linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Qi Zheng <qi.zheng@linux.dev>
To: Shakeel Butt <shakeel.butt@linux.dev>
Cc: hannes@cmpxchg.org, hughd@google.com, mhocko@suse.com,
	roman.gushchin@linux.dev, muchun.song@linux.dev,
	david@kernel.org, lorenzo.stoakes@oracle.com, ziy@nvidia.com,
	harry.yoo@oracle.com, yosry.ahmed@linux.dev,
	imran.f.khan@oracle.com, kamalesh.babulal@oracle.com,
	axelrasmussen@google.com, yuanchu@google.com, weixugc@google.com,
	chenridong@huaweicloud.com, mkoutny@suse.com,
	akpm@linux-foundation.org, hamzamahfooz@linux.microsoft.com,
	apais@linux.microsoft.com, lance.yang@linux.dev, bhe@redhat.com,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	cgroups@vger.kernel.org, Qi Zheng <zhengqi.arch@bytedance.com>
Subject: Re: [PATCH v4 29/31] mm: memcontrol: prepare for reparenting non-hierarchical stats
Date: Tue, 10 Feb 2026 14:47:51 +0800	[thread overview]
Message-ID: <0673b72c-8d7c-4bfb-a8b2-da5ae5bb5f00@linux.dev> (raw)
In-Reply-To: <aYabQii_-9EVdgub@linux.dev>



On 2/7/26 10:19 AM, Shakeel Butt wrote:
> On Thu, Feb 05, 2026 at 05:01:48PM +0800, Qi Zheng wrote:
>> From: Qi Zheng <zhengqi.arch@bytedance.com>
>>
>> To resolve the dying memcg issue, we need to reparent LRU folios of child
>> memcg to its parent memcg. This could cause problems for non-hierarchical
>> stats.
>>
>> As Yosry Ahmed pointed out:
>>
>> ```
>> In short, if memory is charged to a dying cgroup at the time of
>> reparenting, when the memory gets uncharged the stats updates will occur
>> at the parent. This will update both hierarchical and non-hierarchical
>> stats of the parent, which would corrupt the parent's non-hierarchical
>> stats (because those counters were never incremented when the memory was
>> charged).
>> ```
>>
>> Now we have the following two types of non-hierarchical stats, and they
>> are only used in CONFIG_MEMCG_V1:
>>
>> a. memcg->vmstats->state_local[i]
>> b. pn->lruvec_stats->state_local[i]
>>
>> To ensure that these non-hierarchical stats work properly, we need to
>> reparent these non-hierarchical stats after reparenting LRU folios. To
>> this end, this commit makes the following preparations:
>>
>> 1. implement reparent_state_local() to reparent non-hierarchical stats
>> 2. make css_killed_work_fn() to be called in rcu work, and implement
>>     get_non_dying_memcg_start() and get_non_dying_memcg_end() to avoid race
>>     between mod_memcg_state()/mod_memcg_lruvec_state()
>>     and reparent_state_local()
>> 3. change these non-hierarchical stats to atomic_long_t type to avoid race
>>     between mem_cgroup_stat_aggregate() and reparent_state_local()
>>
>> Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
> 
> Overall looks good just a couple of comments.
> 
>> ---
>>   include/linux/memcontrol.h |   4 ++
>>   kernel/cgroup/cgroup.c     |   8 +--
>>   mm/memcontrol-v1.c         |  16 ++++++
>>   mm/memcontrol-v1.h         |   3 +
>>   mm/memcontrol.c            | 113 ++++++++++++++++++++++++++++++++++---
>>   5 files changed, 132 insertions(+), 12 deletions(-)
>>
>> diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
>> index 3970c102fe741..a4f6ab7eb98d6 100644
>> --- a/include/linux/memcontrol.h
>> +++ b/include/linux/memcontrol.h
>> @@ -957,12 +957,16 @@ static inline void mod_memcg_page_state(struct page *page,
>>   
>>   unsigned long memcg_events(struct mem_cgroup *memcg, int event);
>>   unsigned long memcg_page_state(struct mem_cgroup *memcg, int idx);
>> +void reparent_memcg_state_local(struct mem_cgroup *memcg,
>> +				struct mem_cgroup *parent, int idx);
> 
> Put the above in mm/memcontrol-v1.h file.

