Naive question about grant/map

[Kevin Elphinstone]

Volkmar Uhlig <mailto:volkmar at ira.uka.de> scribbled on Thursday, 11 December
2003 5:59 PM:

>> sense of the word.
> 
> This also depends on your point of view.  Your statement that the
> kernel 
> mapping can be a superset of user-level pagers is wrong.  The
> user-level 
> pagers have full knowledge about the replacement policy (since it is
> implemented in user land via map and unmap) and therefore can rely on
> the fact that the kernel won't replace entries in the cache.  If it
> would be a superset then the kernel would have to enter values itself.

I believe I am not wrong, just mis-interpreted :-). The superset I refer it
is where pager effective discards the information required to re-establish
the mapping after establishing it and relies on the kernel to hold the only
copy, i.e. the kernel posseses a superset of the mapping stored by the
pager.

Older versions of sigma0 (maybe the current version too) did not keep track
of mappings they handed out, only the fact that a page had been mapped. This
is a concrete example of what I meant by a superset. Only the kernel has the
actual mapping established and it was lost if a client of sigma zero flushed
it.

I'm not arguing that it is a sensible thing to do in all but the most
controlled situations (and maybe not there as well), but it can be done.

	- Kevin