[7473] in Kerberos

home help back first fref pref prev next nref lref last post

des_string_to_key() anomoly?

daemon@ATHENA.MIT.EDU (Larry J. Hughes Jr.)
Thu Jun 13 01:42:44 1996

To: kerberos@MIT.EDU
Date: 13 Jun 1996 00:58:56 GMT
From: hughes@bodhi.it.iupui.edu (Larry J. Hughes Jr.)

I've built the latest CNS V4 distribution on four different
architectures: two LSBFIRST (DEC MIPS and DEC AXP) and two MSBFIRST
(Sun Sparc and HP 9000).  

I've noticed that the des_string_to_key() function produces different 
results on these two groups of systems.  That is to say, the two LSBFIRST 
systems produce the same hash for the same password, and the two MSBFIRST 
systems produce the same hash for the same password, but the password
hashes don't match between the LSBFIRST and MSBFIRST architectures.

Is this to be expected, or am I missing something obvious?  I've
tracked through the provided DES code and have discovered a few
places where the byte ordering obviously differs, but I would have
expected the hash algorithm to work identically across all systems.

(I have an application for which I'd like to use the des_string_to_key
function, but to do so I need it to work homogeneously.)

---
Larry J. Hughes, Jr.    hughes@indiana.edu
Indiana University      http://copper.ucs.indiana.edu/~hughes/
Author, "Actually Useful Internet Security Techniques," ISBN 1-56205-508-9

home help back first fref pref prev next nref lref last post