1480277 - No audio for track added to recvonly m=line

Status: RESOLVED DUPLICATE of bug 1472321

(Core :: WebRTC: Audio/Video, defect, P2)

Description

[mroberts]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/67.0.3396.99 Safari/537.36

Steps to reproduce:

Please see these two JSFiddles. They represent a real scenario my team encountered (pc1 represents a user, pc2 represents our SFU) replicated using Firefox's RTCPeerConnection. The basic scenario is this:

1. pc1 adds an audio track (track1), offering with offerToReceiveAudio/offerToReceiveVideo set to true.
2. pc2 answers without adding any tracks.
3. pc2 adds an audio track (track2) in a new m=line.
4. pc1 answers, setting the new m=line to recvonly.
5. pc1 adds an audio track (track3), reusing the newly added m=line, and offers, setting this m=line to sendrecv.
6. pc2 answers.

At this point, pc1 and pc2 have negotiated three m=lines. The third m=line is transmitting two tracks. From pc1's perspective, the third m=line receives "remoteTrack2", representing "track2". From pc2's perspective, the third m=line receives "remoteTrack3", representing "track3". Attempting to playback "remoteTrack2" succeeds for me in the current Firefox Stable through Nightly. Attempting to playback "remoteTrack3" fails in Firefox 61.0.1 and 62.0a1, but succeeds in 63.0a1.

- Playback of "remoteTrack2": https://jsfiddle.net/9b34c1hr/
- Playback of "remoteTrack3": https://jsfiddle.net/c51sb043/


Actual results:

No playback of "remoteTrack3".


Expected results:

Playback of "remoteTrack3" succeeds.

Comment 1

[mroberts]

I have tested in Firefox 60.0 and playback of "remoteTrack3" (https://jsfiddle.net/c51sb043/) succeeds.

Comment 2

[Nils Ohlmeier [:drno]]

Byron, does this look like the regression from Peer Identity which you fixed a couple of days ago?

Comment 3

[Nils Ohlmeier [:drno]]

Mark, could you check with Firefox Beta 61 if it's fixed?

Comment 4

[Byron Campen [:bwc]]

(In reply to Nils Ohlmeier [:drno] from comment #2)
> Byron, does this look like the regression from Peer Identity which you fixed
> a couple of days ago?

It does, especially considering that it works in nightly. I can check in a bit.

Comment 5

[Byron Campen [:bwc]]

Yeah, verified that the changeset from bug 1472321 fixed this.