Re: KERNEL: Swapping Kernel Pages

Randal Koene (
Sat, 28 Jun 1997 21:26:56 -0400 (EDT)

On Sat, 28 Jun 1997, Maynard Handley wrote:

> In article
> <>, Randal
> Koene <> wrote:
> > Hi there,
> >

> >
> > Theoretically of course anything but the paging-code&page-fault can be paged
> > out (and since Kernel code is usually clean, it doesn't even need to be
> > swapped out as such, just reloaded). Is it just an issue of speed? If so,
> > I'd think that a cunningly efficient method could be devised to offer just
> > that extra little bit of physical memory at the minimum possible swapping
> > speed cost.
> >
> This is a solved problem. That's part of the theory behind microKernels,
> to get this most essential stuff in one place, then allow everything else
> to run on top without worrying about it.
> Microkernels, like every fucking thing in the UNIX world are the object of
> immense partisanship from people will not conceded a single thing on
> either side about whether or not they are easier to understand,
> increase/decrease performance etc. The practical reality is that one group
> has branched off to do mkLinux based ona microkernel which has the
> potential to do what you ask. (Of course, IMHO running a monolithic kernel
> on top of a microkernel is using only 10% of the power you have available.
> Look at the design of GNU Hurd for something that makes much more
> interesting use of these ideas.)
> Maynard

Thanks. Being interested and all, I sure will have a look. I have come to
the realization though (recently) that most of what I thought of as
inefficient can already be dealt with simply by compiling a minimal
kernel, compiling as many of the drivers as possible as modules and using
the Kerneld daemon to dynamically load/unload drivers at runtime. (It
reafirms my belief in the infinite configurability and supremacy of Linux
:-D !)

PhD2, Neural Modeling Lab, Department of Psychology - McGill University