I still try to keep an eye on Linux kernel development, after about 6 years without any active involvement. It remains an interesting community, project, and codebase to watch, for a number of reasons. One of the ways I follow what's going on is by reading the kernelplanet feed, which aggregates blogs by a number of kernel developers, including Pete Zaitcev's livejournal.

Recently he wrote a couple of entries (part 1, part 2) concerning “relative pointers” and “absolute pointers” under the Xen hypervisor. I spent a while staring at these posts, wondering what kind of magical VM techniques Xen might be using that could meaningfully expose absolute and relative addresses to a guest Linux kernel and X server. Eventually I gave up on understanding the posts themselves, and turned to Google to find out more. I expected my searches to yield pages describing implementation details of the Xen hypervisor.

Instead, I quickly realized my mistake. Pete wasn't writing about addresses, but the mouse pointer. Mark McLoughlin gives the context.