Re: [patch 7/8] vmscan: memcg-aware unevictable page rescue scanner

From: Ying Han
Date: Mon Aug 29 2011 - 03:28:32 EST


On Tue, Jul 19, 2011 at 5:36 PM, Johannes Weiner <hannes@xxxxxxxxxxx> wrote:
> On Tue, Jul 19, 2011 at 03:47:43PM -0700, Ying Han wrote:
>> On Tue, May 31, 2011 at 11:25 PM, Johannes Weiner <hannes@xxxxxxxxxxx>wrote:
>>
>> > Once the per-memcg lru lists are exclusive, the unevictable page
>> > rescue scanner can no longer work on the global zone lru lists.
>> >
>> > This converts it to go through all memcgs and scan their respective
>> > unevictable lists instead.
>> >
>> > Signed-off-by: Johannes Weiner <hannes@xxxxxxxxxxx>
>> > ---
>> >  include/linux/memcontrol.h |    2 +
>> >  mm/memcontrol.c            |   11 +++++++++
>> >  mm/vmscan.c                |   53
>> > +++++++++++++++++++++++++++----------------
>> >  3 files changed, 46 insertions(+), 20 deletions(-)
>> >
>> > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
>> > index cb02c00..56c1def 100644
>> > --- a/include/linux/memcontrol.h
>> > +++ b/include/linux/memcontrol.h
>> > @@ -60,6 +60,8 @@ extern void mem_cgroup_cancel_charge_swapin(struct
>> > mem_cgroup *ptr);
>> >
>> >  extern int mem_cgroup_cache_charge(struct page *page, struct mm_struct
>> > *mm,
>> >                                        gfp_t gfp_mask);
>> > +struct page *mem_cgroup_lru_to_page(struct zone *, struct mem_cgroup *,
>> > +                                   enum lru_list);
>> >
>>
>> Did we miss a #ifdef case for this function? I got compile error by
>> disabling memcg.
>
> I assume it's because the call to it is not optimized away properly in
> the disabled case.  I'll have it fixed in the next round, thanks for
> letting me know.
>

Hi Johannes:

This is the change for the hierarchy_walk() sent on the other patch,
also including a fix. Please consider to fold in your patch:

Fix the hierarchy_walk() in the unevictable page rescue scanner

the patch including changes
1. adjust the change in hierarchy_walk() which needs to hold the reference to
the first mem_cgroup.
2. add stop_hierarchy_walk() at the end which is missed on the original patch.

Signed-off-by: Ying Han <yinghan@xxxxxxxxxx>

Change-Id: I72fb5d351faf0f111c8c99edd90b6cfee6281d3f
---
mm/memcontrol.c | 3 +++
mm/vmscan.c | 7 ++++---
2 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 9bcd429..426092b 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1514,6 +1514,9 @@ void mem_cgroup_stop_hierarchy_walk(struct
mem_cgroup *target,
>------>------->------->------- struct mem_cgroup *first,
>------>------->------->------- struct mem_cgroup *mem)
{
+>------if (!target)
+>------>-------target = root_mem_cgroup;
+
>------if (mem && mem != target)
>------>-------css_put(&mem->css);
·
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 290998e..fd9593b 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -4110,9 +4110,9 @@ static struct page *lru_tailpage(struct zone
*zone, struct mem_cgroup *mem,
#define SCAN_UNEVICTABLE_BATCH_SIZE 16UL /* arbitrary lock hold batch size */
static void scan_zone_unevictable_pages(struct zone *zone)
{
->------struct mem_cgroup *first, *mem = NULL;
+>------struct mem_cgroup *first, *mem;
·
->------first = mem = mem_cgroup_hierarchy_walk(NULL, mem);
+>------first = mem = mem_cgroup_hierarchy_walk(NULL, NULL, NULL);
>------do {
>------>-------unsigned long nr_to_scan;
·
@@ -4139,8 +4139,9 @@ static void scan_zone_unevictable_pages(struct zone *zone)
>------>------->-------spin_unlock_irq(&zone->lru_lock);
>------>------->-------nr_to_scan -= batch_size;
>------>-------}
->------>-------mem = mem_cgroup_hierarchy_walk(NULL, mem);
+>------>-------mem = mem_cgroup_hierarchy_walk(NULL, first, mem);
>------} while (mem != first);
+>------mem_cgroup_stop_hierarchy_walk(NULL, first, mem);
}

--Ying
--
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/