On Thu, 2007-06-28 at 15:24 +0200, Rob van der Heij wrote:
> A higher value of swappiness means that when Linux memory management
> needs some free memory, it is more willing to swap out a process than
> to purge data in cache.
The definition of swappiness is murky. The relevant code snippets and
comments from the source:
/*
* `distress' is a measure of how much trouble we're having
* reclaiming pages. 0 -> no problems. 100 -> great trouble.
*/
distress = 100 >> min(zone->prev_priority, priority);
/*
* The point of this algorithm is to decide when to start
* reclaiming mapped memory instead of just pagecache. Work out
* how much memory
* is mapped.
*/
mapped_ratio = ((global_page_state(NR_FILE_MAPPED) +
global_page_state(NR_ANON_PAGES)) * 100) /
vm_total_pages;
...
/*
* Now decide how much we really want to unmap some pages. The
* mapped ratio is downgraded - just because there's a lot of
* mapped memory doesn't necessarily mean that page reclaim
* isn't succeeding.
*
* The distress ratio is important - we don't want to start
* going oom.
*
* A 100% value of vm_swappiness overrides this algorithm
* altogether.
*/
swap_tendency = mapped_ratio / 2 + distress + sc->swappiness;
...
/*
* Now use this metric to decide whether to start moving mapped
* memory onto the inactive list.
*/
if (swap_tendency >= 100)
force_reclaim_mapped:
reclaim_mapped = 1;
If the distress -- which is defined in terms of zone priorities --
reaches a certain limit, the system starts to swap. try_to_free_pages
repeatedly scans the active/inactive list with a priority value that
starts at 12. The priority is decreased from run to run if not enough
pages can be freed. Black magic.
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
