The patch titled
Subject: slab: remove synchronous rcu_barrier() call in memcg cache release path
has been removed from the -mm tree. Its filename was
slab-remove-synchronous-rcu_barrier-call-in-memcg-cache-release-path.patch
This patch was dropped because it was merged into mainline or a subsystem tree
------------------------------------------------------
From: Tejun Heo <tj@xxxxxxxxxx>
Subject: slab: remove synchronous rcu_barrier() call in memcg cache release path
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.
SLAB_DESTORY_BY_RCU caches need to flush all RCU operations before
destruction because slab pages are freed through RCU and they need to be
able to dereference the associated kmem_cache. Currently, it's done
synchronously with rcu_barrier(). As rcu_barrier() is expensive
time-wise, slab implements a batching mechanism so that rcu_barrier() can
be done for multiple caches at the same time.
Unfortunately, the rcu_barrier() is in synchronous path which is called
while holding cgroup_mutex and the batching is too limited to be actually
helpful.
This patch updates the cache release path so that the batching is
asynchronous and global. All SLAB_DESTORY_BY_RCU caches are queued
globally and a work item consumes the list. The work item calls
rcu_barrier() only once for all caches that are currently queued.
* release_caches() is removed and shutdown_cache() now either directly
release the cache or schedules a RCU callback to do that. This
makes the cache inaccessible once shutdown_cache() is called and
makes it impossible for shutdown_memcg_caches() to do memcg-specific
cleanups afterwards. Move memcg-specific part into a helper,
unlink_memcg_cache(), and make shutdown_cache() call it directly.
Link: http://lkml.kernel.org/r/20170117235411.9408-4-tj@xxxxxxxxxx
Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
Reported-by: Jay Vana <jsvana@xxxxxx>
Acked-by: Vladimir Davydov <vdavydov@xxxxxxxxxxxxx>
Cc: Christoph Lameter <cl@xxxxxxxxx>
Cc: Pekka Enberg <penberg@xxxxxxxxxx>
Cc: David Rientjes <rientjes@xxxxxxxxxx>
Cc: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
mm/slab_common.c | 102 ++++++++++++++++++++++++++-------------------
1 file changed, 60 insertions(+), 42 deletions(-)
diff -puN mm/slab_common.c~slab-remove-synchronous-rcu_barrier-call-in-memcg-cache-release-path mm/slab_common.c
--- a/mm/slab_common.c~slab-remove-synchronous-rcu_barrier-call-in-memcg-cache-release-path
+++ a/mm/slab_common.c
@@ -30,6 +30,11 @@ LIST_HEAD(slab_caches);
DEFINE_MUTEX(slab_mutex);
struct kmem_cache *kmem_cache;
+static LIST_HEAD(slab_caches_to_rcu_destroy);
+static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work);
+static DECLARE_WORK(slab_caches_to_rcu_destroy_work,
+ slab_caches_to_rcu_destroy_workfn);
+
/*
* Set of flags that will prevent slab merging
*/
@@ -215,6 +220,11 @@ int memcg_update_all_caches(int num_memc
mutex_unlock(&slab_mutex);
return ret;
}
+
+static void unlink_memcg_cache(struct kmem_cache *s)
+{
+ list_del(&s->memcg_params.list);
+}
#else
static inline int init_memcg_params(struct kmem_cache *s,
struct mem_cgroup *memcg, struct kmem_cache *root_cache)
@@ -225,6 +235,10 @@ static inline int init_memcg_params(stru
static inline void destroy_memcg_params(struct kmem_cache *s)
{
}
+
+static inline void unlink_memcg_cache(struct kmem_cache *s)
+{
+}
#endif /* CONFIG_MEMCG && !CONFIG_SLOB */
/*
@@ -461,33 +475,59 @@ out_unlock:
}
EXPORT_SYMBOL(kmem_cache_create);
-static int shutdown_cache(struct kmem_cache *s,
- struct list_head *release, bool *need_rcu_barrier)
+static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
{
- if (__kmem_cache_shutdown(s) != 0)
- return -EBUSY;
+ LIST_HEAD(to_destroy);
+ struct kmem_cache *s, *s2;
- if (s->flags & SLAB_DESTROY_BY_RCU)
- *need_rcu_barrier = true;
+ /*
+ * On destruction, SLAB_DESTROY_BY_RCU kmem_caches are put on the
+ * @slab_caches_to_rcu_destroy list. The slab pages are freed
+ * through RCU and and the associated kmem_cache are dereferenced
+ * while freeing the pages, so the kmem_caches should be freed only
+ * after the pending RCU operations are finished. As rcu_barrier()
+ * is a pretty slow operation, we batch all pending destructions
+ * asynchronously.
