Diving in headfirst
By nilsn on May 24, 2005
Welcome to the bungle, we've got fun 'n' games.
I'm a staph engineer in the kernel group. I've been with Sun since they acquired me and the rest of the software development group at Thinking Machines about 9 years ago. Prior to the kernel group I was part of the software team in the high-end hardware organization that brought you such hits as Starfire (the E10k) and Starcat (the F15k).
I should note that I'm starting this blog, if not under duress, then at least under pressure. The woman in the office next to me (who conveniently doesn't have a blog of her own) suggested that I start blogging, presumably because she was tired of having me stand in her doorway, ranting and raving about whatever had gotten under my skin that morning.
More to the point, Bryan Cantrill has been pushing all the Solaris engineers to start blogging. In addition to being one of the big brains behind Dtrace, Bryan is a little bit nuts. Not drive-you-off-the-road-in-his-SUV nuts, but I-know-without-looking-at-the-code-exactly-why-the-CPU-clocks-are-out-of-sync-and-I'm-going-to-spend-the-next-nine-hours-fixing-it nuts. Think Rainman meets Nicholson in the Shining, and you'll be on the right track.
A few years ago they created an orientation DVD for engineers who were new to the Solaris group, and I reviewed it to see if it would be useful for the new engineers in the systems group I was in at the time. It had several engineers and managers giving presentations about the Solaris organization, architecture review processes, and so on. Bryan was the presenter for the portion that dealt with the process of putting back to Solaris. To this day, all I remember of the entire DVD is Bryan staring balefully out of the screen saying, "You did read the README...didn't you?" and thinking he was about to go for my throat.
All of which leads me to back to this point. While visiting the Menlo Park campus this week, Bryan chased me down in the parking lot on my way to lunch and cheerfully asked: "How's the work going on that first blog entry, Nils?" As pleasant as he is in person, I'm still stuck with that image from the DVD. He gives a friendly smile, and I see teeth being bared. So, here I am, trying to save my neck.
My first project in this group was the port of Solaris to the Opteron processor. During the course of the project I managed to get my fingers in just about everything in the kernel, but the vast bulk of my time was spent working with Joe Bonasera, rewriting the VM system's hardware translation layer for Intel CPUs. I also spent an inordinate amount of time reworking most of the boot procedure, about which "I'll talk with you more anon." I understand that Joe has rewritten most of that again for his new project. We'll see which makes it to the world first: my writeup of the current boot process or his implementation of the new process.
I hesitate to bring this up, for fear of not being able to keep up with the email, but I'm currently the project lead for the Linux Application Environment. If I were feeling vengeful I would name my co-conspirators on the project, but since I'm pretty cheerful at the moment I'll let them out themselves. As time goes on, I'll talk more about what we plan to deliver with LAE, how it works, how it's implemented, and why it's not already in OpenSolaris.