From: Jamal Hadi Salim <jhs@mojatatu•com>
To: Neil Horman <nhorman@tuxdriver•com>
Cc: Jiri Pirko <jiri@resnulli•us>,
Florian Fainelli <f.fainelli@gmail•com>,
netdev <netdev@vger•kernel.org>,
David Miller <davem@davemloft•net>,
andy@greyhouse•net, tgraf@suug•ch, dborkman@redhat•com,
ogerlitz@mellanox•com, jesse@nicira•com, pshelar@nicira•com,
azhou@nicira•com, Ben Hutchings <ben@decadent•org.uk>,
Stephen Hemminger <stephen@networkplumber•org>,
jeffrey.t.kirsher@intel•com, vyasevic <vyasevic@redhat•com>,
Cong Wang <xiyou.wangcong@gmail•com>,
John Fastabend <john.r.fastabend@intel•com>,
Eric Dumazet <edumazet@google•com>,
Scott Feldman <sfeldma@cumulusnetworks•com>,
Lennert Buytenhek <buytenh@wantstofly•org>
Subject: Re: [patch net-next RFC 0/4] introduce infrastructure for support of switch chip datapath
Date: Tue, 25 Mar 2014 16:46:44 -0400 [thread overview]
Message-ID: <5331EB34.4060500@mojatatu.com> (raw)
In-Reply-To: <20140325173927.GE8102@hmsreliant.think-freely.org>
On 03/25/14 13:39, Neil Horman wrote:
> No, but it would be really nice if these smaller devices could take advantage of
> this infrastructure.
Indeed.
> Looking at it, I don't see why thats not possible. The
> big trick (as we've discussed in the past), is using a net_device structure to
> take advantage of all the features that net_devices offer while not enabling the
> device specific features that some hardware doesn't allow.
>
Exactly. And i dont think thats hard to do. I do think for capabilities,
netdev->features is insufficient (example I cant export to user space
the size of my h/w fdb table etc). But those things can be easily
ironed out.
> For instance the broadcom chips that live in many wireless routers would be well
> served by the model jiri has here as far as Media level interface control is
> concerned (i.e. ifup/down/speed/duplex/etc), but its a bit lacking in that
> net_devices are assumed to support L3 protocol configuration (i.e. they can have
> ip addresses assigned to them), which you can't IIRC do on these chips.
>
This is part of the challenge i was talking about and why the lowest
common denominator is just ports and L2 bridging.
> Would it be worth considering a private interface model? That is to say:
>
> 1) Ports on a switch chip are accessed using net_device structures, but
> registered to a private list contained within the switch device, rather than to
> the net namespaces device list.
>
> 2) Access to the switch ports via user space is done through the master switch
> interface with additional netlink attributes specifying the port on the switch
> to access (or not to access the master switch device directly)
>
>
> Such a model I think might fit well with Jiri's code here and provide greater
> flexibility for a wider range of devices. It would of course require
> augmentation for user space, but the changes would be additive, so I think they
> would be reasonable. This would also allow the switch device to have a hook in
> the control path to block or allow features that the hardware may or may not
> support while still being able to use the existing net_device infrastructure to
> support these operations as they are normally carried out.
>
I think Jiri's model is upside down (Yes, I was on that boat as well
earlier)
What needs to be exposed are ports. Something like #1 above which is not
a netdev but rather the conduit to the chip.
Note: We already an above working model with bridging today. If i attach
a port to a bridge I can infact get/set the fdb entries from/to the
bridge as well as ones offloaded on the chip/hware.
I should be able to do the same with stats etc.
Seems to make sense we to extend it to other features.
The litmus test is: Can i have my iproute2 please? If you can do that
then you are allowing me to do bridges, routes, ports, vxlan, tunnels
qos etc. Whatever the chips capabilities allow for otherwise I am
terminating at the CPU level.
