Reproducer:

$ java -XX:StartFlightRecording:filename=r.jfr -jar jdk/demo/jfc/J2Ddemo/J2Ddemo.jar
Click around on the different tabs.
$ jfr print --events ThreadStart r.jfr

jdk.ThreadStart {
  startTime = 19:50:04.836 (2023-03-12)
  thread = N/A
  parentThread = N/A
}

Note, might be OK for parent thread to be null (if it's the main thread), both event thread (hidden if null) and the thread field is N/A. Missing thread can also be observed in ExecutionSample.

jdk.ExecutionSample {
  startTime = 16:02:06.365 (2023-03-05)
  sampledThread = N/A
  state = "STATE_RUNNABLE"
  stackTrace = [
    sun.lwawt.macosx.LWCToolkit.getPlatformWindowUnderMouse() line: 1038
    sun.lwawt.LWWindowPeer.notifyMouseEvent(int, long, int, int, int, int, int, int, int, boolean, byte[]) line: 856
    sun.lwawt.macosx.CPlatformResponder.handleMouseEvent(int, int, int, int, int, int, int, int) line: 92
    sun.lwawt.macosx.CPlatformView.deliverMouseEvent(NSEvent) line: 195
  ]
}

The missing threads seems to happen in the first 3-4 seconds of the recording or chunk, but I can't rule out that it happens later as well. Only looked at recording/chunk that lasted for a minute (one chunk) running on Mac.

I = High, information missing
L = M, happens sometimes with JFR
W = H, no known workaround

HMH -> P1