Message-ID: <CABCjUKBk=RTCoH34XOQHxRsSp0G5iwtgeBKtdNEUeTE5kx07Vg@mail.gmail.com>
Date:	Mon, 30 Apr 2012 13:56:46 -0700
From:	Suleiman Souhlal <suleiman@...gle.com>
To:	Glauber Costa <glommer@...allels.com>
Cc:	cgroups@...r.kernel.org, linux-mm@...ck.org,
	linux-kernel@...r.kernel.org, devel@...nvz.org,
	kamezawa.hiroyu@...fujitsu.com, Michal Hocko <mhocko@...e.cz>,
	Johannes Weiner <hannes@...xchg.org>, fweisbec@...il.com,
	Greg Thelen <gthelen@...gle.com>,
	Christoph Lameter <cl@...ux.com>,
	Pekka Enberg <penberg@...helsinki.fi>
Subject: Re: [PATCH 17/23] kmem controller charge/uncharge infrastructure

On Sun, Apr 22, 2012 at 4:53 PM, Glauber Costa <glommer@...allels.com> wrote:
> With all the dependencies already in place, this patch introduces
> the charge/uncharge functions for the slab cache accounting in memcg.
>
> Before we can charge a cache, we need to select the right cache.
> This is done by using the function __mem_cgroup_get_kmem_cache().
>
> If we should use the root kmem cache, this function tries to detect
> that and return as early as possible.
>
> The charge and uncharge functions comes in two flavours:
>  * __mem_cgroup_(un)charge_slab(), that assumes the allocation is
>   a slab page, and
>  * __mem_cgroup_(un)charge_kmem(), that does not. This later exists
>   because the slub allocator draws the larger kmalloc allocations
>   from the page allocator.
>
> In memcontrol.h those functions are wrapped in inline acessors.
> The idea is to later on, patch those with jump labels, so we don't
> incur any overhead when no mem cgroups are being used.
>
> Because the slub allocator tends to inline the allocations whenever
> it can, those functions need to be exported so modules can make use
> of it properly.
>
> I apologize in advance to the reviewers. This patch is quite big, but
> I was not able to split it any further due to all the dependencies
> between the code.
>
> This code is inspired by the code written by Suleiman Souhlal,
> but heavily changed.
>
> Signed-off-by: Glauber Costa <glommer@...allels.com>
> CC: Christoph Lameter <cl@...ux.com>
> CC: Pekka Enberg <penberg@...helsinki.fi>
> CC: Michal Hocko <mhocko@...e.cz>
> CC: Kamezawa Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
> CC: Johannes Weiner <hannes@...xchg.org>
> CC: Suleiman Souhlal <suleiman@...gle.com>
> ---
>  include/linux/memcontrol.h |   68 ++++++++
>  init/Kconfig               |    2 +-
>  mm/memcontrol.c            |  373 +++++++++++++++++++++++++++++++++++++++++++-
>  3 files changed, 441 insertions(+), 2 deletions(-)
>

> +
> +static struct kmem_cache *memcg_create_kmem_cache(struct mem_cgroup *memcg,
> +                                                 struct kmem_cache *cachep)
> +{
> +       struct kmem_cache *new_cachep;
> +       int idx;
> +
> +       BUG_ON(!mem_cgroup_kmem_enabled(memcg));
> +
> +       idx = cachep->memcg_params.id;
> +
> +       mutex_lock(&memcg_cache_mutex);
> +       new_cachep = memcg->slabs[idx];
> +       if (new_cachep)
> +               goto out;
> +
> +       new_cachep = kmem_cache_dup(memcg, cachep);
> +
> +       if (new_cachep == NULL) {
> +               new_cachep = cachep;
> +               goto out;
> +       }
> +
> +       mem_cgroup_get(memcg);
> +       memcg->slabs[idx] = new_cachep;
> +       new_cachep->memcg_params.memcg = memcg;
> +out:
> +       mutex_unlock(&memcg_cache_mutex);
> +       return new_cachep;
> +}
> +
> +struct create_work {
> +       struct mem_cgroup *memcg;
> +       struct kmem_cache *cachep;
> +       struct list_head list;
> +};
> +
> +/* Use a single spinlock for destruction and creation, not a frequent op */
> +static DEFINE_SPINLOCK(cache_queue_lock);
> +static LIST_HEAD(create_queue);
> +static LIST_HEAD(destroyed_caches);
> +
> +static void kmem_cache_destroy_work_func(struct work_struct *w)
> +{
> +       struct kmem_cache *cachep;
> +       char *name;
> +
> +       spin_lock_irq(&cache_queue_lock);
> +       while (!list_empty(&destroyed_caches)) {
> +               cachep = container_of(list_first_entry(&destroyed_caches,
> +                   struct mem_cgroup_cache_params, destroyed_list), struct
> +                   kmem_cache, memcg_params);
> +               name = (char *)cachep->name;
> +               list_del(&cachep->memcg_params.destroyed_list);
> +               spin_unlock_irq(&cache_queue_lock);
> +               synchronize_rcu();

