Re: [hybi] AES-128-CTR not much safer, but not fast either

[Willy Tarreau <[e-mail]>]

On Wed, Jan 12, 2011 at 05:56:28PM +0900, Yuta Kitamura wrote:
> Hi,
> 
> On Mon, Jan 10, 2011 at 9:09 AM, Willy Tarreau <[e-mail]> wrote:
> 
> > willy@pcw:~/c$ time ./aes-128-ctr-get
> > Found the 'GET\n' pattern on the wire after 1608425803 bytes
> >
> >
> The probability that you get some four-byte data from a random byte sequence
> is 1/(256^4) = 1/4294967296 for each byte. Please note that the order of
> 256^4 is almost same as the above number. This means that, statistically,
> it's quite possible to find a four-byte 'GET\n' sequence (and any other
> four-byte sequence) in a few gigabytes of random bytes.

That's what I've been explaining from the beginning !

> I think your experiment has just shown that the output of AES-128-CTR is
> random enough.

I agree it's random enough, just as the simple 32-bit random XOR is ! And
that was my point : there is no vulnerability that would remain with the 
simple XOR that the AES would solve since in the end, we're down to just
a 32-bit randomness.

Willy