Interview with Kernel Hacker Robert Love

An anonymous reader writes: "An interview with the ever interesting Robert Love is at KernelTrap. Robert is author of the kernel preemption patch which has been merged into the 2.5 development kernel. In this interview, Robert discusses the status of Linux kernel preemption, talks about his recent involvement with the O(1) scheduler and explains his recent VM overcommit work. He also reflects upon Linus' use of Bitkeeper, the future of Linux, and the recent Kernel Summit in Ottawa. A Good Read."

Comments

Re:Can someone explain the preemptible kernel?

[4of12]

I'm not a kernel developer, either, but I have just enough ignorance to be dangerous....

AFAICT, if you have some process that's really gotta run, then you don't even want the kernel taking up valuable slices of time because of the potential impact on performance. Hence, allow your process to be effectively negatively niced beyond what is possible without the preemption patch.

I think some purists would argue that you might as well just start using a Real Time version of the kernel in that case. But I think that real time requirements can cause you to lose other good things that you might want to have around.

I've heard anecdotes that using the preemption patch and running X with an effective high priority makes things seem snappier for desktop use. Of course, there are probably plenty of potential applications in the server arena as well.

[I apologize to people that know what's really going on since I only know Jack Shit by reputation.]

Re:Can someone explain the preemptible kernel?

[Lionel Hutts]

It was discussed here. Briefly, it means that the kernel itself, just like user programs, can be preempted, preventing high latency.

The alternative, of course, is just to attack all sources of latency in the kernel, which should be both better and harder.