VLAN in kernel?

VLAN in kernel?

[Holger Kiehl]

Hello

Some time ago Ben Greear has posted a patch to include VLAN support into
the 2.4 kernel. I and many others are using this patch with great success
and without any problems for a very long time. What is the reason that
this patch is not included into the kernel?

Thanks,
Holger

Should VLANs be devices or something else?

[Ben Greear]

I have had a good discussion with Dave Miller today, and there
is one outstanding issue to clear up before my 802.1Q VLAN patch may
be considered for acceptance into the kernel:

Should VLANs be devices or some other thing?

I strongly feel that they should be devices for many reasons.

1)  It makes integration with user-space tools (ip, ifconfig, arp...) a non-issue.

2)  It is logically correct, a VLAN is a (net_)device and in all ways acts like one.

3)  It introduces no fast-path performance degradation that I know of.  The one
    slow path involves the linear lookup of a device by name (or id??).  This can
    be fixed by hashing the list, if needed.

4)  Both VLAN patches have used VLANs-as-devices from the beginning, and have
    seen no ill affects to this approach that would be mitigated by some other
    architecture.

However, we need the community as a whole to agree more-or-less that my
(and others who share them) arguments are sound.  So please, bring your
complaints fowards now...or forever patch by hand!

Also, any other complaints or suggestions for the VLAN code should be
mentioned too, of course!

If you wish to view the patch, get the 1.0.1 release from my vlan page:
http://scry.wanfear.com/~greear/vlan.html
I will release a new one shortly with the fast-dev-lookup code
(which is already #ifdef'd out) completely removed, as per Dave's
wish.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>          <Ben_Greear@excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear

Re: Should VLANs be devices or something else?

[Dax Kelson]

On Tue, 19 Jun 2001, Ben Greear wrote:

> I have had a good discussion with Dave Miller today, and there
> is one outstanding issue to clear up before my 802.1Q VLAN patch may
> be considered for acceptance into the kernel:
>
> Should VLANs be devices or some other thing?

I would vote that VLANs be devices.

Conceptually, VLANs as network devices is a no brainer.

Dax

Re: Should VLANs be devices or something else?

[David S. Miller]

Dax Kelson writes:
 > On Tue, 19 Jun 2001, Ben Greear wrote:
 > > Should VLANs be devices or some other thing?
 > 
 > I would vote that VLANs be devices.
 > 
 > Conceptually, VLANs as network devices is a no brainer.

Conceptually, svr4 streams are a beautiful and elegant
mechanism. :-)

Technical implementation level concerns need to be considered
as well as "does it look nice".

Later,
David S. Miller
davem@redhat.com

Re: Should VLANs be devices or something else?

[Dax Kelson]

On Tue, 19 Jun 2001, David S. Miller wrote:

> Technical implementation level concerns need to be considered
> as well as "does it look nice".

Ok, let's hear the other side of the story.

What are the alternatives?

Dax

Re: Should VLANs be devices or something else?

[Ben Greear]

"David S. Miller" wrote:
> 
> Dax Kelson writes:
>  > On Tue, 19 Jun 2001, Ben Greear wrote:
>  > > Should VLANs be devices or some other thing?
>  >
>  > I would vote that VLANs be devices.
>  >
>  > Conceptually, VLANs as network devices is a no brainer.
> 
> Conceptually, svr4 streams are a beautiful and elegant
> mechanism. :-)
> 
> Technical implementation level concerns need to be considered
> as well as "does it look nice".

I found it to be the easiest way to implement things.  It allowed
me to not have to touch any of layer 3, and I did not have to patch
any user-space program like ip or ifconfig.

I'm not even sure if the nay-sayers ever had another idea, they
just didn't like having lots of interfaces.  Originally, there
were claims of inefficiency, but it seems that other than things
like 'ip' and ifconfig, there are no serious performance problems
I am aware of.


Adding the hashed lookup for devices took the exponential curve out of
ip and ifconfig's performance, btw.


-- 
Ben Greear <greearb@candelatech.com>          <Ben_Greear@excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear

Re: Should VLANs be devices or something else?

[Marcell Gal]

Hi,

Ben Greear wrote:

> >  > > Should VLANs be devices or some other thing?
> I found it to be the easiest way to implement things.  It allowed
> me to not have to touch any of layer 3, and I did not have to patch
> any user-space program like ip or ifconfig.

