cacert-sysadm AT lists.cacert.org

Subject: CAcert System Admins discussion list

List archive

Re: (something about roots)

From: dirk astrath <dastrath AT gmx.de>
To: cacert-board AT lists.cacert.org,cacert-sysadm AT lists.cacert.org
Cc: Wytze van der Raay <wytze AT deboca.net>,Nick Bebout <nick AT bebout.net>
Subject: Re: (something about roots)
Date: Mon, 14 Feb 2011 13:00:00 +0100

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 hiya,

since every change has something to do with software, i asked ulrich to put this on the agenda for our weekly software-telephone-meeting (tuesday 21:00 UTC).

our plan is NOT to work out a solution but to find out, which changes, tests etc. have to be done ...

to be done, if we want to install completely new roots, resign class 1, recreate class 3, stop issuing class 3, ... ....

this should make the decision for the next board easier, because the effort, circumstances etc. will then be available ...

have a nice day

Ian G <iang AT cacert.org> schrieb:

On 14/02/11 10:19 PM, Wytze van der Raay wrote: > On 13.02.2011 00:51, Nick Bebout wrote: >> Mozilla has announced that they will disable the use of MD5 as a hash >> algorithm in all of their products on June 30, 2011. Our class 3 >> intermediate root uses MD5, thus I believe we should stop issuing >> certificates off of that root as soon as possible. I have filed the >> following motion. >> >> Nick >> >> m20110212.1 >> Discontinue issuing certificates from Class 3 root >> >> RESOLVED, that effective as of the date of the close of this vote (or as >> soon as this change may be implemented), CAcert shall no longer issue >> certificates from the Class 3 root. >> >> Background information: >> >> Our Class 3 root uses MD5 as a hash algorithm. >> >> June 30, 2011 – Mozilla will stop accepting MD5 as a hash algorithm for >> intermediate and end-entity certificates. After this date software >> published by Mozilla will return an error when a certificate with an >> MD5-based signature is used. >> >> >> Due: 2011-02-19 23:59:59 UTC > > > This seems a rather over-drastic step to me. With similar arguments you > might as well decide to stop issuing all CAcert certificates :-( Yes. > Someone please correct me if I'm wrong, but isn't it possible to simply > recreate the published Class 3 root certificate with a SHA1 digest rather > than an MD5 digest? The public and private keys don't change under that > transformation, so all issued certs (and new issued certs) can still be > validated properly. We only need to urge people to replace the old > MD5-signed root cert by the new SHA1-signed one in their browsers, e-mail > clients, or web servers. That would seem a much more positive strategy > to me than simply discontinuing issuing certs. My understanding is that re-signing is not possible under PKI. It's not the crypto, but complicated things inside the x.509 rules that are invoked by browsers (etc) to reject stuff that looks odd. In practice, the better measure would be to make a new class 3 subroot entirely. I suppose the occam's razor on this is that if this were done such that we were all agreed that the result was a short term issue, and therefore the roots remained "fail" then ... go ahead. Another possibility would be to re-self-sign the current top root using SHA1. Or SHA256? It matters not if that one changes because one either uses the old root or the "new-old" root? And then roll a new subroot signed by the "new-old" root? Either way .. it would be good to test this with multiple browsers, etc to make sure that they didn't reject it in strange ways? (I'm just whiteboarding here, not really thinking it through...) iang

[no subject], Nick Bebout, 02/12/2011
- Re:, Wytze van der Raay, 02/14/2011
  - Re: (something about roots), Ian G, 02/14/2011
    - Re: (something about roots), dirk astrath, 02/14/2011
      - Re: (something about roots), Guillaume ROMAGNY, 02/14/2011
- Re: MD5, Philipp Gühring, 02/16/2011