From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-we0-x22e.google.com (mail-we0-x22e.google.com [IPv6:2a00:1450:400c:c03::22e]) by dpdk.org (Postfix) with ESMTP id A9A045320 for ; Fri, 14 Jun 2013 22:44:59 +0200 (CEST) Received: by mail-we0-f174.google.com with SMTP id q58so867721wes.19 for ; Fri, 14 Jun 2013 13:45:09 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type; bh=kwW6zLeJ1WrryzvCxJgMP8eh/oj6ZNXhQml4gYsIl/k=; b=r7hZqg593QMstHo3991otdwOt2nkvW38GqJO+YiJUGMuCFYvBHQxA6UfEHNGMeghZS DtAvp1Y6enhYGAYmN72pcBTtiy6LFiuW9BRrt+bpPSYt60C0d8bmaCTsX8hr+mHXrG0d 6GoCah2tm9KL0H7dwcgNlbxwck0K27pyrxZOTHYBdj9IZg/m6SHioMrxRfAwqvgS/tNu YYDdvFZXaziTJeo+Lv2sIXP8RTzasWrzOlfR7exXzemnlRVp3Uhxx48rVvz+BUbPGnEU VzJUySX+mUNcsXF4J7pTXr1EN2B7pVIe45rinCk7TvFa0vrv4pvi8YvmVSEe7trkd4IP n6fA== X-Received: by 10.180.189.68 with SMTP id gg4mr2305592wic.27.1371242709565; Fri, 14 Jun 2013 13:45:09 -0700 (PDT) MIME-Version: 1.0 Received: by 10.216.73.193 with HTTP; Fri, 14 Jun 2013 13:44:49 -0700 (PDT) In-Reply-To: References: From: Peter Chen Date: Fri, 14 Jun 2013 13:44:49 -0700 Message-ID: To: dev@dpdk.org Content-Type: multipart/alternative; boundary=001a11c353009ac75504df23527e Subject: Re: [dpdk-dev] testpmd errors 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: Fri, 14 Jun 2013 20:45:00 -0000 --001a11c353009ac75504df23527e Content-Type: text/plain; charset=ISO-8859-1 Hi, it seems the above error is due to the fact I don't have NUMA enabled even though half of my cores use socket 0 and half of my cores use socket 1. For some reason all the hugepage allocations are on socket 0 (it makes sense as I don't have NUMA enabled). Also I am unable to use: sudo ./testpmd -c ff -n 4 -r 2 --socket-mem=1024,1024 command to allocate memory to socket 1 as it is not part of dpdk usage which is weird (do I have to have NUMA enabled to use this feature?) I have two Intel quad-core Xeon 5506. I am not sure how to enable NUMA on them or if these Intel chips even support them. Does anyone know? Peter On Tue, Jun 11, 2013 at 4:23 PM, Peter Chen wrote: > Just a bit more of information, I am using ubuntu 12.04 server. > > > On Tue, Jun 11, 2013 at 3:56 PM, Peter Chen wrote: > >> Hi All, >> >> I am trying to install dpdk. Since it says to run test and testpmd to >> see if everything works, I ran those apps. >> >> So far test works fine for me. However when I run testpmd, I get the >> following: >> >> *probe driver: 8086:10c8 rte_ixgbe_pmd >> * >> * unbind kernel driver /sys/bus/pci/devices/0000:04:00.0/driver/unbind >> * >> * bind PCI device 0000:04:00.0 to igb_uio driver >> * >> * Device bound >> * >> * map PCI resource for device 0000:04:00.0 >> * >> * PCI memory mapped at 0x7f0e680e6000 >> * >> * rte_memzone_reserve_aligned(): No appropriate segment found >> * >> * pci_uio_map_resource(): cannot store uio mmap details >> * >> * Error - exiting with code: 1 >> * >> * No probed ethernet devices - check that CONFIG_RTE_LIBRTE_IGB_PMD >> =y and that CONFIG_RTE_LIBRTE_IXGBE_PMD=y in your configuration file * >> >> Below is what I see in dmesg: >> >> *igb_uio 0000:04:00.0: irq 79 for MSI/MSI-X >> * >> * uio device registered with irq 4f >> >> * >> * *My setup is a dell R510 with two quad-core Xeons (E5506), 32 GB >> memory. It has 5 NICs of which two are up (eth0: BCM5716 and eth4 Intel >> 82598EB 10 Gigabit). Both NICs are on the same subnet before I had dpdk >> attach to eth4 by running the following command: >> >> *sudo ./testpmd -c ff -n 4 -r 2 -m 2048 -b 0000:01:00.0* >> >> 0000:04:00.0 is the Intel and 0000:01:00.0 is the Broadcom. >> >> Originally I was getting the "No probed ethernet devices" because I >> had a non-supported NIC, however 82598 is a supported NIC and it seems that >> dpdk did bind to the Intel NIC. However, rte_eth_dev_count() somehow still >> returns 0. Any help would be great! Thanks! >> >> Peter >> > > --001a11c353009ac75504df23527e Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable
Hi, it seems the above error is due to the fact I don= 't have NUMA enabled even though half of my cores use socket 0 and half= of my cores use socket 1. For some reason all the hugepage allocations are= on socket 0 (it makes sense as I don't have NUMA enabled). Also I am u= nable to use:
=A0=A0
=A0=A0 sudo ./testpmd -c ff -n 4 -r 2 --socket-mem=3D1024,1024 c= ommand to allocate memory to socket 1 as it is not part of dpdk usage which= is weird (do I have to have NUMA enabled to use this feature?)

