Re: Hashed password secure?

From: Joe C (jkc8289_at_bellsouth.net)
Date: 06/30/04 

Date: Wed, 30 Jun 2004 01:02:44 -0400

"Matthijs Hebly" <heeb@iname.com> wrote in message
news:bimEc.26162$HY4.23980@amsnews03.chello.com...
> Tom St Denis wrote:
> > Question remains: Why are you inventing your own crypto? If it's for
fun
> > you realize that you shouldn't use it in the field right?
> Yes.
>
> > I mean as responsible adults we have to take best measures [even if they
turn out to
> > be wrong later on].
> I admit to be an absolute crypto-newbie, but that doesn't mean I'm
> stupid, and it doesn't mean I can't have interesting ideas. Of course, I
> *can* have stupid ideas like the rest of us... That's the reason why I'm
> posting here and asking your professional opinions. I want to
> understand, learn, know. Like the rest of us.
> I was investigating the topic of "salt" some time ago, doing research on
> the internet, as I think it is important for the application I'm
> writing. I noticed that the salt is always stored, and I started
> wondering: why is this? Why store the salt, if you can actually use a
> *non-persistent* salt to your advantage? Namely in making it very much
> harder for an attacker to do a succesful dictionary attack.
> Maybe my idea is rediculous. But until now, I'm not convinced that it
> is. I think is could be, in some cases, a useful idea. Plz convince me
> otherwise!
>
> Thanx, to all of you here,
>
> Matthijs.
>
> P.S. I'm in europe, where it is *very* late, so: good night to you all :-)

G'd night, Mathis...

I, too, have written some crypto software that no one will ever be
interested in ;-) I've basically tried to learn the
vulnerabilities/problems that arise and circumvent them, without really
learning that much about other methods that have been used to defeat them.
I'm a moron by the sci.crypt standards, I'm sure... What I do is...

seed a PRNG with a hash of the (password + salt) + hash of the file to be
encrypted. Then I step the PRNG forward for ~1-second on the local machine,
before using the PRNG output to encrypt the file. I then prepend the
encrypted file with the unencrypted salt + unencrypted number of iterations
that were used to seed the PRNG + an encrypted (using a second PRNG seeded
with the hash of (pass + salt) and encrypted file hash. The reciever then
uses the hash of (password + salt) and the hash of the encrypted portion of
the file to seed a PRNG used to unencrypt the encrypted hash sent with the
file. He then seeds a second PRNG with the hash of (password + salt) and
hash of unencrypted file, which he steps forward the number of iterations
that required ~ 1 second on the encrypting platform, and uses the output of
the second PRNG to unencrypt the original file.

My rational/logic is that a pre-attack cannot be mounted for any given
message/password pair...and that each brute-force attack try will take ~1
second equivalent computation for the encrypting machine...which should
yield ordinary-length passwords sufficient iff the attacking computer has
similar (order of magnitude) computing capabilities when compared to the
encrypting machine.

Your method is to me an interesting case...where you force the unencrypting
computer to perform a brute-force of a particular difficulty. I see one
potential problem with your method as I understand it...you have
low-resolution control...the thing you are brute-forcing will be 0, 1, 2 3,
4, 5, 6, 7, 8...etc bytes long...whereas I have better resolution with the
number of iterations I choose.

Cheers

Relevant Pages

  • Re: rsa implementation question
    ... > There is a notion of blocks in many public-key ciphers, ... It's not about decrypting to sign, encrypting to ... as it would mean that you'd have to find hash collisions. ... I generate a signature for a string "some string" with SHA. ...
    (comp.lang.python)
  • Re: Decrypting on different platforms
    ... > wprintf, ... > // Hash in the password data. ... > encrypting Unicode.) ...
    (microsoft.public.platformsdk.security)
  • Re: ADVERT: Secure communications
    ... Hash: SHA1 ... security analysis on it in existence were by its author. ... I'm assuming this means RSA without hybridizing with a symmetric cipher. ... than encrypting a whole message block-by-block with a 2048-bit ...
    (sci.crypt)
  • Re: Decrypting on different platforms
    ... The code is lifted from the eMbedded Visual C++ help for encrypting and ... I have looked at the hash and found the hash was different ... > "Frank Perry" wrote in message ... >>> I have been using the sample code for encrypting and decrypting files. ...
    (microsoft.public.platformsdk.security)
  • Encrypted Files
    ... encrypting the original key. ... >I exported out a .pfx file of the security certificate ... unencrypt the files I had previously unencrypted. ...
    (microsoft.public.windowsxp.security_admin)