[rbridge] inner tag always an s-tag?

[Anoop Ghanwani]

This is a long overdue followup on whether
or not it is OK to have the inner tag always
be an s-tag.  The issue was brought up by 
Donald at the TRILL meeting in Chicago:
http://www3.ietf.org/proceedings/07jul/slides/trill-3/sld7.htm

The argument in favor of doing this was that
it allows us to get a DE (discard eligible) bit.

I was opposed to this.  The reason is that we
should preserve whatever inner tag was received
from the attached LAN.  RBridges could be attached
to LANs that actually use s-tags, or they could
be attached to LANs that use c-tags.  We need
to preserve what was originally received.

Also, if in future, IEEE comes up with some 
other way to use the CFI bit and if we need to 
use that within the RBridge network, we would have
to find a new place to carry that bit.

Finally, even with the c-tag format, we do 
have the capability to signal drop eligibility.
It is just encoded within the PCP (what used
to be called the priority and is now the priority
code point) so it reduces the number of priorities,
but if it's OK for bridges, it's probably going
to be OK for RBridges as well. 

Anoop