From: Stephen Hemminger <stephen@networkplumber.org>
To: Nilesh <nileshunhale@cse.iitb.ac.in>
Cc: dev@dpdk.org, users@dpdk.org
Subject: Re: [dpdk-users] [dpdk-dev] Unable to redirect packets into kernel using dtap/dtun
Date: Fri, 2 Aug 2019 08:37:26 -0700 [thread overview]
Message-ID: <20190802083726.3dfde31f@hermes.lan> (raw)
In-Reply-To: <db6a6b8e-4faa-0ede-cfea-54e643e4eccd@cse.iitb.ac.in>
On Fri, 2 Aug 2019 12:45:55 +0530
Nilesh <nileshunhale@cse.iitb.ac.in> wrote:
> Hello,
>
> We are building an application where a majority of the packets on the
> fast path are handled by DPDK, but a small subset of packets on the slow
> path need to be redirected to a legacy userspace socket program that
> runs over the kernel TCP/IP stack. Therefore, we are trying to inject
> the slow path packets received by DPDK into the kernel via TUN/TAP device.
>
> We have done the following: we created a tun device, assigned it an IP
> address, and bound the legacy socket program to this address. The DPDK
> program reads all packets from the NIC and writes the slow path packets
> into this tun device. These packets are then successfully received by
> the legacy socket program. However, we are unable to get this to work
> with DPDK's tun/tap devices (dtun/dtap).
>
> When we use dtap then packet is received in dtap0 (can be seen in
> wireshark) but not reached the socket program.
> When we use dtun then packet is received in dtun0 (rx value increases
> but is not captured by wireshark) and not reached the socket program.
> (We enabled ip_forwarding and disabled rp_filter for both "all" and
> "dtun0" iface)
> (Reference:https://doc.dpdk.org/guides/nics/tap.html)
> <https://doc.dpdk.org/guides/nics/tap.html>
>
> Any pointers regarding how to do this using dtun/dtap will be helpful.
> Are we missing anything due to which linux tun device is working but
> dtun/dtap isn't?
>
> System specification :
> 2 machines with
> Distributor ID: Ubuntu
> Description: Ubuntu 18.04.2 LTS
> Release: 18.04
> Codename: bionic
>
> Kernel: 4.15.0-55-generic
> *DPDK* version: 19.05.0
>
> Hardware:
> CPU: Intel(R) Xeon(R) CPU E5-2650 v4 @ 2.20GHz
> NIC 1: Intel Corporation Ethernet Controller XL710 for 40GbE QSFP+ (i40e)
>
> (If any extra setup configuration information required will be provided)
>
Make sure tap device in kernel is up?
You might try virtio as faster exception path.
https://doc.dpdk.org/guides-18.08/howto/virtio_user_as_exceptional_path.html
prev parent reply other threads:[~2019-08-02 15:37 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-08-02 7:15 [dpdk-users] " Nilesh
2019-08-02 15:37 ` Stephen Hemminger [this message]
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=20190802083726.3dfde31f@hermes.lan \
--to=stephen@networkplumber.org \
--cc=dev@dpdk.org \
--cc=nileshunhale@cse.iitb.ac.in \
--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).