multithreaded kernel

[Bill Frantz]

At 15:10 7/13/95 -0400, Jonathan Shapiro wrote:
>   Perhaps you will be forced to allocate 2 core frames for each pinned page. 
>   If you get backed into that corner, then pinned/non-checkpointed pages
>   become much more interesting.  However, you probably need some mechanism to
>   notify a domain which trys to use the data in such a page that the data has
>   been lost.
>
>
>Such an application almost certainly needs to be exposed to restarts
>in any case, either through a rescinded device key or some equivalent.

What I worry about is that the sudden occurance of an out of range
(presumably zero) value will cause the domain to trap and it will never
discover that the device key is indicating "restart".


-----------------------------------------------------------------
Bill Frantz                   Periwinkle  --  Computer Consulting
(408)356-8506                 16345 Englewood Ave.
frantz@netcom.com             Los Gatos, CA 95032, USA