Re: VMM

[Niels Möller]

address@hidden (Neal H. Walfield) writes:

> Most recently used is a farce: it does not reflect the "big picture."
> Many applications will benefit from the ability to make decisions
> about the memory they use: there are many papers covering this topic.

What are the alternatives to most recently used, if one is looking for
a default "one-size-fits-all" algorithm? I guess libc would have to
implement most recently used for proceses that don't provide any more
information or hints to the momory manager. And the physical memory
server has to export the needed information.

> Nope.  If the client fails to evict the require pages in the amount of
> time given, the physical memory server revokes all of the clients
> contracts and reclaims the memory.

I wonder how hard it is to make this robust under heavy system load.
Some things that might help: Make sure that the paging code itself,
and its data structures, are never paged out. Design the contracts so
that the time proceses are given to respond are specified as cpu time
of the process in question, not as real time.

/Niels