Le 2013-04-24 18:22, Scott Kitterman a écrit :
on Linux, if you set bindv6only to 0 and set up a socket listening on
AF_INET6 you are able to receive IPv4 connections to that IPv6 socket. The
source IPs will be mapped into IPv4-mapped IPv6 space. This means that you
only need to setup one socket instead of one for v4 and one for v6.
That sounds right. Apparently I fail at describing it though. Going back to
the current text in the document:
Section 5 of draft-ietf-spfbis-4408bis-14:
'When any mechanism fetches host addresses to compare with <ip>, when
<ip> is an IPv4, "A" records are fetched; when <ip> is an IPv6
address, "AAAA" records are fetched. SPF implementations on IPv6
servers need to handle both "AAAA" and "A" secords, for clients on
IPv4 mapped IPv6 addresses [RFC4291]. IPv4 <ip> addresses are only
listed in an SPF record using the "ip4" mechanism.'
This is very wrong. IPv4-mapped IPv6 addresses resulting from the use of
AF_INET6 sockets with bindv6only=0 are an *implementation detail*. The
draft should focus exclusively on externally-observable behaviour. Since
there is no IPv4-mapped IPv6 address anywhere in the packets, you should
not say anything at all. Doing otherwise is just creating confusion.
Simon
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------