There's nothing especially special in either of those. I think it's likely the wrong lock in msleep() that I've just posted a patch for to the mailing list. Please test that out and let me know how it goes. (Ok, ok, I'll acquire another eeepc just to use with legacy non-11n nics..) adrian