How to report 100% CPU zombie process

[forthrin]

Which files are necessary for devs to investigate these?

• Spindump
• Sample
• Sysdiagnose (which files, there are hundreds)
• Other?

[cookesan]

For macOS, I would start with the normal macOS bug report template plus one focused hang/CPU sample.

Useful attachments:

mpv --version
system_profiler SPSoftwareDataType | sed -rn 's/^ *System Version: (.*$)/\1/p'

Run the same file/command that triggers it with a log:

mpv --no-config --gpu-debug --log-file=output.txt <same file/options>

Then, while the process is burning CPU, capture a sample:

sample <pid> 10 -file mpv-sample.txt

Activity Monitor’s “Sample Process” output is fine too. A spindump can also help for a hang, but I would attach the smaller process sample first because it is easier to inspect. I would not attach a full sysdiagnose unless a maintainer asks for it; it is large, noisy, and may include unrelated private data.

Also worth clarifying in the report: if the process is using 100% CPU, it probably is not a Unix zombie in the strict sense. A true zombie is already exited and should not be executing. Calling it a hang/spin with 100% CPU will make the report easier to triage.