cheers,
jamal
next prev parent reply other threads:[~2014-03-25 20:46 UTC|newest]
Thread overview: 125+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-03-19 15:33 [patch net-next RFC 0/4] introduce infrastructure for support of switch chip datapath Jiri Pirko
2014-03-19 15:33 ` [patch net-next RFC 1/4] openvswitch: split flow structures into ovs specific and generic ones Jiri Pirko
2014-03-20 13:04 ` Thomas Graf
2014-03-19 15:33 ` [patch net-next RFC 2/4] net: introduce switchdev API Jiri Pirko
2014-03-20 13:59 ` Thomas Graf
2014-03-20 14:18 ` Jiri Pirko
2014-03-20 14:43 ` Nikolay Aleksandrov
2014-03-20 15:42 ` Jiri Pirko
2014-03-19 15:33 ` [patch net-next RFC 3/4] openvswitch: Introduce support for switchdev based datapath Jiri Pirko
2014-03-19 15:33 ` [patch net-next RFC 4/4] net: introduce dummy switch Jiri Pirko
2014-03-20 11:49 ` [patch net-next RFC 0/4] introduce infrastructure for support of switch chip datapath Jamal Hadi Salim
2014-03-20 12:40 ` Jiri Pirko
2014-03-20 17:21 ` Florian Fainelli
2014-03-21 12:04 ` Jamal Hadi Salim
2014-03-22 9:48 ` Jiri Pirko
2014-03-24 23:07 ` Jamal Hadi Salim
2014-03-25 17:39 ` Neil Horman
2014-03-25 18:00 ` Thomas Graf
2014-03-25 19:35 ` Neil Horman
2014-03-25 20:11 ` Florian Fainelli
2014-03-25 20:31 ` Neil Horman
2014-03-25 21:22 ` Jamal Hadi Salim
2014-03-25 21:26 ` Thomas Graf
2014-03-25 21:42 ` Florian Fainelli
2014-03-25 21:54 ` Thomas Graf
2014-03-26 10:55 ` Neil Horman
2014-03-26 5:37 ` Roopa Prabhu
2014-03-26 10:54 ` Jamal Hadi Salim
2014-03-26 15:31 ` John W. Linville
2014-03-26 16:54 ` Roopa Prabhu
2014-03-26 16:59 ` Jiri Pirko
2014-03-26 17:29 ` Florian Fainelli
2014-03-26 17:35 ` Jiri Pirko
2014-03-26 17:58 ` Florian Fainelli
2014-03-26 18:14 ` Jiri Pirko
2014-03-26 18:29 ` Hannes Frederic Sowa
2014-03-26 18:30 ` Florian Fainelli
2014-03-26 21:51 ` Jamal Hadi Salim
2014-03-26 22:22 ` Florian Fainelli
2014-03-26 22:53 ` Jamal Hadi Salim
2014-03-26 23:16 ` Florian Fainelli
2014-03-27 6:56 ` Jiri Pirko
2014-03-27 10:39 ` Jamal Hadi Salim
2014-03-27 10:50 ` Jiri Pirko
2014-03-27 11:12 ` Jamal Hadi Salim
2014-03-27 11:16 ` Jiri Pirko
2014-03-27 14:10 ` Sergey Ryazanov
2014-03-27 16:41 ` Florian Fainelli
2014-03-27 16:57 ` Jiri Pirko
2014-03-27 16:59 ` Thomas Graf
2014-03-27 20:32 ` Sergey Ryazanov
2014-03-27 21:20 ` Florian Fainelli
2014-03-27 21:55 ` Jamal Hadi Salim
2014-03-28 6:28 ` Jiri Pirko
2014-03-30 12:08 ` Alon Harel
2014-03-27 21:41 ` Jamal Hadi Salim
2014-03-27 16:55 ` Jiri Pirko
2014-03-27 19:58 ` Sergey Ryazanov
2014-03-27 20:01 ` Florian Fainelli
2014-03-27 20:04 ` Sergey Ryazanov
2014-03-27 21:47 ` Jamal Hadi Salim
2014-03-27 21:54 ` Florian Fainelli
2014-03-27 21:59 ` Jamal Hadi Salim
2014-03-27 22:19 ` Florian Fainelli
2014-03-27 23:42 ` Thomas Graf
2014-03-27 23:46 ` Florian Fainelli
2014-03-26 17:57 ` Roopa Prabhu
2014-03-26 18:09 ` Florian Fainelli
