Re: Porting the Hurd to VK (Was: running hurd in linux)

[Adam Olsen]

On Mon, Aug 13, 2001 at 11:15:48AM +0200, Neal H Walfield wrote:
> > I think the solution here is that all device access should be done
> > through hurdish interfaces.  Only the parent would have real access to
> > the mach devices.  The subhurd would have to have access given to it,
> > eg via mach ports.  This also contributes to portability, since you'd
> > have defined interfaces for the device drivers to export.
> 
> You appear not to understand is what a subhurd is; do not consider it a
> child, it is a peer to the original Hurd.

I understand that.  But having two things trying to play directly with
the hardware is bound to cause conflicts.  It makes more sense to
export an interface that can handle conflicts, so why not one of a
hurd server?


-- 
Adam Olsen, aka Rhamphoryncus