On Thu, 26 Jan 2012 14:59:58 -0500
Rik van Riel <riel@xxxxxxxxxx> wrote:
> With CONFIG_COMPACTION enabled, kswapd does not try to free
> contiguous free pages, even when it is woken for a higher order
> request.
>
> This could be bad for eg. jumbo frame network allocations, which
> are done from interrupt context and cannot compact memory themselves.
> Higher than before allocation failure rates in the network receive
> path have been observed in kernels with compaction enabled.
>
> Teach kswapd to defragment the memory zones in a node, but only
> if required and compaction is not deferred in a zone.
>
> ...
>
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -2673,6 +2673,7 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
> int priority;
> int i;
> int end_zone = 0; /* Inclusive. 0 = ZONE_DMA */
> + int zones_need_compaction = 1;
> unsigned long total_scanned;
> struct reclaim_state *reclaim_state = current->reclaim_state;
> unsigned long nr_soft_reclaimed;
> @@ -2937,9 +2938,17 @@ out:
> goto loop_again;
> }
>
> + /* Check if the memory needs to be defragmented. */
> + if (zone_watermark_ok(zone, order,
> + low_wmark_pages(zone), *classzone_idx, 0))
> + zones_need_compaction = 0;
> +
> /* If balanced, clear the congested flag */
> zone_clear_flag(zone, ZONE_CONGESTED);
> }
> +
> + if (zones_need_compaction)
> + compact_pgdat(pgdat, order);
> }
Nicer:
From: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Subject: vmscan-kswapd-carefully-call-compaction-fix
reduce scope of zones_need_compaction
Cc: Andrea Arcangeli <aarcange@xxxxxxxxxx>
Cc: Hillf Danton <dhillf@xxxxxxxxx>
Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx>
Cc: Mel Gorman <mel@xxxxxxxxx>
Cc: Minchan Kim <minchan.kim@xxxxxxxxx>
Cc: Rik van Riel <riel@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
mm/vmscan.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
--- a/mm/vmscan.c~vmscan-kswapd-carefully-call-compaction-fix
+++ a/mm/vmscan.c
@@ -2672,7 +2672,6 @@ static unsigned long balance_pgdat(pg_da
int priority;
int i;
int end_zone = 0; /* Inclusive. 0 = ZONE_DMA */
- int zones_need_compaction = 1;
unsigned long total_scanned;
struct reclaim_state *reclaim_state = current->reclaim_state;
unsigned long nr_soft_reclaimed;
@@ -2920,6 +2919,8 @@ out:
* and it is potentially going to sleep here.
*/
if (order) {
+ int zones_need_compaction = 1;
+
for (i = 0; i <= end_zone; i++) {
struct zone *zone = pgdat->node_zones + i;
_
(could have given it type "bool", but that seems unnecessary when it
has "needs" in the name)
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>