Is this synchronize_rcu() still needed, now that we don't use RCU to
protect memcgs from disappearing during allocation anymore?

Also, should we drop the memcg reference we got in
memcg_create_kmem_cache() here?

> +               kmem_cache_destroy(cachep);
> +               kfree(name);
> +               spin_lock_irq(&cache_queue_lock);
> +       }
> +       spin_unlock_irq(&cache_queue_lock);
> +}
> +static DECLARE_WORK(kmem_cache_destroy_work, kmem_cache_destroy_work_func);
> +
> +void mem_cgroup_destroy_cache(struct kmem_cache *cachep)
> +{
> +       unsigned long flags;
> +
> +       BUG_ON(cachep->memcg_params.id != -1);
> +
> +       /*
> +        * We have to defer the actual destroying to a workqueue, because
> +        * we might currently be in a context that cannot sleep.
> +        */
> +       spin_lock_irqsave(&cache_queue_lock, flags);
> +       list_add(&cachep->memcg_params.destroyed_list, &destroyed_caches);
> +       spin_unlock_irqrestore(&cache_queue_lock, flags);
> +
> +       schedule_work(&kmem_cache_destroy_work);
> +}
> +
> +
> +/*
> + * Flush the queue of kmem_caches to create, because we're creating a cgroup.
> + *
> + * We might end up flushing other cgroups' creation requests as well, but
> + * they will just get queued again next time someone tries to make a slab
> + * allocation for them.
> + */
> +void mem_cgroup_flush_cache_create_queue(void)
> +{
> +       struct create_work *cw, *tmp;
> +       unsigned long flags;
> +
> +       spin_lock_irqsave(&cache_queue_lock, flags);
> +       list_for_each_entry_safe(cw, tmp, &create_queue, list) {
> +               list_del(&cw->list);
> +               kfree(cw);
> +       }
> +       spin_unlock_irqrestore(&cache_queue_lock, flags);
> +}
> +
> +static void memcg_create_cache_work_func(struct work_struct *w)
> +{
> +       struct kmem_cache *cachep;
> +       struct create_work *cw;
> +
> +       spin_lock_irq(&cache_queue_lock);
> +       while (!list_empty(&create_queue)) {
> +               cw = list_first_entry(&create_queue, struct create_work, list);
> +               list_del(&cw->list);
> +               spin_unlock_irq(&cache_queue_lock);
> +               cachep = memcg_create_kmem_cache(cw->memcg, cw->cachep);
> +               if (cachep == NULL)
> +                       printk(KERN_ALERT
> +                       "%s: Couldn't create memcg-cache for %s memcg %s\n",
> +                       __func__, cw->cachep->name,
> +                       cw->memcg->css.cgroup->dentry->d_name.name);

We might need rcu_dereference() here (and hold rcu_read_lock()).
Or we could just remove this message.

