DPDK usage discussions
 help / color / mirror / Atom feed
From: Andrew Wang <andrw.wng@gmail.com>
To: users@dpdk.org
Subject: [dpdk-users] Route packets to DPDK interface or setup Bifurcated Flow
Date: Tue, 13 Aug 2019 11:57:39 -0400	[thread overview]
Message-ID: <CANEosPDTtD885N2R2GyctarQUUerV3Bzu9cKo+dmZkF64hFkbg@mail.gmail.com> (raw)

Hi


We have the following setup: a linux box running Ubuntu 18.04 with 2
interfaces, one controlled by the kernel (eno49), from which we receive
incoming packets from the switch, and another interface (eno50), which is
taken over by our DPDK app, and would ideally only process packets coming
from eno49 for a specific IPv6 subnet and forward them to the backend
server. See the schematic below:



+-------------+

|   Switch    |

+-------------+

    |

    |

+---+---------+

|    eno49    |  eno49 - interface controlled by kernel

|             |

|    eno50    |  eno50 - interface taken over by DPDK

+---+---------+

    |

    |

+-------------+

|    Backend  |

+-------------+



However, when the DPDK app takes over the interface, we can no longer see
the interface via commands like "ip link show". So it seems we can no
longer set commands like "ip -6 route add <IPv6 subnet> dev eno50`. Our
DPDK app currently does not implement Neighbor Discovery Protocol so we
cannot set route via an IPv6 address. Is there a way of forwarding incoming
packets from eno49 to our eno50 interface?


We tried setting up flow bifurcation, but got the following error:


# dmesg |grep Virtualization

[    1.537848] DMAR: Intel(R) Virtualization Technology for Directed I/O


# cat /proc/cmdline

BOOT_IMAGE=/boot/vmlinuz-4.15.0-55-generic
root=UUID=023bc8a4-513f-4ddb-977e-9e49ea2f8477 ro maybe-ubiquity
default_hugepagesz=1G hugepagesz=1G hugepages=4 intel_iommu=on iommu=pt


# cat /sys/bus/pci/devices/0000:04:00.1/sriov_numvfs

2


# ip link show enp4s16f1

15: enp4s16f1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode
DEFAULT group default qlen 1000

    link/ether 2a:82:46:01:0d:ce brd ff:ff:ff:ff:ff:ff


# ip link show eno50

7: eno50: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode
DEFAULT group default qlen 1000

    link/ether 5c:b9:01:98:7f:d5 brd ff:ff:ff:ff:ff:ff

    vf 0 MAC 2a:82:46:01:0d:ce, spoof checking on, link-state auto, trust
off, query_rss off

    vf 1 MAC 72:bc:92:d8:81:75, spoof checking on, link-state auto, trust
off, query_rss off


# ethtool -i enp4s16f1

driver: ixgbevf

version: 4.6.1

firmware-version:

expansion-rom-version:

bus-info: 0000:04:10.1

supports-statistics: yes

supports-test: yes

supports-eeprom-access: no

supports-register-dump: yes

supports-priv-flags: yes


# ethtool -K enp4s16f1 ntuple on

Cannot change ntuple-filters

Could not change any device features


Any suggestions on how we can proceed? Either help with directing flow or
getting flow bifurcation to work would be highly appreciated.


Thanks in advance

Andrew

                 reply	other threads:[~2019-08-13 15:58 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CANEosPDTtD885N2R2GyctarQUUerV3Bzu9cKo+dmZkF64hFkbg@mail.gmail.com \
    --to=andrw.wng@gmail.com \
    --cc=users@dpdk.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).