OK.

> 
>>   unsigned long memcg_page_state_output(struct mem_cgroup *memcg, int item);
>>   bool memcg_stat_item_valid(int idx);
>>   bool memcg_vm_event_item_valid(enum vm_event_item idx);
>>   unsigned long lruvec_page_state(struct lruvec *lruvec, enum node_stat_item idx);
>>   unsigned long lruvec_page_state_local(struct lruvec *lruvec,
>>   				      enum node_stat_item idx);
>> +void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
>> +				       struct mem_cgroup *parent, int idx);
> 
> Put the above in mm/memcontrol-v1.h file.

OK.

> 
>>   
>>   void mem_cgroup_flush_stats(struct mem_cgroup *memcg);
>>   void mem_cgroup_flush_stats_ratelimited(struct mem_cgroup *memcg);
>> diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
>> index 94788bd1fdf0e..dbf94a77018e6 100644
>> --- a/kernel/cgroup/cgroup.c
>> +++ b/kernel/cgroup/cgroup.c
>> @@ -6043,8 +6043,8 @@ int cgroup_mkdir(struct kernfs_node *parent_kn, const char *name, umode_t mode)
>>    */
>>   static void css_killed_work_fn(struct work_struct *work)
>>   {
>> -	struct cgroup_subsys_state *css =
>> -		container_of(work, struct cgroup_subsys_state, destroy_work);
>> +	struct cgroup_subsys_state *css = container_of(to_rcu_work(work),
>> +				struct cgroup_subsys_state, destroy_rwork);
>>   
>>   	cgroup_lock();
>>   
>> @@ -6065,8 +6065,8 @@ static void css_killed_ref_fn(struct percpu_ref *ref)
>>   		container_of(ref, struct cgroup_subsys_state, refcnt);
>>   
>>   	if (atomic_dec_and_test(&css->online_cnt)) {
>> -		INIT_WORK(&css->destroy_work, css_killed_work_fn);
>> -		queue_work(cgroup_offline_wq, &css->destroy_work);
>> +		INIT_RCU_WORK(&css->destroy_rwork, css_killed_work_fn);
>> +		queue_rcu_work(cgroup_offline_wq, &css->destroy_rwork);
>>   	}
>>   }
>>   
>> diff --git a/mm/memcontrol-v1.c b/mm/memcontrol-v1.c
>> index c6078cd7f7e53..a427bb205763b 100644
>> --- a/mm/memcontrol-v1.c
>> +++ b/mm/memcontrol-v1.c
>> @@ -1887,6 +1887,22 @@ static const unsigned int memcg1_events[] = {
>>   	PGMAJFAULT,
>>   };
>>   
>> +void reparent_memcg1_state_local(struct mem_cgroup *memcg, struct mem_cgroup *parent)
>> +{
>> +	int i;
>> +
>> +	for (i = 0; i < ARRAY_SIZE(memcg1_stats); i++)
>> +		reparent_memcg_state_local(memcg, parent, memcg1_stats[i]);
>> +}
>> +
>> +void reparent_memcg1_lruvec_state_local(struct mem_cgroup *memcg, struct mem_cgroup *parent)
>> +{
>> +	int i;
>> +
>> +	for (i = 0; i < NR_LRU_LISTS; i++)
>> +		reparent_memcg_lruvec_state_local(memcg, parent, i);
>> +}
>> +
>>   void memcg1_stat_format(struct mem_cgroup *memcg, struct seq_buf *s)
>>   {
>>   	unsigned long memory, memsw;
>> diff --git a/mm/memcontrol-v1.h b/mm/memcontrol-v1.h
>> index eb3c3c1056574..45528195d3578 100644
>> --- a/mm/memcontrol-v1.h
>> +++ b/mm/memcontrol-v1.h
>> @@ -41,6 +41,7 @@ static inline bool do_memsw_account(void)
>>   
>>   unsigned long memcg_events_local(struct mem_cgroup *memcg, int event);
