From: David x Callaway <supersaurus@...>
Date: 2007-03-26T20:44:02+09:00
Subject: Re: Ruby and Cryptography

thanks, that was very illuminating, see comments below.

... snip ...

>> 1) is pkcs5_keyivgen() the way to go rather than hashing  (password 
>> string +  salt) and then using that as the key with 
>> OpenSSL::Cipher::Cipher as EzCrypto does?
> 
> Maybe.  In most cases the pkcs5_keyivgen method is not pkcs5 compliant. 

I'm running ruby 1.8.5 and comments pertain to that source.

pkcs5_keyivgen eventually calls down to  EVP_BytesToKey, passing the
passphrase, salt (must be exactly 8 bytes if given), iteration count
(defaults to 2048) and digest;  it returns the key and IV.

the man page for EVP_BytesToKey says:

    If the total key and IV length is less than the digest length and
    MD5 is used then the derivation algorithm is compatible with
    PKCS#5 v1.5 otherwise a non standard extension is used to derive
    the extra data.

    Newer applications should use more standard algorithms such as
    PKCS#5 v2.0 for key derivation.

however there is no suggestion about how to get v2.0 output.

> You must be using MD2, MD5, or SHA1 with RC2 or DES.  Using any other 
> cipher (like AES) will generate your key/iv in an OpenSSL specific 
> format.
> 
> You probably don't want to use your own key generation method since you 
> are fairly likely to make design mistakes.

I definitely *don't* want to do this, hence I was trying to figure out
how to match "enc".

> 
> Ideally you would use PKCS5 v2.  Unfortunately the ruby OpenSSL module 
> doesn't have a hook into any of the PKCS5 v2 password generating 
> functions.

right, see comment above.

> Below is a pure ruby PKCS5 v2 password -> key method.

thanks, but I don't want to get ahead of the ruby openssl library.  of
course this *does* raise the question of how it will eventually be
incorporated.  it wouldn't be a nice thing if the addition caused all
my encrypted files to be inaccessible.

... snip ...

> 
> If you are designing your own encrypted storage format there are a
> number of things you should do (this is not an exhaustive list of
> everything you should do, just a few pointers).

right, but I don't want to do this at all ;-).

... snip ...

> 
> Test, test, test, test and test.
> 
> There's nothing worse than finding out that you used the wrong
> variable for the key and now everything is encrypted with the same
> key.  Or finding out later than you didn't store the key properly
> and can't retrieve anything.

right, that's precisely why I was attempting to test against openssl
enc.

... snip ...

> 
> Here's a quick and dirty bit of code to encrypt/decrypt the openssl 
> program output.

the key in the code you provided, besides the obvious biz with the
header, is the iteration count is 1, something I wouldn't have
guessed.  inspecting the openssl src for "enc" shows the same (I
should have gone there first).

the other thing I didn't figure out by messing around with "enc" is if
you don't provide it a salt, and if you don't specify -nosalt, then it
will do as you did, provide a random salt string when encrypting.  I
understand the idea of salt, but specifying it on the command line
each time didn't make sense to me, and now I see that's because it
doesn't make sense ;-).

... snip ...


-- 
Posted via http://www.ruby-forum.com/.