Multicast MAC in Source MAC Address Field

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Hi,

is it allowed/forbidden to send an Ethernet Source MAC address field
with a multicast MAC address in it?

Any references? (RFCs, IEEE doks, other??)

I've been searching for a while now and the only thing I found out is
that Cisco doesn't like it and calls it illegal but I did not find the
law yet...

Thanks
Mat

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Matthias Schaerer wrote:
> Hi,
>
> is it allowed/forbidden to send an Ethernet Source MAC address field
> with a multicast MAC address in it?
>
> Any references? (RFCs, IEEE doks, other??)
>
> I've been searching for a while now and the only thing I found out is
> that Cisco doesn't like it and calls it illegal but I did not find the
> law yet...

This is discussed in Radia Perlman's Interconnections (ISBN
0-201-56332-0), Chapter 4, and in IEEE 802.1D Annex C.

In short, an address that looks like a MAC multicast source address
*can* be used, but it's meant for a very special purpose. It is used by
a host to denote that "source routing" instructions are included in the
MAC frame. (Note: this is Layer 2 source "routing" we're talking about,
not the IP option). The "multicast bit" is set in the MAC source
address, to indicate source routing.

But instead of being called G/I bit, as it is when it's associated with
the MAC destination address, in this case it is called RII, for
Routing Information Indicator.

Thing is, though, this applies to IEEE 802.5 and FDDI (respectively
sections C.2.5.1 and C.2.5.2). No mention of Ethernet support.

(If you want to use source routing with Ethernet, don't despair. You
don't code the RII bit in the MAC source address. Instead, you use the
IEEE 802.1Q "tagged frame" format, and set the CFI bit. When the CFI
bit is set in Ethernet tagged frames, that means that there is E-RIF
included in the extended Ethernet header, i.e. routing information
field.)

So that was a bit of a side trip. I think, in short, the RII bit is not
set in Ethernet MAC source addresses. But it can be set in Token Ring
or FDDI, and it makes the MAC source address look like a multicast
address.

Bert

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Albert Manfredi wrote:
(snip regarding multicast bit in source addresses)

> So that was a bit of a side trip. I think, in short, the RII bit is not
> set in Ethernet MAC source addresses. But it can be set in Token Ring
> or FDDI, and it makes the MAC source address look like a multicast
> address.

And what about packets bridged between FDDI or TR and ethernet with
such bit set?

-- glen

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

In article <43147782$1_1@news.bluewin.ch>,
Matthias Schaerer <matthias.schaerer@anyweb.ch> wrote:

> Hi,
>
> is it allowed/forbidden to send an Ethernet Source MAC address field
> with a multicast MAC address in it?
>
> Any references? (RFCs, IEEE doks, other??)
>
> I've been searching for a while now and the only thing I found out is
> that Cisco doesn't like it and calls it illegal but I did not find the
> law yet...
>

IEEE 802.3-2002, Section 3.2.3(b):

The first bit (LSB) shall be used in the Destination Address field as an
address type designation bit to identify the Destination Address either
as an individual or as a group address. If this bit is 0, it shall
indicate that the address field contains an individual address. If this
bit is 1, it shall indicate that the address eld contains a group
address that identies none, one or more, or all of the stations connected
to the LAN. In the Source Address field, the first bit is reserved and
set to 0.


The last sentence says it all; the first bit of the SA must be set to
zero; multicast SAs are prohibited.


--
Rich Seifert Networks and Communications Consulting
21885 Bear Creek Way
(408) 395-5700 Los Gatos, CA 95033
(408) 228-0803 FAX

Send replies to: usenet at richseifert dot com

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

glen herrmannsfeldt wrote:
> Albert Manfredi wrote:
> (snip regarding multicast bit in source addresses)
>
> > So that was a bit of a side trip. I think, in short, the RII bit is not
> > set in Ethernet MAC source addresses. But it can be set in Token Ring
> > or FDDI, and it makes the MAC source address look like a multicast
> > address.
>
> And what about packets bridged between FDDI or TR and ethernet with
> such bit set?

