Closed Bug 1400488 Opened 7 years ago Closed 3 years ago

NotReadableError sharing mic in multiple tabs in the same process (regression)

Categories

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

Product:
Core
Component:
WebRTC: Audio/Video
Version:
57 Branch
Type:
defect

Tracking

()

Status:
RESOLVED DUPLICATE of bug 1404977
Tracking Flags:
Tracking Status
firefox56 --- wontfix
firefox57 - wontfix
firefox58 - wontfix
firefox59 --- wontfix
firefox60 --- wontfix
firefox61 --- wontfix
firefox62 --- fix-optional

People

(Reporter: jib, Unassigned)

References

Details

(Keywords: dev-doc-needed, regression)

+++ This bug was initially created as a clone of Bug #1381638 +++ This is a follow-up to fix bug 1381638 properly. STRs: 1. Open https://jsfiddle.net/jib1/6oy8dmg4/ + share mic, in 5 tabs 2. Snap fingers Expected result: Responsive feedback in speakers. Actual result: After 5th tab opens, "NotReadableError: Concurrent mic process limit." Workaround: Close other tabs using mic, or increase dom.ipc.processCount.
[Tracking Requested - why for this release]: Purposely failing concurrent tab use of microphone based on process (bug 1381638) was a stop-gap to avoid severe audio distortion in 56. We should work to find a better solution for 57, if possible. We may not discover a way until 58 (some re-architecting may be needed), but let's track it in case we do.
tracking-firefox57: --- → ?
That sounds like a dangerous thing to try and do in beta. Can we use the same stop-gap in 57 as in 56?
Flags: needinfo?(jib)
Is it the same problem below ? 1. use the demo https://webrtc.github.io/samples/src/content/devices/input-output/ to getusermedia. 2. open a new tab and use the demo https://webrtc.github.io/samples/src/content/devices/input-output/ to getusermedia again , it will fail to getusermedia: navigator.getUserMedia error: MediaStreamError { name: "NotReadableError", message: "Concurrent mic process limit.", constraint: "", stack: "" }
(In reply to xpeng from comment #4) > Is it the same problem below ? > 1. use the demo > https://webrtc.github.io/samples/src/content/devices/input-output/ to > getusermedia. > 2. open a new tab and use the demo > https://webrtc.github.io/samples/src/content/devices/input-output/ to > getusermedia again , it will fail to getusermedia: > navigator.getUserMedia error: MediaStreamError { name: "NotReadableError", > message: "Concurrent mic process limit.", constraint: "", stack: "" } Yes. I'm actively working on fixing this: allowing any number of microphone access in a tab or in any number of tabs.
Paul tells me this isn't going in 58.
status-firefox58: affected → wontfix
tracking-firefox58: ? → -
Depends on: 1404977
Rank: 15
(In reply to Jan-Ivar Bruaroey [:jib] (needinfo? me) from comment #0) > Actual result: After 5th tab opens, "NotReadableError: Concurrent mic process limit." FYI. This bug still happens, except I now get a different error message: "NotReadableError: Failed to allocate audiosource" Looks like which error message is thrown changed sometime this spring. I don't know that we need a separate bug to track that, since everything else looks the same.

Bump.

This issue is causing problems for us improving Firefox support at https://app.stageten.tv/. Our use case is that we have a live video mixer that allows the user to connect multiple local audio and video sources into a live show. This works correctly in Chrome, i.e. if I have two cameras and two mics connected to my computer then I can add both sources in the app. In Firefox (as of 77.0.1 (64-bit)) I can add a single camera and mic but when I try to add the second mic I get NotReadableError: Concurrent mic process limit

Bump.

I did notice the same case as @Sanborn where changing the mic input to something else caused a concurrent mic process limit error. In our application at least, it happens when we have new AudioContext() initialized (we use it to try and get volume levels). Once we completely stop the AudioContext from being initialized in Firefox, the error goes away. So now we only have volume levels in Chrome, Safari and Edge, but disabled completely in FF.

Hope that helps narrow down the issue. Might be related to the AudioContext worklet?

We also use AudioContext extensively in our product and hit this Firefox-specific bug. I would really appreciate bump in priority here.

Hi jan-Ivar!
I'm trying to reproduce old bugs to see if we can resolve some. I was not able to reproduce this issue on latest Nightly 96.0a1 (2021-11-02)(64-bit) on Windows 10. Despite the error message is not shown, I've noticed a huge echo sound.
Could you tell us if the issue still is reproducible?
Thanks!

Flags: needinfo?(jib)

WFM. I suspect this old issue (which is solely about the same-process case) was fixed in bug 1404977. Comment 9 and on appear to be about two mics, which is bug 1238038.

Status: NEW → RESOLVED
Closed: 3 years ago
Flags: needinfo?(jib)
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.