All of lore.kernel.org
 help / color / mirror / Atom feed
From: Vladimir Davydov <vdavydov@tarantool.org>
To: Tejun Heo <tj@kernel.org>
Cc: cl@linux.com, penberg@kernel.org, rientjes@google.com,
	iamjoonsoo.kim@lge.com, akpm@linux-foundation.org, jsvana@fb.com,
	hannes@cmpxchg.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org, cgroups@vger.kernel.org, kernel-team@fb.com
Subject: Re: [PATCH 8/9] slab: remove synchronous synchronize_sched() from memcg cache deactivation path
Date: Sat, 14 Jan 2017 16:57:27 +0300	[thread overview]
Message-ID: <20170114135727.GG2668@esperanza> (raw)
In-Reply-To: <20170114055449.11044-9-tj@kernel.org>

On Sat, Jan 14, 2017 at 12:54:48AM -0500, Tejun Heo wrote:
> With kmem cgroup support enabled, kmem_caches can be created and
> destroyed frequently and a great number of near empty kmem_caches can
> accumulate if there are a lot of transient cgroups and the system is
> not under memory pressure.  When memory reclaim starts under such
> conditions, it can lead to consecutive deactivation and destruction of
> many kmem_caches, easily hundreds of thousands on moderately large
> systems, exposing scalability issues in the current slab management
> code.  This is one of the patches to address the issue.
> 
> slub uses synchronize_sched() to deactivate a memcg cache.
> synchronize_sched() is an expensive and slow operation and doesn't
> scale when a huge number of caches are destroyed back-to-back.  While
> there used to be a simple batching mechanism, the batching was too
> restricted to be helpful.
> 
> This patch implements slab_deactivate_memcg_cache_rcu_sched() which
> slub can use to schedule sched RCU callback instead of performing
> synchronize_sched() synchronously while holding cgroup_mutex.  While
> this adds online cpus, mems and slab_mutex operations, operating on
> these locks back-to-back from the same kworker, which is what's gonna
> happen when there are many to deactivate, isn't expensive at all and
> this gets rid of the scalability problem completely.
> 
> Signed-off-by: Tejun Heo <tj@kernel.org>
> Reported-by: Jay Vana <jsvana@fb.com>
> Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
> Cc: Christoph Lameter <cl@linux.com>
> Cc: Pekka Enberg <penberg@kernel.org>
> Cc: David Rientjes <rientjes@google.com>
> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>

I don't think there's much point in having the infrastructure for this
in slab_common.c, as only SLUB needs it, but it isn't a show stopper.

Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>

WARNING: multiple messages have this Message-ID (diff)
From: Vladimir Davydov <vdavydov@tarantool.org>
To: Tejun Heo <tj@kernel.org>
Cc: cl@linux.com, penberg@kernel.org, rientjes@google.com,
	iamjoonsoo.kim@lge.com, akpm@linux-foundation.org, jsvana@fb.com,
	hannes@cmpxchg.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org, cgroups@vger.kernel.org, kernel-team@fb.com
Subject: Re: [PATCH 8/9] slab: remove synchronous synchronize_sched() from memcg cache deactivation path
Date: Sat, 14 Jan 2017 16:57:27 +0300	[thread overview]
Message-ID: <20170114135727.GG2668@esperanza> (raw)
In-Reply-To: <20170114055449.11044-9-tj@kernel.org>

