Generic Linux architectural discussions
 help / color / mirror / Atom feed
From: xu.xin@linux.dev
To: david@kernel.org
Cc: akpm@linux-foundation.org, aneesh.kumar@kernel.org,
	apopple@nvidia.com, axelrasmussen@google.com, baohua@kernel.org,
	baolin.wang@linux.alibaba.com, baoquan.he@linux.dev,
	chengming.zhou@linux.dev, chrisl@kernel.org, dev.jain@arm.com,
	gourry@gourry.net, hannes@cmpxchg.org, harry@kernel.org,
	hughd@google.com, jack@suse.cz, jannh@google.com, jgg@ziepe.ca,
	jhubbard@nvidia.com, josef@toxicpanda.com,
	joshua.hahnjy@gmail.com, kas@kernel.org, kasong@tencent.com,
	kees@kernel.org, kexec@lists.infradead.org, lance.yang@linux.dev,
	liam@infradead.org, linux-arch@vger.kernel.org,
	linux-cxl@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	linux-kernel@vger.kernel.org, linux-mm@kvack.org, ljs@kernel.org,
	matthew.brost@intel.com, mhocko@kernel.org, mhocko@suse.com,
	minchan@kernel.org, muchun.song@linux.dev, nico.pache@linux.dev,
	nphamcs@gmail.com, npiggin@gmail.com, osalvador@suse.de,
	pasha.tatashin@soleen.com, peterx@redhat.com,
	peterz@infradead.org, pfalcato@suse.de, pratyush@kernel.org,
	qi.zheng@linux.dev, rakie.kim@sk.com, riel@surriel.com,
	rppt@kernel.org, ryan.roberts@arm.com, shakeel.butt@linux.dev,
	shikemeng@huaweicloud.com, stable@vger.kernel.org,
	surenb@google.com, usama.arif@linux.dev, vbabka@kernel.org,
	weixugc@google.com, will@kernel.org, willy@infradead.org,
	xu.xin@linux.dev, ying.huang@linux.alibaba.com,
	youngjun.park@lge.com, yuanchu@google.com, ziy@nvidia.com
Subject: Re: [PATCH 33/46] mm/madvise: fix data loss from MADV_FREE on a KSM page
Date: Thu, 29 Oct 2026 12:53:07 +0800	[thread overview]
Message-ID: <20261029045307.131675-1-xu.xin@linux.dev> (raw)
In-Reply-To: <20261027-fixes_core-v1-33-6f74bec9a5bb@kernel.org>

> MADV_FREE can clear the dirty flag of a singly mapped KSM folio. The
> folio remains in the stable tree and can gain new mappings without being
> dirtied again. Once marked lazyfree, reclaim can discard data that those
> mappings still need.
> 
> KSM always marks a folio dirty when converting it to a KSM folio in
> try_to_merge_one_page(). Marking a folio lazyfree clears PG_swapbacked
> but leaves PG_dirty unchanged. Preserving PG_dirty makes reclaim restore
> PG_swapbacked instead of discarding the folio. This also protects folios
> that were already lazyfree when KSM converted them.
> 
> Skip KSM folios before clearing their dirty flag, checking under the
> folio lock to serialize with KSM conversion.
> 
> Fixes: 854e9ed09ded ("mm: support madvise(MADV_FREE)")
> Cc: stable@vger.kernel.org
> Reported-by: Josef Bacik <josef@toxicpanda.com>
> Reviewed-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
> Signed-off-by: David Hildenbrand (Arm) <david@kernel.org>
> ---
>  mm/madvise.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/mm/madvise.c b/mm/madvise.c
> index 75b409e97dbf2..a16e3cb256b7b 100644
> --- a/mm/madvise.c
> +++ b/mm/madvise.c
> @@ -756,6 +756,15 @@ static int madvise_free_pte_range(pmd_t *pmd, unsigned long addr,
>  		if (folio_test_swapcache(folio) || folio_test_dirty(folio)) {
>  			if (!folio_trylock(folio))
>  				continue;
> +			/*
> +			 * KSM folios can gain new mappings at any time, so we must
> +			 * not clear their dirty flag.
> +			 */
> +			if (folio_test_ksm(folio)) {
> +				folio_unlock(folio);
> +				continue;
> +			}
> +
>  			/*
>  			 * If we have a large folio at this point, we know it is
>  			 * fully mapped so if its mapcount is the same as its
> 
> -- 
> 2.43.0

The fix looks correct to me: checking folio_test_ksm() under the
folio lock properly serializes with KSM conversion, and skipping
the folio avoids clearing PG_dirty on a KSM folio.