>>   unsigned long memcg_page_state_local(struct mem_cgroup *memcg, int idx);
>> +void mod_memcg_page_state_local(struct mem_cgroup *memcg, int idx, unsigned long val);
>>   unsigned long memcg_page_state_local_output(struct mem_cgroup *memcg, int item);
>>   bool memcg1_alloc_events(struct mem_cgroup *memcg);
>>   void memcg1_free_events(struct mem_cgroup *memcg);
>> @@ -73,6 +74,8 @@ void memcg1_uncharge_batch(struct mem_cgroup *memcg, unsigned long pgpgout,
>>   			   unsigned long nr_memory, int nid);
>>   
>>   void memcg1_stat_format(struct mem_cgroup *memcg, struct seq_buf *s);
>> +void reparent_memcg1_state_local(struct mem_cgroup *memcg, struct mem_cgroup *parent);
>> +void reparent_memcg1_lruvec_state_local(struct mem_cgroup *memcg, struct mem_cgroup *parent);
>>   
>>   void memcg1_account_kmem(struct mem_cgroup *memcg, int nr_pages);
>>   static inline bool memcg1_tcpmem_active(struct mem_cgroup *memcg)
>> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
>> index c9b5dfd822d0a..e7d4e4ff411b6 100644
>> --- a/mm/memcontrol.c
>> +++ b/mm/memcontrol.c
>> @@ -225,6 +225,26 @@ static inline struct obj_cgroup *__memcg_reparent_objcgs(struct mem_cgroup *memc
>>   	return objcg;
>>   }
>>   
>> +#ifdef CONFIG_MEMCG_V1
>> +static void __mem_cgroup_flush_stats(struct mem_cgroup *memcg, bool force);
>> +
>> +static inline void reparent_state_local(struct mem_cgroup *memcg, struct mem_cgroup *parent)
>> +{
>> +	if (cgroup_subsys_on_dfl(memory_cgrp_subsys))
>> +		return;
>> +
>> +	__mem_cgroup_flush_stats(memcg, true);
>> +
>> +	/* The following counts are all non-hierarchical and need to be reparented. */
>> +	reparent_memcg1_state_local(memcg, parent);
>> +	reparent_memcg1_lruvec_state_local(memcg, parent);
>> +}
>> +#else
>> +static inline void reparent_state_local(struct mem_cgroup *memcg, struct mem_cgroup *parent)
>> +{
>> +}
>> +#endif
>> +
>>   static inline void reparent_locks(struct mem_cgroup *memcg, struct mem_cgroup *parent)
>>   {
>>   	spin_lock_irq(&objcg_lock);
>> @@ -407,7 +427,7 @@ struct lruvec_stats {
>>   	long state[NR_MEMCG_NODE_STAT_ITEMS];
>>   
>>   	/* Non-hierarchical (CPU aggregated) state */
>> -	long state_local[NR_MEMCG_NODE_STAT_ITEMS];
>> +	atomic_long_t state_local[NR_MEMCG_NODE_STAT_ITEMS];
>>   
>>   	/* Pending child counts during tree propagation */
>>   	long state_pending[NR_MEMCG_NODE_STAT_ITEMS];
>> @@ -450,7 +470,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
>>   		return 0;
>>   
>>   	pn = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
>> -	x = READ_ONCE(pn->lruvec_stats->state_local[i]);
>> +	x = atomic_long_read(&(pn->lruvec_stats->state_local[i]));
>>   #ifdef CONFIG_SMP
>>   	if (x < 0)
>>   		x = 0;
>> @@ -458,6 +478,27 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
>>   	return x;
>>   }
>>   
> 
> Please put the following function under CONFIG_MEMCG_V1. Just move it in
> the same block as reparent_state_local().

OK, will try to do it.

