IPv6 Link Local Addresses [was Re: Is 1111 1110 10 equal to 0xfe80 or 0x3fa?]

Hi,

RFC4291 defines Link Local Address as:

 2.5.6.  Link-Local IPv6 Unicast Addresses

   Link-Local addresses are for use on a single link.  Link-Local
   addresses have the following format:

   |   10     |
   |  bits    |         54 bits         |          64 bits           |
   +----------+-------------------------+----------------------------+
   |1111111010|           0             |       interface ID         |
   +----------+-------------------------+----------------------------+

   Link-Local addresses are designed to be used for addressing on a
   single link for purposes such as automatic address configuration,
   neighbor discovery, or when no routers are present.

   Routers must not forward any packets with Link-Local source or
   destination addresses to other links.

This means that link local addresses have a 10 bit prefix (1111111010) followed by 54 bits of zeros.  That is it, nothing more.   Address with different prefixes or with a 1111111010 prefix followed by non-zero 54 bits are not link local addresses.

This is not ambiguous.  

Bob

> On Jun 10, 2019, at 9:08 AM, Mudric, Dusan (Dusan) <dmudric@avaya.com> wrote:
> 
>  
> 
>  
> 
> From: Yucel Guven <yucel.guven@gmail.com> 
> Sent: Monday, June 10, 2019 12:47 PM
> To: Mudric, Dusan (Dusan) <dmudric@avaya.com>
> Cc: Tom Herbert <tom@herbertland.com>; ipv6@ietf.org
> Subject: Re: Is 1111 1110 10 equal to 0xfe80 or 0x3fa?
> 
>  
> 
> No need to define as a range.
> 
> When you specify the prefix-length, it already defines a range.
> 
> e.g. FE80::/10 (absolutely not  FE8::/10)  has the range of 
> 
> fe80:0000:0000:0000::/10 - febf:ffff:ffff:ffff::/10
> 
> [Dusan] I agree that FE80::/10 is the FE80::/10 – FEBF::/10 range. However, it is not define like a range in RFC 4291 and if often misinterpreted as just one value of FE80:0000:0000:0000. The question is how to write it to make it clear this definition is a range?
> 
>  
> 
> https://tools.ietf.org/html/rfc4291#section-2.5.6 defines LL as the address with 64 bit FE80:0000:0000:0000 prefix, not as the fe80:0000:0000:0000::/10 - febf:ffff:ffff:ffff::/10 range. There are applications that need more flexibility for the LL prefix, like draft-petrescu-6man-ll-prefix-len. For these applications, FE80::/10 would be defined as LL identifier, not a prefix. The LL prefix would start with LL identifier and can have a variable length. 
> 
>  
> 
> Is there any strong reason to keep 54 bits of zeros in this definition, other than backward compatibility?
> 
>    |   10     |
> 
>    |  bits    |         54 bits         |          64 bits           |
> 
>    +----------+-------------------------+----------------------------+
> 
>    |1111111010|           0             |       interface ID         |
> 
>    +----------+-------------------------+----------------------------+
> 
>  
> 
>  
> 
> Reg.'s
> 
> Yucel
> 
>  
> 
> On Mon, Jun 10, 2019 at 3:58 PM Mudric, Dusan (Dusan) <dmudric@avaya.com> wrote:
> 
> > On Friday, June 7, 2019 at 3:48 PM Tom Herbert
> > <tom@herbertland.com> wrote
> > >
> > > >
> > > > Message: 3
> > > > Date: Fri, 7 Jun 2019 10:53:28 -0700
> > > > From: Fred Baker <fredbaker.ietf@gmail.com>
> > > > To: Alexandre Petrescu <alexandre.petrescu@gmail.com>
> > > > Cc: IPv6 <ipv6@ietf.org>
> > > > Subject: Re: Is 1111 1110 10 equal to 0xfe80 or 0x3fa?
> > > > Message-ID: <A722E202-7671-4111-BA92-8A67B3D3B924@gmail.com>
> > > > Content-Type: text/plain; charset="us-ascii"
> > > >
> > > >
> > > >
> > > > If I have prefix fe80::/10, as described in RFC 4291, the next bit
> > > > is bit 11. Doing the same subdivision of the prefix is fe80::/11 and
> > fea0::/11.
> > > [Dusan] The hexadecimal definition for LL address is not syntactically
> > correct. The binary 10 bit prefix 1111111010 cannot be presented as
> > hexadecimal FE80::/10. It is rather a range FE80::/10 - FEBF::/10. In this
> > notation, FE80::/10 = FEBF::/10,  because the first 10 bits are equal and other
> > 6 should be ignored. 111 1111010 can be defined as FE80::/10 only if every
> > time it is also mentioned that the trailing 6 bits are all zero.
> > 
> > By that logic, we'd have to mention that the trailing 118 bits are zero.  E.g.
> > FE80::/10 == FEBD:F676:BBBB:C654:FEBD:F676:BBBB:C654/10
> > also. It's obviously convenient canonical notication to express all the trailing
> > bits as zeroes for a prefix, but not required. For instance, ifconfig shows my
> > host address as fe80::ac2f:ea58:94a:438/64 which in one string indicates both
> > a fully qualified address and it's prefix bits.
> [Dusan] In this example fe80::ac2f:ea58:94a:438/64 is LL address with 0xfe80 0x0000 0x0000 0x0000 prefix. Based on LL prefix definition, this LL address can have a value of feab::ac2f:ea58:94a:438/64 and still have the binary 10 bit prefix 1111111010
> 
> May be LL address can be defined in hex notation as a range FE8::/10 - FEB::/10? This range always has the same well know LL identifier, the binary 10 bit prefix 1111111010.
> 
> > 
> > Tom
> > 
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> ipv6@ietf.org
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------
> 
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> ipv6@ietf.org
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------