Re: EROS/Coyotos fault handers vs l4 pager hierarchy

[Espen Skoglund]

[Marcus Brinkmann]
> [...]  However, please note: The L4 people consider the in-kernel
> mapping database to be a _cache_ of the real mapping data, which is
> distributed over the various pagers in the system (in our case:
> physmem plus each individual pager in each task).

Just a small note: You should be a bit careful about generalizing
statements like this.  I was only saying that the mapping database
*can* be viewed as a cache for the real mapping information kept in
the pagers.  In particular I was saying this in the context where
someone forcibly revokes memory because of, e.g., memory presure.

There's nothing stopping you from designing a system where there's a
contract between mapper and mappee saying that the mapping will not be
revoked (of course, in this case the mappee must trust the mapper).

        eSk