On Sat, Jan 14, 2017 at 12:54:48AM -0500, Tejun Heo wrote:
> With kmem cgroup support enabled, kmem_caches can be created and
> destroyed frequently and a great number of near empty kmem_caches can
> accumulate if there are a lot of transient cgroups and the system is
> not under memory pressure.  When memory reclaim starts under such
> conditions, it can lead to consecutive deactivation and destruction of
> many kmem_caches, easily hundreds of thousands on moderately large
> systems, exposing scalability issues in the current slab management
> code.  This is one of the patches to address the issue.
> 
> slub uses synchronize_sched() to deactivate a memcg cache.
> synchronize_sched() is an expensive and slow operation and doesn't
> scale when a huge number of caches are destroyed back-to-back.  While
> there used to be a simple batching mechanism, the batching was too
> restricted to be helpful.
> 
> This patch implements slab_deactivate_memcg_cache_rcu_sched() which
> slub can use to schedule sched RCU callback instead of performing
> synchronize_sched() synchronously while holding cgroup_mutex.  While
> this adds online cpus, mems and slab_mutex operations, operating on
> these locks back-to-back from the same kworker, which is what's gonna
> happen when there are many to deactivate, isn't expensive at all and
> this gets rid of the scalability problem completely.
> 
> Signed-off-by: Tejun Heo <tj@kernel.org>
> Reported-by: Jay Vana <jsvana@fb.com>
> Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
> Cc: Christoph Lameter <cl@linux.com>
> Cc: Pekka Enberg <penberg@kernel.org>
> Cc: David Rientjes <rientjes@google.com>
> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>

I don't think there's much point in having the infrastructure for this
in slab_common.c, as only SLUB needs it, but it isn't a show stopper.

Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

WARNING: multiple messages have this Message-ID (diff)
From: Vladimir Davydov <vdavydov-mJ5StSI9eqKXDw4h08c5KA@public.gmane.org>
To: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
Cc: cl-vYTEC60ixJUAvxtiuMwx3w@public.gmane.org,
	penberg-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org,
	rientjes-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org,
	iamjoonsoo.kim-Hm3cg6mZ9cc@public.gmane.org,
	akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org,
	jsvana-b10kYP2dOMg@public.gmane.org,
	hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org,
	linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	linux-mm-Bw31MaZKKs3YtjvyW6yDsg@public.gmane.org,
	cgroups-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	kernel-team-b10kYP2dOMg@public.gmane.org
Subject: Re: [PATCH 8/9] slab: remove synchronous synchronize_sched() from memcg cache deactivation path
Date: Sat, 14 Jan 2017 16:57:27 +0300	[thread overview]
Message-ID: <20170114135727.GG2668@esperanza> (raw)
In-Reply-To: <20170114055449.11044-9-tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>

On Sat, Jan 14, 2017 at 12:54:48AM -0500, Tejun Heo wrote:
> With kmem cgroup support enabled, kmem_caches can be created and
> destroyed frequently and a great number of near empty kmem_caches can
> accumulate if there are a lot of transient cgroups and the system is
> not under memory pressure.  When memory reclaim starts under such
> conditions, it can lead to consecutive deactivation and destruction of
> many kmem_caches, easily hundreds of thousands on moderately large
> systems, exposing scalability issues in the current slab management
> code.  This is one of the patches to address the issue.
> 
> slub uses synchronize_sched() to deactivate a memcg cache.
> synchronize_sched() is an expensive and slow operation and doesn't
> scale when a huge number of caches are destroyed back-to-back.  While
> there used to be a simple batching mechanism, the batching was too
> restricted to be helpful.
> 
> This patch implements slab_deactivate_memcg_cache_rcu_sched() which
> slub can use to schedule sched RCU callback instead of performing
> synchronize_sched() synchronously while holding cgroup_mutex.  While
> this adds online cpus, mems and slab_mutex operations, operating on
> these locks back-to-back from the same kworker, which is what's gonna
> happen when there are many to deactivate, isn't expensive at all and
> this gets rid of the scalability problem completely.
> 
> Signed-off-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
> Reported-by: Jay Vana <jsvana-b10kYP2dOMg@public.gmane.org>
> Cc: Vladimir Davydov <vdavydov.dev-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Cc: Christoph Lameter <cl-vYTEC60ixJUAvxtiuMwx3w@public.gmane.org>
> Cc: Pekka Enberg <penberg-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
> Cc: David Rientjes <rientjes-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
> Cc: Joonsoo Kim <iamjoonsoo.kim-Hm3cg6mZ9cc@public.gmane.org>
> Cc: Andrew Morton <akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org>

I don't think there's much point in having the infrastructure for this
in slab_common.c, as only SLUB needs it, but it isn't a show stopper.

