Muli Ben-Yehuda's journal

Hacking on shpte (shared page tables) now, after a long hiatus. Tonight’s loot is a brute-force forward port to 2.6.2-rc1, running under UML. Surprisingly enough, it boots and runs userspace with Tom’s rootboot root file system. I’ll try the debian root filesystem tomorrow, and then give it a spin on real hardware. Last time I worked on it, it didn’t even boot (l)userspace, so this is progress, of sorts.

Playing with orkut is fun, especially as new people I know keep joining. Two very useful features would be the ability to see when your friends add new friends (because many times friends of your friends are your friends as well – got that, Joe?) and the ability to see people’s RSS feeds. Because orkut to me is all about putting real people behind names, and what better way to get to know someone than read what they have to say? I better send this to feedback@orkut, me’thinks.

I finished reading rml’s Linux Kernel Development book. Overall, I liked it very much. It’s a great introduction, and I constantly recommend it to people who want to start kernel hacking. I do have a couple of nits to pick, mostly to do with the fact that it is overly simplistic at times and doesn’t mention the underline complexity of things. Kernel hacking is fun, true, but it’s also hard. Also, several of the structures had “comments” added to them which weren’t particularly instructive. Other than that, rock on, rml!

Ranny, ladypine’s officemate and our friend, went to the U.S. a couple of weeks ago to work with his thesis advisor who’s on a sabbatical. I gave him tea, the faithful thinkpad R30 laptop that serves as my main kernel crash box usually, for the trip. Today he stopped by and dropped off tea. He also brought me a bottle of 15 year old Glenmorangie Burgundy Wood Finish 15-year-old Single Malt Scotch Whisky. Tasted marvelous. Thanks, Ranny, and cheers!

Plans for next week: on Sunday and Monday Paul McKenney is visiting IBM HRL, on Monday evening choo will give Haifux the second part of an introductory talk talk on Linux drivers that I should attend, on Thursday there’s the developers meeting that Shachar is hosting in Herzliya, that I should attend as well. Also the usual stuff, work, study, hack, bla bla. Life is good 🙂

I’m playing with orkut now, thanks to gby. Leave a comment or email me if you want me to add or invite you and I haven’t yet.

laptop porn. I wonder how well this beastie supports linooks?

Rusty posted a string of hilarious emails from lkml and elsewhere that were shown during the LCA2004 dinner. Recommended reading, provided you’re not drinking anything at the same time…

I gave the first run of the “Intro to Linux Device Drivers” talk yesterday to the study group at work, and today to the students of the Technion’s CS faculty’s Operating Systems course. It went quite well yesterday, with lots of audience participation and excellent feedback afterwards. Today it didn’t go quite as well (much tougher crowd…), but I think the objectives were achieved and they have an idea of what writing a driver entails.

After the talk, I went to the office, feeling rather miserable. Had lunch and then gave up and went home, feeling like I’m about to hurl at any moment. Got home and climbed into bed with James Clavell’s Shogun.

It’s one AM now, and I have a backlog of work stuff to take care of before tomorrow. Let me just put on some music and let the fun begin. No more talks for me in the near future! It’s time to write some free software.

Had one of those days that make me wish I could get better acquainted with every nearby computer, preferably with a chainsaw. I spent 7.5 hours(!) trying to get 2.6.1 to boot on a thinkpad running Fedora Core I with the root file system on an LVM volume. Eventually, it did.

The rest of the day was spent alternatively writing code and slides, for the “Introduction to Linux Device Drivers” talk. I can’t say I’m too thrilled with the way it turned out, but hopefully, it will do. The code, klife-0.02, is available here. It’s does mmap and hooking into the timer interrupt, but the races are abundant. To be cleaned up after I get some sleep (or maybe left as an exercise to the readers ;-))

Quick catchup time:
– da_x announced coLinux which he’s been telling me about for a while now and the reception was terrific. Well done, da_x!
– UML with 2.6 woes (posted earlier) were solved thanks to the nice guys on #uml. Turns out that UML does not play well with exec-shield, which FC1 comes with. Running ‘i386 ./linux …’ instead of plain ‘./linux’ allows it to run. No idea what the i386 utility does yet, but disabling exec-shield seems like a pretty safe bet.
– I could swear there was more I needed to write about, but I can’t remember what.

When writing slides for a talk that includes a lot of code, is it better to accompany each slide of code with a slide of explanation, or just intermingle the explanation with the code, in the form of copious comments? I’m leaning towards the comments form, but I’m open to persuasion either way. What do y’all think?

Interesting talk, gave me a lot of pointers on what to talk and what not to talk about on Wednesday. Lecture slides available here (zippped .ps) and code samples here. Very rough notes available here, to be cleaned up and reposted as time permits.

I am about to head for the Technion to listen to a lecture on developing Windows device drivers. I will give the corresponding lecture on developing Linux device drivers on Wednesday. Impressions and ponderings to appear when I’ll be back.

Paul McKenney, IBM Distinguished Enginneer and Linux Kernel hacker, will be visiting IBM’s Haifa Research Lab next week. He’ll give a talk on “An Analysis of Read-Copy-Update Techniques in Operating System Kernels”, which is open to the public (details will show up here). I’m definitely excited to meet him again!

Abstract:

Although large-scale shared-memory multiprocessing hardware and software reached the mainstream in the past decade, their synchronization mechanisms make use of costly operations that inherently limit both single-CPU performance and shared-memory-multiprocessor scalability. The key problem with these mechanismsis that they do nothing to decrease the intensity of communication required by conventional algorithms, and this high intensity of communication, or tight coupling, in turn requires heavy use of the expensive hardware synchronization mechanisms that impose performance and scalability limitations.

Although there have been some high-performance and highly scalable algorithms developed for some important special cases, such as memory allocation and statistical counters, one would wish for a more general approach. Recently, a wide-ranging set of specific solutions to particular synchronization algorithms have come to light which use a common implementation of some support functions and some design patterns. This set of solutions has been loosely termed “read-copy update” or RCU.

This talk demonstrates the performance problems of previous approaches, and analyzes the use of RCU techniques in earlier operating-system kernels in order to derive the needed design patterns. I used these patterns to architect the implementation and use of RCU in the Linux 2.6 kernel, which was instrumental in improving the performance and scalability over that of the Linux 2.4 kernel. Empirical and analytic techniques are used to analyze the performance and simplicity benefits of RCU.

About the speaker:

Paul McKenney is a Distinguished Engineer in the Storage Software Architecture group and the Linux Technology Center, and is a member of the IBM Academy of Technology. He joined IBM in 2000, working with AIX, Linux, and storage. Prior to that, he worked at Sequent doing SMP and NUMA algorithms. The work described in this talk stems from his work at Sequent and with the Linux 2.6 kernel.