Re: [PATCH] bridge: Forward EAPOL Kconfig option BRIDGE_PAE_FORWARD

["Michał Mirosław" <mirqus@gmail.com>]

2011/7/1 David Lamparter <equinox@diac24.net>:
> On Wed, Jun 29, 2011 at 04:34:23PM -0700, Stephen Hemminger wrote:
>> The problem is that the damn 802.1 committees keep loading up protocols
>> on the same multicast address range. Trying to solve a design committee
>> problem in the kernel is not going to make anybody happy.
>>
>> I am happy with the simple solution of:
>>   no STP == Hub
>>   STP    == Bridge
>> These are both well know configurations and are blessed by standards.
>
> I agree, that is how we should behave by default, and we'll match most
> admin's expectations.
>
> Regarding multicast groups, I would summarise like this:
> 1. any multicast gets forwarded by default,
>  2. unless it is 01:80:c2:00:00:01 or :02 (pause/bonding)
>    (this rule applies regardless of STP state)
>  3. if STP is on:
>  4. 01:80:c2:00:00:00 (STP) never gets forwarded
>  5. 01:80:c2:00:00:03-0f don't get forwarded by default
>
> What we can do is add a switch to disable the #5 rule. The way I see it
> is that that switch would remove an exception from the rule and turn it
> back to the default #1; that's acceptable for making a new knob in my
> eyes.
>
> (Adding an 802.1X knob would be an exception to the exception for me,
> which is why I'm against it.)
>
> I'll cook up a patch in a few minutes, we really need to get rule #2
> right anyway. We _MUST_NOT_ pass bonding frames in any case, but we
> currently do that if STP is off. (cf. my earlier patch 1/2)

If you use linux box as a (invisible) L2 network tap, then you want to
pass everything in the hub mode (including LACP/whatever).

Best Regards,
Michał Mirosław