> > Most processes are blocked in the bowels of the kernel and have far too much
> > kernel state to keep anywhere else.
> 
> Read up on what was done in Mach, please. You may want to start here:
> 
> http://citeseer.nj.nec.com/draves94control.html

I've got as far as page 58...

> In particular, your statement "too much kernel state to keep anywhere
> else" is carefully analyzed, and found to be wrong.

I guess the long-term sleepers (poll, select, tty/socket read etc)
have limited state.  For short term sleeps (eg in the filesystem code)
you do have a lot more state.

Removing the kernel stacks from LWPs sleeping in certain states is
rather different from only having one stack per cpu.

FWIW I would have used the 'continuation function' technique to avoid
proliferating system processes.

	David

-- 
David Laight: david@l8s.co.uk