> 
>> +void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
>> +				       struct mem_cgroup *parent, int idx)
>> +{
>> +	int i = memcg_stats_index(idx);
>> +	int nid;
>> +
>> +	if (WARN_ONCE(BAD_STAT_IDX(i), "%s: missing stat item %d\n", __func__, idx))
>> +		return;
>> +
>> +	for_each_node(nid) {
>> +		struct lruvec *child_lruvec = mem_cgroup_lruvec(memcg, NODE_DATA(nid));
>> +		struct lruvec *parent_lruvec = mem_cgroup_lruvec(parent, NODE_DATA(nid));
>> +		struct mem_cgroup_per_node *parent_pn;
>> +		unsigned long value = lruvec_page_state_local(child_lruvec, idx);
>> +
>> +		parent_pn = container_of(parent_lruvec, struct mem_cgroup_per_node, lruvec);
>> +
>> +		atomic_long_add(value, &(parent_pn->lruvec_stats->state_local[i]));
>> +	}
>> +}
>> +
> 
> [...]
> 
>>   
>> +#ifdef CONFIG_MEMCG_V1
>> +/*
>> + * Used in mod_memcg_state() and mod_memcg_lruvec_state() to avoid race with
>> + * reparenting of non-hierarchical state_locals.
>> + */
>> +static inline struct mem_cgroup *get_non_dying_memcg_start(struct mem_cgroup *memcg)
>> +{
>> +	if (cgroup_subsys_on_dfl(memory_cgrp_subsys))
>> +		return memcg;
>> +
>> +	rcu_read_lock();
>> +
>> +	while (memcg_is_dying(memcg))
>> +		memcg = parent_mem_cgroup(memcg);
>> +
>> +	return memcg;
>> +}
>> +
>> +static inline void get_non_dying_memcg_end(void)
>> +{
>> +	if (cgroup_subsys_on_dfl(memory_cgrp_subsys))
>> +		return;
>> +
>> +	rcu_read_unlock();
>> +}
>> +#else
>> +static inline struct mem_cgroup *get_non_dying_memcg_start(struct mem_cgroup *memcg)
>> +{
>> +	return memcg;
>> +}
>> +
>> +static inline void get_non_dying_memcg_end(void)
>> +{
>> +}
>> +#endif
> 
> Add the usage of these start and end functions in mod_memcg_state() and
> mod_memcg_lruvec_state() in this patch.

Using these two function will change the behavior of mod_memcg_state()
and mod_memcg_lruvec_state(), but LRU folios has not yet been
reparented.

To ensure the patch itself is error-free, I chose to place the usage of
these two function in patch #30.

Thanks,
Qi

> 



  reply	other threads:[~2026-02-10  6:49 UTC|newest]

