My idea isn't fully formed yet, but...

Wildcard certs are more risky than normal certs since the CA doesn't know 
exactly what they are securing.  All they know is the secured base level 
domain.  Therefore, I think the public has a strong interest in knowing when a 
wildcard cert was issued v. a standard FQDN cert.  However, I'm not sure 
there's much more risk to end certificate requester - they still know 
everything that's been issued for their domain.  It certainly doesn't make life 
easier for the CT operator or CA, but it gives important information to the 
relying parties looking at certs.  If they look up a cert in the CT log, 
they'll be able to easily identify if the entire domain is secured by the same, 
logged cert.  

Jeremy

-----Original Message-----
From: Trans [mailto:[email protected]] On Behalf Of Daniel Kahn Gillmor
Sent: Friday, January 30, 2015 3:07 PM
To: trans
Subject: Re: [Trans] [trans] #54 (rfc6962-bis): Simplify name redaction

On Fri 2015-01-30 15:40:38 -0500, Jeremy Rowley wrote:
> Although I just realized you're probably wondering if you can 
> substitute "?" of "*" - so *.example.com would become ?.example.com.
>
> It's a good question. I think there's real value in knowing whether a 
> wildcard cert was issued over a non-wildcard.  I'd actually like the 
> rfc to say you can't substitute ? for a wildcard character as you're 
> essentially substituting ? for an unlimited number of based domain 
> names.

But isn't that kind of the point?  If i control foo.example, and i'm monitoring 
the logs to make sure no one impersonates anything in my zone (including the 
apex), then i'm looking for all of the following literal matches as suffixes:

 foo.example
 ?.example
 *.example

In particular, of course, i'm looking for entries in the logs that are 
associated with public keys not under my control (i can ignore my own entries).

Note that the first entry there also covers literal strings like:

 ?.foo.example
 *.foo.example

If i find anything in the logs matching those suffixes, i know that my domain 
name may have been hijacked, because a matching form could be there.

It wouldn't matter to me whether it was being hijacked by someone with a cert 
that says "foo.example" instead of "*.example", or "*.foo.example"
instead of "www.foo.example" -- either one is sufficient to raise an alarm, and 
(because of the redaction) i wouldn't know if "?.foo.example"
was issued to "www.foo.example" or "bar.foo.example" anyway.

I still don't see why we shouldn't allow redaction of a wildcard label.
What does this restriction defend against?  how does it make life easier or 
better for any of the players in the CT ecosystem?

        --dkg

_______________________________________________
Trans mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/trans

Reply via email to