Ted Unangst wrote:
> 
> Unfortunately, our original assessment was not entirely correct. Some of the
> features exploited by FREAK were not deleted until after 5.6, although this
> was not known until testing tools became available. We've corrected libssl
> by backporting the necessary changes to 5.6.

A little background to explain how the mixup happened.

Jul 9, 2014: jsing removed most of the export cipher handling.
http://marc.info/?l=openbsd-cvs&m=140490515728102&w=2

Dec 6, 2014: jsing removes temporary RSA key support.
http://marc.info/?l=openbsd-cvs&m=141787588308711&w=2

The first of these commits was the one in most developers mind when asked,
When was export cipher support removed? The second diff seemed like a cleanup
at the time.

Jan 8, 2015: OpenSSL advisory announcing CVE-2015-0204
https://www.openssl.org/news/secadv_20150108.txt

Looking at the OpenSSL changes, the affected code had already been deleted
from current in cvs. From the description of the bug, EXPORT_RSA, the earlier
of the two commits above is what stuck in most developer's minds.

It was an oversight to assume that because the bug was already fixed, we
didn't need to look back and verify *when* it was fixed. And, at the time,
this was a low priority issue. It was checked off as done.

Then it was renamed to the FREAK attack and the consequences demonstrated.
The original description "A server could present a weak temporary key
and downgrade the security of the session." did not mention that the server
in question could be a malicious man in the middle, not the end point server.

The natural question was if we fixed it. Of course, it was already checked off
as done. I even went back to verify the dates when FREAK was announced, but I
only double checked the date of the first commit, forgetting about the second
commit.

Reply via email to