Producer sent receiver report, local ssrc(packet sender ssrc) is zero.
It should defaults to 1, the same with webrtc magic default recvonly ssrc.
And chromium’s webrtc code, will update receiver_report_local_ssrc_ to first sender’s ssrc.
Related chromium discussion:
Issue 51099004: Set local SSRCs on receivers added before senders. - Code Review (webrtc.org)
Receiver SSRC in RR packets (google.com)
Please. Specify how this affects to mediasoup. We are not gonna change 0 to 1 just because libwebrtc does it. BTW the issue you referenced is super old and we are not aware of any related issue when using Chrome and mediasoup together.
I’m testing with Linphone client, it will ignore this RTCP-RR, so RTT computation and other RTP stats is broken.
And this RTCP-RR will be silently skipped by current webrtc code.
See line 414, it compares the sender ssrc.
webrtc/modules/rtp_rtcp/source/rtcp_receiver.cc - external/webrtc - Git at Google (googlesource.com)
SORRY, the actual RR process line is 456, RTCPReceiver::HandleReportBlock(), remote_ssrc arg is zero.
If RTCP generated by mediasoup was ignored by libwebrtc we would have noticed it already. That doesn’t happen.
Regarding Liphone, that seems to be a bug in Liphone. mediasoup follows the standards here.
Thanks, though i already fix it in forked linphone and mediasoup repo.
I will revert the mediasoup fix and re-test it.