I faced the same issue when implementing RFC2684 (formerly 1483)
Ethernet over ATM-AAL5. Since users want to do the same thing
(ifconfig, tcpdump, rfc 2514 pppoe, dhcp, ipx) as with traditional eth0
using register_netdev was 'the right thing'.
However having the possibility of many devices annoyed
some people. (upto appr. 4095/ATM-VC in case of vlan over rfc2684 over
atm ;-)

My answer to the (old) 'long ifconfig listing' argument:
Users do not have more interfaces in the ifconfig listing than those they
create for themselves.
That's ok, exactly what they want. Those who do not like many interfaces
do not
create many.
The real thrill would be maintaining new (or patched) tools just because
we want to
avoid having the _possibility_ of long listings at any cost...

I remember
/proc/sys/net/ipv4/conf/
was broken for about >300 devices. I do not know how's it today.

> Adding the hashed lookup for devices took the exponential curve out of
> ip and ifconfig's performance, btw.

n^2 for creating n devices (in the unfortunate increasing or random
order),
(not 2^n), I guess.

    Cell

--
You'll never see all the places, or read all the books, but fortunately,
they're not all recommended.

Re: [VLAN] Re: Should VLANs be devices or something else?

[Ben Greear]

Marcell Gal wrote:
> 

> I remember
> /proc/sys/net/ipv4/conf/
> was broken for about >300 devices. I do not know how's it today.

My VLAN code creates an entry for every vlan in /proc/net/vlan/
too, and it seems to quit creating entries after about 250 or so.
It's read-only info, so it doesn't seem to do too much harm.

I think there must be an 8-bit limit somewhere in the proc-fs.

> > Adding the hashed lookup for devices took the exponential curve out of
> > ip and ifconfig's performance, btw.
> 
> n^2 for creating n devices (in the unfortunate increasing or random
> order),
> (not 2^n), I guess.

It definately isn't fast, but then again, it is fast enough when you
only have 500 or so interfaces.  For the wierdos that want more, we
can just wait a little longer... :)

I'll offer the hashed-device-lookup
patch separately on my web site so it can be used if needed...

Ben

-- 
Ben Greear <greearb@candelatech.com>          <Ben_Greear@excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear

Re: [VLAN] Re: Should VLANs be devices or something else?

[Eran Man]

Ben Greear wrote:
> 
> "David S. Miller" wrote:

> > Conceptually, svr4 streams are a beautiful and elegant
> > mechanism. :-)
> >
> > Technical implementation level concerns need to be considered
> > as well as "does it look nice".
> 
> I found it to be the easiest way to implement things.  It allowed
> me to not have to touch any of layer 3, and I did not have to patch
> any user-space program like ip or ifconfig.
> 
> I'm not even sure if the nay-sayers ever had another idea, they
> just didn't like having lots of interfaces.  Originally, there
> were claims of inefficiency, but it seems that other than things
> like 'ip' and ifconfig, there are no serious performance problems
> I am aware of.

There is the issue with netlink notification of large number of events.
See the mail thread starting from:
http://oss.sgi.com/projects/netdev/mail/netdev/msg01879.html

Re: Should VLANs be devices or something else?

[Gleb Natapov]

On Tue, Jun 19, 2001 at 03:49:13PM -0700, David S. Miller wrote:
> 
> Dax Kelson writes:
>  > On Tue, 19 Jun 2001, Ben Greear wrote:
>  > > Should VLANs be devices or some other thing?
>  > 
>  > I would vote that VLANs be devices.
>  > 
>  > Conceptually, VLANs as network devices is a no brainer.
> 
> Conceptually, svr4 streams are a beautiful and elegant
> mechanism. :-)
> 
> Technical implementation level concerns need to be considered
> as well as "does it look nice".

How can I implement intermediate layer between L3 and L2 in the current 
kernel? This is what VLAN is all about. The only way to do it today is to 
pretend to be a network device for L3, do your job (adding VLAN header) and 
the job of L2 (build ethernet header) and queue packet to master device for 
transition. This is what ipip module does, this is what bonding module does 
and many others. And this is because L1 and L2 coupled too tightly together in 
the kernel now. In fact it is almost impossible to implement new L2 protocol without 
changing net_device structure. Something should be done about L1+L2 design till then 
pretend to be the net_device is the only solution if you want VLAN to be transparent 
for L3 protocols. If you want to implement VLANs only for IP layer this can be done
differently of course.

P.S: This topic was already discussed on netdev list before (and not once I think :)). 

--
			Gleb.

Re: [VLAN] Should VLANs be devices or something else?

[Sander Steffann]

Hi Ben & all,

> Should VLANs be devices or some other thing?

VLANs should be devices IMHO. It 'feels' right, and I think it's what most
(if not all) users expect them to be.

Bye,
Sander.

Re: [VLAN] Should VLANs be devices or something else?

[Peter C. Norton]

I agree with the device vote.  Also, if the interface to the vlan devices
are similar enough to ethernet that the bonding driver can easily
incorperate them, then you can get bonded (a.k.a. redundant) trunks for
cheap.  This would lead to linux becoming a more robust router in practice.

