# Re: Modulo instead of XOR in CFB

**From:** Gregory G Rose (*ggr_at_qualcomm.com*)

**Date:** 08/10/05

**Next message:**Jan Panteltje: "Re: Using The Internet To Store Data"**Previous message:**Scott \: "Re: Modulo instead of XOR in CFB"**In reply to:**Thomas: "Modulo instead of XOR in CFB"**Next in thread:**Sebastian Gottschalk: "Re: Modulo instead of XOR in CFB"**Messages sorted by:**[ date ] [ thread ] [ subject ] [ author ] [ attachment ]

Date: 10 Aug 2005 08:24:55 -0700

In article <42f9f6c0$0$8411$79720d31@newsreader.inode.at>,

Thomas <arot.ts@gmx.net> wrote:

*>In CFB mode usually the plain text is XORed with the encrypted queue
*

*>bit. However, I want encrypt only bytes below 0xFF and the cipher should
*

*>not contain any 0xFF.
*

*>Therefore I thought of replacing the XOR with (mod 255).
*

*>Are there any security concerns I might hav overseen?
*

Assuming that the generator is good, you'll

introduce a bias by doing this. That's because

the generator will generate 0 and 255 with equal

probability, both of which will encrypt the

plaintext the same way (in fact, by not changing

it). That means that a ciphertext byte is twice

as likely to be the same as the plaintext as it

is to be any other byte. This is a staggeringly

large bias, by crypto standards.

Greg.

-- Greg Rose 232B EC8F 44C6 C853 D68F E107 E6BF CD2F 1081 A37C Qualcomm Australia: http://www.qualcomm.com.au

**Next message:**Jan Panteltje: "Re: Using The Internet To Store Data"**Previous message:**Scott \: "Re: Modulo instead of XOR in CFB"**In reply to:**Thomas: "Modulo instead of XOR in CFB"**Next in thread:**Sebastian Gottschalk: "Re: Modulo instead of XOR in CFB"**Messages sorted by:**[ date ] [ thread ] [ subject ] [ author ] [ attachment ]