Natty SRU: vmscan: fix a livelock in kswapd
On Wed, Jul 20, 2011 at 03:47:14PM -0600, Tim Gardner wrote:
> From 64ba82a4b63a45fd0c102dd97ea81c59fc522b76 Mon Sep 17 00:00:00 2001
> From: Shaohua Li <firstname.lastname@example.org>
> Date: Tue, 19 Jul 2011 08:49:26 -0700
> Subject: [PATCH] vmscan: fix a livelock in kswapd
> BugLink: http://bugs.launchpad.net/bugs/813797
> I'm running a workload which triggers a lot of swap in a machine with 4
> nodes. After I kill the workload, I found a kswapd livelock. Sometimes
> kswapd3 or kswapd2 are keeping running and I can't access filesystem,
> but most memory is free.
> This looks like a regression since commit 08951e545918c159 ("mm: vmscan:
> correct check for kswapd sleeping in sleeping_prematurely").
> Node 2 and 3 have only ZONE_NORMAL, but balance_pgdat() will return 0
> for classzone_idx. The reason is end_zone in balance_pgdat() is 0 by
> default, if all zones have watermark ok, end_zone will keep 0.
> Later sleeping_prematurely() always returns true. Because this is an
> order 3 wakeup, and if classzone_idx is 0, both balanced_pages and
> present_pages in pgdat_balanced() are 0. We add a special case here.
> If a zone has no page, we think it's balanced. This fixes the livelock.
> Signed-off-by: Shaohua Li <email@example.com>
> Acked-by: Mel Gorman <firstname.lastname@example.org>
> Cc: Minchan Kim <email@example.com>
> Cc: <firstname.lastname@example.org>
> Signed-off-by: Andrew Morton <email@example.com>
> Signed-off-by: Linus Torvalds <firstname.lastname@example.org>
> (cherry picked from commit 4746efded84d7c5a9c8d64d4c6e814ff0cf9fb42)
> Signed-off-by: Tim Gardner <email@example.com>
> mm/vmscan.c | 3 ++-
> 1 files changed, 2 insertions(+), 1 deletions(-)
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 4b8b37c..1e0eefe 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -2245,7 +2245,8 @@ static bool pgdat_balanced(pg_data_t *pgdat, unsigned long balanced_pages,
> for (i = 0; i <= classzone_idx; i++)
> present_pages += pgdat->node_zones[i].present_pages;
> - return balanced_pages > (present_pages >> 2);
> + /* A special case here: if zone has no page, we think it's balanced */
> + return balanced_pages >= (present_pages >> 2);
> /* is kswapd sleeping prematurely? */
Yeeks that is subtle. Looks to do what it claims.
Acked-by: Andy Whitcroft <firstname.lastname@example.org>
kernel-team mailing list