Re: The hall of bloat

[Alan Cox <alan lxorguk ukuu org uk>]

On Llu, 2005-05-16 at 06:45, Mark McLoughlin wrote:
> On Sun, 2005-05-15 at 13:57 -0500, William Lovaton wrote:
> 
> > Reading your little chat with the kernel guys it seems that there is
> > little benefit (or none) in reducing VM Size, why is that?
> 
> 	Because the kernel won't back this mapping with physical memory until
> the application actually tries to use it - i.e. only when the app tries
> to write to any these pages does the kernel get a page fault and

There is a cost if the mapping has just a few random pages filled as you
have to fill in the whole page table heirarchy down to the page in
question. It will also hurt you badly on a system running with
overcommit disabled as is often done on server systems. 

In overcommit disable modes you must have enough swap + 50% of ram to
back all virtual address space allocated (so that all out of memory is
reliably detected). Gnome currently runs ok in that mode although
evolution 2.* didn't last time I tried it.

Alan