DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Gaëtan Rivet" <gaetan.rivet@6wind.com>
To: Thomas Monjalon <thomas@monjalon.net>
Cc: dev@dpdk.org, Ferruh Yigit <ferruh.yigit@intel.com>, users@dpdk.org
Subject: Re: [dpdk-dev] [PATCH v2 0/4] Introduce net bus
Date: Thu, 8 Jun 2017 22:31:09 +0200	[thread overview]
Message-ID: <20170608203109.GA29091@bidouze.vm.6wind.com> (raw)
In-Reply-To: <34816509.TmcHNXKEUC@xps>

Hi Thomas,

On Thu, Jun 08, 2017 at 09:31:23AM +0200, Thomas Monjalon wrote:
> I'm sorry, I do not understand the description.
> Please start with a real problem statement and explain how it is solved.
> Thanks

Sure.

Currently, we probe devices by using

./usertools/dpkd-devbind.py -b igb_uio 00:02.0
./build/app/whatever -w 00:02.0
./usertools/dpkd-devbind.py -b ixgbe 00:02.0

Wouldn't it be nice to be able to:

./build/app/whatever -w ens2

Instead?

Many things are missing for this. This PoC only shows the simplest
case with a bifurcated driver, but the concept is there.

This bus is an intepretation layer to transform a kernel netdevice name
into a usable DPDK device, which should give enough info about its
dependencies: bus, IO module, resource management policies,
etc... to prepare it.

I plan to add binding / unbinding afterward, and I wanted to
propose this early enough as there were talks of kernel control plane,
and this could be a building block to this general direction.

Auto-unbind assumes a clean exit from the DPDK application. This might
be a big departure from current assumptions, where buses very seldom
plan for ending operations. i.e. rte_bus has scan and probe, but no
remove API. While unplug should soon exist, this is a crude workaround.

-- 
Gaëtan Rivet
6WIND

      reply	other threads:[~2017-06-08 20:31 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-06-01 10:44 [dpdk-dev] [PATCH 0/3] " Gaetan Rivet
2017-06-01 10:44 ` [dpdk-dev] [PATCH 1/3] bus/net: introduce " Gaetan Rivet
2017-06-01 10:44 ` [dpdk-dev] [PATCH 2/3] net/mlx4: net bus support Gaetan Rivet
2017-06-01 10:44 ` [dpdk-dev] [PATCH 3/3] net/mlx5: " Gaetan Rivet
2017-06-08  0:00 ` [dpdk-dev] [PATCH v2 0/4] Introduce net bus Gaetan Rivet
2017-06-08  0:00   ` [dpdk-dev] [PATCH v2 1/4] bus/net: introduce " Gaetan Rivet
2017-06-08  0:00   ` [dpdk-dev] [PATCH v2 2/4] bus/net: implement hotplug bus operations Gaetan Rivet
2017-06-08  0:01   ` [dpdk-dev] [PATCH v2 3/4] net/mlx4: net bus support Gaetan Rivet
2017-06-08  0:01   ` [dpdk-dev] [PATCH v2 4/4] net/mlx5: " Gaetan Rivet
2017-06-08  7:31   ` [dpdk-dev] [PATCH v2 0/4] Introduce net bus Thomas Monjalon
2017-06-08 20:31     ` Gaëtan Rivet [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=20170608203109.GA29091@bidouze.vm.6wind.com \
    --to=gaetan.rivet@6wind.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=thomas@monjalon.net \
    --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).