[PATCH] mm: vmscan: deactivate isolated pages with lru lock released

[Hillf Danton]

Spinners on other CPUs, if any, could take the lru lock and do their jobs while
isolated pages are deactivated on the current CPU if the lock is released
actively. And no risk of race raised as pages are already queued on locally
private list.


Signed-off-by: Hillf Danton <dhillf@xxxxxxxxx>
---

--- a/mm/vmscan.c	Thu Dec 29 20:20:16 2011
+++ b/mm/vmscan.c	Wed Jan 11 20:40:40 2012
@@ -1464,6 +1464,7 @@ update_isolated_counts(struct mem_cgroup
 	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(mz);

 	nr_active = clear_active_flags(isolated_list, count);
+	spin_lock_irq(&zone->lru_lock);
 	__count_vm_events(PGDEACTIVATE, nr_active);

 	__mod_zone_page_state(zone, NR_ACTIVE_FILE,
@@ -1482,6 +1483,7 @@ update_isolated_counts(struct mem_cgroup

 	reclaim_stat->recent_scanned[0] += *nr_anon;
 	reclaim_stat->recent_scanned[1] += *nr_file;
+	spin_unlock_irq(&zone->lru_lock);
 }

 /*
@@ -1577,15 +1579,13 @@ shrink_inactive_list(unsigned long nr_to
 			__count_zone_vm_events(PGSCAN_DIRECT, zone,
 					       nr_scanned);
 	}
+	spin_unlock_irq(&zone->lru_lock);

-	if (nr_taken == 0) {
-		spin_unlock_irq(&zone->lru_lock);
+	if (nr_taken == 0)
 		return 0;
-	}

 	update_isolated_counts(mz, sc, &nr_anon, &nr_file, &page_list);

-	spin_unlock_irq(&zone->lru_lock);

 	nr_reclaimed = shrink_page_list(&page_list, mz, sc, priority,
 						&nr_dirty, &nr_writeback);
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/