I have two Intel quad-core Xeon 5506. I am not sure how to enable NUMA on= them or if these Intel chips even support them. Does anyone know?

Peter


On Tue, Jun 11, 2013 at 4:23 PM, Peter Chen <peter.feifan.= chen@gmail.com> wrote:
Just a bit more of informat= ion, I am using ubuntu 12.04 server.


On Tue, Jun 1= 1, 2013 at 3:56 PM, Peter Chen <peter.feifan.chen@gmail.com&= gt; wrote:
Hi All,

=A0=A0=A0 I am trying to install dpdk. Si= nce it says to run test and testpmd to see if everything works, I ran those= apps.

=A0=A0=A0 So far test works fine for me. However when I run testp= md, I get the following:

=A0=A0=A0 probe driver: 8086:10c8 rte_ixgbe_pmd
<= b>=A0=A0=A0 unbind kernel driver /sys/bus/pci/devices/0000:04:00.0/driver/u= nbind
=A0=A0=A0 bind PCI device 0000:04:00.0 to igb_uio dri= ver
=A0=A0=A0 Device bound
=A0=A0=A0 map PCI resource for de= vice 0000:04:00.0
=A0=A0=A0 PCI memory mapped at 0x7f0e680e= 6000
=A0=A0=A0 rte_memzone_reserve_align= ed(): No appropriate segment found
=A0=A0=A0 pci_uio_map_resource(): cannot store uio mmap d= etails
=A0=A0=A0 Error - exiting with code: 1
<= /div>
=A0=A0=A0 No probed ethernet devices - check that CONFIG_RTE_L= IBRTE_IGB_PMD =3Dy and that CONFIG_RTE_LIBRTE_IXGBE_PMD=3Dy in your configu= ration file

=A0=A0=A0=A0 Below is what I see in dmesg:

=A0=A0=A0=A0 igb_uio 0000:04:00.0: irq 79 for MSI/MSI-X
=A0=A0=A0=A0 uio device registered with irq 4f

=A0=A0=A0=A0 My setup is a dell R510 with two quad-core Xeons (E550= 6), 32 GB memory. It has 5 NICs of which two are up (eth0: BCM5716 and eth4= Intel 82598EB 10 Gigabit). Both NICs are on the same subnet before I had d= pdk attach to eth4 by running the following command:

=A0=A0=A0=A0 sudo ./testpmd -c ff -n 4 -r 2 -m 2048 -b 00= 00:01:00.0
=A0=A0=A0=A0
=A0=A0=A0=A0 0000= :04:00.0 is the Intel and 0000:01:00.0 is the Broadcom.

=A0=A0=A0= =A0 Originally I was getting the "No probed ethernet devices" bec= ause I had a non-supported NIC, however 82598 is a supported NIC and it see= ms that dpdk did bind to the Intel NIC. However, rte_eth_dev_count() someho= w still returns 0. Any help would be great! Thanks!
=A0
Peter


--001a11c353009ac75504df23527e--