[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CALCETrWmCK-xRjMFtHRg4kSSz3m6N2MDSeOyv7NiJkz5xO1hSQ@mail.gmail.com>
Date: Mon, 26 May 2014 17:59:42 -0700
From: Andy Lutomirski <luto@...capital.net>
To: discussions <discussions@...sword-hashing.net>
Subject: Re: [PHC] Re: [Cryptography] The proper way to hash password files
On Mon, May 26, 2014 at 5:51 PM, Peregrine <peregrinebf@...il.com> wrote:
> Of course, one can simply use an HMAC construction with any normal password
> hashing function. If you have that in a dedicated hardware chip then you may
> gain security, otherwise one should assume that any attack capable of
> getting the hashed-password database is capable of getting any other secret
> data, like private keys. Some of the submissions explicitly support keyed
> password hashing, and with others any secure HMAC scheme should work. Using
> a strong password hashing scheme with a key protects weak passwords, and
> protects strong passwords even if the key is somehow leaked.
I think it's sad that, even on a crypto list, people still talk about
new uses of HMAC (that's a Hash-based Message Authentication Code) for
things that require properties beyond what a MAC provides and do not
require message authentication.
In this particular case, using HMAC-fancy-password-hash is extra
silly, because HMAC invokes the hash *twice*.
I agree that there's some value in using a keyed password hashing
function, and similar benefits can probably be obtained with a simple
composition of a password hash with a keyed PRF or even just a simple
encrypted password hash database.
--Andy
Powered by blists - more mailing lists