v3.14-rc1~94^2~564 > > > > > > > > Is there an example of code in DPDK that requires specific kernel > > > > versions? What is the preferred method for coding such dependencies? > > > > > > > > John > > > > -- > > > > John W. Linville Someday the world will need a hero, and you > > > > linville@tuxdriver.com might be all we have. Be ready. > > > > > > > -- > > John W. Linville Someday the world will need a hero, and you > > linville@tuxdriver.com might be all we have. Be ready. > -- John W. Linville Someday the world will need a hero, and you linville@tuxdriver.com might be all we have. Be ready.