Reviewed-by: Xu Xin <xu.xin@linux.dev> 

  reply	other threads:[~2026-10-29  4:53 UTC|newest]

Thread overview: 187+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-26 23:17 [PATCH 00/46] mm: various (urgent?) fixes David Hildenbrand (Arm)
2026-10-26 23:17 ` [PATCH 01/46] mm/mglru: don't use stack pointers when walking PUD and PMD tables David Hildenbrand (Arm)
2026-10-27  1:32   ` Rik van Riel
2026-10-27  3:05   ` Qi Zheng
2026-10-27  7:20   ` Dev Jain
2026-10-27  8:06   ` Lorenzo Stoakes (ARM)
2026-10-27 19:06     ` David Hildenbrand (Arm)
2026-10-28 16:54       ` Kairui Song
2026-10-27 10:22   ` Kairui Song
2026-10-27 20:07   ` Harry Yoo
2026-10-27 20:24   ` Barry Song
2026-10-28  8:24   ` Baolin Wang
2026-10-28 10:29   ` Pedro Falcato
2026-10-28 16:09   ` Kiryl Shutsemau
2026-10-26 23:17 ` [PATCH 02/46] fork: take mmap_lock before copying mm_struct David Hildenbrand (Arm)
2026-10-27  1:38   ` Rik van Riel
2026-10-27  8:18   ` Vlastimil Babka (SUSE)
2026-10-27  9:45   ` Barry Song
2026-10-27 10:38   ` Pedro Falcato
2026-10-27 19:07     ` David Hildenbrand (Arm)
2026-10-27 19:48       ` Barry Song
2026-10-28  8:20         ` David Hildenbrand (Arm)
2026-10-28 12:55         ` Pedro Falcato
2026-10-28 21:11           ` Barry Song
2026-10-29 10:57             ` David Hildenbrand (Arm)
2026-10-27 13:56   ` Harry Yoo
2026-10-28 16:19   ` Kiryl Shutsemau
2026-10-26 23:17 ` [PATCH 03/46] mm/memfd: check for real shmem inodes before accessing shmem state David Hildenbrand (Arm)
2026-10-27  2:04   ` Rik van Riel
2026-10-27  8:13   ` Mike Rapoport
2026-10-27 13:42     ` David Hildenbrand (Arm)
2026-10-27 14:41   ` Pratyush Yadav
2026-10-28 16:53   ` Pasha Tatashin
2026-10-26 23:17 ` [PATCH 04/46] mm/memfd: make f_pos accesses atomic David Hildenbrand (Arm)
2026-10-27  2:06   ` Rik van Riel
2026-10-27  3:34   ` Gregory Price
2026-10-27  8:44     ` Lorenzo Stoakes (ARM)
2026-10-27 19:09     ` David Hildenbrand (Arm)
2026-10-27  8:15   ` Mike Rapoport
2026-10-27  8:46   ` Lorenzo Stoakes (ARM)
2026-10-27 10:39   ` Pedro Falcato
2026-10-28 16:40   ` Kiryl Shutsemau
2026-10-28 16:55   ` Pasha Tatashin
2026-10-26 23:17 ` [PATCH 05/46] mm/truncate: look again if a folio left the mapping David Hildenbrand (Arm)
2026-10-27  2:17   ` Rik van Riel
2026-10-28 16:54   ` Kiryl Shutsemau
2026-10-28 21:45     ` David Hildenbrand (Arm)
2026-10-29 11:37       ` Kiryl Shutsemau
2026-10-29  3:00   ` Zi Yan
2026-10-26 23:17 ` [PATCH 06/46] mm/filemap: read mapping->a_ops->free_folio under i_lock David Hildenbrand (Arm)
2026-10-27  2:43   ` Rik van Riel
2026-10-27 19:10     ` David Hildenbrand (Arm)
2026-10-27 15:47   ` Pedro Falcato
2026-10-26 23:17 ` [PATCH 07/46] mm/filemap: don't lose a folio lock wakeup to a signal David Hildenbrand (Arm)
2026-10-27  3:03   ` Rik van Riel
2026-10-27 16:10   ` Pedro Falcato
2026-10-27 19:18     ` David Hildenbrand (Arm)
2026-10-28  2:17       ` Andrew Morton
2026-10-27 22:02     ` Rik van Riel
2026-10-26 23:17 ` [PATCH 08/46] mm/filemap: finish fault cleanup before dropping the fault lock David Hildenbrand (Arm)
2026-10-28 19:26   ` Thadeu Lima de Souza Cascardo
2026-10-28 20:56     ` David Hildenbrand (Arm)
2026-10-26 23:17 ` [PATCH 09/46] mm/rmap: handle device-exclusive entries in folio_referenced_one() David Hildenbrand (Arm)
