Linux has support for 6rd, but unfortunately it is based on earlier versions of the pre-specification. Later on, WAN address compression was added to the pre-spec which allows the ISP to configure the number of leading bits in the WAN IPv4 address to be dropped. (The v4 WAN address is tacked onto the end of the ISP's 6rd ipv6 prefix; any bits in the v4 WAN address that are site specific are useless since they in no way help identify the WAN client. Dropping these bits just means more bits for the customer to subnet with)
I did up a perl wrapper for myself that configures a linux 6rd tunnel based on the dynamic dhcpv4 address. I can't test the address compression since the kernel doesn't support it. Point being, if someone's going to add it to FreeBSD, put that in too so I can switch to FreeBSD
Anyways, the implementation is trivial so I'm sure once freebsd has the 6rd structure in the kernel, patching the tunnel tool will be easy.
DHCPv4 will definitely be the mechanism transferring the isp prefix, prefix length, compression length of the wan address, and 6rd relay server.
One of the really nice things about 6rd is that you can test it without having 6rd support from your ISP. You can test the machinery out by using the 6to4 prefix and anycast relay... it won't know the difference, since 6to4 is a subset of 6rd.
S.F, Chicago, and Philly are getting Comcast dual-stack trials. Everyone else is going to be 6rd, since there's no physical restraints to test it. 6rd, dual stack, then CNAT, basically a transition wave: 6rd=v6 over v4, dual stack = v4 and v6, CNAT = v4 over v6. Then by about the year 2079 we can turn off the last native v4 device.