On Sat, Dec 31, 2011 at 02:18:43PM +0800, Huang Shijie wrote:We get cc->order is -1 when user echos to /proc/sys/vm/compact_memory. In this case, we should check that if we have enough pages for the compaction in the zone. If we do not check this, in our MX6Q board(arm), i ever observed COMPACT_CLUSTER_MAX pages were compaction failed in per migrate_pages(). That's mean we can not alloc any pages by the free scanner in the zone. This patch checks the watermark to avoid this problem. Signed-off-by: Huang Shijie<b32955@xxxxxxxxxxxxx> --- mm/compaction.c | 14 ++++++++++++-- 1 files changed, 12 insertions(+), 2 deletions(-) diff --git a/mm/compaction.c b/mm/compaction.c index 899d956..0f12cc9 100644 --- a/mm/compaction.c +++ b/mm/compaction.c @@ -442,8 +442,13 @@ static int compact_finished(struct zone *zone, * order == -1 is expected when compacting via * /proc/sys/vm/compact_memory */ - if (cc->order == -1) + if (cc->order == -1) { + /* Check if we have enough pages now. */ + watermark = low_wmark_pages(zone) + COMPACT_CLUSTER_MAX * 2; + if (!zone_watermark_ok(zone, 0, watermark, 0, 0)) + return COMPACT_SKIPPED; return COMPACT_CONTINUE; + }We already do the watermark check in compact_finished. Would moving the cc->order == -1 check below it not be functionally equivalent? This
Now, I think these code for compact_finish() is not needed. These lines can be removed.
The watermark check in the compaction_suitable() is enough to avoid the problem.
Huang Shijie
would be preferable to duplicating the code for the watermark check. Same for the second check in your patch.
-- 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>