At 1:58 AM 3/29/96 -0500, Jonathan S. Shapiro wrote:
>I've been wrestling with device configuration. My (admittedly
>sketchy) understanding of KeyKOS is that each device is handled more
>or less specially, and no attempt is made to have a common interface
>across devices beyond the degree of commonality that is implicit in
>the IBM channel architecture.
That is basically correct. The only device the kernel handled directly was the disk. (And the Tymnet adapeter, but that was a mistake.)
>Roughly speaking, where I'm going here is to take the virtualization
>provided by the 370 channel subsystem and implement that in the
>supervisor.
I think that is a good way to go. Anything more than that should be done outside the kernel. (Back in the 60s, one of the big deal features of IBM's OS/360 was the ability for a application program to have its sequental output directed to either a tape or a disk transparently to the application. That level of device independence belongs in domain interface standards.)
Bill Frantz | The CDA means | Periwinkle -- Computer Consulting (408)356-8506 | lost jobs and | 16345 Englewood Ave. frantz@netcom.com | dead teenagers | Los Gatos, CA 95032, USA