Other Directory Sites: SeekWonder | Directory Owners Forum
The following archive was created by hippie-mail 7.98617-22 on Fri Aug 21 1998 - 17:20:04 ADT
Hamdi Tounsi (hamdi.tounsi@ati.tn)
Thu, 2 Jul 1998 11:05:54 -0100 (GMT+1)
>
> Hamdi Tounsi writes:
> >do you think that this schema is acceptable ?
>
> NO!
>
> In fact Bruce has already said in this thread why not. There is only a
> small probability that these modified values will even be used.
>
> Greg.
>
thanks greg. but as i said in my previous message. i don't care about
the key len if it is > 128. my question is deeper than that
i'm not trying to extend the max key len from 448 to 576 by
using the last 2 long words.
this is precisely my question :
is there a security risk in xoring the last 2 long words with zeros (i.e
leaving them unchanged) instead of xoring them with some key bytes.
1- in the normal blf i have :
a 128-bit key, cyclically xored though the p-array
the last two longs in the p-array WILL BE xored with some key bytes
2- in the schema i'm using, it's the same, except that i have
a max len key which is :
128-bit hash from some password + zeros until 448 bits
my question (again !) is :
can i assume an extra long key and xor the last 2
long words in the p-array with zeros ? does xoring the last 2 longs with
zeros and not with some key bytes introduces some risk (even small) ?
in this case, it's true that i'm using a 576-bit key, with unknown 128-bit
and zeros until 576. But 14 elements out of 18 are known (as they were left
unchanged by the first xor loop in the key setup) before the key setup
goes into the Pbox initialization
i hope i was more precise this time.
Hamdi
The following archive was created by hippie-mail 7.98617-22 on Fri Aug 21 1998 - 17:20:04 ADT