Re: [PEN-TEST] Oracle USER$ password hashes

[Pete Krawczyk <petek () BSOD NET>]

Date: Thu, 9 Nov 2000 14:04:38 -0800
From: "Edwards, Steve" <sedwards () SEDWARDS COM>
Subject: Re: Oracle USER$ password hashes

}Here are some example pairs. Only the name is changed, the password is
}the same -- "ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ" (30 Z's.)
}
}NAME                           PASSWORD
}------------------------------ ------------------------------
}ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ C829CE9D9DAE5F3B

I have also done this with one Z to 30 Z's in the username.  Note the 30 Z
username/password hash is the same as in the example above:

NAME                           PASSWORD
------------------------------ ------------------------------
Z                              BDA7BD93C02D9685
ZZ                             9B82D0C5D7238869
ZZZ                            4D249563D9CEB1DA
ZZZZ                           36E5D13056E74735
ZZZZZ                          7551931A3868EC4B
ZZZZZZ                         9E1813B857CBC46B
ZZZZZZZ                        74214B2B5710D4DB
ZZZZZZZZ                       92E348C06ACDB1DF
ZZZZZZZZZ                      093BD63A47C95704
ZZZZZZZZZZ                     84FE29884943296F
ZZZZZZZZZZZ                    6A2B1094F05FF16D
ZZZZZZZZZZZZ                   778E4F7D76A6D44B
ZZZZZZZZZZZZZ                  DB5831A4530E8EB1
ZZZZZZZZZZZZZZ                 2D07D64E854F2286
ZZZZZZZZZZZZZZZ                0986719BAFB7E58E
ZZZZZZZZZZZZZZZZ               62901F14E96D651A
ZZZZZZZZZZZZZZZZZ              509902F4498B917A
ZZZZZZZZZZZZZZZZZZ             B3F8311B73473501
ZZZZZZZZZZZZZZZZZZZ            0C40E63F2E9B6C67
ZZZZZZZZZZZZZZZZZZZZ           A1C5A0833056DE3A
ZZZZZZZZZZZZZZZZZZZZZ          47B97C7B1FAB25DD
ZZZZZZZZZZZZZZZZZZZZZZ         92FE731355463CED
ZZZZZZZZZZZZZZZZZZZZZZZ        99465277D42480E5
ZZZZZZZZZZZZZZZZZZZZZZZZ       A13A46DEE200FC4D
ZZZZZZZZZZZZZZZZZZZZZZZZZ      CFB8E1170DC38C6E
ZZZZZZZZZZZZZZZZZZZZZZZZZZ     2DE7D2C11006D90C
ZZZZZZZZZZZZZZZZZZZZZZZZZZZ    E25994E69997E29D
ZZZZZZZZZZZZZZZZZZZZZZZZZZZZ   50FB00980AFFAB8C
ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ  93D1C5948CDC40FC
ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ C829CE9D9DAE5F3B

So, apparently only the username and password are used in the calculation
of the encrypted password hash.  If anything else were used, the hash
would perhaps differ, since my database is different from the other one
used to create the same hashes.

I've also done some off-the-cuff calculations; if there are only the
characters 0-9A-F contained within, there are only 64 bits of
entropy.  There are 30 chars username, 30 chars password.  With 64 bits of
entropy max, this creates a problem.  Assuming no punctuation in the
passwords (and punctuation is allowed; I'm making math easier) you have
   36^30 = 48873677980689257489322752273774603865660850176
combinations of passwords alone - just for the 30 character passwords.
And 64 bits of entropy is many orders of magnitude less than that.  So
it's very possible that for a single username, many cleartext passwords
would work.

Why 36, not 62, for A-Za-z0-9?  Usernames and passwords are apparently
case-insensitive.  The username is always uppercased (although in a test
database, I have a lowercase username somehow and the hash is the same as
the uppercase username right next to it).  Also, check this out:

create user ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ identified by zzzzzzzzzzzzzzzzzzzzzzzzzzzzzz;

NAME                           PASSWORD
------------------------------ ------------------------------
ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ C829CE9D9DAE5F3B

It's the same as the all-uppercase version.

-Pete K
--
Pete Krawczyk
  petek at bsod dot net or pkrawczy at uiuc dot edu
  http://www.uiuc.edu/ph/www/pkrawczy/