On 04/03/2014 10:57 AM, Heesub Shin wrote:
Free scanner does not works well on systems having zones which do not
span to pageblock-aligned boundary.
zone->compact_cached_free_pfn is reset when the migration and free
scanner across or compaction restarts. After the reset, if end_pfn of
the zone was not aligned to pageblock_nr_pages, free scanner tries to
isolate free pages from the middle of pageblock to the end, which can
be very small range.
Hm good catch. I think that the same problem can happen (at least
theoretically) through zone->compact_cached_free_pfn with
CONFIG_HOLES_IN_ZONE enabled. Then compact_cached_free_pfn could be set
to a non-aligned-to-pageblock pfn and spoil scans. I'll send a patch
that solves it on isolate_freepages() level, which allows further
simplification of the function.
Vlastimil
Signed-off-by: Heesub Shin <heesub.s...@samsung.com>
Cc: Dongjun Shin <d.j.s...@samsung.com>
Cc: Sunghwan Yun <sunghwan....@samsung.com>
---
mm/compaction.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mm/compaction.c b/mm/compaction.c
index 1ef9144..fefe1da 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -983,7 +983,7 @@ static int compact_zone(struct zone *zone, struct
compact_control *cc)
*/
cc->migrate_pfn = zone->compact_cached_migrate_pfn;
cc->free_pfn = zone->compact_cached_free_pfn;
- if (cc->free_pfn < start_pfn || cc->free_pfn > end_pfn) {
+ if (cc->free_pfn < start_pfn || cc->free_pfn >= end_pfn) {
cc->free_pfn = end_pfn & ~(pageblock_nr_pages-1);
zone->compact_cached_free_pfn = cc->free_pfn;
}
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
