Re: [U]VM subsystem for L4 X.2 (and later)

[Farid Hajji]

Hi Espen,

> > So, the issue of an L4 pmap seems resolvable, both for current L4
> > (say: hazelnut) and for the upcoming Version 4 spec. It is just a
> > matter of using L4's memory mapping primitives inside a new pmap.c
> > to implement the functions required in pmap(9).
> 
> Question: how does UVM deal with multiple page sizes?  Can it deal
> with multiple page sizes at all?  Being able to do so can prove very
> beneficial since the TLB working sets are effectively reduced and such
> reductions can probably have a high impact on system performance (in
> particual in multiserver systems where one has to take into account
> the working sets of all system components).

hmmm... it looks like vm_pages currently refer to fixed-size pages
throughout UVM. I doubt that multiple page sizes would be possible,
at least not if they are to be freely mixed. I could imagine that
multiple pools of pages with fixed sizes could be maintained separatly
though. Take this with a grain of salt: I'm not the expert here; we'll
have to ask Chuck for an authoritative answer on UVM multiple page
sizes first.

> I can't really tell how much reductions in the TLB working set really
> helps (maybe the UNSW people can shed some light on that), but one
> should at least try not to use a design which rules out such
> optimizations.
> 
>       eSk

Regards,

-Farid.

-- 
Farid Hajji -- Unix Systems and Network Admin | Phone: +49-2131-67-555
Broicherdorfstr. 83, D-41564 Kaarst, Germany  | address@hidden
- - - - - - - - - - - - - - - - - - - - - - - + - - - - - - - - - - - -
One OS To Rule Them All And In The Darkness Bind Them... --Bill Gates.