[PLUG] Five year old unpatched vulnerable in code? That's OK, we're Seagate.

Tim tim-pdxlug at sentinelchicken.org
Fri Mar 6 18:52:28 UTC 2015


>    Apparently a lot of ancient vulnerabilities are still present in current
> software releases. Yesterday's news was that Google and Apple are among the
> Web sites still using the 512-bit encryption madated by the feds as
> export-acceptable in the early- mid-1990s. Apparently, a moderately competent
> cryptanalyst with the power of ~15 servers (which can be accessed at AWS and
> other cloud computing providers) can extract the key in about 7 hours. Oops!
> For whatever the reason, they (and a bunch of others) never upgraded their
> encryption to 1024-bit or 2048-bit. Chrome is apparently not vulnerable, but
> Safari is.
> 
>    I don't have the URLs to the articles available or I'd post them here.


Well, it's not *quite* that blatant.  You see, SSL/TLS supports many
different crypto algorithms grouped into "cipher suites".  When the
server and client do their handshake, they agree on the "most secure"
or "most preferred" cipher suite that they both support.  So weak
ciphers *shouldn't* be used, even if both sides support it and an
attacker modifies the handshake.  The handshake modification is
supposed to be detected and rejected if that happens.

I believe this gimmick-y named FREAK vuln is simply a cipher suite
downgrade attack, which allows one to force a weaker cipher suite to
be used.  Cipher suite downgrade vulnerabilities have happened many
times before in many SSL/TLS libs.  If you disable the legacy/weak
cipher suites, then you avoid the issue, but the downgrade is still
the main vulnerability.  SSL/TLS is designed to prevent it, afterall.

In summary, yes, you could argume that those library authors should
have disabled the weak cipher suites a long time ago, but there are
always pressures to minimize compatibility problems.  If the downgrade
vuln didn't exist, it shouldn't be much of a risk to leave them
enabled, provided you have stronger preferred suites, since they'll
almost never be used.  But then the downgrade did happen, didn't it?

tim



More information about the PLUG mailing list