[asterisk-dev] Re: AstriDevCon Recap - IAX2 RENEW for encryption

[Russell Bryant]

Kevin P. Fleming wrote:
> Tony Mountifield wrote:
> 
>> In the IAX scenario, having updated the even key, it would continue using
>> the odd key for encryption until it received an ack of the change to the
>> even key from the peer. It would then be free to start using the even key
>> at some point, and at any time later to update the odd key.
> 
> I think the issue here is that in IAX2, the *entire* packet is
> encrypted, there is no unencrypted header that can indicate which key
> was used for encryption.

Well, there is a small part of the header that is unencrypted, but it is 
just enough to know which call the packet is associated with.  In any 
case, there is no space in the header to put the information unless we 
defined the frames to be a different format in encrypted sessions, which 
I don't think we really want to do.

-- 
Russell Bryant
Software Engineer
Digium, Inc.