2014-03-27 13:46 ` John W. Linville
2014-03-26 17:47 ` Roopa Prabhu
2014-03-26 18:03 ` Jiri Pirko
2014-03-26 21:27 ` Roopa Prabhu
2014-03-26 21:31 ` Jiri Pirko
2014-03-27 15:35 ` Roopa Prabhu
2014-03-27 16:10 ` Jiri Pirko
2014-04-01 19:13 ` Scott Feldman
2014-04-02 6:41 ` Jiri Pirko
2014-04-02 15:37 ` Scott Feldman
2014-04-02 14:32 ` Andy Gospodarek
2014-04-02 15:25 ` John W. Linville
2014-04-02 16:15 ` Scott Feldman
2014-04-02 16:47 ` Florian Fainelli
2014-04-02 21:52 ` Thomas Graf
2014-04-02 19:29 ` John W. Linville
2014-04-02 19:54 ` Scott Feldman
2014-04-02 20:06 ` John W. Linville
2014-04-02 20:04 ` Stephen Hemminger
2014-04-02 20:23 ` Jiri Pirko
2014-04-02 20:38 ` John W. Linville
2014-04-02 21:36 ` Thomas Graf
2014-03-25 20:56 ` Jamal Hadi Salim
2014-03-25 21:19 ` Thomas Graf
2014-03-25 21:24 ` Jamal Hadi Salim
2014-03-26 7:21 ` Jiri Pirko
2014-03-26 11:00 ` Jamal Hadi Salim
2014-03-26 11:06 ` Jamal Hadi Salim
2014-03-26 11:31 ` Jamal Hadi Salim
2014-03-26 13:20 ` Jiri Pirko
2014-03-26 13:23 ` Jamal Hadi Salim
2014-03-26 13:17 ` Jiri Pirko
2014-03-26 11:10 ` Neil Horman
2014-03-26 11:29 ` Thomas Graf
2014-03-26 12:58 ` Jamal Hadi Salim
2014-03-26 15:22 ` John W. Linville
2014-03-26 21:36 ` Jamal Hadi Salim
2014-03-26 18:21 ` Neil Horman
2014-03-26 19:11 ` Florian Fainelli
2014-03-26 22:44 ` Jamal Hadi Salim
2014-03-26 23:15 ` Thomas Graf
2014-03-26 23:21 ` Florian Fainelli
2014-03-27 15:26 ` Neil Horman
2014-03-27 21:33 ` Jamal Hadi Salim
2014-03-26 19:24 ` Hannes Frederic Sowa
2014-03-27 13:43 ` John W. Linville
2014-03-26 12:19 ` Jamal Hadi Salim
2014-03-26 15:27 ` John W. Linville
2014-03-25 18:33 ` Florian Fainelli
2014-03-25 19:40 ` Neil Horman
2014-03-25 20:00 ` Florian Fainelli
2014-03-25 21:39 ` tgraf
2014-03-25 22:08 ` Jamal Hadi Salim
2014-03-26 5:48 ` Roopa Prabhu
2014-03-25 20:46 ` Jamal Hadi Salim [this message]
2014-03-26 7:24 ` Jiri Pirko
2014-03-22 9:40 ` Jiri Pirko
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=5331EB34.4060500@mojatatu.com \
--to=jhs@mojatatu$(echo .)com \
--cc=andy@greyhouse$(echo .)net \
--cc=azhou@nicira$(echo .)com \
--cc=ben@decadent$(echo .)org.uk \
--cc=buytenh@wantstofly$(echo .)org \
--cc=davem@davemloft$(echo .)net \
--cc=dborkman@redhat$(echo .)com \
--cc=edumazet@google$(echo .)com \
--cc=f.fainelli@gmail$(echo .)com \
--cc=jeffrey.t.kirsher@intel$(echo .)com \
--cc=jesse@nicira$(echo .)com \
--cc=jiri@resnulli$(echo .)us \
--cc=john.r.fastabend@intel$(echo .)com \
--cc=netdev@vger$(echo .)kernel.org \
--cc=nhorman@tuxdriver$(echo .)com \
--cc=ogerlitz@mellanox$(echo .)com \
--cc=pshelar@nicira$(echo .)com \
--cc=sfeldma@cumulusnetworks$(echo .)com \
--cc=stephen@networkplumber$(echo .)org \
--cc=tgraf@suug$(echo .)ch \
--cc=vyasevic@redhat$(echo .)com \
--cc=xiyou.wangcong@gmail$(echo .)com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
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