Current memcg_caches arrays relocation path looks rather awkward: on
setting kmem limit we call memcg_update_all_caches located on slab's
side, which walks over all root caches and calls back to memcontrol.c
through memcg_update_cache_size and memcg_update_array_size, which in
turn reallocate the memcg_caches array for a particular cache and update
the arrays' size respectively.
The first call from memcontrol.c to slab_common.c, which is
memcg_update_all_caches, is justified, because to iterate over root
caches we have to take the slab_mutex. However, I don't see any reason
why we can't reallocate a memcg_caches array on the slab's size or why
we should call memcg_update_cache_size under the slab_mutex. So let's
clean up this mess.
Signed-off-by: Vladimir Davydov <vdavydov@xxxxxxxxxxxxx>
---
include/linux/memcontrol.h | 3 --
include/linux/slab.h | 3 +-
mm/memcontrol.c | 125 +++++++++++++-------------------------------
mm/slab_common.c | 36 ++++++++-----
4 files changed, 61 insertions(+), 106 deletions(-)
diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index dfc2929a3877..6e59393e03f9 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -496,9 +496,6 @@ int memcg_alloc_cache_params(struct mem_cgroup *memcg, struct kmem_cache *s,
struct kmem_cache *root_cache);
void memcg_free_cache_params(struct kmem_cache *s);
-int memcg_update_cache_size(struct kmem_cache *s, int num_groups);
-void memcg_update_array_size(int num_groups);
-
struct kmem_cache *
__memcg_kmem_get_cache(struct kmem_cache *cachep, gfp_t gfp);
diff --git a/include/linux/slab.h b/include/linux/slab.h
index 095ce8e47a36..c437be67917b 100644
--- a/include/linux/slab.h
+++ b/include/linux/slab.h
@@ -120,6 +120,7 @@ struct kmem_cache *kmem_cache_create_memcg(struct mem_cgroup *,
struct kmem_cache *,
const char *);
int kmem_cache_init_memcg_array(struct kmem_cache *, int);
+int kmem_cache_memcg_arrays_grow(int);
#endif
void kmem_cache_destroy(struct kmem_cache *);
int kmem_cache_shrink(struct kmem_cache *);
@@ -545,8 +546,6 @@ struct memcg_cache_params {
};
};
-int memcg_update_all_caches(int num_memcgs);
-
struct seq_file;
int cache_show(struct kmem_cache *s, struct seq_file *m);
void print_slabinfo_header(struct seq_file *m);
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 08937040ed75..714d7bd7f140 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -3027,84 +3027,52 @@ int memcg_cache_id(struct mem_cgroup *memcg)
return memcg ? memcg->kmemcg_id : -1;
}
-static size_t memcg_caches_array_size(int num_groups)
+static int memcg_init_cache_id(struct mem_cgroup *memcg)
{
- ssize_t size;
- if (num_groups <= 0)
- return 0;
+ int err = 0;
+ int id, size;
+
+ lockdep_assert_held(&activate_kmem_mutex);
+
+ id = ida_simple_get(&kmem_limited_groups,
+ 0, MEMCG_CACHES_MAX_SIZE, GFP_KERNEL);
+ if (id < 0)
+ return id;
- size = 2 * num_groups;
+ if (id < memcg_limited_groups_array_size)
+ goto out_setid;
+
+ /*
+ * We don't have enough space for the new id in the arrays that store
+ * per memcg data. Let's try to grow them then.
+ */
+ size = id * 2;
if (size < MEMCG_CACHES_MIN_SIZE)
size = MEMCG_CACHES_MIN_SIZE;
else if (size > MEMCG_CACHES_MAX_SIZE)
size = MEMCG_CACHES_MAX_SIZE;
- return size;
-}
-
-/*
- * We should update the current array size iff all caches updates succeed. This
- * can only be done from the slab side. The slab mutex needs to be held when
- * calling this.
- */
-void memcg_update_array_size(int num)
-{
- if (num > memcg_limited_groups_array_size)
- memcg_limited_groups_array_size = memcg_caches_array_size(num);
-}
-
-int memcg_update_cache_size(struct kmem_cache *s, int num_groups)
-{
- struct memcg_cache_params *cur_params = s->memcg_params;
-
- VM_BUG_ON(!is_root_cache(s));
-
- if (!cur_params)
- return 0;
-
- if (num_groups > memcg_limited_groups_array_size) {
- int i;
- struct memcg_cache_params *new_params;
- ssize_t size = memcg_caches_array_size(num_groups);
-
- size *= sizeof(void *);
- size += offsetof(struct memcg_cache_params, memcg_caches);
-
- new_params = kzalloc(size, GFP_KERNEL);
- if (!new_params)
- return -ENOMEM;
+ mutex_lock(&memcg_slab_mutex);
+ err = kmem_cache_memcg_arrays_grow(size);
+ mutex_unlock(&memcg_slab_mutex);
- new_params->is_root_cache = true;
+ if (err)
+ goto out_rmid;
- /*
- * There is the chance it will be bigger than
- * memcg_limited_groups_array_size, if we failed an allocation
- * in a cache, in which case all caches updated before it, will
- * have a bigger array.
