memcg: fix rmdir, force_empty with THP

Now, when THP is enabled, memcg's rmdir() function is broken because
move_account() for THP page is not supported.

This will cause account leak or -EBUSY issue at rmdir().
This patch fixes the issue by supporting move_account() THP pages.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 7a94ef6..5b562b3 100644 (file)
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2197,8 +2197,11 @@ void mem_cgroup_split_huge_fixup(struct page *head, struct page *tail)
  */
 
 static void __mem_cgroup_move_account(struct page_cgroup *pc,
-       struct mem_cgroup *from, struct mem_cgroup *to, bool uncharge)
+       struct mem_cgroup *from, struct mem_cgroup *to, bool uncharge,
+       int charge_size)
 {
+       int nr_pages = charge_size >> PAGE_SHIFT;
+
        VM_BUG_ON(from == to);
        VM_BUG_ON(PageLRU(pc->page));
        VM_BUG_ON(!page_is_cgroup_locked(pc));
@@ -2212,14 +2215,14 @@ static void __mem_cgroup_move_account(struct page_cgroup *pc,
                __this_cpu_inc(to->stat->count[MEM_CGROUP_STAT_FILE_MAPPED]);
                preempt_enable();
        }
-       mem_cgroup_charge_statistics(from, PageCgroupCache(pc), -1);
+       mem_cgroup_charge_statistics(from, PageCgroupCache(pc), -nr_pages);
        if (uncharge)
                /* This is not "cancel", but cancel_charge does all we need. */
-               mem_cgroup_cancel_charge(from, PAGE_SIZE);
+               mem_cgroup_cancel_charge(from, charge_size);
 
        /* caller should have done css_get */
        pc->mem_cgroup = to;
-       mem_cgroup_charge_statistics(to, PageCgroupCache(pc), 1);
+       mem_cgroup_charge_statistics(to, PageCgroupCache(pc), nr_pages);
        /*
         * We charges against "to" which may not have any tasks. Then, "to"
         * can be under rmdir(). But in current implementation, caller of
@@ -2234,15 +2237,19 @@ static void __mem_cgroup_move_account(struct page_cgroup *pc,
  * __mem_cgroup_move_account()
  */
 static int mem_cgroup_move_account(struct page_cgroup *pc,
-               struct mem_cgroup *from, struct mem_cgroup *to, bool uncharge)
+               struct mem_cgroup *from, struct mem_cgroup *to,
+               bool uncharge, int charge_size)
 {
        int ret = -EINVAL;
        unsigned long flags;
 
+       if ((charge_size > PAGE_SIZE) && !PageTransHuge(pc->page))
+               return -EBUSY;
+
        lock_page_cgroup(pc);
        if (PageCgroupUsed(pc) && pc->mem_cgroup == from) {
                move_lock_page_cgroup(pc, &flags);
-               __mem_cgroup_move_account(pc, from, to, uncharge);
+               __mem_cgroup_move_account(pc, from, to, uncharge, charge_size);
                move_unlock_page_cgroup(pc, &flags);
                ret = 0;
        }
@@ -2267,6 +2274,8 @@ static int mem_cgroup_move_parent(struct page_cgroup *pc,
        struct cgroup *cg = child->css.cgroup;
        struct cgroup *pcg = cg->parent;
        struct mem_cgroup *parent;
+       int charge = PAGE_SIZE;
+       unsigned long flags;
        int ret;
 
        /* Is ROOT ? */
@@ -2278,17 +2287,23 @@ static int mem_cgroup_move_parent(struct page_cgroup *pc,
                goto out;
        if (isolate_lru_page(page))
                goto put;
+       /* The page is isolated from LRU and we have no race with splitting */
+       charge = PAGE_SIZE << compound_order(page);
 
        parent = mem_cgroup_from_cont(pcg);
-       ret = __mem_cgroup_try_charge(NULL, gfp_mask, &parent, false,
-                                     PAGE_SIZE);
+       ret = __mem_cgroup_try_charge(NULL, gfp_mask, &parent, false, charge);
        if (ret || !parent)
                goto put_back;
 
-       ret = mem_cgroup_move_account(pc, child, parent, true);
+       if (charge > PAGE_SIZE)
+               flags = compound_lock_irqsave(page);
+
+       ret = mem_cgroup_move_account(pc, child, parent, true, charge);
        if (ret)
-               mem_cgroup_cancel_charge(parent, PAGE_SIZE);
+               mem_cgroup_cancel_charge(parent, charge);
 put_back:
+       if (charge > PAGE_SIZE)
+               compound_unlock_irqrestore(page, flags);
        putback_lru_page(page);
 put:
        put_page(page);
@@ -4868,7 +4883,7 @@ retry:
                                goto put;
                        pc = lookup_page_cgroup(page);
                        if (!mem_cgroup_move_account(pc,
-                                               mc.from, mc.to, false)) {
+                                       mc.from, mc.to, false, PAGE_SIZE)) {
                                mc.precharge--;
                                /* we uncharge from mc.from later. */
                                mc.moved_charge++;