Comment 7 for bug 1021471

I'm quite surprised that with all of these tests I haven't got the mutex_lock bug again though, it was definitely happening on that machine... maybe some other fixes fixed it or I'm just not exercising the exact code path that's triggering it.