Re: Mistake in FORMAT; question about N.

[Colin Percival <cperciva@tarsnap.com>]

On 04/27/12 06:44, Dmitry Chestnykh wrote:
> 1.
> 
> I believe there's a mistake in FORMAT file. It says that the last HMAC is for
> data only, while in reality it includes the header as well:
> 
> 96+X    32      HMAC-SHA256(bytes 96 .. 96 + (X - 1))
> 
> must be
> 
> 96+X    32      HMAC-SHA256(bytes 0 .. 96 + (X - 1))

Yep.

> 2.
> 
> Regarding the N parameter in scrypt KDF.
> 
> The comment in crypto_scrypt.h says:
> 
>    The parameter N must be a power of 2 greater than 1.
> 
> The code accepts N == 1:
> 
>    if (((N & (N - 1)) != 0) || (N == 0)) {
>         // .. error ..
>    }
> 
> 1 & (1-1) == 0, so it passes the check.
> 
> My question is, should N be greater than 1 or >= 1?
> (I realize that in practice it should be much larger).

The documentation is right; the code is wrong.

Do you have a Tarsnap account?  This is worth a bug bounty. :-)

> 3.
> 
> I've implemented scrypt in Go:
> 
> KDF -- https://github.com/dchest/scrypt
> file encryption tool -- https://github.com/dchest/scryptutil

Cool!

> PPS Someone just dropped nuclear bombs on my cities. How do I bring people back
> to life?

That's what good backups are for. :-)

-- 
Colin Percival
Security Officer, FreeBSD | freebsd.org | The power to serve
Founder / author, Tarsnap | tarsnap.com | Online backups for the truly paranoid