AudioStreamController may stuck after flushing buffer

[mad-gooze]

What version of Hls.js are you using?

0.12.2

What browser and OS are you using?

Latests Chrome

Test stream:

Test page -- http://jsfiddle.net/0hzpgc7a/1/

Checklist

• The issue observed is not already reported by searching on Github under https://github.com/video-dev/hls.js/issues
• The issue occurs in the stable client on https://hls-js.netlify.com/demo and not just on my page

• The issue occurs in the latest client on https://hls-js-latest.netlify.com/demo and not just on my page

• The stream has correct Access-Control-Allow-Origin headers (CORS)
• There are no network errors such as 404s in the browser console when trying to play the stream

Steps to reproduce

This issue may happen in the wild when video stream is switched by abr controller or by user and BUFFER_FLUSHING event is triggered.

I made a synthetic example with a sample stream with a single audio and video level.

In this example BUFFER_FLUSHING event is triggered manually after FRAG_LOADING event for the 3rd audio chunk (the exact number does not matter here).

After BUFFER_FLUSHING playback may stall (this happens most of the times, try reloading the page). The problem is that FragementTracker marks fragCurrent as APPENDING https://github.com/video-dev/hls.js/blob/v0.12.2/src/controller/audio-stream-controller.js#L314 because it was loaded during IDLE state and was not parsed so it never gets loaded and parsed again.

Expected behavior

Playback continues

Actual behavior

Playback stalls

Console output

https://gist.github.com/mad-gooze/f77d1381f571049b6e44777771a98800

[mad-gooze]

please see possible solution in supplied PR

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[mad-gooze]

still an issue

[johnBartos]

👍

…

On Sat, Mar 16, 2019 at 16:25 Alexey Gusev ***@***.***> wrote: still an issue — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2079 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEzwZFewg9vG2pmfYRh5VrzD0YBTIKljks5vXVMogaJpZM4Z_12N> .

[itsjamie]

It's possible this is a byproduct of #2143. If the controller believes the buffer was flushed but it wasn't actually behaviour would definitely be undefined.

The synthetic example doesn't pass a type to flush to the event, which would result in logs being output that a flush occured, but no actual source buffer flush would occur.

Nope, tried the jsfiddle with the fix from the PR for the mentioned issue, still get the cycling.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[mad-gooze]

still an issue here

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[mad-gooze]

ping

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[itsjamie]

ping

[robwalch]

Hi @mad-gooze,

This appears to be working v0.14.5 with @latest: http://jsfiddle.net/0hzpgc7a/1/

Can you confirm? (I'll reopen if I missed something in reproducing the issue)

[robwalch]

There are also additional improvements to audio track sync and state handling coming in #2919