DTLS-SRTP roles

[Mamadou Diop]

Hello,

I'm using Nightly "22.0a1 (2013-02-20)" and making call from chrome to
FF through a gateway. The problem is that there is a role conflict in
DTLS. FF is the called party but is sends "Client Hello" message. Our
gateway uses rfc4145 to determine roles. If the remote party doesn't
support this rfc, we just consider that the called party is the server
and the calling the client.
Could someone explain how FF determines the roles?

Regards,

[Eric Rescorla]

Firefox currently treats the caller as the server and the callee as
the client. (This is the recommended configuration from 5763).
Eventually we will do RFC 4572 roles as defined in
RFC 5763/5764.

-Ekr

> _______________________________________________
> dev-media mailing list
> dev-...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-media
>

[Eric Rescorla]

I should mention that we reversed this order about a month ago,
which may have been confusing to people. It was announced, but
maybe not as widely as one might have liked.

-Ekr

On Fri, Feb 22, 2013 at 8:06 AM, Eric Rescorla <e...@rtfm.com> wrote:

> Firefox currently treats the caller as the server and the callee as
> the client. (This is the recommended configuration from 5763).
> Eventually we will do RFC 4572 roles as defined in
> RFC 5763/5764.
>
> -Ekr
>
>
> On Fri, Feb 22, 2013 at 5:17 AM, Mamadou Diop <bos...@yahoo.fr> wrote:
>

[lmin...@gmail.com]

Hi Eric,

apologies for this very late reply to a post you sent a couple of months ago, but I only recently started trying to get DTLS working in Asterisk as well. I'm still stuck with a few other issues, but I'd like to focus on this one at first and then try and address the other ones in other posts.

I also noticed the missing setup attribute in the signaling coming from both Chrome and Firefox, which did confuse me (and Asterisk) a bit. A similar question on the discuss-webrtc group was answered by Justin Uberti, who told that right now the active/passive roles are basically determined out of the ICE roles: "Chrome uses the ICE roles to determine active vs passive; the CONTROLLED party takes the active role". Is this the same in Firefox as of now, or does your latest post in here still apply?

Just to clarify, since as you say that order has been reversed, does this mean that at the moment Firefox expects the caller to take the active role in DTLS, and the callee to take the passive role? I'm using the 23.0a1 nightly build from today.

Thanks,
Lorenzo

[Eric Rescorla]

On Tue, Apr 30, 2013 at 8:40 AM, <lmin...@gmail.com> wrote:

> Hi Eric,
>
> apologies for this very late reply to a post you sent a couple of months
> ago, but I only recently started trying to get DTLS working in Asterisk as
> well. I'm still stuck with a few other issues, but I'd like to focus on
> this one at first and then try and address the other ones in other posts.
>
> I also noticed the missing setup attribute in the signaling coming from
> both Chrome and Firefox, which did confuse me (and Asterisk) a bit. A
> similar question on the discuss-webrtc group was answered by Justin Uberti,
> who told that right now the active/passive roles are basically determined
> out of the ICE roles: "Chrome uses the ICE roles to determine active vs
> passive; the CONTROLLED party takes the active role". Is this the same in
> Firefox as of now, or does your latest post in here still apply?
>
> Just to clarify, since as you say that order has been reversed, does this
> mean that at the moment Firefox expects the caller to take the active role
> in DTLS, and the callee to take the passive role? I'm using the 23.0a1
> nightly build from today.
>

I'm not sure what my latest post said, but Firefox assumes that the caller
(who is generally the ICE controller) is the DTLS server (passive). This
matches Chrome's behavior except in ICE conflicts.

-Ekr

[lmin...@gmail.com]

Thanks for the clarification. The latest posts were, respectively, "Firefox currently treats the caller as the server and the callee as the client" (which is what you just confirmed again, IIUC), and "I should mention that we reversed this order about a month ago, which may have been confusing to people.", which is what confused me as well :-)

L.