Commit ae3abae64f177586be55b04a7fb7047a34b21a3e

Authored by Daisuke Nishimura
Committed by Linus Torvalds
1 parent 0816178638

memcg: fix mem_cgroup_shrink_usage()

Current mem_cgroup_shrink_usage() has two problems.

1. It doesn't call mem_cgroup_out_of_memory and doesn't update
   last_oom_jiffies, so pagefault_out_of_memory invokes global OOM.

2. Considering hierarchy, shrinking has to be done from the
   mem_over_limit, not from the memcg which the page would be charged to.

mem_cgroup_try_charge_swapin() does all of these things properly, so we
use it and call cancel_charge_swapin when it succeeded.

The name of "shrink_usage" is not appropriate for this behavior, so we
change it too.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.cn>
Cc: Paul Menage <menage@google.com>
Cc: Dhaval Giani <dhaval@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Showing 3 changed files with 20 additions and 25 deletions Side-by-side Diff

include/linux/memcontrol.h
... ... @@ -56,7 +56,7 @@
56 56 enum lru_list from, enum lru_list to);
57 57 extern void mem_cgroup_uncharge_page(struct page *page);
58 58 extern void mem_cgroup_uncharge_cache_page(struct page *page);
59   -extern int mem_cgroup_shrink_usage(struct page *page,
  59 +extern int mem_cgroup_shmem_charge_fallback(struct page *page,
60 60 struct mm_struct *mm, gfp_t gfp_mask);
61 61  
62 62 extern unsigned long mem_cgroup_isolate_pages(unsigned long nr_to_scan,
... ... @@ -155,7 +155,7 @@
155 155 {
156 156 }
157 157  
158   -static inline int mem_cgroup_shrink_usage(struct page *page,
  158 +static inline int mem_cgroup_shmem_charge_fallback(struct page *page,
159 159 struct mm_struct *mm, gfp_t gfp_mask)
160 160 {
161 161 return 0;
... ... @@ -1617,37 +1617,28 @@
1617 1617 }
1618 1618  
1619 1619 /*
1620   - * A call to try to shrink memory usage under specified resource controller.
1621   - * This is typically used for page reclaiming for shmem for reducing side
1622   - * effect of page allocation from shmem, which is used by some mem_cgroup.
  1620 + * A call to try to shrink memory usage on charge failure at shmem's swapin.
  1621 + * Calling hierarchical_reclaim is not enough because we should update
  1622 + * last_oom_jiffies to prevent pagefault_out_of_memory from invoking global OOM.
  1623 + * Moreover considering hierarchy, we should reclaim from the mem_over_limit,
  1624 + * not from the memcg which this page would be charged to.
  1625 + * try_charge_swapin does all of these works properly.
1623 1626 */
1624   -int mem_cgroup_shrink_usage(struct page *page,
  1627 +int mem_cgroup_shmem_charge_fallback(struct page *page,
1625 1628 struct mm_struct *mm,
1626 1629 gfp_t gfp_mask)
1627 1630 {
1628 1631 struct mem_cgroup *mem = NULL;
1629   - int progress = 0;
1630   - int retry = MEM_CGROUP_RECLAIM_RETRIES;
  1632 + int ret;
1631 1633  
1632 1634 if (mem_cgroup_disabled())
1633 1635 return 0;
1634   - if (page)
1635   - mem = try_get_mem_cgroup_from_swapcache(page);
1636   - if (!mem && mm)
1637   - mem = try_get_mem_cgroup_from_mm(mm);
1638   - if (unlikely(!mem))
1639   - return 0;
1640 1636  
1641   - do {
1642   - progress = mem_cgroup_hierarchical_reclaim(mem,
1643   - gfp_mask, true, false);
1644   - progress += mem_cgroup_check_under_limit(mem);
1645   - } while (!progress && --retry);
  1637 + ret = mem_cgroup_try_charge_swapin(mm, page, gfp_mask, &mem);
  1638 + if (!ret)
  1639 + mem_cgroup_cancel_charge_swapin(mem); /* it does !mem check */
1646 1640  
1647   - css_put(&mem->css);
1648   - if (!retry)
1649   - return -ENOMEM;
1650   - return 0;
  1641 + return ret;
1651 1642 }
1652 1643  
1653 1644 static DEFINE_MUTEX(set_limit_mutex);
... ... @@ -1340,8 +1340,12 @@
1340 1340 shmem_swp_unmap(entry);
1341 1341 spin_unlock(&info->lock);
1342 1342 if (error == -ENOMEM) {
1343   - /* allow reclaim from this memory cgroup */
1344   - error = mem_cgroup_shrink_usage(swappage,
  1343 + /*
  1344 + * reclaim from proper memory cgroup and
  1345 + * call memcg's OOM if needed.
  1346 + */
  1347 + error = mem_cgroup_shmem_charge_fallback(
  1348 + swappage,
1345 1349 current->mm,
1346 1350 gfp);
1347 1351 if (error) {