Steps: 1. Request audio from gUM with an integrated mic and a USB mic 2. When the permission prompt appears, select the USB mic 3. Put the MediaStream given into an audio tag and start playing it 4. Generate sounds into the integrated mic and USB mic Expected: The device's speakers should only be sending sound out from the USB mic. Actual: The device's speakers send sound out from both the integrated and USB mics.

Sounds like bug 797796 isn't working as expected, unless the front-end is sending bogus data to the back-end.

(In reply to Dão Gottwald [:dao] from comment #1) > Sounds like bug 797796 isn't working as expected, unless the front-end is > sending bogus data to the back-end. Okay. I was debating on this one specifically if it's front-end or core. Anant - Can you confirm this is a core issue?

Looks like a core issue, we'll take a look.

I think this is fixed from my local testing

Further testing shows we were selecting the last device in the list always. This used to always be the first device on linux and mac, which is the 'default' device, due to another bug.

Comment on attachment 681845 [details] [diff] [review] set capture index for audio input when device is Allocate()ed I left the SetRecordingDevice() in Init() to verify it can record; may not really be needed.

Hmm, this might be tricky. Since we're using the same engine, if I call Allocate() on another object, then the device index gets changed. Would it be better if we move to SetRecordingDevice call to Start()?

Even worse: capture microphone 1 in tab 1. Capture microphone 2 in tab 2. What microphone data does each see? (was just trying this to test another bug; it doesn't work of course)

Comment on attachment 681845 [details] [diff] [review] set capture index for audio input when device is Allocate()ed Review of attachment 681845 [details] [diff] [review]: ----------------------------------------------------------------- Stealing review at jesup's request. ::: content/media/webrtc/MediaEngineWebRTCAudio.cpp @@ +52,5 @@ > if (mState != kReleased) { > return NS_ERROR_FAILURE; > } > > + webrtc::VoEHardware* ptrVoEHw = webrtc::VoEHardware::GetInterface(mVoiceEngine); If you call GetInterface(), you need to call Release().

Also plugs an existing leak of ptrVoEHw in ::Init()

https://hg.mozilla.org/integration/mozilla-inbound/rev/5444aff38dfb FF17/18 are unaffected by this bug because it was introduced shortly after last uplift (other than the leak)

Verified on 11/19.

If we could fake the devices, we might be able to automate this. Although as it stands in the current framework, we don't have this ability. Putting in-testsuite- for now, although if device faking becomes possible at some point, then we should re-evaluate this one for a test.