Hi,

Jeff Pyle wrote on 09/18/2014 08:28 PM:
> I fixed some next_hop options and SIP signaling is good now.  Before
> it was not.  Still, the UDPTL flow after the T.38 reinvite completes
> isn't right.  Sems reports the following errors:
> 
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=6) !
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=6) !
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=6) !
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=8) !
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=8) !
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=8) !
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=10) !
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=10) !
>  [parse, AmRtpPacket.cpp:112] ERROR:  bad rtp packet
> (hdr-size=12;pkt-size=10) !
> 
> Coming from the gateway routing to Sems, Wireshark decodes the traffic
> asnormal UDPTLPacket(s).  After Sems' media relay, Wireshark decodes
> it as "Unknown RTP version 0".  It's as if Sems is trying to handle
> the UDPTL traffic as RTP and getting confused in the process.
> 
> What can I do from here?
check why AmRtpStream::relay_raw is not enabled for the UDPTL stream.

maybe try something like this:

diff --git a/core/AmB2BMedia.cpp b/core/AmB2BMedia.cpp
index b83d5aa..5c76def 100644
--- a/core/AmB2BMedia.cpp
+++ b/core/AmB2BMedia.cpp
@@ -495,7 +495,10 @@ void AudioStreamData::setReceiving(bool r) {
 AmB2BMedia::RelayStreamPair::RelayStreamPair(AmB2BSession *_a,
AmB2BSession *_b)
 : a(_a, _a ? _a->getRtpInterface() : -1),
   b(_b, _b ? _b->getRtpInterface() : -1)
-{ }
+{
+  a.enableRawRelay();
+  b.enableRawRelay();
+}

 AmB2BMedia::AmB2BMedia(AmB2BSession *_a, AmB2BSession *_b):
   ref_cnt(0), // everybody who wants to use must add one reference
itselves


Normally, it should be set in AmB2BMedia::updateRelayStream.

hth
Stefan

_______________________________________________
Sems mailing list
[email protected]
http://lists.iptel.org/mailman/listinfo/sems

Reply via email to