> Let me guess, using fsck as an example:

> 	fsck gets "/dev/disk/rsd2d" as its device
> 	fsck calls open("/dev/disk/rsd2d")		CS #1
> 	Eventually the vnode gets resolved, and
> 		the open routine for the vnode goes
> 		out to userland to talk to this daemon
> 		which handles all the device mappings	CS #2

Whoa.  I think this is the first time anyone has even suggested that
userland would be involved on *every open*.  Some schemes have a
userland program that shoves wedge info into the kernel and dies; some
have a daemon that notices new disks appearing and does something
useful with them.  Some have both.  But I don't think I've seen
anything that doesn't even load the wedge info into the kernel.

> Microkernels don't work -- context switching is not cheap.

Heh.

	Why don't microkernels work?  Too expensive - all those context
	switches.

	Why are context switches expensive?  Because they happen
	rarely, so nobody bothers to optimize them.

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B