To repeat this bug, just execute the below code(I've run it with ruby185 and 186, you may need increase the complexity
of regexp if you has a high performance machine):

require "rfuzz/session"
include RFuzz

c = HttpClient.new("www.ruby-lang.org", 80)
r = c.get("/en/")
p = r.http_body
p.gsub!(/\r\n/, '')
s = "<div .*?span.*?<h3>.*?<dd>.*?<dl>.*?</dd>.*?</dl>"

timeout(1) {
  p.gsub(Regexp.new(s)) {
  }
}

This snippet code will be executed more than 1 second under windows with my old hp  notebook,  and it won't raise any
exception. I've traced the Ruby source, and I found it is because monitor thread(y thread in below code) won't get woken
up:

timeout.rb:

  def timeout(sec, exception=Error)
    return yield if sec == nil or sec.zero?
    raise ThreadError, "timeout within critical session" if Thread.critical
    begin
      x = Thread.current
      y = Thread.start {
        sleep sec
        x.raise exception, "execution expired" if x.alive?
      }
      yield sec
      #    return true
    ensure
      y.kill if y and y.alive?
    end
  end

The reason why y thread won't be woken up is because rb_thread_tick in CHECK_INTS isn't decreased THREAD_TICK(500) times,
so Y thread won't get scheduled until x finished:

rubysig.h

#define CHECK_INTS do {\
    if (!(rb_prohibit_interrupt || rb_thread_critical)) {\
	if (rb_thread_tick-- <= 0) {\
	    rb_thread_tick = THREAD_TICK;\
            rb_thread_schedule();\
	}\
    }\
    if (rb_trap_pending) rb_trap_exec();\
} while (0)

So, in fact, this is a thread schedule bug. 

I hope I understand correctly and the above analytics can give you a little help.