Re: is Linux obsolete?

Oscar Levi (elf@buici.com)
Fri, 28 May 1999 18:05:33 -0700


On Tue, May 25, 1999 at 05:28:41PM -0700, doctor@fruitbat.org wrote:
> Stefan Monnier said ...
> >
> > >>>>> "David" == David Leal <lorien@esoterica.pt> trolled:
> > > developed today are microkernel based, e. g. Windows NT, GNU HURD, BeOS...
> >
> > Nitpick: WinNT is *not* based on a microkernel.
>
> Just out of curiosity, what makes you say that?

The quintessential microkernel operating system is Mach. It is based
on passing messages between independent services. The memory
allocator is a services as is the process scheduler and the disk IO
subsystem. Windows NT is based on VMS. They made some gestures
toward a layered kernel design, but it fails when you examine it
closely. Their interfaces are overloaded with special-purpose
parameters. All kernel services, except for graphics drivers in 4.0,
share the same process space and can therefore stomp on each other.
In spite of moving display drivers out of the normal kernel space,
they can still crash the OS. Whoops.

The NT kernel API *looks* like it could be for a microkernel, but
their execution is sloppy. Mach did a good job of designing an
appealing microkernel, but the service transitions sap kernel
performance. The NT designers foundered between a clean microkernel
design and striving for performance. AFAICT, they lost on both
fronts.

-O

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/