2015-11-23 12:22:45

by Vladimir Davydov

[permalink] [raw]
Subject: [PATCH] memcg: fix memory.high target

When the memory.high threshold is exceeded, try_charge() schedules a
task_work to reclaim the excess. The reclaim target is set to the number
of pages requested by try_charge(). This is wrong, because try_charge()
usually charges more pages than requested (batch > nr_pages) in order to
refill per cpu stocks. As a result, a process in a cgroup can easily
exceed memory.high significantly when doing a lot of charges w/o
returning to userspace (e.g. reading a file in big chunks).

Fix this issue by assuring that when exceeding memory.high a process
reclaims as many pages as were actually charged (i.e. batch).

Signed-off-by: Vladimir Davydov <[email protected]>
---
mm/memcontrol.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 648cc9f02437..06c476ab0f2c 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2133,7 +2133,7 @@ done_restock:
*/
do {
if (page_counter_read(&memcg->memory) > memcg->high) {
- current->memcg_nr_pages_over_high += nr_pages;
+ current->memcg_nr_pages_over_high += batch;
set_notify_resume(current);
break;
}
--
2.1.4


2015-11-23 16:18:16

by Johannes Weiner

[permalink] [raw]
Subject: Re: [PATCH] memcg: fix memory.high target

On Mon, Nov 23, 2015 at 03:22:31PM +0300, Vladimir Davydov wrote:
> When the memory.high threshold is exceeded, try_charge() schedules a
> task_work to reclaim the excess. The reclaim target is set to the number
> of pages requested by try_charge(). This is wrong, because try_charge()
> usually charges more pages than requested (batch > nr_pages) in order to
> refill per cpu stocks. As a result, a process in a cgroup can easily
> exceed memory.high significantly when doing a lot of charges w/o
> returning to userspace (e.g. reading a file in big chunks).
>
> Fix this issue by assuring that when exceeding memory.high a process
> reclaims as many pages as were actually charged (i.e. batch).
>
> Signed-off-by: Vladimir Davydov <[email protected]>

Acked-by: Johannes Weiner <[email protected]>

2015-11-24 16:39:03

by Michal Hocko

[permalink] [raw]
Subject: Re: [PATCH] memcg: fix memory.high target

On Mon 23-11-15 15:22:31, Vladimir Davydov wrote:
> When the memory.high threshold is exceeded, try_charge() schedules a
> task_work to reclaim the excess. The reclaim target is set to the number
> of pages requested by try_charge(). This is wrong, because try_charge()
> usually charges more pages than requested (batch > nr_pages) in order to
> refill per cpu stocks. As a result, a process in a cgroup can easily
> exceed memory.high significantly when doing a lot of charges w/o
> returning to userspace (e.g. reading a file in big chunks).
>
> Fix this issue by assuring that when exceeding memory.high a process
> reclaims as many pages as were actually charged (i.e. batch).

Good point. This will not affect the single page load because the
reclaim is done in SWAP_CLUSTER_MAX chunks anyway.

> Signed-off-by: Vladimir Davydov <[email protected]>

Acked-by: Michal Hocko <[email protected]>

> ---
> mm/memcontrol.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 648cc9f02437..06c476ab0f2c 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -2133,7 +2133,7 @@ done_restock:
> */
> do {
> if (page_counter_read(&memcg->memory) > memcg->high) {
> - current->memcg_nr_pages_over_high += nr_pages;
> + current->memcg_nr_pages_over_high += batch;
> set_notify_resume(current);
> break;
> }
> --
> 2.1.4

--
Michal Hocko
SUSE Labs