Viktor Dukhovni <[email protected]> (Do 08 Mai 2014 17:23:43 CEST):
> On Thu, May 08, 2014 at 04:09:21PM +0100, Jeremy Harris wrote:
> 
> > Since a simple expansion can call a custom ACL, which might be
> > given visibility of that certificate, we can get as complex
> > as we want.   It would be able to do the checking of
> > SN/SAN against the peer name (hmm.  Which one?  The MX?) if
> > we didn't get around to hardwiring that (and apparently if
> > we did hardwire it we'd need an exception for DANE... is that
> > true?)
> 
> You'd be inviting non-experts to mess with some of murkiest and
> dragon-infested parts of PKI.  Don't go there.
> 
> In DANE for SMTP the (RFC 6125 jargon) "reference identity" for
> the server is a set of up 3 names.  Each of which may match any of
> the server's subjecAltName DNS names with support for single label
> wildcards, but this is ONLY for certificate usage 2 and not 3.
> With usage 3 non-leaf certs are ignored, as is the cert expiration
> and trust chain.
> 
> DO NOT delegate this logic to users, they will get wrong, causing
> both security and interoperability issues.

I fully agree.
I'm thinking of something about that way (not sure, if I
got the ${acl{}} feature right…

    begin acl

        acl_check_dane:
            accept verify = dane
            deny


    begin transports

        remote_smtps:
            driver = smtp
            hosts_require_tls = *
            tls_continue = ${acl{acl_check_dane}}

This give the user the power to implement whatever he wants
as the condition.

-- 
Heiko

Attachment: signature.asc
Description: Digital signature

-- 
## List details at https://lists.exim.org/mailman/listinfo/exim-dev Exim 
details at http://www.exim.org/ ##

Reply via email to