Thread overview: 50+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-02-05  8:54 [PATCH v4 00/31] Eliminate Dying Memory Cgroup Qi Zheng
2026-02-05  8:54 ` [PATCH v4 01/31] mm: memcontrol: remove dead code of checking parent memory cgroup Qi Zheng
2026-02-05  8:54 ` [PATCH v4 02/31] mm: workingset: use folio_lruvec() in workingset_refault() Qi Zheng
2026-02-05  8:54 ` [PATCH v4 03/31] mm: rename unlock_page_lruvec_irq and its variants Qi Zheng
2026-02-05  8:54 ` [PATCH v4 04/31] mm: vmscan: prepare for the refactoring the move_folios_to_lru() Qi Zheng
2026-02-05  8:54 ` [PATCH v4 05/31] mm: vmscan: refactor move_folios_to_lru() Qi Zheng
2026-02-05  8:54 ` [PATCH v4 06/31] mm: memcontrol: allocate object cgroup for non-kmem case Qi Zheng
2026-02-05  8:54 ` [PATCH v4 07/31] mm: memcontrol: return root object cgroup for root memory cgroup Qi Zheng
2026-02-05  9:01 ` [PATCH v4 08/31] mm: memcontrol: prevent memory cgroup release in get_mem_cgroup_from_folio() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 09/31] buffer: prevent memory cgroup release in folio_alloc_buffers() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 10/31] writeback: prevent memory cgroup release in writeback module Qi Zheng
2026-02-05  9:01 ` [PATCH v4 11/31] mm: memcontrol: prevent memory cgroup release in count_memcg_folio_events() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 12/31] mm: page_io: prevent memory cgroup release in page_io module Qi Zheng
2026-02-05  9:01 ` [PATCH v4 13/31] mm: migrate: prevent memory cgroup release in folio_migrate_mapping() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 14/31] mm: mglru: prevent memory cgroup release in mglru Qi Zheng
2026-02-05  9:01 ` [PATCH v4 15/31] mm: memcontrol: prevent memory cgroup release in mem_cgroup_swap_full() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 16/31] mm: workingset: prevent memory cgroup release in lru_gen_eviction() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 17/31] mm: thp: prevent memory cgroup release in folio_split_queue_lock{_irqsave}() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 18/31] mm: zswap: prevent memory cgroup release in zswap_compress() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 19/31] mm: workingset: prevent lruvec release in workingset_refault() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 20/31] mm: zswap: prevent lruvec release in zswap_folio_swapin() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 21/31] mm: swap: prevent lruvec release in lru_gen_clear_refs() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 22/31] mm: workingset: prevent lruvec release in workingset_activation() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 23/31] mm: do not open-code lruvec lock Qi Zheng
2026-02-05  9:01 ` [PATCH v4 24/31] mm: memcontrol: prepare for reparenting LRU pages for " Qi Zheng
2026-02-05 15:02   ` kernel test robot
2026-02-05 15:02   ` kernel test robot
2026-02-06  6:13     ` Qi Zheng
2026-02-06 23:34       ` Shakeel Butt
2026-02-05  9:01 ` [PATCH v4 25/31] mm: vmscan: prepare for reparenting traditional LRU folios Qi Zheng
2026-02-07  1:28   ` Shakeel Butt
2026-02-05  9:01 ` [PATCH v4 26/31] mm: vmscan: prepare for reparenting MGLRU folios Qi Zheng
2026-02-12  8:46   ` Harry Yoo
2026-02-15  7:28     ` Qi Zheng
2026-02-05  9:01 ` [PATCH v4 27/31] mm: memcontrol: refactor memcg_reparent_objcgs() Qi Zheng
2026-02-05  9:01 ` [PATCH v4 28/31] mm: workingset: use lruvec_lru_size() to get the number of lru pages Qi Zheng
2026-02-07  1:48   ` Shakeel Butt
2026-02-07  3:59   ` Muchun Song
2026-02-05  9:01 ` [PATCH v4 29/31] mm: memcontrol: prepare for reparenting non-hierarchical stats Qi Zheng
2026-02-07  2:19   ` Shakeel Butt
2026-02-10  6:47     ` Qi Zheng [this message]
2026-02-11  0:38       ` Shakeel Butt
2026-02-05  9:01 ` [PATCH v4 30/31] mm: memcontrol: eliminate the problem of dying memory cgroup for LRU folios Qi Zheng
2026-02-07 19:59   ` Usama Arif
2026-02-07 22:25   ` Shakeel Butt
2026-02-09  3:49     ` Qi Zheng
2026-02-09 17:53       ` Shakeel Butt
2026-02-10  3:11         ` Qi Zheng
2026-02-05  9:01 ` [PATCH v4 31/31] mm: lru: add VM_WARN_ON_ONCE_FOLIO to lru maintenance helpers Qi Zheng
2026-02-07 22:26   ` Shakeel Butt

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=0673b72c-8d7c-4bfb-a8b2-da5ae5bb5f00@linux.dev \
    --to=qi.zheng@linux.dev \
    --cc=akpm@linux-foundation.org \
    --cc=apais@linux.microsoft.com \
    --cc=axelrasmussen@google.com \
    --cc=bhe@redhat.com \
    --cc=cgroups@vger.kernel.org \
    --cc=chenridong@huaweicloud.com \
    --cc=david@kernel.org \
    --cc=hamzamahfooz@linux.microsoft.com \
    --cc=hannes@cmpxchg.org \
    --cc=harry.yoo@oracle.com \
    --cc=hughd@google.com \
    --cc=imran.f.khan@oracle.com \
    --cc=kamalesh.babulal@oracle.com \
    --cc=lance.yang@linux.dev \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=lorenzo.stoakes@oracle.com \
    --cc=mhocko@suse.com \
    --cc=mkoutny@suse.com \
    --cc=muchun.song@linux.dev \
    --cc=roman.gushchin@linux.dev \
    --cc=shakeel.butt@linux.dev \
    --cc=weixugc@google.com \
    --cc=yosry.ahmed@linux.dev \
    --cc=yuanchu@google.com \
    --cc=zhengqi.arch@bytedance.com \
    --cc=ziy@nvidia.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox