[472] in Kerberos-V5-bugs
DES
daemon@ATHENA.MIT.EDU (Theodore Ts'o)
Tue May 17 14:35:20 1994
Date: Tue, 17 May 94 14:35:07 EDT
From: tytso@MIT.EDU (Theodore Ts'o)
To: pau@watson.ibm.com (Pau-Chen Cheng)
Cc: krb5-bugs@MIT.EDU
In-Reply-To: "[466] in Kerberos-V5-bugs"
Date: Thu, 5 May 1994 18:34:36 -0400
From: pau@watson.ibm.com (Pau-Chen Cheng)
The MIT code failed during "verify" CBC test. It passed the ECB test.
Also, I fed both AIX and MIT routines an 8-byte input to further test
the ECB mode; using the same key ("0123456789abcdef") as in the MIT
verify program. The AIX routines can encrypt and decrypt and recover
the input. The MIT routines generate a different cipher than AIX
routines do.
I believe it must have been something worng with my built process.
However, since I am not a DES expert. I would like to make sure
if anybody else has reported a similar bug.
Yes, this sounds like a porting/building problem. No one else has
reported a similar bug, though. The obvious thing to check is whether
the little-endian vs. big-endian configuration was set correctly.
- Ted
