Justin Karneges wrote:
> On Wednesday 14 January 2009 13:16:10 Peter Saint-Andre wrote:
>> Justin Karneges wrote:
>>> On Tuesday 13 January 2009 15:16:23 Peter Saint-Andre wrote:
>>>> According to my reading of RFC 4568, SDP Security Descriptions MUST NOT
>>>> be used unless the signalling channel (that's XMPP for us) can "provide
>>>> strong message authentication and packet-payload encryption, as well as
>>>> effective replay protection". Because we don't provide those services in
>>>> XMPP out of the box, I don't think we can securely use a=crypto (or our
>>>> XMLish flavor of a=crypto as currently described in XEP-0167). But we
>>>> might be able to use it if we negotiate XTLS (or some other e2e method)
>>>> first.
>>> I'm of the opinion that requiring e2e encryption to bootstrap secure oob
>>> sessions is perfectly acceptable.  Relatedly, I'm of the opinion that
>>> having oob sessions inherit the security properties of XMPP helps avoid
>>> confusion.
>>>
>>> With the way XEP-167 is currently written, RTP is just as secure as your
>>> text chat.  It's as simple as that.  If both participants use the same
>>> server, and use TLS to that server, then RTP is protected within that
>>> domain, just as text chats are.  If both participants use e2e encryption
>>> to protect the Jingle iq stanzas, then RTP is fully protected end-to-end.
>> It's just as secure, but I think the layering is wrong. IMHO we want to
>> pull the security negotiation out of the transport negotiation. At a
>> high level, this:
>>
>> <iq>
>>   <jingle>
>>     <content>
>>       <description/>
>>       <transport/>
>>       <security/>
>>     </content>
>>   </jingle>
>> </iq>
> 
> Well, yes and no.  Yes, I agree generic transport security should be offered 
> at the Jingle level.  

I don't know that there is such a thing as generic transport security,
but I need to ponder that further.

> However, SRTP (the topic of this thread, or at least 
> where I branched it) should not be offered at the Jingle level because it is 
> application-specific.

That seems sensible.

> So, to be clear, I think Jingle should offer exactly two generic security 
> options: TLS (for use with stream-based transports/apps) and DTLS (for use 
> with datagram-based transports/apps).

I'm not sure about that yet.

> If SRTP (or ZRTP, etc) are used, these would be details of Jingle RTP, and 
> their associated data exchanges would occur within <description/>.

That part seems agreeable -- it is what we have today in XEP-0167,
except that we haven't yet defined how to include the zrtp-hash.

Peter

Reply via email to