From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga01.intel.com (mga01.intel.com [192.55.52.88]) by dpdk.org (Postfix) with ESMTP id 775111B247 for ; Mon, 30 Oct 2017 23:39:30 +0100 (CET) Received: from fmsmga005.fm.intel.com ([10.253.24.32]) by fmsmga101.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 30 Oct 2017 15:39:29 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.44,321,1505804400"; d="scan'208";a="169172783" Received: from fyigit-mobl1.ger.corp.intel.com (HELO [10.241.225.254]) ([10.241.225.254]) by fmsmga005.fm.intel.com with ESMTP; 30 Oct 2017 15:39:27 -0700 To: =?UTF-8?Q?Ga=c3=abtan_Rivet?= , "Xu, Qian Q" , Matej Vido Cc: "Lin, Xueqin" , "Zhang, Helin" , Thomas Monjalon , "Pei, Yulong" , "Peng, Yuan" , "dev@dpdk.org" , "Glynn, Michael J" , "Liu, Yu Y" References: <82F45D86ADE5454A95A89742C8D1410E3B7F050C@shsmsx102.ccr.corp.intel.com> <20171030081408.GG10890@bidouze.vm.6wind.com> From: Ferruh Yigit Message-ID: Date: Mon, 30 Oct 2017 15:39:27 -0700 User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101 Thunderbird/52.4.0 MIME-Version: 1.0 In-Reply-To: <20171030081408.GG10890@bidouze.vm.6wind.com> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 8bit Subject: Re: [dpdk-dev] vfio-pci blocking issue X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 30 Oct 2017 22:39:31 -0000 On 10/30/2017 1:14 AM, Gaëtan Rivet wrote: > Hi Xueqin, > > Thank you for the thorough test report. > Multiple filings for this bug, please refer to the thread > > http://dpdk.org/ml/archives/dev/2017-October/080746.html > > Which will focus discussion. A first fix and a new proposition are > available, please tell us if it works for you or if there are > any other problem with either of the solutions. Hi Gaetan, Another issue related moving pci, szedata2 driver build seems broken [1] because it uses a pci function which moved now. Can you please check it? Also the pci_get_sysfs_path API moved from eal seems still in the eal .map file. [1] rte_eth_szedata2.o: In function `szedata2_eth_pci_probe': /tmp/dpdk_maintain/self/dpdk/drivers/net/szedata2/rte_eth_szedata2.c:(.text+0x432): undefined reference to `pci_get_sysfs_path' Thanks, ferruh > > On Mon, Oct 30, 2017 at 07:25:20AM +0000, Xu, Qian Q wrote: >> We have found the blocking issue about vfio-pci in RC2, which blocked many test cases. It is a blocking issue for 17.11 release. Please help check. >> >> Add more failed case as below: it will also cause the VF port can't be probed. >> >> ? Components version >> dpdk 17.11 rc2 >> ? Defect component version (release version or repo/branch/commit ID) >> ? Environment components version (QEMU, OVS, etc) >> ? HW platform info >> fvl10g >> ? Test Steps and Test Result >> 1. Created VFs and bind to vfio-pci >> ./usertools/dpdk-devbind.py --bind=igb_uio 81:00.0 81:00.1 >> echo 1 > /sys/bus/pci/devices/0000:81:00.0/max_vfs >> echo 1 > /sys/bus/pci/devices/0000:81:00.1/max_vfs >> ./usertools/dpdk-devbind.py -s >> 0000:05:02.0 'XL710/X710 Virtual Function 154c' if= drv=i40evf unused=igb_uio >> 0000:05:06.0 'XL710/X710 Virtual Function 