Chapter 1. 10h00 - A 2.6 Kernel Roadmap

A 2.6 Kernel Roadmap

Jonathan Corbet

slides: http://lwn.net/talks/ols2005/

To be honest, most of this roadmap was actually spent looking backwards instead of forwards. After the customary token quotes, Jonathan launched into an extended recap into the major developments along the road of Linux kernel development, mostly over the last year. By the end of the talk there were very few actual predictions made.

The talk provided a very good overview of the current state of Linux. If you weren't familiar with Linux, or haven't been reading lwn.net lately, or were off getting your MBA, then this talk would have provided a very nice way to bring yourself up-to-date.

The notes from Jonathan's talk are available and very well done. Reading through those notes will give you just about everything that was said in the talk. The major points covered included:

• the new kernel numbering scheme (and how it came about)

• the bitkeeper/git issue (mostly to praise bitkeeper as being the application that finally got Linus to start using a code revision system)

• what has changed in the last year.

Like I said, Jonathan spent the second half of his presentation going over all the things that have changed in the kernel in the last year including:

• latency and real-time

• virtualization

• filesystems

• the desktop effort

• security

• time-related code

• memory management

In addition to what was presented in his slides I detected three other main themes running through his talk that are sort of in his slides but which Jonathan made reference to more than what was written down:

• professionalism: it's good to see the kernel become more professional through the use of debugging tools, bug reporting, the Linux Test Project effort, and regression testing.

• fragmentation: Jonathan has started to notice how different kernel subgroups have begun to fragment from each other. It has been a very slow, but continuous process. We now have different groups entirely isolated from each other and therefore coming up with their own libraries and debugging aides. Jonathan sees this as a bad development and hopes that these subgroups start coming back together.

• tools: the very sore lack of tools, especially kernel debugging tools, available to developers to help them create new code and debug existing problems.

It was good to see Jonathan's talk isolated so that everyone could attend. It provided a very good kick-off to OLS; mostly because as he would bring up a topic he would cross-reference it with "see so-and-so's talk about this at such-and-such time for more information". Jonathan is a very valuable resource to the Linux community and a very interesting speaker; it was a very good idea of the OLS planning committee to put is talk first and by itself.