lkml.org 
[lkml]   [2011]   [Aug]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 8/8] mm: vmscan: Do not writeback filesystem pages from kswapd
On Thu, Jul 21, 2011 at 05:28:50PM +0100, Mel Gorman wrote:
> Assuming that flusher threads will always write back dirty pages promptly
> then it is always faster for reclaimers to wait for flushers. This patch
> prevents kswapd writing back any filesystem pages.
>
> Signed-off-by: Mel Gorman <mgorman@suse.de>

Relying on the flushers may mean that every dirty page in the system
has to be written back before the pages from the zone of interest are
clean.

De-facto we have only one mechanism to stay on top of the dirty pages
from a per-zone perspective, and that is single-page writeout from
reclaim.

While we all agree that this sucks, we can not remove it unless we
have a replacement that makes zones reclaimable in a reasonable time
frame (or keep them reclaimable in the first place, what per-zone
dirty limits attempt to do).

As such, please include

Nacked-by: Johannes Weiner <jweiner@redhat.com>


\
 
 \ /
  Last update: 2011-08-03 13:39    [from the cache]
©2003-2011 Jasper Spaans