154c' if= drv=i40evf unused=igb_uio >> ./usertools/dpdk-devbind.py --bind=vfio-pci 81:02.0 81:06.0 >> ./usertools/dpdk-devbind.py -s >> 0000:81:00.0 'Ethernet Controller X710 for 10GbE SFP+ 1572' drv=igb_uio unused=vfio-pci >> 0000:81:00.1 'Ethernet Controller X710 for 10GbE SFP+ 1572' drv=igb_uio unused=vfio-pci >> 0000:81:02.0 'XL710/X710 Virtual Function 154c' drv=vfio-pci unused=igb_uio >> 0000:81:06.0 'XL710/X710 Virtual Function 154c' drv=vfio-pci unused=igb_uio >> 2. up PF with testpmd, >> ./x86_64-native-linuxapp-gcc/app/testpmd -l 23-27 -n 4 --socket-mem=1024,1024 --file-prefix=pf -w 81:00.0 -w 81:00.1 - -i --rxq=4 --txq=4 >> 3. start VFs with testpmd, >> ./x86_64-native-linuxapp-gcc/app/testpmd -l 28-32 -n 4 --socket-mem=1024,1024 --file-prefix=vf -w 81:02.0 -w 81:06.0 - -i --rxq=4 --txq=4 >> EAL: Detected 88 lcore(s) >> EAL: Some devices want iova as va but pa will be used because.. EAL: vfio-noiommu mode configured >> EAL: Probing VFIO support... >> EAL: VFIO support initialized >> EAL: PCI device 0000:81:02.0 on NUMA socket 1 >> EAL: probe driver: 8086:154c net_i40e_vf >> EAL: Requested device 0000:81:02.0 cannot be used >> EAL: PCI device 0000:81:06.0 on NUMA socket 1 >> EAL: probe driver: 8086:154c net_i40e_vf >> EAL: Requested device 0000:81:06.0 cannot be used >> EAL: No probed ethernet devices >> Interactive-mode selected >> USER1: create a new mbuf pool : n=179456, size=2176, socket=0 >> USER1: create a new mbuf pool : n=179456, size=2176, socket=1 >> Done >> testpmd> >> it can be seen "No probed ethernet devices" >> >> >> From: Lin, Xueqin >> Sent: Monday, October 30, 2017 1:59 PM >> To: gaetan.rivet@6wind.com; Zhang, Helin >> Cc: Xu, Qian Q ; Pei, Yulong ; Peng, Yuan >> Subject: vfio-pci >> >> Hi Rivet, >> >> Some cases need to bind port to vfio-pci, for example, test inline inpsec. >> >> 1. modprobe vfio-pci >> >> 2. bind port to vfio-pci >> >> ./usertools/dpdk-devbind.py -b vfio-pci 81:00.0 81:00.1 >> >> 3. Run ipsec >> sudo ./build/ipsec-secgw -l 20,21 -w 81:00.0 -w 81:00.1 --vdev "crypto_null" --log-level 8 --socket-mem 1024,1 -- -p 0xf -P -u 0x2 -j 9000\ >> --config="(0,0,20),(1,0,21)" -f ./enc.cfg >> Find error as below, vfio-pci can't be used. >> >> EAL: probe driver: 8086:10fb net_ixgbe >> EAL: Requested device 0000:81:00.0 cannot be used >> EAL: PCI device 0000:81:00.1 on NUMA socket 1 >> EAL: probe driver: 8086:10fb net_ixgbe >> EAL: Requested device 0000:81:00.1 cannot be used >> EAL: Search driver crypto_null to probe device crypto_null >> CRYPTODEV: [crypto_null] - Creating cryptodev crypto_null >> >> Find your patch results in the error, block lots of vfio-pci cases, which is urgent. >> DPDK-4026 should meet same issue. >> If any wrong to use vfio-pci, could you give a guide? >> >> commit 279b581c897d4bf01f2b35c7327e767400c3ee80 >> Author: Gaetan Rivet > >> Date: Thu Oct 26 12:05:52 2017 +0200 >> >> vfio: expose functions >> >> The following symbols are used by vfio implementations within the PCI bus. >> They need to be publicly available for the PCI bus to be outside the >> EAL. >> >> >> Best regards, >> Xueqin >> >