On Thu, Jan 27, 2011 at 11:26:19AM +0200, Michael S. Tsirkin wrote:

On Thu, Jan 27, 2011 at 11:21:47AM +0200, Avi Kivity wrote:

quoted

On 01/26/2011 02:08 PM, Michael S. Tsirkin wrote:

quoted

quoted

 >
 >I just mean that once you fault you map sptes and then you can use them
 >without exits.  mmio will cause exits each time. Right?

 The swapper scanning sptes, ksmd, khugepaged, and swapping can all
 cause a page to be unmapped.  Though it should certainly happen with
 a much lower frequency than mmio.

Right. That's why I say that sorting by size might not be optimal.
Maybe a cache ...

Why would it not be optimal?

If you have 16GB RAM in two slots and a few megabytes here and there
scattered in some slots, you have three orders of magnitudes to
spare.

Yes but you might not be accessing all that RAM.
Maybe your workset is just tens of megabytes.

Anyway, what I am saying this is all just guesswork.
Some kind of cache sounds, to me, like a better
approach if we have a ton of slots.

quoted

-- 
error compiling committee.c: too many arguments to function