Re: [keyassure] crypto hash alg deprecation is a myth

[Phillip Hallam-Baker <[e-mail]>]

On another algorithm issue, certain people were waiting for CAs to
support the new algorithm before implementing them in application
software.

By support, they meant issue certificates that would break with their software.


The reason that there is concern about algorithm support in protocols
is that it takes a very very long time to get changes through the
system. It can take five years to persuade vendors to make changes and
then another ten for the old software to work through the system.

Auto update helps, but that may only prove to be a temporary change.
There are enough applications out there that do auto update very badly
that people are being taught to turn them off.


One of the very few signals available that is listened to is to stop
supporting old algorithms in new protocols.


On Thu, Mar 3, 2011 at 2:46 AM, Rob Stradling <[e-mail]> wrote:
> On Thursday 03 Mar 2011 05:09:15 Martin Rex wrote:
>> Phillip Hallam-Baker wrote:
>> > Further, even though we stopped using MD5 in the mid 90s
>>
> <snip>
>> Firefox happily verifies md5WithRsaEncryption signatures on TLS server
>> certs.
>
> Not for much longer.
>
> https://wiki.mozilla.org/CA:MD5and1024 says:
> "    * 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.
>          o This change is being tracked in
> https://bugzilla.mozilla.org/show_bug.cgi?id=590364"
>
> <snip>
>> MSIE 8 on Windows 7 (and prior) will happily talk to TLS servers
>> using certs signed not only with md5WithRsaSignature, but also
>> md2WithRsaSignature and even (**cough**) md4WithRsaSignature!
>
> http://technet.microsoft.com/en-us/library/cc751157.aspx#EMAA says:
> "In the event of an imminent MD5 pre-image attack
> Microsoft may update Windows to reject all MD2, MD4 or MD5 end-entity and
> subordinate CA certificates when it has reasons to believe that successful MD5
> pre-image attacks are imminent."
>
> Perhaps somebody should ask them to consider switching off MD2 and MD4 sooner
> than that.
>
> <snip>
>> (If it is possible to disable md5WithRsaEncryption signature verification
>>  in Firefox 3.5 -- then it is sufficiently well hidden that I don't
>>  see it.)
>
> https://bugzilla.mozilla.org/show_bug.cgi?id=590364 suggests setting the
> following environment variable:
>
> NSS_HASH_ALG_SUPPORT=-MD2,-MD5
>
> <snip>
>
> Rob Stradling
> Senior Research & Development Scientist
> COMODO - Creating Trust Online
>



-- 
Website: http://hallambaker.com/