The 2009 Linux Kernel Summit

Day 1

The sessions held on the first day of the summit were:

• Mini-summit readouts; reports from various mini-summit meetings which have happened over the last six months.

• The state of the scheduler, the kernel subsystem that everybody loves to complain about.

• The end-user panel, wherein Linux users from the enterprise and embedded sectors talk about how Linux could serve them better.

• Regressions. Nobody likes them; are the kernel developers doing better at avoiding and fixing them?

• The future of perf events; a discussion of where this new subsystem is likely to go next.

• LKML volume and related issues. A session slot set aside for lightning talks was really mostly concerned with the linux-kernel mailing list and those who post there.

• Generic device trees. The device tree abstraction has proved helpful in the creation of generic kernels for embedded hardware. This session talked about what a device tree is and why it's useful.

Day 2

The discussions on the second day were:

• Legal issues; a lawyer visits the summit to talk about the software patent threat and how to respond to it.

• How Google uses Linux: the challenges faced by one of our largest and most secretive users.

• Performance regressions: is the kernel getting slower? How do we know and where are the problems coming from?

• Realtime: issues related to the merging of the realtime preemption tree into the mainline.

• Generic architecture support: making it easier to port Linux to new processor architectures.

• Development process issues, including linux-next, staging, merge window rules, and more.

The kernel summit closed with a general feeling that the discussions had gone well. It was also noted that our Japanese hosts had done an exceptional job in supporting the summit and enabling everything to happen; it would not be surprising to see developers agitating for the summit to return to Japan in the near future.

See also: the obligatory kernel summit group photo.