netdev - Re: [PATCH RFC 0/2] Control VF link state

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1368551556.2741.8.camel@bwh-desktop.uk.solarflarecom.com>
Date:	Tue, 14 May 2013 18:12:36 +0100
From:	Ben Hutchings <bhutchings@...arflare.com>
To:	Or Gerlitz <or.gerlitz@...il.com>
CC:	Or Gerlitz <ogerlitz@...lanox.com>, <netdev@...r.kernel.org>,
	<amirv@...lanox.com>, <ronye@...lanox.com>
Subject: Re: [PATCH RFC 0/2] Control VF link state

On Mon, 2013-05-13 at 00:48 +0300, Or Gerlitz wrote:
> On Fri, May 10, 2013 at 3:34 AM, Ben Hutchings
> <bhutchings@...arflare.com> wrote:
> > On Fri, 2013-05-10 at 02:17 +0300, Or Gerlitz wrote:
> > > On Thu, May 9, 2013 at 6:24 PM, Ben Hutchings <bhutchings@...arflare.com> wrote:
> > > >
> > > > On Wed, 2013-05-08 at 16:45 +0300, Or Gerlitz wrote:
> > > > > Here's a suggestion for API and implementation that lets the admin to
> > > > > configure
> > > > > the link state of the VF / SRIOV eSwitch vPORT. Basically, it has three
> > > > > states
> > > > >
> > > > > Auto -        the VF link state will reflect the PF link state
> > > > >
> > > > > Enable - VF link stat will be always up, traffic from VF to VF can
> > > > >        work even if PF link is down.
> > > >
> > > > It seems like it would be useful to implement these two options on the PF as well.
> > >
> > > You mean that PF <--> VF communication on the same node can be made to
> > > work even when the physical link is down? this is a bit problematic to
> > > model/implement I think. Generally speaking it makes things easier to
> > > grasp if PF is considered to be the uplink of the eSwitch whos link is
> > > 1:1 as the physical link, but need to think that a bit more.
> >
> > Yeah.  In some ways it could be better for a PF driver to create two net
> > devices, one which acts as a vswitch port and one which bypasses it (if possible).
> 
> That's interesting approach, any rough thoughts what you think it would by us?

There are many attributes that could differ between the external port
(or ports - there could be more than one on the same integrated switch)
and the PF's switch port, including at least:

- Link state
- Packet counters
- MTU
- VLAN filtering

So long as we conflate these two (or more) ports into a single net
device, there will be confusion about what the attributes mean.

> > But then that's going to confuse people too.  I don't think we can win...
> >
> > > > Perhaps the default should also be specified?
> > >
> > > mmm, not sure if we can require/enforce the same default for all drivers.
> >
> > I know. :-/  But arbitrary differences between drivers are no fun for sysadmins.
> 
> yes, basically, on a production cloud environment, I would say that
> before a vport (VF) interface
> is configured, e.g ndo_set_vf_yyy is applied for it, we would want the
> VF link to be down, but for non
> production environments it could be problematic to have down/disabled
> as the default... not sure, maybe
> the default should be auto and for smart env they would set it to down
> before the VF is probed on the VM?

Not sure about that.

Ben.

-- 
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html