Re: [TLS] Prohibiting RC4 Cipher Suites

["Paterson, Kenny" <[e-mail]>]

Andrei,

Your intro says:

"Recent cryptanalysis results [ALF] exploit biases in the RC4 keystream to
recover early portions of plaintexts."


The attacks can recover repeated plaintext from ANYWHERE in the plaintext
stream, so they are more flexible in application than your text suggests.

Another (better?) link for the attacks by AlFardan et al. is
www.isg.rhul.ac.uk/tls. The "official" USENIX link, which should be
long-lasting, is:

https://www.usenix.org/conference/usenixsecurity13/security-rc4-tls


Best wishes

Kenny

On 21/08/2013 13:59, "Andrei Popov" <[e-mail]> wrote:

>Hello All,
> 
>RC4 is a widely deployed cipher, which is commonly preferred by TLS
>servers: our tests show ~40% of the high-traffic HTTPS sites pick RC4 if
>IE offers this cipher. A significant percentage of web sites and e-mail
>servers have only RC4 enabled,
> so a client cannot altogether disable RC4 without breaking
>interoperability. At the same time, attacks on RC4 are improving (e.g.
>http://www.isg.rhul.ac.uk/tls/), to the point that practical exploits are
>possible.
> 
>I have posted a new Internet-Draft ³Prohibiting RC4 Cipher Suites²
>(draft-popov-tls-prohibiting-rc4-00
><http://datatracker.ietf.org/doc/draft-popov-tls-prohibiting-rc4/>) to
>deprecate the use of RC4 cipher suites in TLS.
> 
>Looking forward to comments and feedback on the draft,
> 
>Andrei Popov
>