stupid keykos question
Thu, 24 Sep 1998 11:08:02 -0700
At 10:54 AM 9/24/98 -0400, Jonathan S. Shapiro wrote:
>In EROS, the segment key LSS is 1, and invoking a red segment key
>causes the invocation to be passed through to the keeper.
>Should the same logic apply if the invoked key is a **node** key?
>That is, should opacity be the only difference for purposes of
>I believe that the answer should be yes.
>What did KeyKOS do?
The real answer is at:
A quick look at that stuff, and rusty memory, leaves me with this impression:
A "segmode" key includes both node and segment keys. There are two ways to
encounter a segmode key. (1) It can be the direct target of an invocation,
or (2) It can be encountered while attempting to resolve a memory reference.
If you encounter it through (1), then if it is a node key, it responds to
the node key operations, and if it is a segment key with a keeper (etc.
etc.) the call goes to the keeper.
In case (2), a node key always acts to define a segment, which may be red
or black. If it is red, it may define a keeper, which will be called to
resolve memory faults given that the NO_CALL bits permit the call.
I hope this response helps answer the question.