[Asterisk-Dev] Seeking a clue RE: how * does the right thing with NAT'ed SIP/SDP devices

[David Pollak]

Folks,

I'm looking to understand how Asterisk determines if it's going to 
re-write an SDP contained in a SIP message from an ill-behaving 
application behind a firewall.

Specifically, the Microsoft RTC client running behind a NAT puts the 
wrong IP address (the local non-routable address, rather than the NAT's 
address) in the SDP header.  However, Asterisk does the right thing and 
send media streams to the IP address of the NAT.

I've looked through chan_sip.c:process_sdp and don't see anything that 
remotely looks like logic to re-write the host.

If anyone can lend a clue to me, I'd be grateful to the tune of a couple 
of beers.

Thanks,

David