> +               /* Drop the reference gotten when we enqueued. */
> +               css_put(&cw->memcg->css);
> +               kfree(cw);
> +               spin_lock_irq(&cache_queue_lock);
> +       }
> +       spin_unlock_irq(&cache_queue_lock);
> +}
> +
> +static DECLARE_WORK(memcg_create_cache_work, memcg_create_cache_work_func);
> +
> +/*
> + * Enqueue the creation of a per-memcg kmem_cache.
> + * Called with rcu_read_lock.
> + */
> +static void memcg_create_cache_enqueue(struct mem_cgroup *memcg,
> +                                      struct kmem_cache *cachep)
> +{
> +       struct create_work *cw;
> +       unsigned long flags;
> +
> +       spin_lock_irqsave(&cache_queue_lock, flags);
> +       list_for_each_entry(cw, &create_queue, list) {
> +               if (cw->memcg == memcg && cw->cachep == cachep) {
> +                       spin_unlock_irqrestore(&cache_queue_lock, flags);
> +                       return;
> +               }
> +       }
> +       spin_unlock_irqrestore(&cache_queue_lock, flags);
> +
> +       /* The corresponding put will be done in the workqueue. */
> +       if (!css_tryget(&memcg->css))
> +               return;
> +
> +       cw = kmalloc_no_account(sizeof(struct create_work), GFP_NOWAIT);
> +       if (cw == NULL) {
> +               css_put(&memcg->css);
> +               return;
> +       }
> +
> +       cw->memcg = memcg;
> +       cw->cachep = cachep;
> +       spin_lock_irqsave(&cache_queue_lock, flags);
> +       list_add_tail(&cw->list, &create_queue);
> +       spin_unlock_irqrestore(&cache_queue_lock, flags);
> +
> +       schedule_work(&memcg_create_cache_work);
> +}
> +
> +/*
> + * Return the kmem_cache we're supposed to use for a slab allocation.
> + * If we are in interrupt context or otherwise have an allocation that
> + * can't fail, we return the original cache.
> + * Otherwise, we will try to use the current memcg's version of the cache.
> + *
> + * If the cache does not exist yet, if we are the first user of it,
> + * we either create it immediately, if possible, or create it asynchronously
> + * in a workqueue.
> + * In the latter case, we will let the current allocation go through with
> + * the original cache.
> + *
> + * This function returns with rcu_read_lock() held.
> + */
> +struct kmem_cache *__mem_cgroup_get_kmem_cache(struct kmem_cache *cachep,
> +                                            gfp_t gfp)
> +{
> +       struct mem_cgroup *memcg;
> +       int idx;
> +
> +       gfp |=  cachep->allocflags;
> +
> +       if ((current->mm == NULL))
> +               return cachep;
> +
> +       if (cachep->memcg_params.memcg)
> +               return cachep;
> +
> +       idx = cachep->memcg_params.id;
> +       VM_BUG_ON(idx == -1);
> +
> +       memcg = mem_cgroup_from_task(current);
> +       if (!mem_cgroup_kmem_enabled(memcg))
> +               return cachep;
> +
> +       if (rcu_access_pointer(memcg->slabs[idx]) == NULL) {
> +               memcg_create_cache_enqueue(memcg, cachep);
> +               return cachep;
> +       }
> +
> +       return rcu_dereference(memcg->slabs[idx]);

Is it ok to call rcu_access_pointer() and rcu_dereference() without
holding rcu_read_lock()?

> +}
> +EXPORT_SYMBOL(__mem_cgroup_get_kmem_cache);
> +
> +void mem_cgroup_remove_child_kmem_cache(struct kmem_cache *cachep, int id)
> +{
> +       rcu_assign_pointer(cachep->memcg_params.memcg->slabs[id], NULL);
> +}
> +
> +bool __mem_cgroup_charge_kmem(gfp_t gfp, size_t size)
> +{
> +       struct mem_cgroup *memcg;
> +       bool ret = true;
> +
> +       rcu_read_lock();
> +       memcg = mem_cgroup_from_task(current);
> +
> +       if (!mem_cgroup_kmem_enabled(memcg))
> +               goto out;
> +
> +       mem_cgroup_get(memcg);

Why do we need to get a reference to the memcg for every charge?
How will this work when deleting a memcg?

