[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: lutil_entropy()

On Tue, 26 Oct 1999, Kurt D. Zeilenga wrote:

> I commited what routine that should be good enough for one
> type use keys (such as those needed for DIGEST-MD5
> implementation).
> 
> The code uses, if available, /dev/urandom (or equivalent).
> Otherwise is fallbacks to an algorithm suggested Phil Karn in
> "practical randomness" discussions.  Paraphasing:  The worse
> case is if that each 16-bytes is a MD5 hash of a counter.
> As long as MD5 is a cryptograhically strong hash, this
> should be fairly resistent to attack.
> 

- It doesn't compile on Solaris 2.5.1 ( see ITS 337 ). 

- Booker C. Bense