+ */
+ mutex_lock(&slab_mutex);
+ list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
+ mutex_unlock(&slab_mutex);
- list_move(&s->list, release);
- return 0;
+ if (list_empty(&to_destroy))
+ return;
+
+ rcu_barrier();
+
+ list_for_each_entry_safe(s, s2, &to_destroy, list) {
+#ifdef SLAB_SUPPORTS_SYSFS
+ sysfs_slab_release(s);
+#else
+ slab_kmem_cache_release(s);
+#endif
+ }
}
-static void release_caches(struct list_head *release, bool need_rcu_barrier)
+static int shutdown_cache(struct kmem_cache *s)
{
- struct kmem_cache *s, *s2;
+ if (__kmem_cache_shutdown(s) != 0)
+ return -EBUSY;
- if (need_rcu_barrier)
- rcu_barrier();
+ list_del(&s->list);
+ if (!is_root_cache(s))
+ unlink_memcg_cache(s);
- list_for_each_entry_safe(s, s2, release, list) {
+ if (s->flags & SLAB_DESTROY_BY_RCU) {
+ list_add_tail(&s->list, &slab_caches_to_rcu_destroy);
+ schedule_work(&slab_caches_to_rcu_destroy_work);
+ } else {
#ifdef SLAB_SUPPORTS_SYSFS
sysfs_slab_release(s);
#else
slab_kmem_cache_release(s);
#endif
}
+
+ return 0;
}
#if defined(CONFIG_MEMCG) && !defined(CONFIG_SLOB)
@@ -602,22 +642,8 @@ void memcg_deactivate_kmem_caches(struct
put_online_cpus();
}
-static int __shutdown_memcg_cache(struct kmem_cache *s,
- struct list_head *release, bool *need_rcu_barrier)
-{
- BUG_ON(is_root_cache(s));
-
- if (shutdown_cache(s, release, need_rcu_barrier))
- return -EBUSY;
-
- list_del(&s->memcg_params.list);
- return 0;
-}
-
void memcg_destroy_kmem_caches(struct mem_cgroup *memcg)
{
- LIST_HEAD(release);
- bool need_rcu_barrier = false;
struct kmem_cache *s, *s2;
get_online_cpus();
@@ -631,18 +657,15 @@ void memcg_destroy_kmem_caches(struct me
* The cgroup is about to be freed and therefore has no charges
* left. Hence, all its caches must be empty by now.
*/
- BUG_ON(__shutdown_memcg_cache(s, &release, &need_rcu_barrier));
+ BUG_ON(shutdown_cache(s));
}
mutex_unlock(&slab_mutex);
put_online_mems();
put_online_cpus();
-
- release_caches(&release, need_rcu_barrier);
}
-static int shutdown_memcg_caches(struct kmem_cache *s,
- struct list_head *release, bool *need_rcu_barrier)
+static int shutdown_memcg_caches(struct kmem_cache *s)
{
struct memcg_cache_array *arr;
struct kmem_cache *c, *c2;
@@ -661,7 +684,7 @@ static int shutdown_memcg_caches(struct
c = arr->entries[i];
if (!c)
continue;
- if (__shutdown_memcg_cache(c, release, need_rcu_barrier))
+ if (shutdown_cache(c))
/*
* The cache still has objects. Move it to a temporary
* list so as not to try to destroy it for a second
@@ -684,7 +707,7 @@ static int shutdown_memcg_caches(struct
*/
list_for_each_entry_safe(c, c2, &s->memcg_params.list,
memcg_params.list)
- __shutdown_memcg_cache(c, release, need_rcu_barrier);
+ shutdown_cache(c);
list_splice(&busy, &s->memcg_params.list);
@@ -697,8 +720,7 @@ static int shutdown_memcg_caches(struct
return 0;
}
#else
-static inline int shutdown_memcg_caches(struct kmem_cache *s,
- struct list_head *release, bool *need_rcu_barrier)
+static inline int shutdown_memcg_caches(struct kmem_cache *s)
{
return 0;
}
@@ -714,8 +736,6 @@ void slab_kmem_cache_release(struct kmem
void kmem_cache_destroy(struct kmem_cache *s)
{
- LIST_HEAD(release);
- bool need_rcu_barrier = false;
int err;
if (unlikely(!s))
@@ -731,9 +751,9 @@ void kmem_cache_destroy(struct kmem_cach
if (s->refcount)
goto out_unlock;
- err = shutdown_memcg_caches(s, &release, &need_rcu_barrier);
+ err = shutdown_memcg_caches(s);
if (!err)
- err = shutdown_cache(s, &release, &need_rcu_barrier);
+ err = shutdown_cache(s);
if (err) {
pr_err("kmem_cache_destroy %s: Slab cache still has objects\n",
@@ -745,8 +765,6 @@ out_unlock:
put_online_mems();
put_online_cpus();
-
- release_caches(&release, need_rcu_barrier);
}
EXPORT_SYMBOL(kmem_cache_destroy);
_
Patches currently in -mm which might be from tj@xxxxxxxxxx are
--
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