> +       ret = memcg_charge_kmem(memcg, gfp, size) == 0;
> +       if (ret)
> +               mem_cgroup_put(memcg);
> +out:
> +       rcu_read_unlock();
> +       return ret;
> +}
> +EXPORT_SYMBOL(__mem_cgroup_charge_kmem);
> +
> +void __mem_cgroup_uncharge_kmem(size_t size)
> +{
> +       struct mem_cgroup *memcg;
> +
> +       rcu_read_lock();
> +       memcg = mem_cgroup_from_task(current);
> +
> +       if (!mem_cgroup_kmem_enabled(memcg))
> +               goto out;
> +
> +       mem_cgroup_put(memcg);
> +       memcg_uncharge_kmem(memcg, size);
> +out:
> +       rcu_read_unlock();
> +}
> +EXPORT_SYMBOL(__mem_cgroup_uncharge_kmem);
> +
> +bool __mem_cgroup_charge_slab(struct kmem_cache *cachep, gfp_t gfp, size_t size)
> +{
> +       struct mem_cgroup *memcg;
> +       bool ret = true;
> +
> +       rcu_read_lock();
> +       memcg = cachep->memcg_params.memcg;
> +       if (!mem_cgroup_kmem_enabled(memcg))
> +               goto out;
> +
> +       ret = memcg_charge_kmem(memcg, gfp, size) == 0;
> +out:
> +       rcu_read_unlock();
> +       return ret;
> +}
> +EXPORT_SYMBOL(__mem_cgroup_charge_slab);
> +
> +void __mem_cgroup_uncharge_slab(struct kmem_cache *cachep, size_t size)
> +{
> +       struct mem_cgroup *memcg;
> +
> +       rcu_read_lock();
> +       memcg = cachep->memcg_params.memcg;
> +
> +       if (!mem_cgroup_kmem_enabled(memcg)) {
> +               rcu_read_unlock();
> +               return;
> +       }
> +       rcu_read_unlock();
> +
> +       memcg_uncharge_kmem(memcg, size);
> +}
> +EXPORT_SYMBOL(__mem_cgroup_uncharge_slab);
> +
> +static void memcg_slab_init(struct mem_cgroup *memcg)
> +{
> +       int i;
> +
> +       for (i = 0; i < MAX_KMEM_CACHE_TYPES; i++)
> +               rcu_assign_pointer(memcg->slabs[i], NULL);
> +}
>  #endif /* CONFIG_CGROUP_MEM_RES_CTLR_KMEM */
>
>  static void drain_all_stock_async(struct mem_cgroup *memcg);
> @@ -4790,7 +5103,11 @@ static struct cftype kmem_cgroup_files[] = {
>
>  static int memcg_init_kmem(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
>  {
> -       return mem_cgroup_sockets_init(memcg, ss);
> +       int ret = mem_cgroup_sockets_init(memcg, ss);
> +
> +       if (!ret)
> +               memcg_slab_init(memcg);
> +       return ret;
>  };
>
>  static void kmem_cgroup_destroy(struct mem_cgroup *memcg)
> @@ -5805,3 +6122,57 @@ static int __init enable_swap_account(char *s)
>  __setup("swapaccount=", enable_swap_account);
>
>  #endif
> +
> +#ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
> +int memcg_charge_kmem(struct mem_cgroup *memcg, gfp_t gfp, s64 delta)
> +{
> +       struct res_counter *fail_res;
> +       struct mem_cgroup *_memcg;
> +       int may_oom, ret;
> +       bool nofail = false;
> +
> +       may_oom = (gfp & __GFP_WAIT) && (gfp & __GFP_FS) &&
> +           !(gfp & __GFP_NORETRY);
> +
> +       ret = 0;
> +
> +       if (!memcg)
> +               return ret;
> +
> +       _memcg = memcg;
> +       ret = __mem_cgroup_try_charge(NULL, gfp, delta / PAGE_SIZE,
> +           &_memcg, may_oom);
> +       if (ret == -ENOMEM)
> +               return ret;
> +       else if ((ret == -EINTR) || (ret && (gfp & __GFP_NOFAIL)))  {
> +               nofail = true;
> +               /*
> +                * __mem_cgroup_try_charge() chose to bypass to root due
> +                * to OOM kill or fatal signal.
> +                * Since our only options are to either fail the
> +                * allocation or charge it to this cgroup, force the
> +                * change, going above the limit if needed.
> +                */
> +               res_counter_charge_nofail(&memcg->res, delta, &fail_res);

We might need to charge memsw here too.

> +       }
> +
> +       if (nofail)
> +               res_counter_charge_nofail(&memcg->kmem, delta, &fail_res);
> +       else
> +               ret = res_counter_charge(&memcg->kmem, delta, &fail_res);
> +
> +       if (ret)
> +               res_counter_uncharge(&memcg->res, delta);
> +
> +       return ret;
> +}
> +
> +void memcg_uncharge_kmem(struct mem_cgroup *memcg, s64 delta)
> +{
> +       if (!memcg)
> +               return;
> +
> +       res_counter_uncharge(&memcg->kmem, delta);
> +       res_counter_uncharge(&memcg->res, delta);

Might need to uncharge memsw.

> +}
> +#endif /* CONFIG_CGROUP_MEM_RES_CTLR_KMEM */
> --
> 1.7.7.6
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists