From: Stephen Hemminger <stephen@networkplumber.org>
To: Ramakrishnan G <gramakr@gmail.com>
Cc: aaron.f.brown@intel.com, dev@dpdk.org, saravana3@gmail.com
Subject: Re: Flow Bifurcation of splitting the traffic between kernel space and user space (DPDK)
Date: Mon, 31 Oct 2022 17:12:55 -0700 [thread overview]
Message-ID: <20221031171255.3e221d2c@shemminger-XPS-13-9360> (raw)
In-Reply-To: <CAFBoL6piCSsgJ3tYFCZTp704RuMKXEpQNkxxvN=WQJUK7FaFmg@mail.gmail.com>
On Sat, 29 Oct 2022 02:39:01 +0530
Ramakrishnan G <gramakr@gmail.com> wrote:
> From: Ramakrishnan G <gramakr@gmail.com>
> To: aaron.f.brown@intel.com, dev@dpdk.org, saravana3@gmail.com
> Subject: Flow Bifurcation of splitting the traffic between kernel space and user space (DPDK)
> Date: Sat, 29 Oct 2022 02:39:01 +0530
>
> Dear Aaron and DPDK Dev Team,
>
> Thanks for the Article talks about the Traffic Flow bifurcation
> between kernel space and user space (DPDK) (3. Flow Bifurcation How-to
> Guide — Data Plane Development Kit 16.07.2 documentation (dpdk.org)
> <https://doc.dpdk.org/guides-16.07/howto/flow_bifurcation.html>)
That DPDK release is over 6 years old. That feature is no longer supported on Intel
NIC's. You are better off using AF_XDP.
>
> We are trying to test this functionality for sending only the SSH (port 22)
> traffic to kernel and all the other traffic to be transferred to the user
> space (DPDK) by assigning same IP for both the virtual interface (one
> virtual interface is owned by the DPDK and another virtual interface is
> owned by the DPDK )
>
> Using the igb driver with max_vfs setting, we were able to create the
> virtual link and map it to user space (DPDK) and another link into kernel
> space. we assigned different IP addresses and we were able to reach from
> other host.
>
> But when we are trying to configure the flow-type for port 22
>
> Ubuntu# ethtool -K eth9 ntuple on
> Ubuntu## ethtool -N eth9 flow-type ip4 dst-port 22 action 0
> rmgr: Cannot insert RX class rule: Invalid argument
> Ubuntu## ethtool -N eth9 flow-type ip4 dst-port 22 action 1
> rmgr: Cannot insert RX class rule: Invalid argument
> Ubuntu## ethtool -N eth9 flow-type ip4 dst-port 22 action 2
> rmgr: Cannot insert RX class rule: Invalid argument
>
> We tried to apply the patch that was given in the following link,
> (
> https://patchwork.ozlabs.org/project/intel-wired-lan/patch/1451456399-13353-1-git-send-email-gangfeng.huang@ni.com/#1236040
> )
>
> But we couldn't patch any of the latest igb driver and we tried to patch
> with the 2016 igb driver.
>
> please help us in sharing the info where can we apply the patch for igb
> driver in Ubuntu.
The igb NIC does not have an flow direction.
The bifurcation for Intel NIC is based off of kernel flow director.
next prev parent reply other threads:[~2022-11-01 0:13 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-28 21:09 Ramakrishnan G
2022-10-31 17:27 ` Ramakrishnan G
2022-11-01 0:12 ` Stephen Hemminger [this message]
2022-11-01 2:30 ` Ramakrishnan G
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=20221031171255.3e221d2c@shemminger-XPS-13-9360 \
--to=stephen@networkplumber.org \
--cc=aaron.f.brown@intel.com \
--cc=dev@dpdk.org \
--cc=gramakr@gmail.com \
--cc=saravana3@gmail.com \
/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).