Re: [Tls-reg-review] [IANA #1135278] Re: Request to register value in TLS bar registry (tls-parameters)

[Смышляев Станислав Витальевич <[e-mail]>]

Dear Benjamin,

I haven't seen this SSL_RSA_WITH_3DES_EDE_CBC_MD5 = 0xFF83 in real life myself - found it in https://opensource.apple.com/source/Security/Security-55163.44/libsecurity_ssl/Security/CipherSuite.h?txt. And it is marked as "for SSL 2 cipher kinds which are not specified for SSL 3". So it seems that it's old as hell and may not be worth specifying it. 


P.S.: Regarding particular numbers for GOSTs: I don't think that it will be a real problem if that 0xFF83 goes to a GOST suite defined for TLS 1.2. I just mentioned that, other things being equal, I would prefer {0xFF, 0x89} and {0xFF, 0x88}, since it is used in preliminary implementations - but not in widely available ones yet, so there won't be a huge problem to change to {0xFF, 0x83} and {0xFF, 0x84}.

And {0xFF, 0x85} for TLS_GOSTR341112_256_WITH_28147_CNT_IMIT is actually widely used in Russia, so this particular number is quite important.

Best regards,
Stanislav Smyshlyaev, Ph.D.
CISO, CryptoPro LLC

-----Original Message-----
From: Benjamin Kaduk <[e-mail]> 
Sent: Saturday, February 2, 2019 7:33 AM
To: Смышляев Станислав Витальевич <[e-mail]>
Cc: Yoav Nir <[e-mail]>; [e-mail]; [e-mail]; Коллегин Максим Дмитриевич <[e-mail]>; Алексеев Евгений Константинович <[e-mail]>; [e-mail]; Смышляева Екатерина Сергеевна <[e-mail]>; Белявский Дмитрий <[e-mail]>
Subject: Re: [Tls-reg-review] [IANA #1135278] Re: Request to register value in TLS bar registry (tls-parameters)

On Fri, Feb 01, 2019 at 09:30:06PM +0000, Смышляев Станислав Витальевич wrote:
>    Dear Yoav, Dear Amanda,
>    Thank you very much for your prompt reply!
> 
>        * The others (TBD7, TBD8, TBD9) are finite field, so they should start
>          at 261.
> 
>    Let me correct you a little here, if you don't mind: these three are also
>    elliptic curve ones (they are defined in RFC 4357 along with finite field
>    ones, but these three are elliptic curves, unlike GOST R 34.10-94
>    parameters, which were finite field ones). So all TBD6-TBD12 should start
>    at 34. 
> 
>      None of these are IETF-endorsed algorithms, so the Recommended column is
>      'N'.  The numbers are fine, but I believe that no specific numbers were
>      requested, so any numbers should be OK.
> 
>      {0xFF, 0x85} for TLS_GOSTR341112_256_WITH_28147_CNT_IMIT
> 
>    is perfect, but another one ({0xFF, 0x83} collides with a privately used
>    (by someone) old SSL_RSA_WITH_3DES_EDE_CBC_MD5. 

Do we know anything more about this private use?  It may be worth writing a brief document noting it and marking it as such in the registry.

-Ben

>    So, if possible, it would be great if you could assign the following
>    numbers, for example:
>    {0xFF, 0x89} to TLS_GOSTR341112_256_WITH_KUZNYECHIK_CTR_OMAC, 
>    {0xFF, 0x88} to TLS_GOSTR341112_256_WITH_MAGMA_CTR_OMAC,
>    (and 
>    {0xFF, 0x85} for TLS_GOSTR341112_256_WITH_28147_CNT_IMIT, as you have
>    proposed)
>    Of course, "N" in "Recommended" column for all of them. 
>    Thank you very much!
>    Best regards,
>    Stanislav