deck stopped playing on 2.1.0 git master

[Be]

As I was playing with current git master 2.1.0, a deck stopped playing. I can't identify anything in particular that triggered this. The waveforms kept moving forward for the deck but there was no sound output from the deck and the level meters showed no signal from the deck. I quit Mixxx and played the track that was loaded at the time again and it played fine. In the log I saw:

Warning [Engine]: DEBUG ASSERT: "pChunk->getState() != CachingReaderChunkForOwner::READ_PENDING" in file src/cachingreader.cpp, line 83
Warning [Engine]: DEBUG ASSERT: "READ_PENDING != m_state" in file src/cachingreaderchunk.cpp, line 146
Warning [Engine]: DEBUG ASSERT: "READ_PENDING == m_state" in file src/cachingreaderchunk.h, line 132
Warning [Engine]: DEBUG ASSERT: "pChunk->getState() != CachingReaderChunkForOwner::READ_PENDING" in file src/cachingreader.cpp, line 83
Warning [Engine]: DEBUG ASSERT: "READ_PENDING != m_state" in file src/cachingreaderchunk.cpp, line 146
Warning [Engine]: DEBUG ASSERT: "READ_PENDING == m_state" in file src/cachingreaderchunk.h, line 132
Warning [Engine]: DEBUG ASSERT: "pChunk->getState() != CachingReaderChunkForOwner::READ_PENDING" in file src/cachingreader.cpp, line 83
Warning [Engine]: DEBUG ASSERT: "READ_PENDING != m_state" in file src/cachingreaderchunk.cpp, line 146
Warning [Engine]: DEBUG ASSERT: "READ_PENDING == m_state" in file src/cachingreaderchunk.h, line 132
Debug [Engine]: ERROR: No LRU chunk to free in allocateChunkExpireLRU.
Debug [Engine]: ERROR: Couldn't allocate spare CachingReaderChunk to make CachingReaderChunkReadRequest.

The last 2 lines kept repeating until I forcibly quit Mixxx (Control + C).

[Uwe Klotz (uklotzde-deactivatedaccount)]

I've never seen this before, but will have a look at the code.

Any information about what you did before the error happened? If the file is not corrupt then there must be an edge case in the CachingReader that we might have missed.

[Be (be.ing)]

I was working on the Hercules P32 mapping (https://github.com/be-ing/mixxx/tree/hercules_p32_mapping ) when this happened. I suppose it's possible something in the script triggered this, but I doubt it. I don't recall doing anything unusual when this started. However, I did notice some periodic pops that sounded like xruns for 30-60 seconds before this happened and when I looked at Mixxx I saw some unusual spikes in the audio buffer meter that I couldn't explain.

[Uwe Klotz (uklotzde-deactivatedaccount)]

I think I found the cause: We must not reuse chunks in the state READ_PENDING while they are still accessed by a worker. Those chunks must not be added to the MRU/LRU list while the worker is reading.

[Uwe Klotz (uklotzde-deactivatedaccount)]

https://github.com/mixxxdj/mixxx/pull/914

[Swiftb0y (swiftb0y)]

Mixxx now uses GitHub for bug tracking. This bug has been migrated to:
https://github.com/mixxxdj/mixxx/issues/8506