* Zachary Amsden (zach@xxxxxxxxxx) wrote: > Done. Looks like you want empty_zero_page write protected too. That > seems like a fine idea to me unless something really wants to do atomic > 64-bit reads on it. Thanks, I added this set (minus the 3/5, which I already have) to the virt-2.6 tree. > The -1 is quite useless when you're going to shift >> 12 anyways to get > a frame number to index into mem_map, which is why they are not there. > Plus, it just seems scary if you got it wrong - let's say you had a not > present page - not that you could, but now you are freeing a misaligned > address in the _previous_ page. I really don't like that -1 at all. I > will clean it up, but it does certainly deserve another patch. Heh, that's exactly what I was concerned with from the patch (being on the wrong page). Thanks for clearing up my confusion.