2026-10-27 20:25   ` Harry Yoo
2026-10-27 21:22   ` Barry Song
2026-10-28  7:51   ` Dev Jain
2026-10-28  8:08     ` David Hildenbrand (Arm)
2026-10-28  9:09       ` Lorenzo Stoakes (ARM)
2026-10-26 23:17 ` [PATCH 10/46] mm/huge_memory: fix list corruption when deferred_split_scan() races with deferred_split_folio() David Hildenbrand (Arm)
2026-10-27 14:10   ` Johannes Weiner
2026-10-27 15:15     ` David Hildenbrand (Arm)
2026-10-27 18:03       ` Mani Tofigh
2026-10-27 19:19         ` David Hildenbrand (Arm)
2026-10-28  0:14   ` Mani Tofigh
2026-10-28 10:41   ` Lance Yang
2026-10-26 23:17 ` [PATCH 11/46] mm/huge_memory: properly handle refcounted PUD mappings on fork David Hildenbrand (Arm)
2026-10-27 21:10   ` Zi Yan
2026-10-28 13:05   ` Pedro Falcato
2026-10-28 13:12   ` Lance Yang
2026-10-26 23:17 ` [PATCH 12/46] mm/huge_memory: fix livelock on device private folio David Hildenbrand (Arm)
2026-10-28  2:08   ` Zi Yan
2026-10-28 15:04   ` Lance Yang
2026-10-26 23:17 ` [PATCH 13/46] mm/huge_memory: fix truncate racing with PMD zapping David Hildenbrand (Arm)
2026-10-28  2:54   ` Zi Yan
2026-10-28  8:16     ` David Hildenbrand (Arm)
2026-10-28 12:43   ` Pedro Falcato
2026-10-28 15:54   ` Lance Yang
2026-10-26 23:17 ` [PATCH 14/46] mm/huge_memory: fix zapping migration entries for device-private THPs David Hildenbrand (Arm)
2026-10-28  2:58   ` Zi Yan
2026-10-28  8:25   ` Dev Jain
2026-10-26 23:17 ` [PATCH 15/46] mm/khugepaged: always recheck the pmd in try_collapse_pte_mapped_thp() David Hildenbrand (Arm)
2026-10-29  1:20   ` Gregory Price
2026-10-26 23:17 ` [PATCH 16/46] mm/khugepaged: fix PMD table lock leak " David Hildenbrand (Arm)
2026-10-29  1:25   ` Gregory Price
2026-10-26 23:17 ` [PATCH 17/46] mm/khugepaged: preserve dirty bits when collapsing PTE-mapped THPs David Hildenbrand (Arm)
2026-10-28 13:30   ` Pedro Falcato
2026-10-29  1:27   ` Gregory Price
2026-10-26 23:17 ` [PATCH 18/46] mm/khugepaged: check if PTE table is empty in try_collapse_pte_mapped_thp() David Hildenbrand (Arm)
2026-10-29  1:48   ` Gregory Price
2026-10-29 11:01     ` David Hildenbrand (Arm)
2026-10-26 23:17 ` [PATCH 19/46] mm/gup: recheck the PUD under PTL in follow_pud_mask() David Hildenbrand (Arm)
2026-10-26 23:17 ` [PATCH 20/46] mm/gup: fix clearing folios in memfd_pin_folios() David Hildenbrand (Arm)
2026-10-27 15:00   ` Pratyush Yadav
2026-10-27 19:20     ` David Hildenbrand (Arm)
2026-10-28 11:40       ` Pratyush Yadav
2026-10-28 20:57         ` David Hildenbrand (Arm)
2026-10-27 19:43   ` Samiullah Khawaja
2026-10-26 23:17 ` [PATCH 21/46] mm/gup: fix out-of-bounds write for nr_pages == 0 in GUP-fast David Hildenbrand (Arm)
2026-10-27  9:37   ` Lorenzo Stoakes (ARM)
2026-10-27  9:40   ` Muchun Song
2026-10-26 23:17 ` [PATCH 22/46] mm/gup: check for overflow when converting nr_pages to bytes " David Hildenbrand (Arm)
2026-10-27  9:34   ` Lorenzo Stoakes (ARM)
2026-10-27  9:44   ` Muchun Song
2026-10-27 21:31   ` Barry Song
2026-10-26 23:17 ` [PATCH 23/46] mm/userfaultfd: write-lock the VMA before waking waiters when unregistering David Hildenbrand (Arm)
2026-10-27  8:16   ` Mike Rapoport
2026-10-27  8:43   ` Lorenzo Stoakes (ARM)
