ipaddress: Two more IPv6 ranges (RFC 9602, 9780) missing from _private_networks

[mjbommar]

Bug report

Bug description:

Description

This is a peer issue to gh-151049 / PR 151050, which is addressing RFC 9665 ranges.

In this case, we similarly have the following examples return True:

• ipaddress.ip_address('5f00::1').is_global
• ipaddress.ip_address('100:0:0:1::1').is_global

Both blocks are registered by IANA with Globally Reachable = False.

The companion is_private returns False, so the documented-safe not ip.is_global check treats these addresses as ordinary public addresses.

IANA registration

The IANA IPv6 Special-Purpose Address Registry lists both blocks as Globally Reachable = False:

Address Block	Name	RFC	Globally Reachable
5f00::/16	Segment Routing (SRv6) SIDs	RFC 9602	False
100:0:0:1::/64	Dummy IPv6 Prefix	RFC 9780	False

Registry: https://www.iana.org/assignments/iana-ipv6-special-registry/iana-ipv6-special-registry.xhtml

Both were added after the table was last revised: RFC 9602 (Segment Routing over IPv6 SIDs) registered 5f00::/16 in 2024, and RFC 9780 registered the 100:0:0:1::/64 Dummy IPv6 Prefix. Note that 100:0:0:1::/64 is disjoint from the already-listed 100::/64 (RFC 6666 Discard-Only), so the existing 100::/64 entry does not cover it.

The bug

_IPv6Constants._private_networks enumerates the not-globally-reachable blocks. It lists, among others, 100::/64, 2001:db8::/32, 2002::/16, 3fff::/20, but is missing 5f00::/16 and 100:0:0:1::/64, so addresses in those blocks fall through to globally-reachable.

This is the same class as gh-124217 (which added 3fff::/20) and the IPv4 case in bpo-42937 (the 192.0.0.8 dummy address), and the GR=False complement of the in-flight #151050 (which adds the GR=True 2001:1::3/128 to the exceptions list).

Reproduce

On current main:

>>> import ipaddress
>>> ipaddress.ip_address('5f00::1').is_global
True          # expected: False (RFC 9602, IANA GR=False)
>>> ipaddress.ip_address('5f00::1').is_private
False         # expected: True
>>> ipaddress.ip_address('100:0:0:1::1').is_global
True          # expected: False (RFC 9780, IANA GR=False)
>>> ipaddress.ip_address('100:0:0:1::1').is_private
False         # expected: True
>>> ipaddress.ip_address('3fff::1').is_global      # already-fixed sibling, correct
False
>>> ipaddress.ip_address('100::1').is_global       # adjacent in-table block, correct
False

Fix

Add the two blocks to _IPv6Constants._private_networks, with RFC reference comments, e.g.:

        IPv6Network('5f00::/16'),       # RFC 9602 (SRv6 SIDs)
        IPv6Network('100:0:0:1::/64'),  # RFC 9780 (Dummy IPv6 Prefix)

and extend testReservedIpv6 to assert that 5f00::1 and 100:0:0:1::1 are not global and are private, and that an adjacent unregistered address (e.g. 100:0:0:2::1) stays unchanged so the carve-out is exact. A Misc/NEWS entry as usual. The exceptions framework and the related entries landed in GH-113179 (first shipped in 3.13), so this applies to 3.13+.

CPython versions tested on:

3.15, CPython main branch

Operating systems tested on:

Linux

Linked PRs

• gh-151749: Add 5f00::/16 (RFC 9602) and 100:0:0:1::/64 (RFC 9780) to ipaddress IPv6 private networks #151764

[lul-cas]

Hi, can I work on this?

[mjbommar]

Hi, can I work on this?

Fine with me! I think the bigger question for maintainers is whether they just want to roll this into the pending PR #151050. If so, it will be for @tonghuaroot

[StanFromIreland]

Please wait, I think @tonghuaroot can take care of this one.

[lul-cas]

Ok! Happy to help if needed anyway :D

[StanFromIreland]

Hi Lucas! There are a good few other open issues ;-) If you're interested in contributing I also recommend checking out the devguide, if you haven't already.