Re: [dane] PGP security models, was Summary of IETF LC for draft-ietf-dane-openpgpkey

Paul Wouters <paul@nohats.ca> Mon, 21 September 2015 19:14 UTC

Date: Mon, 21 Sep 2015 15:14:36 -0400
From: Paul Wouters <paul@nohats.ca>
To: ietf@ietf.org
In-Reply-To: <20150921172109.19893.qmail@ary.lan>
Message-ID: <alpine.LFD.2.20.1509211455150.420@bofh.nohats.ca>
References: <20150921172109.19893.qmail@ary.lan>
User-Agent: Alpine 2.20 (LFD 67 2015-01-07)
MIME-Version: 1.0
Content-Type: text/plain; format="flowed"; charset="US-ASCII"
Archived-At: <http://mailarchive.ietf.org/arch/msg/dane/HEDtKseGGx_goU8Q9oDnaMkRJmg>
Cc: dane WG list <dane@ietf.org>
Subject: Re: [dane] PGP security models, was Summary of IETF LC for draft-ietf-dane-openpgpkey
Precedence: list

On Mon, 21 Sep 2015, John Levine wrote:

>> OPENPGP is a data format, WoT is one way to employ that format to
>> exchange messages.   It is not a *required* way to use OPENPGP.
>
> Sure, but it's the way that everyone has used PGP for 20 years,
> and it's the security model that everyone I know expects when they
> use PGP keys.

Actually, nmost people I know never use the WoT. They only use keys
obtained directly from the person they want to exchange encrypted email
with.

> This draft uses a model in which the key is bound to a mailbox

openpgp keys are bound to ID's, which can ultimately end up in a
mailbox but is not required to do so.

For instance, the gpg key used to sign fedora21 packages with an openpgp
key ID containing "fedora21@fedoraproject.org" might not have any mailbox
associated with it. It is merely shared in the DNS under an email address,
without a mailbox or valid local-part.

> any stronger identity, and you have to trust that the domain's
> management fairly represents its users

Correct, the domain's management that controls either DNS or SMTP servers,
can steal a users email.

> That's not a ridiculous model, but if
> that's the model, the draft and draft-ietf-dane-openpgpkey-usage need
> to say so.  At this point, neither does.

>From the Introduction:

    This document specifies a method for publishing and
    locating OpenPGP public keys in DNS for a specific email address
    using a new OPENPGPKEY DNS Resource Record.  Security is provided via
    DNSSEC.

So your point is made already pretty clear in the introduction
already. Security comes from DNSSEC, so whoever controls the domain,
controls the publishing of openpgp keys.

Section 5.2 also contains some advise. Section 7.4 also mentions this,
but not under a section title that makes that very clear.

Some clarifications will be made, especially in the security
considerations section, to clarify this, based on the IETF LC comments.

Thank you,

Paul

Re: [dane] Summary of IETF LC for draft-ietf-dane… Viktor Dukhovni
Re: [dane] PGP security models, was Summary of IE… John Levine
Re: [dane] PGP security models, was Summary of IE… Paul Wouters
Re: [dane] PGP security models, was Summary of IE… manning
Re: [dane] PGP security models, was Summary of IE… manning
Re: [dane] PGP security models, was Summary of IE… Scott Kitterman
Re: [dane] PGP security models, was Summary of IE… John C Klensin
Re: [dane] PGP security models, was Summary of IE… Joe Abley
Re: [dane] PGP security models, was Summary of IE… Paul Wouters
Re: [dane] provisioning assumptions, was PGP secu… John Levine
Re: [dane] provisioning assumptions, was PGP secu… Paul Wouters
Re: [dane] PGP security models, was Summary of IE… Randy Bush
Re: [dane] PGP security models, was Summary of IE… Viktor Dukhovni
Re: [dane] PGP security models, was Summary of IE… Randy Bush
Re: [dane] PGP security models, was Summary of IE… Sam Hartman
Re: [dane] PGP security models, was Summary of IE… Dave Crocker
Re: [dane] PGP security models, was Summary of IE… Paul Wouters
Re: [dane] PGP security models, was Summary of IE… Sam Hartman
Re: [dane] PGP security models, was Summary of IE… Dave Crocker