[Smcwg-public] Proposed method for "validating applicant as operator of associated mail servers"

Fri Jan 21 14:52:57 UTC 2022

Hello Dimitris,

Thank you very much for reviewing the proposed text. Responses can be found
inline.

On Wed, Jan 19, 2022 at 6:37 PM Dimitris Zacharopoulos (HARICA) via
Smcwg-public <smcwg-public at cabforum.org> wrote:

>
> This method has several challenges some of which have already been
> discussed in the validation subcommittee and the server certificate working
> group. It's basically the fact that the control of the domain name is
> performed by a different entity, which is different than the actual
> "Applicant". It is like the Applicant is delegating control over to a
> different entity.
>

I appreciate the concerns on the correctness of the entity that applies for
a certificate. I believe that this case is the same as the case of
"control" vs "operate" the device in the WebPKI BRs. The CA/B Forum has
agreed that both of these cases are valid and acceptable cases for the
issuance of a TLS certificate.

In addition, in this case the Applicant has the same properties as the
Applicant in the Enterprise RA validation method, which this WG has already
agreed is an acceptable validation method, and one of the major use cases
of all CAs. Note that the entity that the CA is validating in the S/MIME
BRs is the mailbox and not the domain, therefore in the Enterprise RA case
the CAs still don't have the entity that controls the mailbox as an
Applicant, but the operator entity that controls the domain, and therefore
controls the MX records that point to the mail server. Having said that, I
believe that if the WG raises these concerns, they should be raised for all
similar cases, both control of the mail server and control of the domain
(Enterprise RA case) since both of them are based on the same principles.

>
> Assuming we overcome these issues, if we accept the fact that the operator
> of a server in the MX records of a Domain Name is by virtue *authorized *to
> issue any S/MIME certificate that contains any email address that contains
> the domain part of the validated domain name, this validation evidence *should
> not be allowed to be reused*. That is because the Domain Name owner could
> decide to switch to another mail provider, change the MX records and
> wouldn't want the previous mail provider to be authorized to issue
> certificates under their Domain Name.
>
> This means that, similarly to the CAA mandatory checking requirement in
> the TLS BRs, the CA would need to always check the MX records for every
> Domain Name being validated using this method.
>

Similarly to the previous concern, this concern can be raised for most
other types of validation, both in the S/MIME BRs and the TLS BRs. For
example:

- For the Enterprise RA case in the S/MIME BRs: the entity that applied for
the issuance of a certificate decides to stop using this email address, and
therefore the domain administrator should not be authorized to issue
certificates under their domain name for the old email address.
- For the Enterprise RA case in the S/MIME BRs (second case): the entity
that controls the domain changes the NS record in their registrar, and
therefore the domain administrator should not be authorized to issue
certificates under their domain name.
- For the Agreed‑Upon Change to Website in the TLS BRs (more clear use
case): the entity that controls the domain decides to change the A/AAAA
record, and therefore the operator should not be authorized to issue
certificates.
- For the Email to DNS CAA Contact in the TLS BRs (exactly the same use
case): the entity that controls the domain decides to change the email
contact in the CAA record.

In practice, in all those cases, reuse of validation information is allowed
based on section 4.2.1.

The way the CA/B Forum mitigates this risk in the TLS BRs is not by
requiring mandatory checking of the validation for every single issuance,
but with the following additions:

4.9.1.1 Reasons for Revoking a Subscriber Certificate
The CA SHALL revoke a Certificate within 24 hours if one or more of the
following occurs:
…
5. The CA obtains evidence that the validation of domain authorization or
control for any Fully‐Qualified Domain Name or IP address in the
Certificate should not be relied upon.
…
9.6.3 Subscriber representations and warranties
…
5. Reporting and Revocation: An obligation and warranty to:
…
b. promptly request revocation of the Certificate, and cease using it, if
any information in the Certificate is or becomes incorrect or inaccurate;

I would be glad to work on sections 4.9.1.1 and 9.6.3 once the WG gets
there, in order to address your concerns, but at the moment I feel that the
validation method should be accepted as it has the exact same properties as
the Enterprise RA validation method.

With regards to the CAA comment, I think that this is a completely
different use case related to the authorization of the CA and not the
validation of the Applicant.

Best regards,
Fotis

>
>
> Dimitris.
>
> On 19/1/2022 7:17 μ.μ., Stephen Davidson via Smcwg-public wrote:
>
> Hello all:
>
>
>
> Per our discussion today of the draft text of section 3.2.2 “Validation
> of mailbox authorization or control
> <https://github.com/cabforum/smime/blob/preSBR/SBR.md#322-validation-of-mailbox-authorization-or-control>”
> Fotis Loukos has proposed a new method for "validating applicant as
> operator of associated mail servers."  This would apply in cases where I
> have my own domain but redirect/outsource the operation of my entire email
> domain to a service.
>
>
>
> I believe that we should accommodate this common use case in the S/MIME
> BR, but know that it’s different from our previous discussions on mailbox
> verification which centered mainly on familiar methods from the TLS BR.  As
> such, I attach the proposed text below and hope that WG members can review
> the associated RFC 5321 Section 5.1 and provide feedback on list.  For
> example, this may tie in with another agenda item on the reuse periods for
> different types of verification. If needed we can schedule specific time
> (perhaps at the F2F) to work on this as well.
>
>
>
> It was a useful call today; thank you.
>
> Regards, Stephen
>
>
>
>
>
>
>
> *#### 3.2.2.3 Validating applicant as operator of associated mail
> server(s)*
>
>
>
> Confirming the Applicant's control over the rfc822Name email address by
> confirming control of the SMTP FQDN to which a message delivered to the
> email address should be directed. The SMTP FQDN MUST be identified using
> the address resolution algorithm defined in RFC 5321 Section 5.1 which
> determines which SMTP FQDNs are authoritative for a given email address. If
> more than one SMTP FQDNs have been discovered, the CA MUST verify control
> of an SMTP FQDN following the selection process at RFC 5321 Section 5.1.
>
>
>
> When confirming the Applicant's control of the SMTP FQDN, the CA MUST use
> the methods described in Section 3.2.2.4 of the TLS Baseline Requirements.
>
>
>
> This method is suitable for validating control of all email addresses
> under a single domain.
>
> _______________________________________________
> Smcwg-public mailing listSmcwg-public at cabforum.orghttps://lists.cabforum.org/mailman/listinfo/smcwg-public
>
>
> _______________________________________________
> Smcwg-public mailing list
> Smcwg-public at cabforum.org
> https://lists.cabforum.org/mailman/listinfo/smcwg-public
>

-- 

Fotis Loukos |  Security Engineer |  fotisl at google.com |
Brandschenkestrasse 110, 8002 Zurich, Switzerland
Company Identifikationsnummer: CH-020.4.028.116-1

This email can contain confidential information.If you received this email
by mistake,
do not pass it to third parties and delete all copies and enclosures,
and let us know that it has been delivered to the wrong address.

Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cabforum.org/pipermail/smcwg-public/attachments/20220121/97199578/attachment.html>