sems g729 is not compatible with pjsip g729

[juha-h]

Now that G729 codec support was introduced to sems, I gave it a try by calling sems from pjsip pjsua. Sems reported these errors:

May 17 08:01:42 host sems[7621]: [#7f7ff9363700] [g729_2_pcm16, g729.c:160] ERROR:  g729_2_pcm16: number of blocks should be integral (block size = 10)
May 17 08:01:42 host sems[7621]: [#7f7ff9363700] [receive, AmRtpAudio.cpp:223] ERROR:  decode() returned -1

Perhaps g729.c is not yet compatible with libbcg729?

[lemenkov]

Weird. I've tested it with Cisco 7940 and 7960 phones and they were doing great. I wonder if it has something with g.729 a/b modes? Could you please make a wireshark capture to look at? SIP + RTP if possible.

[juha-h]

Peter Lemenkov writes:

Weird. I've tested it with Cisco 7940 and 7960 phones and they were doing great. I wonder if it has something with g.729 a/b modes? Could you please make a wireshark capture to look at? SIP + RTP if possible.

I haven't made any a/b configurations. I tested between two pjsua's and g.729 worked fine. Then I made a call from pjsua to sems using g.729 and got to syslog same errors as before: ``` May 17 18:47:00 salmon sems[11298]: [#7f540c9d1700] [g729_2_pcm16, g729.c:160] ERROR: g729_2_pcm16: number of blocks should be integral (block size = 10) May 17 18:47:00 salmon sems[11298]: [#7f540c9d1700] [receive, AmRtpAudio.cpp:223] ERROR: decode() returned -1 ``` I took pcap of the call (attached if github allows). It shows shows sems (at 192.168.43.99) sending two g.729 packets to pjsua (at 192.26.134.10) after which it only receives g.729 packets. Both pjsua and sems are using the same g729 lib.

[juha-h]

Looks like the attachement was censored by github. I put the pcap here:

https://tutpro.com/tmp/sems.pcap