Everything makes sense and is clear, except the objection misiak proposed (I actually thought about this before I even saw his response ). It seems (at first sight, anyway) that the percentage of fresh swap remaining should be <= the total amount of swap space remaining reported by free.