On Thu, Jul 21, 2016 at 04:31:56PM +0900, Joonsoo Kim wrote:
> On Wed, Jul 20, 2016 at 04:21:46PM +0100, Mel Gorman wrote:
> > Both Joonsoo Kim and Minchan Kim have reported premature OOM kills on
> > a 32-bit platform. The common element is a zone-constrained high-order
> > allocation failing. Two factors appear to be at fault -- pgdat being
> > considered unreclaimable prematurely and insufficient rotation of the
> > active list.
> >
> > Unfortunately to date I have been unable to reproduce this with a variety
> > of stress workloads on a 2G 32-bit KVM instance. It's not clear why as
> > the steps are similar to what was described. It means I've been unable to
> > determine if this series addresses the problem or not. I'm hoping they can
> > test and report back before these are merged to mmotm. What I have checked
> > is that a basic parallel DD workload completed successfully on the same
> > machine I used for the node-lru performance tests. I'll leave the other
> > tests running just in case anything interesting falls out.
>
> Hello, Mel.
>
> I tested this series and it doesn't solve my problem. But, with this
> series and one change below, my problem is solved.
>
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index f5ab357..d451c29 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1819,7 +1819,7 @@ static void move_active_pages_to_lru(struct lruvec
> *lruvec,
>
> nr_pages = hpage_nr_pages(page);
> update_lru_size(lruvec, lru, page_zonenum(page), nr_pages);
> - list_move(&page->lru, &lruvec->lists[lru]);
> + list_move_tail(&page->lru, &lruvec->lists[lru]);
> pgmoved += nr_pages;
>
> if (put_page_testzero(page)) {
>
> It is brain-dead work-around so it is better you to find a better solution.
I tested below patch roughly and it enhanced performance a lot.
diff --git a/mm/vmscan.c b/mm/vmscan.c
index cd68a18..9061e5a 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1809,7 +1809,8 @@ shrink_inactive_list(unsigned long nr_to_scan, struct
lruvec *lruvec,
static void move_active_pages_to_lru(struct lruvec *lruvec,
struct list_head *list,
struct list_head *pages_to_free,
- enum lru_list lru)
+ enum lru_list lru,
+ bool tail)
{
struct pglist_data *pgdat = lruvec_pgdat(lruvec);
unsigned long pgmoved = 0;
@@ -1825,7 +1826,10 @@ static void move_active_pages_to_lru(struct lruvec
*lruvec,
nr_pages = hpage_nr_pages(page);
update_lru_size(lruvec, lru, page_zonenum(page), nr_pages);
- list_move(&page->lru, &lruvec->lists[lru]);
+ if (!tail)
+ list_move(&page->lru, &lruvec->lists[lru]);
+ else
+ list_move_tail(&page->lru, &lruvec->lists[lru]);
pgmoved += nr_pages;
if (put_page_testzero(page)) {
@@ -1847,6 +1851,47 @@ static void move_active_pages_to_lru(struct lruvec
*lruvec,
__count_vm_events(PGDEACTIVATE, pgmoved);
}
+static bool inactive_list_is_extreme_low(struct lruvec *lruvec, bool file,
+ struct scan_control *sc)
+{
+ unsigned long inactive;
+
+ /*
+ * If we don't have swap space, anonymous page deactivation
+ * is pointless.
+ */
+ if (!file && !total_swap_pages)
+ return false;
+
+ inactive = lruvec_lru_size(lruvec, file * LRU_FILE);
+
+ /*
+ * For global reclaim on zone-constrained allocations, it is necessary
+ * to check if rotations are required for lowmem to be reclaimed. This
+ * calculates the inactive/active pages available in eligible zones.
+ */
+ if (global_reclaim(sc)) {
+ struct pglist_data *pgdat = lruvec_pgdat(lruvec);
+ int zid;
+
+ for (zid = sc->reclaim_idx + 1; zid < MAX_NR_ZONES; zid++) {
+ struct zone *zone = &pgdat->node_zones[zid];
+ unsigned long inactive_zone;
+
+ if (!populated_zone(zone))
+ continue;
+
+ inactive_zone = zone_page_state(zone,
+ NR_ZONE_LRU_BASE + (file * LRU_FILE));
+
+ inactive -= min(inactive, inactive_zone);
+ }
+ }
+
+
+ return inactive <= (SWAP_CLUSTER_MAX * num_online_cpus());
+}
+
static void shrink_active_list(unsigned long nr_to_scan,
struct lruvec *lruvec,
struct scan_control *sc,
@@ -1937,9 +1982,11 @@ static void shrink_active_list(unsigned long nr_to_scan,
* get_scan_count.
*/
reclaim_stat->recent_rotated[file] += nr_rotated;
+ move_active_pages_to_lru(lruvec, &l_active, &l_hold, lru, false);
+ move_active_pages_to_lru(lruvec, &l_inactive,
+ &l_hold, lru - LRU_ACTIVE,
+ inactive_list_is_extreme_low(lruvec, is_file_lru(lru), sc));
- move_active_pages_to_lru(lruvec, &l_active, &l_hold, lru);
- move_active_pages_to_lru(lruvec, &l_inactive, &l_hold, lru -
LRU_ACTIVE);
__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, -nr_taken);
spin_unlock_irq(&pgdat->lru_lock);
