🚀 go-pugleaf

cme@ellisun.sw.stratus.com (Carl Ellison) writes:

> >I need a 32-bit cryptographically strong checksum.

> This is a contradiction in terms.

> If all you keep is 32 bits, then it's too easy to find a collision for any
> given checksum (ie., by brute force search).  Therefore, it's not
> cryptographically strong.

It seems that I didn't express my question well enough. By
"cryptographically strong" I meant that there is no other way to forge
it than brute force.

Anyway, several people replied by privete e-mail. Their opinions can
be grouped into two groups:

1) That any consecutive 32 bits of the result are equally good.

2) That I should use a 32-bit hash function of the result - like a
CRC-32, or divide the result into 32-bit chunks and XOR them together,
and so on.

Interestingly, everybody who supported opinion 2) mentioned in their
message "I am not a crypto expert". :-)

Well, I guess that I'll use the 32 least significant bits of MD4, MD5,
or SHA.

For those of you that are curious, I don't need this checksum for
strong authentication purposes. I am writing a utility that creates
"maps" of the constant areas of the different computer viruses and
keeps the checksums of these areas. The checksums are further used for
exact virus identification. I just wanted to make difficult enough for
the virus writers to forge them, i.e. to create a virus which has the
same map with the same checksums of the constant areas, but the virus
is different. 32 bits are strong enough for my purpose.

Again, thanks to everybody who responded.

Regards,
Vesselin
-- 
Vesselin Vladimirov Bontchev          Virus Test Center, University of Hamburg
Tel.:+49-40-54715-224, Fax: +49-40-54715-226      Fachbereich Informatik - AGN
< PGP 2.0 public key available on request. > Vogt-Koelln-Strasse 30, rm. 107 C
e-mail: bontchev@fbihh.informatik.uni-hamburg.de    D-2000 Hamburg 54, Germany