linux-kernel - Re: [PATCH v2] mm: do not update memcg stats for NR_{FILE/SHMEM}

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <d99b6375-cbae-41f1-9221-f1dd25aab150@redhat.com>
Date: Mon, 6 May 2024 21:50:10 +0200
From: David Hildenbrand <david@...hat.com>
To: Yosry Ahmed <yosryahmed@...gle.com>,
 Andrew Morton <akpm@...ux-foundation.org>
Cc: Shakeel Butt <shakeel.butt@...ux.dev>,
 Johannes Weiner <hannes@...xchg.org>, Michal Hocko <mhocko@...nel.org>,
 Roman Gushchin <roman.gushchin@...ux.dev>,
 Muchun Song <muchun.song@...ux.dev>, linux-mm@...ck.org,
 cgroups@...r.kernel.org, linux-kernel@...r.kernel.org,
 syzbot+9319a4268a640e26b72b@...kaller.appspotmail.com
Subject: Re: [PATCH v2] mm: do not update memcg stats for
 NR_{FILE/SHMEM}_PMDMAPPED

On 06.05.24 21:29, Yosry Ahmed wrote:
> Previously, all NR_VM_EVENT_ITEMS stats were maintained per-memcg,
> although some of those fields are not exposed anywhere. Commit
> 14e0f6c957e39 ("memcg: reduce memory for the lruvec and memcg stats")
> changed this such that we only maintain the stats we actually expose
> per-memcg via a translation table.
> 
> Additionally, commit 514462bbe927b ("memcg: warn for unexpected events
> and stats") added a warning if a per-memcg stat update is attempted for
> a stat that is not in the translation table. The warning started firing
> for the NR_{FILE/SHMEM}_PMDMAPPED stat updates in the rmap code. These
> stats are not maintained per-memcg, and hence are not in the translation
> table.
> 
> Do not use __lruvec_stat_mod_folio() when updating NR_FILE_PMDMAPPED and
> NR_SHMEM_PMDMAPPED. Use __mod_node_page_state() instead, which updates
> the global per-node stats only.
> 
> Reported-by: syzbot+9319a4268a640e26b72b@...kaller.appspotmail.com
> Closes: https://lore.kernel.org/lkml/0000000000001b9d500617c8b23c@google.com
> Fixes: 514462bbe927 ("memcg: warn for unexpected events and stats")
> Acked-by: Shakeel Butt <shakeel.butt@...ux.dev>
> Signed-off-by: Yosry Ahmed <yosryahmed@...gle.com>
> ---
>   mm/rmap.c | 15 +++++++++------
>   1 file changed, 9 insertions(+), 6 deletions(-)
> 
> diff --git a/mm/rmap.c b/mm/rmap.c
> index 12be4241474ab..ed7f820369864 100644
> --- a/mm/rmap.c
> +++ b/mm/rmap.c
> @@ -1435,13 +1435,14 @@ static __always_inline void __folio_add_file_rmap(struct folio *folio,
>   		struct page *page, int nr_pages, struct vm_area_struct *vma,
>   		enum rmap_level level)
>   {
> +	pg_data_t *pgdat = folio_pgdat(folio);
>   	int nr, nr_pmdmapped = 0;
>   
>   	VM_WARN_ON_FOLIO(folio_test_anon(folio), folio);
>   
>   	nr = __folio_add_rmap(folio, page, nr_pages, level, &nr_pmdmapped);
>   	if (nr_pmdmapped)
> -		__lruvec_stat_mod_folio(folio, folio_test_swapbacked(folio) ?
> +		__mod_node_page_state(pgdat, folio_test_swapbacked(folio) ?
>   			NR_SHMEM_PMDMAPPED : NR_FILE_PMDMAPPED, nr_pmdmapped);
>   	if (nr)
>   		__lruvec_stat_mod_folio(folio, NR_FILE_MAPPED, nr);
> @@ -1493,6 +1494,7 @@ static __always_inline void __folio_remove_rmap(struct folio *folio,
>   		enum rmap_level level)
>   {
>   	atomic_t *mapped = &folio->_nr_pages_mapped;
> +	pg_data_t *pgdat = folio_pgdat(folio);
>   	int last, nr = 0, nr_pmdmapped = 0;
>   	bool partially_mapped = false;
>   	enum node_stat_item idx;
> @@ -1540,13 +1542,14 @@ static __always_inline void __folio_remove_rmap(struct folio *folio,
>   	}
>   
>   	if (nr_pmdmapped) {
> +		/* NR_{FILE/SHMEM}_PMDMAPPED are not maintained per-memcg */
>   		if (folio_test_anon(folio))
> -			idx = NR_ANON_THPS;
> -		else if (folio_test_swapbacked(folio))
> -			idx = NR_SHMEM_PMDMAPPED;
> +			__lruvec_stat_mod_folio(folio, NR_ANON_THPS, -nr_pmdmapped);
>   		else
> -			idx = NR_FILE_PMDMAPPED;
> -		__lruvec_stat_mod_folio(folio, idx, -nr_pmdmapped);
> +			__mod_node_page_state(pgdat,

folio_pgdat(folio) should fit here easily. :)

But I would actually suggest something like the following in mm/rmap.c

static void __folio_mod_node_file_state(folio, int nr_pages)
{
	enum node_stat_item idx = NR_FILE_PMDMAPPED;

	if (folio_test_swapbacked(folio))
		idx = NR_SHMEM_PMDMAPPED;

	__mod_node_page_state(folio_pgdat(folio), idx, nr_pages);
}

And then simply calling here

__folio_mod_node_file_state(folio, -nr_pmdmapped);

And likewise in __folio_add_file_rmap()


.. will be cleaner.

In any case

Acked-by: David Hildenbrand <david@...hat.com>

-- 
Cheers,

David / dhildenb