takashikkbn / gmail.com wrote:
> Ah, one thing I forgot to note which was found on watching CI is that looping in `stop_worker()` (calling `rb_postponed_job_flush`) may not flush mjit_copy_job_handler when EC_EXEC_TAG() is not TAG_NONE. That's why I think `stop_worker_p` should be checked inside `copy_cache_from_main_thread()`.
> 
> My current attempt to fix this issue is r65312. I'll watch CI results.

OK, also, I don't think rb_postponed_job_register is safe to
call without GVL because vm->postponed_jobs needs protecting.

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>