I have a crash with a stack trace of like 20 frames of secondary crashes from the one that I'm debugging. One such crash is here:

#5 0x00007fe8db5ff0ff in report_vm_error (
    file=file@entry=0x7fe8dc5f7430 "assert(t->is_Java_thread()) failed",
    detail_fmt=detail_fmt@entry=0x7fe8dc6ddde4 "invariant")
    at src/hotspot/share/runtime/thread.hpp:858
#6 0x00007fe8db9f23f6 in JfrJavaSupport::check_java_thread_in_vm (t=t@entry=0x7fe8d43146d0)
    at src/hotspot/share/jfr/jni/jfrJavaSupport.cpp:54
#7 0x00007fe8db9ded0e in prepare_for_emergency_dump (thread=0x7fe8d43146d0)
    at src/hotspot/share/jfr/recorder/repository/jfrEmergencyDump.cpp:556

In this case, the crash is not the JavaThread, it's the VM thread. It's in a safepoint, so it's probably safe to dump the JFR recording.