threads: avoid deadlock from recursive lock acquire (PR JuliaLang/jul…

…ia#38487) (JuliaLang/julia#38753)

Finalizers can't safely acquire many essential locks (such as the
iolock, to cleanup libuv objects) if they are run inside another lock.
Therefore, inhibit all finalizers on the thread until all locks are
released (previously, this was only true for our internal locks).

(cherry-picked from ec8466b)

src/messages.jl

@@ -147,6 +147,7 @@ function flush_gc_msgs(w::Worker)
  end
 
  # del_msgs gets populated by finalizers, so be very careful here about ordering of allocations
+ # XXX: threading requires this to be atomic
  new_array = Any[]
  msgs = w.del_msgs
  w.del_msgs = new_array
@@ -178,7 +179,7 @@ function send_msg_(w::Worker, header, msg, now::Bool)
  wait(w.initialized)
  end
  io = w.w_stream
- lock(io.lock)
+ lock(io)
  try
  reset_state(w.w_serializer)
  serialize_hdr_raw(io, header)
@@ -191,7 +192,7 @@ function send_msg_(w::Worker, header, msg, now::Bool)
  flush(io)
  end
  finally
- unlock(io.lock)
+ unlock(io)
  end
 end