Re: Heartbleed, clients and Android

[Yves-Alexis Perez <corsac () debian org>]

On Wed, Apr 09, 2014 at 11:30:29AM +0200, Hanno Böck wrote:
> Hi,
>
> I was asking myself some questions and I think others with more insight
> into what heartbleed means may be able to answer quickly:
> How does this affect client software? The PoCs we see send some
> malicous payload to servers and get some memory dumps. That doesn't
> affect clients?

Yes, it does affect clients.
> Is this vulnerability exploitable by a Man-in-the-Middle in any way?
> Can someone send a package with a wrong authentication block inside an
> existing connection and therefore dump any memory from a client?

Yeah, anyone connecting to any TLS server using a libssl client is
affected. Most web browsers actually don't use OpenSSL but rather nss,
but other clients (like, for example, svn/git/etc., wget/curl etc.) are
affected if they are linked against OpenSSL.

> I think it's a very obvious question to be asked if we need to push all
> server users of openssl or ALL users of openssl.

All users, obviously.

People insist on server because that's the urgent problem. If NSS was
vulnerable, it'd be really bad because it's pretty easy to trick any
browser to open a connection to a random TLS website which would try to
steal stuff from the browser memory (although sandboxing might help).
For libssl clients though, I'm not sure it's that easy, but it's still
possible.

> Because the latter
> would include Android. We are all pretty aware that android updates are
> in large parts nonexistent.

I don't have much clue about Android, but I think I heard heartbeat was
disabled in Android, but I don't have a link right now. Also, I'm unsure
what actually use libssl in Android and what uses NSS.

Regards,
-- 
Yves-Alexis Perez

Attachment: signature.asc
Description: Digital signature