The answer must be, the source MAC address lowest bit is returned to 0,
the frame is coded up like a 802.1Q tagged Ethernet frame, and the CFI
bit is set in the extended header, to show that source "routing" info
is also added to the Ethernet header. And the routing info from the
source-routed TR or FDDI frame is copied into the E-RIF of the extended
Ethernet header.

But that might just be a textbook answer. I don't know whether switch
manufacturers bother with this in practice.

Bert

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

> > And what about packets bridged between FDDI or TR and ethernet with
> > such bit set?
>
> The answer must be, the source MAC address lowest bit is returned to 0,

Shucks. That's "the lowest bit of the first byte of the source MAC
address ..."

> the frame is coded up like a 802.1Q tagged Ethernet frame, and the CFI
> bit is set in the extended header, to show that source "routing" info
> is also added to the Ethernet header. And the routing info from the
> source-routed TR or FDDI frame is copied into the E-RIF of the extended
> Ethernet header.
>
> But that might just be a textbook answer. I don't know whether switch
> manufacturers bother with this in practice.
>
> Bert

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

In article <1125518655.823436.325490@g44g2000cwa.googlegroups.com>,
"Albert Manfredi" <bert22306@hotmail.com> wrote:

> glen herrmannsfeldt wrote:
> > Albert Manfredi wrote:
> > (snip regarding multicast bit in source addresses)
> >
> > And what about packets bridged between FDDI or TR and ethernet with
> > such bit set?
>
> The answer must be, the source MAC address lowest bit is returned to 0,
> the frame is coded up like a 802.1Q tagged Ethernet frame, and the CFI
> bit is set in the extended header, to show that source "routing" info
> is also added to the Ethernet header. And the routing info from the
> source-routed TR or FDDI frame is copied into the E-RIF of the extended
> Ethernet header.
>
> But that might just be a textbook answer. I don't know whether switch
> manufacturers bother with this in practice.
>

In the days when bridges between FDDI/TR and Ethernet were being
designed, there was no such thing as IEEE 802.1Q VLAN tagging, so these
devices could not encapsulate source routing information for transport
across an Ethernet network.

In a non-802.1Q environment, source-routed frames (i.e., with the first
bit of the SA set to "1") were discarded at any Ethernet ports of a
bridge; they could not be validly forwarded onto an Ethernet. Source
routing could be used only within FDDI and TR networks.

There is an extensive discussion of this in Chapter 6 of "The Switch
Book."


--
Rich Seifert Networks and Communications Consulting
21885 Bear Creek Way
(408) 395-5700 Los Gatos, CA 95033
(408) 228-0803 FAX

Send replies to: usenet at richseifert dot com

 

[Stephen]

Archived from groups: comp.dcom.lans.ethernet (More info?)

"glen herrmannsfeldt" <gah@ugcs.caltech.edu> wrote in message
news:8dedneQq8poYnIveRVn-vQ@comcast.com...
> Albert Manfredi wrote:
> (snip regarding multicast bit in source addresses)
>
> > So that was a bit of a side trip. I think, in short, the RII bit is not
> > set in Ethernet MAC source addresses. But it can be set in Token Ring
> > or FDDI, and it makes the MAC source address look like a multicast
> > address.
>
> And what about packets bridged between FDDI or TR and ethernet with
> such bit set?

dont know about FDDI, but when i was doing L2 mixed Ethernet and T/R on IBM
networks, the translation bridge has to take care of it.

basically - no source route info across Ethernet unless you use one of the
proprietary Token ring tunnel protocols

FWIW a translation bridge is usually configured with a "ring number" for
Ethernet, but the RIF doesnt propagate - which can be a good way to cause
the loops source routing is supposed to resolve......

The same sort of thing happens with DLSw although the mechanisms are
different.

Also you have to worry about 3 different interacting spanning trees - there
were more good reasons for ditching Token Ring than just expensive NIC
cards.
>
> -- glen
--
Regards

Stephen Hope - return address needs fewer xxs