phc-discussions - Re: [PHC] Why protect against side channel attacks

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Thu, 25 Jun 2015 12:18:24 +0100
From: Peter Maxwell <peter@...icient.co.uk>
To: "discussions@...sword-hashing.net" <discussions@...sword-hashing.net>
Subject: Re: [PHC] Why protect against side channel attacks

On 25 June 2015 at 07:25, Solar Designer <solar@...nwall.com> wrote:

> On Thu, Jun 25, 2015 at 12:35:42AM +0100, Peter Maxwell wrote:
> > Lets assume there is
> > i bits of global secret data, s, and j bits of per-password secret data,
> h,
> > say.  The question then becomes: how much of s and h you can determine in
> > observing the side-channel for a single PDF calculation
>
> With proper design, where the secrets are passed through a fast
> cryptographic hash first (that is not itself susceptible to side-channel
> leaks), the answer to your question above is:
>
> Essentially none, as long as i and j are large enough (e.g. 128 bits
> each) and s and h are cryptographically random.
>
> In fact, to fully defeat the attack, it is sufficient to have s or h;
> it is not necessary to have both.  (In practice, it may be helpful to
> have both for other reasons.)
>

Not sure if we're talking at cross purposes here, or that having had a
night's sleep I'm a bit more awake, but I think the algorithm would still
be in trouble.

If you pre-hash with constant time, say, global secret s and per-password
data, h, to create, q, say then the access pattern will be a function of
both s and h.  Remember that s is constant and h is picked from a space of,
around, 2^40 (or some reasonable reflection of password strength.

The attacker would proceed by first determining, q, in the same manner as a
normal side-channel attack and then brute-force h against result
intermediate result q.

(you can of course apply some per-password secret key, k, say which would
ameliorate these concerns but that's essentially a secret salt)

>
> > how many calculations, n, per hash you can observe
>
> This becomes irrelevant.
>

Statistically, it's rather important in the above scenario, at the moment
at least.

>
> > and, obviously i and j.
>
> Yes.  At least one of these must be large enough.
>

I would posit that j needs to be large enough when any data-dependent
timings are involved, using my argument above.  That's the original
problem: making people choose better passwords.

Or one must start storing per-password secret data too, which is perfectly
feasible to be fair and might produce a nice middle ground here.

Content of type "text/html" skipped