Acked-by: Vladimir Davydov <vdavydov.dev-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>

  reply	other threads:[~2017-01-14 13:57 UTC|newest]

Thread overview: 56+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-01-14  5:54 [PATCHSET] slab: make memcg slab destruction scalable Tejun Heo
2017-01-14  5:54 ` Tejun Heo
2017-01-14  5:54 ` Tejun Heo
2017-01-14  5:54 ` [PATCH 1/9] Revert "slub: move synchronize_sched out of slab_mutex on shrink" Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14  5:54 ` [PATCH 2/9] slab: remove synchronous rcu_barrier() call in memcg cache release path Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14 13:19   ` Vladimir Davydov
2017-01-14 13:19     ` Vladimir Davydov
2017-01-14 15:19     ` Tejun Heo
2017-01-14 15:19       ` Tejun Heo
2017-01-17  0:07       ` Joonsoo Kim
2017-01-17  0:07         ` Joonsoo Kim
2017-01-17  0:07         ` Joonsoo Kim
2017-01-17 16:37         ` Tejun Heo
2017-01-17 16:37           ` Tejun Heo
2017-01-17 17:02           ` Tejun Heo
2017-01-17 17:02             ` Tejun Heo
2017-01-14  5:54 ` [PATCH 3/9] slab: simplify shutdown_memcg_caches() Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14 13:27   ` Vladimir Davydov
2017-01-14 13:27     ` Vladimir Davydov
2017-01-14 15:38     ` Tejun Heo
2017-01-14 15:38       ` Tejun Heo
2017-01-14 15:53       ` Tejun Heo
2017-01-14 15:53         ` Tejun Heo
2017-01-14  5:54 ` [PATCH 4/9] slab: reorganize memcg_cache_params Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14 13:30   ` Vladimir Davydov
2017-01-14 13:30     ` Vladimir Davydov
2017-01-14  5:54 ` [PATCH 5/9] slab: link memcg kmem_caches on their associated memory cgroup Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14 13:33   ` Vladimir Davydov
2017-01-14 13:33     ` Vladimir Davydov
2017-01-14  5:54 ` [PATCH 6/9] slab: don't put memcg caches on slab_caches list Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14 13:39   ` Vladimir Davydov
2017-01-14 13:39     ` Vladimir Davydov
2017-01-14 15:39     ` Tejun Heo
2017-01-14 15:39       ` Tejun Heo
2017-01-14 15:39       ` Tejun Heo
2017-01-14  5:54 ` [PATCH 7/9] slab: introduce __kmemcg_cache_deactivate() Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14 13:42   ` Vladimir Davydov
2017-01-14 13:42     ` Vladimir Davydov
2017-01-14 15:39     ` Tejun Heo
2017-01-14 15:39       ` Tejun Heo
2017-01-14  5:54 ` [PATCH 8/9] slab: remove synchronous synchronize_sched() from memcg cache deactivation path Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14 13:57   ` Vladimir Davydov [this message]
2017-01-14 13:57     ` Vladimir Davydov
2017-01-14 13:57     ` Vladimir Davydov
2017-01-14  5:54 ` [PATCH 9/9] slab: remove slub sysfs interface files early for empty memcg caches Tejun Heo
2017-01-14  5:54   ` Tejun Heo
2017-01-14 14:00   ` Vladimir Davydov
2017-01-14 14:00     ` Vladimir Davydov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20170114135727.GG2668@esperanza \
    --to=vdavydov@tarantool.org \
    --cc=akpm@linux-foundation.org \
    --cc=cgroups@vger.kernel.org \
    --cc=cl@linux.com \
    --cc=hannes@cmpxchg.org \
    --cc=iamjoonsoo.kim@lge.com \
    --cc=jsvana@fb.com \
    --cc=kernel-team@fb.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=penberg@kernel.org \
    --cc=rientjes@google.com \
    --cc=tj@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.