lkml.org 
[lkml]   [2012]   [Feb]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 4/6 v4] memcg: use new logic for page stat accounting
On Mon, 13 Feb 2012 23:22:22 -0800
Greg Thelen <gthelen@google.com> wrote:

> On Mon, Feb 13, 2012 at 7:14 PM, KAMEZAWA Hiroyuki
> <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> > From ad2905362ef58a44d96a325193ab384739418050 Mon Sep 17 00:00:00 2001
> > From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> > Date: Thu, 2 Feb 2012 11:49:59 +0900
> > Subject: [PATCH 4/6] memcg: use new logic for page stat accounting.
> >
> > Now, page-stat-per-memcg is recorded into per page_cgroup flag by
> > duplicating page's status into the flag. The reason is that memcg
> > has a feature to move a page from a group to another group and we
> > have race between "move" and "page stat accounting",
> >
> > Under current logic, assume CPU-A and CPU-B. CPU-A does "move"
> > and CPU-B does "page stat accounting".
> >
> > When CPU-A goes 1st,
> >
> >            CPU-A                           CPU-B
> >                                    update "struct page" info.
> >    move_lock_mem_cgroup(memcg)
> >    see flags
>
> pc->flags?
>
yes.


> >    copy page stat to new group
> >    overwrite pc->mem_cgroup.
> >    move_unlock_mem_cgroup(memcg)
> >                                    move_lock_mem_cgroup(mem)
> >                                    set pc->flags
> >                                    update page stat accounting
> >                                    move_unlock_mem_cgroup(mem)
> >
> > stat accounting is guarded by move_lock_mem_cgroup() and "move"
> > logic (CPU-A) doesn't see changes in "struct page" information.
> >
> > But it's costly to have the same information both in 'struct page' and
> > 'struct page_cgroup'. And, there is a potential problem.
> >
> > For example, assume we have PG_dirty accounting in memcg.
> > PG_..is a flag for struct page.
> > PCG_ is a flag for struct page_cgroup.
> > (This is just an example. The same problem can be found in any
> >  kind of page stat accounting.)
> >
> >          CPU-A                               CPU-B
> >      TestSet PG_dirty
> >      (delay)                        TestClear PG_dirty_
>
> PG_dirty
>
> >                                     if (TestClear(PCG_dirty))
> >                                          memcg->nr_dirty--
> >      if (TestSet(PCG_dirty))
> >          memcg->nr_dirty++
> >
>
> > @@ -141,6 +141,31 @@ static inline bool mem_cgroup_disabled(void)
> >        return false;
> >  }
> >
> > +void __mem_cgroup_begin_update_page_stat(struct page *page,
> > +                                       bool *lock, unsigned long *flags);
> > +
> > +static inline void mem_cgroup_begin_update_page_stat(struct page *page,
> > +                                       bool *lock, unsigned long *flags)
> > +{
> > +       if (mem_cgroup_disabled())
> > +               return;
> > +       rcu_read_lock();
> > +       *lock = false;
>
> This seems like a strange place to set *lock=false. I think it's
> clearer if __mem_cgroup_begin_update_page_stat() is the only routine
> that sets or clears *lock. But I do see that in patch 6/6 'memcg: fix
> performance of mem_cgroup_begin_update_page_stat()' this position is
> required.
>

Ah, yes. Hmm, it was better to move this to the body of function.



> > +       return __mem_cgroup_begin_update_page_stat(page, lock, flags);
> > +}
>
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index ecf8856..30afea5 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -1877,32 +1877,54 @@ bool mem_cgroup_handle_oom(struct mem_cgroup *memcg, gfp_t mask)
> >  * If there is, we take a lock.
> >  */
> >
> > +void __mem_cgroup_begin_update_page_stat(struct page *page,
> > +                               bool *lock, unsigned long *flags)
> > +{
> > +       struct mem_cgroup *memcg;
> > +       struct page_cgroup *pc;
> > +
> > +       pc = lookup_page_cgroup(page);
> > +again:
> > +       memcg = pc->mem_cgroup;
> > +       if (unlikely(!memcg || !PageCgroupUsed(pc)))
> > +               return;
> > +       if (!mem_cgroup_stealed(memcg))
> > +               return;
> > +
> > +       move_lock_mem_cgroup(memcg, flags);
> > +       if (memcg != pc->mem_cgroup || !PageCgroupUsed(pc)) {
> > +               move_unlock_mem_cgroup(memcg, flags);
> > +               goto again;
> > +       }
> > +       *lock = true;
> > +}
> > +
> > +void __mem_cgroup_end_update_page_stat(struct page *page,
> > +                               bool *lock, unsigned long *flags)
>
> 'lock' looks like an unused parameter. If so, then remove it.
>

Ok.

> > +{
> > +       struct page_cgroup *pc = lookup_page_cgroup(page);
> > +
> > +       /*
> > +        * It's guaranteed that pc->mem_cgroup never changes while
> > +        * lock is held
>
> Please continue comment describing what provides this guarantee. I
> assume it is because rcu_read_lock() is held by
> mem_cgroup_begin_update_page_stat(). Maybe it's best to to just make
> small reference to the locking protocol description in
> mem_cgroup_start_move().
>
Ok, I will update this.


> > +        */
> > +       move_unlock_mem_cgroup(pc->mem_cgroup, flags);
> > +}
> > +
> > +
>
> I think it would be useful to add a small comment here declaring that
> all callers of this routine must be in a
> mem_cgroup_begin_update_page_stat(), mem_cgroup_end_update_page_stat()
> critical section to keep pc->mem_cgroup stable.
>

Sure, will do.

Thank you for review.
-Kame


> >  void mem_cgroup_update_page_stat(struct page *page,
> >                                 enum mem_cgroup_page_stat_item idx, int val)
> >  {
>

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2012-02-14 09:47    [W:0.062 / U:0.020 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site