The patch titled
ZVC/zone_reclaim: Leave 1% of unmapped pagecache pages for file I/O
has been added to the -mm tree. Its filename is
zvc-zone_reclaim-leave-1%-of-unmapped-pagecache-pages-for-file-i-o.patch
See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this
------------------------------------------------------
Subject: ZVC/zone_reclaim: Leave 1% of unmapped pagecache pages for file I/O
From: Christoph Lameter <clameter@xxxxxxx>
It turns out that it is advantageous to leave a small portion of unmapped file
backed pages if all of a zone's pages (or almost all pages) are allocated and
so the page allocator has to go off-node.
This allows recently used file I/O buffers to stay on the node and
reduces the times that zone reclaim is invoked if file I/O occurs
when we run out of memory in a zone.
The problem is that zone reclaim runs too frequently when the page cache is
used for file I/O (read write and therefore unmapped pages!) alone and we have
almost all pages of the zone allocated. Zone reclaim may remove 32 unmapped
pages. File I/O will use these pages for the next read/write requests and the
unmapped pages increase. After the zone has filled up again zone reclaim will
remove it again after only 32 pages. This cycle is too inefficient and there
are potentially too many zone reclaim cycles.
With the 1% boundary we may still remove all unmapped pages for file I/O in
zone reclaim pass. However. it will take a large number of read and writes
to get back to 1% again where we trigger zone reclaim again.
The zone reclaim 2.6.16/17 does not show this behavior because we have a 30
second timeout.
Signed-off-by: Christoph Lameter <clameter@xxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxx>
---
mm/vmscan.c | 20 ++++++++++++--------
1 file changed, 12 insertions(+), 8 deletions(-)
diff -puN mm/vmscan.c~zvc-zone_reclaim-leave-1%-of-unmapped-pagecache-pages-for-file-i-o mm/vmscan.c
--- a/mm/vmscan.c~zvc-zone_reclaim-leave-1%-of-unmapped-pagecache-pages-for-file-i-o
+++ a/mm/vmscan.c
@@ -1590,18 +1590,22 @@ int zone_reclaim(struct zone *zone, gfp_
int node_id;
/*
- * Do not reclaim if there are not enough reclaimable pages in this
- * zone that would satify this allocations.
+ * Zone reclaim reclaims unmapped file backed pages.
*
- * All unmapped pagecache pages are reclaimable.
+ * A small portion of unmapped file backed pages is needed for
+ * file I/O otherwise pages read by file I/O will be immediately
+ * thrown out if the zone is overallocated. So we do not reclaim
+ * if less than 1% of the zone is used by unmapped file backed pages.
*
- * Both counters may be temporarily off a bit so we use
- * SWAP_CLUSTER_MAX as the boundary. It may also be good to
- * leave a few frequently used unmapped pagecache pages around.
+ * The division by 128 approximates this and is here because a division
+ * would be too expensive in this hot code path.
+ *
+ * Is it be useful to have a way to set the limit via /proc?
*/
if (zone_page_state(zone, NR_FILE_PAGES) -
- zone_page_state(zone, NR_FILE_MAPPED) < SWAP_CLUSTER_MAX)
- return 0;
+ zone_page_state(zone, NR_FILE_MAPPED) <
+ zone->present_pages / 128)
+ return 0;
/*
* Avoid concurrent zone reclaims, do not reclaim in a zone that does
_
Patches currently in -mm which might be from clameter@xxxxxxx are
origin.patch
zoned-vm-counters-create-vmstatc-h-from-page_allocc-h.patch
zoned-vm-counters-basic-zvc-zoned-vm-counter-implementation.patch
zoned-vm-counters-convert-nr_mapped-to-per-zone-counter.patch
zoned-vm-counters-conversion-of-nr_pagecache-to-per-zone-counter.patch
zoned-vm-counters-remove-nr_file_mapped-from-scan-control-structure.patch
zoned-vm-counters-split-nr_anon_pages-off-from-nr_file_mapped.patch
zoned-vm-counters-zone_reclaim-remove-proc-sys-vm-zone_reclaim_interval.patch
zoned-vm-counters-conversion-of-nr_slab-to-per-zone-counter.patch
zoned-vm-counters-conversion-of-nr_pagetables-to-per-zone-counter.patch
zoned-vm-counters-conversion-of-nr_dirty-to-per-zone-counter.patch
zoned-vm-counters-conversion-of-nr_writeback-to-per-zone-counter.patch
zoned-vm-counters-conversion-of-nr_unstable-to-per-zone-counter.patch
zoned-vm-counters-conversion-of-nr_bounce-to-per-zone-counter.patch
zoned-vm-counters-remove-useless-struct-wbs.patch
use-zoned-vm-counters-for-numa-statistics-v3.patch
light-weight-event-counters-v5.patch
slab-consolidate-code-to-free-slabs-from-freelist.patch
usb-remove-empty-destructor-from-drivers-usb-mon-mon_textc.patch
zvc-zone_reclaim-leave-1%-of-unmapped-pagecache-pages-for-file-i-o.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
