Master out audio stutters during drag & drop

[Sean M. Pappalardo]

On Debian 8.9 with Mixxx git master r6402; built on: Dec 2 2017, 48kHz sample rate, ALSA back-end, master out set to a USB sound card, headphone out set to internal sound card, latency at 2.33ms, dragging and dropping tracks either to a crate or a deck causes the master out audio to stutter during the dragging. (I didn't check the headphone audio.)

Increasing the latency reduces the effect, but it's still present even at 21.3ms. It happens the same with just the master out set to the internal sound card and no headphone output set.

Even at 42.7ms, I still get occasional glitches on the master output dragging a track from the library to a crate or a deck.

I *do* see this problem in the latest 2.0/1.12 branch as well FWIW, so maybe it's just my system?

[Owen Williams (ywwg)]

This usually indicates the latency is set too low. Where do you have it?

[Uwe Klotz (uklotzde-deactivatedaccount)]

I also noticed the following warnings and audible pauses in the sound just today:

ALSA lib pcm.c:8323:(snd_pcm_recover) underrun occurred
Warning [Engine]: underflowHappened code: 6
Warning [Engine]: underflowHappened code: 24

Those warnings and audio cuts appear repeatedly and are unrelated to drag and drop. Restarting Mixxx didn't fix it.

I'm currently running some disk I/O tasks in the background, but will keep an eye on that!

[Uwe Klotz (uklotzde-deactivatedaccount)]

Might the drop outs be caused by theses recent changes?

https://github.com/mixxxdj/mixxx/pull/894

[Uwe Klotz (uklotzde-deactivatedaccount)]

I didn't notice any more drop outs at 10 ms:
https://github.com/mixxxdj/mixxx/pull/1415

[Uwe Klotz (uklotzde-deactivatedaccount)]

By doing a git bisect I was able to locate the cause of the buffer underflows in the networkclockref branch, i.e. https://github.com/mixxxdj/mixxx/pull/894.

Unfortunately the code does not compile or Mixxx crashes with this range, so I was not able to identify single commit.

[Daniel Schürmann (daschuer)]

Unfortunately I cannot confirm this. :-/
My Setup:

    Ubuntu Trusty, ALSA
    internal Souncard: Master
    USB Soundcard: Headpones
    48000 kHz 5.33 ms
    Multi Souncard = Default

I have compared current master 24028f2
with the commit just before merging this f76d5c7

I played a track in deck 1 and analysed a track in deck 2.
No overflows in both version.
I only notice a cpu usage peak in the old version when analysis starts.

I have started Mixxx with:
./mixxx --developer --logLevel debug

And I have no underflow log entries.

2.67 ms does not work at all. There is a kind of deadlock in the alsa code.
This is a known issue https://bugs.launchpad.net/mixxx/+bug/1310553

What is your test case?

[Sean M. Pappalardo (pegasus-renegadetech)]

If this is latency-dependent, why does the CPU meter not indicate a problem? (Mine is very low in the green even while the problem manifests.)

[Daniel Schürmann (daschuer)]

Which engine clock do you use?

[Sean M. Pappalardo (pegasus-renegadetech)]

Default (long delay)

[Sean M. Pappalardo (pegasus-renegadetech)]

Possible duplicate? https://bugs.launchpad.net/mixxx/+bug/1300664

[Be (be.ing)]

What is the status of this? It is marked as In Progress but there is no open pull request. Should we push it back to 2.2?

[Daniel Schürmann (daschuer)]

I have to free that up for now, because I cannot reproduce this.

[Sean M. Pappalardo (pegasus-renegadetech)]

Why is this not appropriate for 2.1.1? We haven't released it yet. :)

[Swiftb0y (swiftb0y)]

Mixxx now uses GitHub for bug tracking. This bug has been migrated to:
https://github.com/mixxxdj/mixxx/issues/9004