- *
- * But if that is the case, the data after
- * memcg_limited_groups_array_size is certainly unused
- */
- for (i = 0; i < memcg_limited_groups_array_size; i++) {
- if (!cur_params->memcg_caches[i])
- continue;
- new_params->memcg_caches[i] =
- cur_params->memcg_caches[i];
- }
+ /*
+ * Update the arrays' size only after we grew them so that readers
+ * walking over such an array won't get an index out of range provided
+ * they use an appropriate mutex to protect the array's elements.
+ */
+ memcg_limited_groups_array_size = size;
- /*
- * Ideally, we would wait until all caches succeed, and only
- * then free the old one. But this is not worth the extra
- * pointer per-cache we'd have to have for this.
- *
- * It is not a big deal if some caches are left with a size
- * bigger than the others. And all updates will reset this
- * anyway.
- */
- rcu_assign_pointer(s->memcg_params, new_params);
- kfree_rcu(cur_params, rcu_head);
- }
+out_setid:
+ memcg->kmemcg_id = id;
return 0;
+
+out_rmid:
+ ida_simple_remove(&kmem_limited_groups, id);
+ return err;
}
int memcg_alloc_cache_params(struct mem_cgroup *memcg, struct kmem_cache *s,
@@ -4950,7 +4918,6 @@ static int __memcg_activate_kmem(struct mem_cgroup *memcg,
unsigned long long limit)
{
int err = 0;
- int memcg_id;
if (memcg_kmem_is_active(memcg))
return 0;
@@ -4980,24 +4947,10 @@ static int __memcg_activate_kmem(struct mem_cgroup *memcg,
if (err)
goto out;
- memcg_id = ida_simple_get(&kmem_limited_groups,
- 0, MEMCG_CACHES_MAX_SIZE, GFP_KERNEL);
- if (memcg_id < 0) {
- err = memcg_id;
- goto out;
- }
-
- /*
- * Make sure we have enough space for this cgroup in each root cache's
- * memcg_params.
- */
- mutex_lock(&memcg_slab_mutex);
- err = memcg_update_all_caches(memcg_id + 1);
- mutex_unlock(&memcg_slab_mutex);
+ err = memcg_init_cache_id(memcg);
if (err)
- goto out_rmid;
+ goto out;
- memcg->kmemcg_id = memcg_id;
INIT_LIST_HEAD(&memcg->memcg_slab_caches);
/*
@@ -5017,10 +4970,6 @@ static int __memcg_activate_kmem(struct mem_cgroup *memcg,
out:
memcg_resume_kmem_account();
return err;
-
-out_rmid:
- ida_simple_remove(&kmem_limited_groups, memcg_id);
- goto out;
}
static int memcg_activate_kmem(struct mem_cgroup *memcg,
diff --git a/mm/slab_common.c b/mm/slab_common.c
index ce5e96aff4e6..801999247619 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -77,10 +77,17 @@ static inline int kmem_cache_sanity_check(const char *name, size_t size)
#endif
#ifdef CONFIG_MEMCG_KMEM
+/*
+ * (Re)allocates the memcg_caches array of the given kmem cache so that it
+ * can hold up to nr_entries. The caller must hold slab_mutex.
+ */
static int __kmem_cache_init_memcg_array(struct kmem_cache *s, int nr_entries)
{
+ int i;
size_t size;
- struct memcg_cache_params *new;
+ struct memcg_cache_params *new, *old;
+
+ old = s->memcg_params;
size = offsetof(struct memcg_cache_params, memcg_caches);
size += nr_entries * sizeof(void *);
@@ -90,10 +97,15 @@ static int __kmem_cache_init_memcg_array(struct kmem_cache *s, int nr_entries)
return -ENOMEM;
new->is_root_cache = true;
+ if (old) {
+ for_each_memcg_cache_index(i)
+ new->memcg_caches[i] = old->memcg_caches[i];
+ }
/* matching rcu_dereference is in cache_from_memcg_idx */
rcu_assign_pointer(s->memcg_params, new);
-
+ if (old)
+ kfree_rcu(old, rcu_head);
return 0;
}
@@ -110,32 +122,30 @@ int kmem_cache_init_memcg_array(struct kmem_cache *s, int nr_entries)
return ret;
}
-int memcg_update_all_caches(int num_memcgs)
+int kmem_cache_memcg_arrays_grow(int nr_entries)
{
struct kmem_cache *s;
int ret = 0;
- mutex_lock(&slab_mutex);
+ mutex_lock(&slab_mutex);
list_for_each_entry(s, &slab_caches, list) {
if (!is_root_cache(s))
continue;
- ret = memcg_update_cache_size(s, num_memcgs);
+ ret = __kmem_cache_init_memcg_array(s, nr_entries);
/*
- * See comment in memcontrol.c, memcg_update_cache_size:
- * Instead of freeing the memory, we'll just leave the caches
- * up to this point in an updated state.
+ * We won't shrink the arrays back to the initial size on
+ * failure, because it isn't a big deal if some caches are left
+ * with a size greater than others. Further updates will reset
+ * this anyway.
*/
if (ret)
- goto out;
+ break;
}
-
- memcg_update_array_size(num_memcgs);
-out:
mutex_unlock(&slab_mutex);
return ret;
}
-#endif
+#endif /* CONFIG_MEMCG_KMEM */
/*
* Figure out what the alignment of the objects will be given a set of
--
1.7.10.4
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>