2026-10-27 17:38     ` Suren Baghdasaryan
2026-10-27 19:21     ` David Hildenbrand (Arm)
2026-10-26 23:17 ` [PATCH 24/46] mm/userfaultfd: wake all waiters on unregister David Hildenbrand (Arm)
2026-10-27  8:17   ` Mike Rapoport
2026-10-27  8:36   ` Lorenzo Stoakes (ARM)
2026-10-26 23:17 ` [PATCH 25/46] mm/userfaultfd: wake waiters when registration modes are removed David Hildenbrand (Arm)
2026-10-27  5:10   ` Gregory Price
2026-10-27  8:24     ` Lorenzo Stoakes (ARM)
2026-10-27 19:31       ` David Hildenbrand (Arm)
2026-10-27  8:21   ` Mike Rapoport
2026-10-27  8:31   ` Lorenzo Stoakes (ARM)
2026-10-26 23:17 ` [PATCH 26/46] mm/mmu_gather: fix flushing ranges with 4 GiB or larger David Hildenbrand (Arm)
2026-10-27  3:06   ` Qi Zheng
2026-10-27  3:40   ` Gregory Price
2026-10-27  7:35   ` Dev Jain
2026-10-27 19:25     ` David Hildenbrand (Arm)
2026-10-27  9:50   ` Muchun Song
2026-10-27 15:42   ` Pedro Falcato
2026-10-26 23:17 ` [PATCH 27/46] mm/mmu_notifier: finish non-blocking interval invalidation on error David Hildenbrand (Arm)
2026-10-27  9:33   ` Lorenzo Stoakes (ARM)
2026-10-26 23:17 ` [PATCH 28/46] mm/mmu_notifier: set ->mm only after mmgrab() David Hildenbrand (Arm)
2026-10-27  3:42   ` Gregory Price
2026-10-27  8:50   ` Lorenzo Stoakes (ARM)
2026-10-26 23:17 ` [PATCH 29/46] mm/mremap: use pudp_huge_get_and_clear() when moving huge PUDs David Hildenbrand (Arm)
2026-10-28 13:28   ` Pedro Falcato
2026-10-29  8:02   ` Dev Jain
2026-10-26 23:17 ` [PATCH 30/46] mm/mremap: recheck PUDs under the page table lock David Hildenbrand (Arm)
2026-10-28 13:26   ` Pedro Falcato
2026-10-26 23:17 ` [PATCH 31/46] mm/mremap: fix moving PUD leaves David Hildenbrand (Arm)
2026-10-28 13:20   ` Pedro Falcato
2026-10-28 21:06     ` David Hildenbrand (Arm)
2026-10-26 23:17 ` [PATCH 32/46] mm/memory_hotplug: avoid races in pfn_to_online_page() with dying pgmap David Hildenbrand (Arm)
2026-10-27 12:13   ` Lance Yang
2026-10-27 19:41     ` David Hildenbrand (Arm)
2026-10-28  4:32       ` Lance Yang
2026-10-28 21:59   ` Gregory Price (Meta)
2026-10-26 23:17 ` [PATCH 33/46] mm/madvise: fix data loss from MADV_FREE on a KSM page David Hildenbrand (Arm)
2026-10-29  4:53   ` xu.xin [this message]
2026-10-29 11:02     ` David Hildenbrand (Arm)
2026-10-26 23:17 ` [PATCH 34/46] mm/memory: reject remote COW faults on PFN mappings David Hildenbrand (Arm)
2026-10-27 16:34   ` Pedro Falcato
2026-10-26 23:17 ` [PATCH 35/46] mm: fix unmap_mapping_range() beyond 4 GiB on 32-bit David Hildenbrand (Arm)
2026-10-27 16:27   ` Pedro Falcato
2026-10-26 23:17 ` [PATCH 36/46] mm: fix unkillable fault loop on MAP_DROPPABLE memory David Hildenbrand (Arm)
2026-10-29 11:06   ` David Hildenbrand (Arm)
2026-10-26 23:17 ` [PATCH 37/46] mm: fix deadlock of mremap() against UFFDIO_MOVE David Hildenbrand (Arm)
2026-10-27  9:23   ` Lorenzo Stoakes (ARM)
2026-10-27  9:46   ` Muchun Song
2026-10-26 23:17 ` [PATCH 38/46] mm/ksm: reset the unstable trees of all nodes David Hildenbrand (Arm)
2026-10-27  9:14   ` Lorenzo Stoakes (ARM)
2026-10-28 16:56   ` Pasha Tatashin
2026-10-26 23:17 ` [PATCH 39/46] mm/ksm: fix use of a freed anon_vma at swap-in David Hildenbrand (Arm)
2026-10-26 23:18 ` [PATCH 40/46] mm/ksm: check large swap-cache folios before mapping them David Hildenbrand (Arm)
