Re: [TLS] SCSV vs RI when both specified. Was: Updated draft

[Steve Dispensa <[e-mail]>]

On Jan 12, 2010, at 9:17 AM, Marsh Ray wrote:

> Martin Rex wrote:
>>
>> There is absolutely _NO_ known security problem and no  
>> vulnerability for
>> a server to accept an initial handshake with the existing SSLv3,  
>> TLSv1.0,
>> TLSv1.1 or TLSv1.2 protocol, even when the ClientHello handshake
>> message does neither contain SCSV nor an empty extension RI.
>
> Baloney!
>
>> The possibility for an "attack" where an old client's renegotiation
>> handshake is proxied into the initial handshake of a server is of
>> no concern to the server.
>
> TLS is a network data security protocol. Both parties have an interest
> in the security of the connection.
>
> If a client has the attacker's data in his input buffer, then a
> legitimate server's handshake provides a certificate which the client
> interprets as authenticating the previous data, the attacker has  
> stolen
> the servers credentials!
>

FWIW, I believed as Martin did in the early days while we were trying  
to work out the implications of the attack, but I've come to believe  
that Marsh is right - letting a client get data into his input buffer  
that is subsequently authenticated by a legitimate server is obviously  
a bad thing, and there's enough vagueness in real implementations (and  
complete silence in documentation) that I would be shocked if someone  
weren't vulnerable to this attack.

Now, of course, it's a client attack at that point, not a server  
attack, so one might reasonably wonder whether it's the server's  
responsibility to defend against it. (A patched client that indicates  
support for safe renegotiation to a server that also supports it will  
defeat this attack, without the server doing anything special.)

Still, I think you can make an argument about total Internet security  
that says that servers shouldn't negotiate with unpatched clients.  
That obviously involves a giant trade-off in the short term, but don't  
be caught thinking that there is no security impact to agreeing to  
negotiate (as opposed to re-negotiate) with an unpatched client.

  -Steve