答复: 答复: 答复: 答复: [PATCH] mm/memcontrol.c: speed up to force empty a memory cgroup

["Li,Rongqing" <lirongqing@baidu.com>]

> -----邮件原件-----
> 发件人: linux-kernel-owner@vger.kernel.org
> [mailto:linux-kernel-owner@vger.kernel.org] 代表 Michal Hocko
> 发送时间: 2018年3月23日 20:29
> 收件人: Li,Rongqing <lirongqing@baidu.com>
> 抄送: linux-kernel@vger.kernel.org; linux-mm@kvack.org;
> cgroups@vger.kernel.org; hannes@cmpxchg.org; Andrey Ryabinin
> <aryabinin@virtuozzo.com>
> 主题: Re: 答复: 答复: 答复: [PATCH] mm/memcontrol.c: speed up to force
> empty a memory cgroup
> 
> On Fri 23-03-18 12:04:16, Li,Rongqing wrote:
> [...]
> > shrink_slab does not reclaim any memory, but take lots of time to
> > count lru
> >
> > maybe we can use the returning of shrink_slab to control if next
> > shrink_slab should be called?
> 
> How? Different memcgs might have different amount of shrinkable memory.
> 

maybe there is not a easy way to implement it

in this case, shrink_slab call many times list_lru_count_one, which is calling list_lru_count_one, which use spinlock, maybe replace spinlock with RCU, to optimize


-RongQing


> > Or define a slight list_lru_empty to check if sb->s_dentry_lru is
> > empty before calling list_lru_shrink_count, like below
> 
> Does it really help to improve numbers?
> 
> > diff --git a/fs/super.c b/fs/super.c
> > index 672538ca9831..954c22338833 100644
> > --- a/fs/super.c
> > +++ b/fs/super.c
> > @@ -130,8 +130,10 @@ static unsigned long super_cache_count(struct
> shrinker *shrink,
> >         if (sb->s_op && sb->s_op->nr_cached_objects)
> >                 total_objects = sb->s_op->nr_cached_objects(sb, sc);
> >
> > -       total_objects += list_lru_shrink_count(&sb->s_dentry_lru, sc);
> > -       total_objects += list_lru_shrink_count(&sb->s_inode_lru, sc);
> > +       if (!list_lru_empty(sb->s_dentry_lru))
> > +               total_objects +=
> list_lru_shrink_count(&sb->s_dentry_lru, sc);
> > +       if (!list_lru_empty(sb->s_inode_lru))
> > +               total_objects +=
> > + list_lru_shrink_count(&sb->s_inode_lru, sc);
> >
> >         total_objects = vfs_pressure_ratio(total_objects);
> >         return total_objects;
> 
> --
> Michal Hocko
> SUSE Labs