2026-10-27  9:00   ` Lorenzo Stoakes (ARM)
2026-10-26 23:18 ` [PATCH 41/46] mm/vma: call ->may_split() before a partial merge David Hildenbrand (Arm)
2026-10-27  3:45   ` Gregory Price
2026-10-26 23:18 ` [PATCH 42/46] mm/huge_memory: split huge PUDs when adjusting VMA boundaries David Hildenbrand (Arm)
2026-10-29 11:45   ` Kiryl Shutsemau
2026-10-26 23:18 ` [PATCH 43/46] mm/vma: don't leak the charge for stack growth David Hildenbrand (Arm)
2026-10-27 16:25   ` Pedro Falcato
2026-10-27 20:32   ` Harry Yoo
2026-10-28  8:18     ` David Hildenbrand (Arm)
2026-10-26 23:18 ` [PATCH 44/46] mm/cma: don't release CMA pages still in use David Hildenbrand (Arm)
2026-10-27  8:55   ` Lorenzo Stoakes (ARM)
2026-10-27  9:48   ` Muchun Song
2026-10-26 23:18 ` [PATCH 45/46] mm/memory-tiers: unregister empty tiers outside memory_tier_lock David Hildenbrand (Arm)
2026-10-28 22:54   ` Gregory Price
2026-10-26 23:18 ` [PATCH 46/46] fs/proc/task_mmu: refuse addresses beyond 32 bits on 32bit David Hildenbrand (Arm)
2026-10-27 16:13   ` Pedro Falcato
2026-10-29 11:08 ` [PATCH 00/46] mm: various (urgent?) fixes David Hildenbrand (Arm)

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=20261029045307.131675-1-xu.xin@linux.dev \
    --to=xu.xin@linux.dev \
    --cc=akpm@linux-foundation.org \
    --cc=aneesh.kumar@kernel.org \
    --cc=apopple@nvidia.com \
    --cc=axelrasmussen@google.com \
    --cc=baohua@kernel.org \
    --cc=baolin.wang@linux.alibaba.com \
    --cc=baoquan.he@linux.dev \
    --cc=chengming.zhou@linux.dev \
    --cc=chrisl@kernel.org \
    --cc=david@kernel.org \
    --cc=dev.jain@arm.com \
    --cc=gourry@gourry.net \
    --cc=hannes@cmpxchg.org \
    --cc=harry@kernel.org \
    --cc=hughd@google.com \
    --cc=jack@suse.cz \
    --cc=jannh@google.com \
    --cc=jgg@ziepe.ca \
    --cc=jhubbard@nvidia.com \
    --cc=josef@toxicpanda.com \
    --cc=joshua.hahnjy@gmail.com \
    --cc=kas@kernel.org \
    --cc=kasong@tencent.com \
    --cc=kees@kernel.org \
    --cc=kexec@lists.infradead.org \
    --cc=lance.yang@linux.dev \
    --cc=liam@infradead.org \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-cxl@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=ljs@kernel.org \
    --cc=matthew.brost@intel.com \
    --cc=mhocko@kernel.org \
    --cc=mhocko@suse.com \
    --cc=minchan@kernel.org \
    --cc=muchun.song@linux.dev \
    --cc=nico.pache@linux.dev \
    --cc=nphamcs@gmail.com \
    --cc=npiggin@gmail.com \
    --cc=osalvador@suse.de \
    --cc=pasha.tatashin@soleen.com \
    --cc=peterx@redhat.com \
    --cc=peterz@infradead.org \
    --cc=pfalcato@suse.de \
    --cc=pratyush@kernel.org \
    --cc=qi.zheng@linux.dev \
    --cc=rakie.kim@sk.com \
    --cc=riel@surriel.com \
    --cc=rppt@kernel.org \
    --cc=ryan.roberts@arm.com \
    --cc=shakeel.butt@linux.dev \
    --cc=shikemeng@huaweicloud.com \
    --cc=stable@vger.kernel.org \
    --cc=surenb@google.com \
    --cc=usama.arif@linux.dev \
    --cc=vbabka@kernel.org \
    --cc=weixugc@google.com \
    --cc=will@kernel.org \
    --cc=willy@infradead.org \
    --cc=ying.huang@linux.alibaba.com \
    --cc=youngjun.park@lge.com \
    --cc=yuanchu@google.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