>>>>> "Bastien" == Bastien Roucaries <ro...@debian.org> writes:

Thanks, Bastien, for preparing this update.
I do have a couple of comments for the stable release team to consider:

    Bastien> [ Reason ] CVE-2025-3576

    Bastien> [ Impact ] CVE-2025-3576 is not fixed.

It's not clear what the impact of this is.
At a minimum, people using krb5-kdc to support S4uProxy PACCs are
vulnerable to an attack on the PACC.
Redhat's CVE talks about an attack against GSS though, and I haven't
seen public discussion of that.
(The original paper noted problems with the cryptographic primitives
used for RC4 GSS; that's clearly true, but the impact of those defects
is unclear.)

    Bastien> [ Tests ] Test suite

    Bastien> [ Risks ] low disabling security hardening is possible

I strongly disagree with this characterization, and am somewhat
frustrated because I asked Bastien to call out risks in the proposed
update request that were not called out.
(On the other hand I am very greatful that Bastien prepared this update
and all I had to do is review).

In my mind, the risk of this update is toward the high end of what we
accept in stable updates.
This change disables two encryption types in the over-the-wire protocol.
That is, it intentionally introduces an incompatibility. If you install
this update, things may stop workin for you.
As Bastien points out, you can disable the security hardening and get
things working again, but you have to take manual action.

In my mind, the PACC attack plus the threat of not-very-public GSS
attacks is worth introducing this incompatibility.
I also think the set of configurations that we will break is low.
So I do recommend this update is accepted.

Attachment: signature.asc
Description: PGP signature

Reply via email to