Re: Review of draft-saintandre-tls-server-id-check

[Shumon Huque <[e-mail]>]

On Mon, Sep 13, 2010 at 10:08:11AM -0600, Peter Saint-Andre wrote:
> On 9/9/10 12:22 PM, Shumon Huque wrote:
> > On Wed, Sep 08, 2010 at 11:08:29PM +0200, Stefan Santesson wrote:
> >> The only thing the client need to do is to verify that the domain name
> >> provided in the input to the lookup matches the host names provided in the
> >> output. It can then safely use the host names in the SRV record as reference
> >> identifiers IF the SRV-ID in the server certificate matches the the
> >> reference identifier.
> > 
> > This only works if the certificate matching rules say something 
> > like "match the SRVName AND also match the DNS resolved target
> > hostname in dNSName". If a client attempts to match _only_ the DNS 
> > resolved hostname without DNSSEC, there is a security problem.
> > 
> > The question is: what should the certificate matching rules say
> > when encountering a certificate with multiple identity types?
> > Right now the draft approximately says "find a match" (ie. find
> > ANY match), rather than match some logically AND'ed combination of 
> > identity types.
> > 
> >   http://tools.ietf.org/html/draft-saintandre-tls-server-id-check-09#section-4
> 
> Hi Shumon,
> 
> As I see it, this I-D is attempting to capture best current practices
> regarding the issuance and checking of certificates containing
> application server identities.

Yes, but whether they are actually "current" best practices is 
debatable. I certainly would like them to become best practices.
For example I don't believe any existing commercial CAs issue 
certificates with the SRVName or URI SAN name forms.

> Do we have evidence that any existing
> certification authorities issue certificates containing both an SRVname
> for the source domain (e.g., example.com) and dNSName for the target
> domain (e.g., apphosting.example.net)? Do we have evidence that any
> existing application clients perform such checks? If not, I would
> consider such complications to be out of scope for this I-D.

I think the question is whether we have examples of applications 
that need to verify "combinations" of subjectaltname name forms in 
certificates. Stefan says there are, but so far no-one has offered
up any public specifications of such apps. So, I think until we 
have them, I agree we can defer considerations of them to future 
documents.

I think it's reasonable for this draft to consider multiple identity
types in certificates (eg. common name, dNSName, SRVName) with the
current matching rules of ANY. This might be needed to gradually
transition an app from validating a host specific identity to an
application specific identity. The current draft allows this.

> That said, we need to be aware that if such usage arises in the future,
> someone might write a document that updates or obsoletes this I-D; in
> fact the present authors very much expect that such documents will
> emerge after the Internet community (specifically certification
> authorities, application service providers, and application client
> developers) have gained more experience with PKIX certificates in the
> context of various application technologies.

Sounds reasonable.

> Peter Saint-Andre
> https://stpeter.im/
> 

-- 
Shumon Huque
University of Pennsylvania.