[asterisk-dev] DTLS-SRTP SDP correction?

[Daniel Pocock]

On 28/01/14 13:34, Joshua Colp wrote:
> On 14-01-28 08:25 AM, Daniel Pocock wrote:
>> Hi,
> Greetings,
>
>> I realize Digium is not necessarily going to fix the DTLS-SRTP issues
>> immediately
>>
>> However, I think it would be really important to clarify the situation
>> with the media sub-field for the protocol, if the current Asterisk
>> behavior is not correct, then it would be good to tell people not to try
>> and duplicate this behavior in other products that are intended to work
>> with Asterisk
>>
>> Specifically, Asterisk is expecting "UDP/TLS/RTP/SAVPF" in the protocol
>> sub-field of an INVITE message
> According to RFC5764 our behavior is correct, specifically the part on
> session description:
>
> This specification defines new tokens to describe the protocol used
>    in SDP media descriptions ("m=" lines and their associated
>    parameters).  The new values defined for the proto field are:
>
>    o  When a RTP/SAVP or RTP/SAVPF [RFC5124] stream is transported over
>       DTLS with the Datagram Congestion Control Protocol (DCCP), then
>       the token SHALL be DCCP/TLS/RTP/SAVP or DCCP/TLS/RTP/SAVPF
>       respectively.
>
>    o  When a RTP/SAVP or RTP/SAVPF stream is transported over DTLS with
>       UDP, the token SHALL be UDP/TLS/RTP/SAVP or UDP/TLS/RTP/SAVPF
>       respectively.
>
>    The "fmt" parameter SHALL be as defined for RTP/SAVP.
>
>> The RFC 5763 examples seem to show that the INVITE should just say
>> "RTP/SAVPF" and the full value of "UDP/TLS/RTP/SAVPF" is only used in
>> the SDP sent with a 200 OK response.
> The examples are including RFC 5939 (SDP capability negotiation) which
> changes things. Asterisk doesn't support this.


Is that what Firefox is trying to do with the SDP it sends on INVITE?

I implemented a quick hack in JSCommunicator that tweaks the SDP from
Firefox into what Asterisk expects.  This makes it work.  Here is the
commit:
https://github.com/opentelecoms-org/jscommunicator/commit/6980f8e1c3311c46154b3840d695f0ddc9c8c8ae

However, I didn't want to keep that in my code in the long term.  It is
just to make the sip5060.net/test-calls work for as many people as possible.

Do you believe this should be fixed by some change in Firefox or is it
something that would potentially have to be implemented in Asterisk?