Re: Why memory is not demand paged?

Shuveb Hussain wrote:
> Hi,
>
> When a process needs to be migrated in 2.5-om, I understand that whole
> process memory   pages are sent over to the remote node. Is there a
> reason why it is done this way? I was wondering, it would be simple
> (kind of :-D) to load memory pages from the deputy in an on-demand
> fashion by extending the Page Fault handler. I guess after running for
> a while, the process will have a comfortable set of pages to run
> smoothly without faulting.
>
>   
Well, processes are sent with their own address space, processes are 
expected to use most often their very own memory ... that's why, but the 
other aproach could be used as well,

I still think performances are best in the way it's done for now. Still 
I'll remember to implement the second approach and allow it to change 
behaviour on runtime (but that's a todo item ;) )

Florian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV