2008-06-22 00:18:48

by Christoph Lameter

[permalink] [raw]
Subject: Re: 2.6.26-rc: nfsd hangs for a few sec

On Sat, 21 Jun 2008, Mel Gorman wrote:

> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1249,15 +1249,13 @@ static unsigned long shrink_zone(int priority, struct zone *zone,
> static unsigned long shrink_zones(int priority, struct zonelist *zonelist,
> struct scan_control *sc)
> {
> + enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
> unsigned long nr_reclaimed = 0;
> - struct zone **zones = zonelist->zones;
> - int i;
> -
> + struct zone **z;
> + struct zone *zone;
>
> sc->all_unreclaimable = 1;
> - for (i = 0; zones[i] != NULL; i++) {
> - struct zone *zone = zones[i];
> -
> + for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
> if (!populated_zone(zone))
> continue;
> /*
>
> Code before - Walk the zonelist for GFP_KERNEL

Before the change we walk all zones of the zonelist.

> Code after - Filter zonelist based on what is allowed for GFP_KERNEL

After the change we walk only zones for GFP_KERNEL. Meaning no HIGHMEM
and MOVABLE zones. Doesnt that mean that reclaim is limited to ZONE_DMA
and ZONE_NORMAL? Is that really intended?

If not then the following patch should return us to old behavior:

---
mm/vmscan.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)

Index: linux-2.6/mm/vmscan.c
===================================================================
--- linux-2.6.orig/mm/vmscan.c 2008-06-21 17:15:45.597627317 -0700
+++ linux-2.6/mm/vmscan.c 2008-06-21 17:17:16.273293260 -0700
@@ -1249,13 +1249,12 @@ static unsigned long shrink_zone(int pri
static unsigned long shrink_zones(int priority, struct zonelist *zonelist,
struct scan_control *sc)
{
- enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
unsigned long nr_reclaimed = 0;
struct zoneref *z;
struct zone *zone;

sc->all_unreclaimable = 1;
- for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
+ for_each_zone_zonelist(zone, z, zonelist, MAX_NR_ZONES - 1) {
if (!populated_zone(zone))
continue;
/*