-Peter

On Wed, Jun 20, 2001 at 09:21:55AM +0200, Sander Steffann wrote:
> Hi Ben & all,
> 
> > Should VLANs be devices or some other thing?
> 
> VLANs should be devices IMHO. It 'feels' right, and I think it's what most
> (if not all) users expect them to be.
> 
> Bye,
> Sander.
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

-- 
The 5 year plan:
In five years we'll make up another plan.
Or just re-use this one.

Re: [VLAN] Should VLANs be devices or something else?

[Albert D. Cahalan]

> Should VLANs be devices or some other thing?

What is good for PPP-over-Ethernet is good for VLANs,
which are basically Ethernet-over-Ethernet.

Re: [VLAN] Should VLANs be devices or something else?

[Ben Greear]

"Albert D. Cahalan" wrote:
> 
> > Should VLANs be devices or some other thing?
> 
> What is good for PPP-over-Ethernet is good for VLANs,
> which are basically Ethernet-over-Ethernet.

So, um, for those fortunate enough not to need PPPoE, are they
devices or what?

-- 
Ben Greear <greearb@candelatech.com>          <Ben_Greear@excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear

Re: [Vlan-devel] Should VLANs be devices or something else?

[Guy Van Den Bergh]

Maybe this has been discussed already, but what about integration
with the bridging code? Is it possible to add add a vlan interface to a 
bridge? In other words, can you bridge between one or more regular 
interfaces and a vlan?

Regards,
Guy

Ben Greear wrote:

> I have had a good discussion with Dave Miller today, and there
> is one outstanding issue to clear up before my 802.1Q VLAN patch may
> be considered for acceptance into the kernel:
> 
> Should VLANs be devices or some other thing?
> 
> I strongly feel that they should be devices for many reasons.
> 
> 1)  It makes integration with user-space tools (ip, ifconfig, arp...) a non-issue.
> 
> 2)  It is logically correct, a VLAN is a (net_)device and in all ways acts like one.
> 
> 3)  It introduces no fast-path performance degradation that I know of.  The one
>     slow path involves the linear lookup of a device by name (or id??).  This can
>     be fixed by hashing the list, if needed.
> 
> 4)  Both VLAN patches have used VLANs-as-devices from the beginning, and have
>     seen no ill affects to this approach that would be mitigated by some other
>     architecture.
> 
> However, we need the community as a whole to agree more-or-less that my
> (and others who share them) arguments are sound.  So please, bring your
> complaints fowards now...or forever patch by hand!
> 
> Also, any other complaints or suggestions for the VLAN code should be
> mentioned too, of course!
> 
> If you wish to view the patch, get the 1.0.1 release from my vlan page:
> http://scry.wanfear.com/~greear/vlan.html
> I will release a new one shortly with the fast-dev-lookup code
> (which is already #ifdef'd out) completely removed, as per Dave's
> wish.
> 
> Thanks,
> Ben
> 
> 

Re: [Vlan-devel] Should VLANs be devices or something else?

[Ben Greear]

Guy Van Den Bergh wrote:
> 
> Maybe this has been discussed already, but what about integration
> with the bridging code? Is it possible to add add a vlan interface to a
> bridge? In other words, can you bridge between one or more regular
> interfaces and a vlan?
> 
> Regards,
> Guy

I hear it does work with the bridging code, just as you would expect
it to.  I have not tried it personally...

Ben

-- 
Ben Greear <greearb@candelatech.com>          <Ben_Greear@excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear

Re: [Vlan-devel] Should VLANs be devices or something else?

[Matthias Welwarsky]

Hi all,

Can I use CBQ with VLans? It should be possible if they are devices, has
anybody tried this yet?

regards,
    Matthias Welwarsky

--
-----------------------------------------------------------------------
Matthias Welwarsky               mail: matthias.welwarsky@frontsite.de
System Engineer

frontsite AG
Gesellschaft für Informationstechnologie  tel.: +49 6151 - 86 00 00
Gutenbergstraße 10                        fax.: +49 6151 - 86 00 499
64331 Weiterstadt                         web: http://www.frontsite.de
-----------------------------------------------------------------------

Re: [Vlan-devel] Should VLANs be devices or something else?

[Chris Wedgwood]

On Fri, Jun 22, 2001 at 05:00:29PM +0200, Matthias Welwarsky wrote:

    Can I use CBQ with VLans? It should be possible if they are
    devices, has anybody tried this yet?

Unless I horribly misunderstand things, anything you can do with a
standard ethernet interface without the presence of vlans, you should
be able to to on a vlan-interface in the presence of vlans.

So, yes, things like CBQ and bridging _should_ work. If they don't,
then arguably its a bug.



  --cw