On 12/03/2015 08:11 AM, Joonsoo Kim wrote:
It doesn't make sense that we reset defer counter
in compaction_defer_reset() when compaction request under the order of
compact_order_failed succeed. Fix it.
Right.
And, it does make sense that giving enough chance for updated failed
order compaction before deferring. Change it.
Sorry, can't understand the meaning here. From the code it seems that
you want to reset defer_shift to 0 instead of increasing it, when the
current order is lower than the failed one? That makes sense, yeah.
How about this?
"On the other hand, when deferring compaction for an order lower than
the current compact_order_failed, we can assume the lower order will
recover more quickly, so we should reset the progress made previously on
compact_defer_shift with the higher order."
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
Acked-by: Vlastimil Babka <vbabka@xxxxxxx>
---
mm/compaction.c | 19 +++++++++++--------
1 file changed, 11 insertions(+), 8 deletions(-)
diff --git a/mm/compaction.c b/mm/compaction.c
index 67b8d90..1a75a6e 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -126,11 +126,14 @@ static struct page *pageblock_pfn_to_page(unsigned long start_pfn,
*/
static void defer_compaction(struct zone *zone, int order)
{
- zone->compact_considered = 0;
- zone->compact_defer_shift++;
-
- if (order < zone->compact_order_failed)
+ if (order < zone->compact_order_failed) {
+ zone->compact_considered = 0;
+ zone->compact_defer_shift = 0;
zone->compact_order_failed = order;
+ } else {
+ zone->compact_considered = 0;
+ zone->compact_defer_shift++;
+ }
if (zone->compact_defer_shift > COMPACT_MAX_DEFER_SHIFT)
zone->compact_defer_shift = COMPACT_MAX_DEFER_SHIFT;
@@ -161,11 +164,11 @@ bool compaction_deferred(struct zone *zone, int order)
/* Update defer tracking counters after successful compaction of given order */
static void compaction_defer_reset(struct zone *zone, int order)
{
- zone->compact_considered = 0;
- zone->compact_defer_shift = 0;
-
- if (order >= zone->compact_order_failed)
+ if (order >= zone->compact_order_failed) {
+ zone->compact_considered = 0;
+ zone->compact_defer_shift = 0;
zone->compact_order_failed = order + 1;
+ }
trace_mm_compaction_defer_reset(zone, order);
}
--
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/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>