home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: alt.security
- Path: sparky!uunet!mcsun!sun4nl!fwi.uva.nl!casper
- From: casper@fwi.uva.nl (Casper H.S. Dik)
- Subject: Re: passwd security check
- Message-ID: <1992Jul24.074114.27345@fwi.uva.nl>
- Sender: news@fwi.uva.nl
- Nntp-Posting-Host: adam.fwi.uva.nl
- Organization: FWI, University of Amsterdam
- References: <1992Jul22.190827.30077@iitmax.iit.edu> <1992Jul22.221222.6185@Princeton.EDU> <1992Jul23.092715.1@zodiac.rutgers.edu> <14mgkaINN1uq@moe.ksu.ksu.edu> <12431@inews.intel.com>
- Date: Fri, 24 Jul 1992 07:41:14 GMT
- Lines: 39
-
- adam@gomez.intel.com (Adam Margulies ~) writes:
-
- >In article <14mgkaINN1uq@moe.ksu.ksu.edu> rjq@phys.ksu.edu (Rob Quinn) writes:
- >>In <1992Jul23.092715.1@zodiac.rutgers.edu> leichter@zodiac.rutgers.edu writes:
- >>]One thing it's important to remember is that there are many passwords that
- >>]hash to the same value. Even if you and I have the same salt and the same
- >>]hash value, it doesn't mean we chose the same password - though it DOES mean
- >>]that either of our passwords will work on either account.
- >>
- >> Can you provide an example? Or is there some mathematical proof? This question
- >>has come up a lot before, and there have been answers on both sides, but no
- >>proof either way that I have seen.
-
- >DES has a theorectical weakness in that for any key there are exactly 7 other keys that will
- >crypt to the same string. I.E. if your password is "batman!" there exist seven other keys which
- >are not "batman!" that will allow access to your account. Fortunately they are almost certainly
- >extremely strange strings like "@gW #s(u", and not likely to match a human generated password.
-
- What's true of DES isn't necessarily true of crypt(3).
- Remember that crypt runs (modified) DES 25 times and that your password is
- used as the key. After the first encryption the eight keys will have encrypted
- to the same value. But the resulting encrypted string is encrypted again,
- resulting in 8 different values. I think that it is pretty much an open
- question whether there are password/salt combinations that yield the
- same encryption string. For this property to be applicable to crypt(3)
- you must have a 0 (..) salt and 2 of the 8 keys must be self-decoding.
-
- >Another interesting thing about DES is that there are 8 keys that crypt to a string of
- >all spaces and there are even keys that when encrypted reproduce themselves in the
- >crypted output. Weird.
-
- Crypt what to all spaces? The standard crypt(3) cleartext?
-
- And try to keep your lines 80 columns.
-
- Casper
- --
- | Casper H.S. Dik
- | casper@fwi.uva.nl
-