FCP A/V Output crashing with BlackMagic UltraStudio 4K Mini

If FCP’s A/V Output crashes with BlackMagic UltraStudio 4K Mini, try these steps:

1. Update Software: Ensure Final Cut Pro, Blackmagic Desktop Video, and macOS are all updated.
2. Check Connections: Verify Thunderbolt/HDMI/SDI connections and cables.
3. Select Correct Output: In FCP, go to Preferences → Playback → A/V Output and choose UltraStudio as the device.
4. Restart Devices: Restart Mac and UltraStudio hardware.
5. Test in Another App: Confirm UltraStudio works in Blackmagic Media Express to rule out hardware issues.

Study of the crash logs and MacOS system logs indicates this is caused by a multi-thread memory management bug in FCP's “DestVideo” subsystem. Objects representing video destinations (the output requests and output devices) are not being synchronized properly across threads. The main/UI thread can tear down or modify these objects (e.g. when stopping playback or switching viewers) while background threads (e.g, “procDestChg” worker or Helium render threads) are still using them. There are fluctuating timing windows where this can happen, so the problem is non-deterministic.

Reproducing it requires either video output hardware and a similar system and timeline, or a test plugin that injects similar delays, exposing the crash condition. I've been working on that for some time, but it's very difficult.

AnujAbhiwan study of the crash logs and MacOS system logs indicates this is caused by a multi-thread memory management bug in FCP's “DestVideo” subsystem. Objects representing video destinations (the output requests and output devices) are not being synchronized properly across threads. The main/UI thread can tear down or modify these objects (e.g. when stopping playback or switching viewers) while background threads (e.g, “procDestChg” worker or Helium render threads) are still using them. There are fluctuating timing windows where this can happen, so the problem is non-deterministic.

Reproducing it requires either video output hardware and a similar system and timeline, or a test plugin that injects similar delays, exposing the crash condition. I've been working on that for some time, but it's very difficult.

This is apparently an issue that began with FCP 11.x.

[Edited by Moderator]