[PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[Lance Yang]

From: Lance Yang <lance.yang@linux.dev>

migrate_folio_move() records the deferred split queue state from src and
replays it on dst. Replaying it after remove_migration_ptes(src, dst, 0)
makes dst visible before it is requeued, so a concurrent rmap-removal path
can mark dst partially mapped and trip the WARN in deferred_split_folio().

Move the requeue before remove_migration_ptes() so dst is back on the
deferred split queue before it becomes visible again.

Because migration still holds dst locked at that point, teach
deferred_split_scan() to requeue a folio when folio_trylock() fails.
Otherwise a fully mapped underused folio can be dequeued by the shrinker
and silently lost from split_queue.

Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
Cc: <stable@vger.kernel.org>
Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
Signed-off-by: Lance Yang <lance.yang@linux.dev>
---

[ Backport note ]
This patch is a follow-up fix for 8a8ca142a488 ("mm: migrate: requeue
destination folio on deferred split queue"), which is currently only in
mm-stable, and should be backported together with it.

Credit for this fix goes to David, thanks!

 mm/huge_memory.c | 12 +++++++-----
 mm/migrate.c     | 18 +++++++++---------
 2 files changed, 16 insertions(+), 14 deletions(-)

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index ff9a42abd1b6..ac6d823e351f 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -4558,7 +4558,7 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
 				goto next;
 		}
 		if (!folio_trylock(folio))
-			goto next;
+			goto requeue;
 		if (!split_folio(folio)) {
 			did_split = true;
 			if (underused)
@@ -4569,11 +4569,13 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
 next:
 		if (did_split || !folio_test_partially_mapped(folio))
 			continue;
+requeue:
 		/*
-		 * Only add back to the queue if folio is partially mapped.
-		 * If thp_underused returns false, or if split_folio fails
-		 * in the case it was underused, then consider it used and
-		 * don't add it back to split_queue.
+		 * Add back partially mapped folios, or underused folios
+		 * that we could not lock this round.  If thp_underused()
+		 * returns false, or if split_folio() succeeds, or if
+		 * split_folio() fails in the case it was underused, then
+		 * consider it used and don't add it back to split_queue.
 		 */
 		fqueue = folio_split_queue_lock_irqsave(folio, &flags);
 		if (list_empty(&folio->_deferred_list)) {
diff --git a/mm/migrate.c b/mm/migrate.c
index 05cb408846f2..8a64291ab5b4 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1385,6 +1385,15 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
 	if (rc)
 		goto out;
 
+	/*
+	 * Requeue the destination folio on the deferred split queue if
+	 * the source was on the queue.  The source is unqueued in
+	 * __folio_migrate_mapping(), so we recorded the state from
+	 * before move_to_new_folio().
+	 */
+	if (src_deferred_split)
+		deferred_split_folio(dst, src_partially_mapped);
+
 	/*
 	 * When successful, push dst to LRU immediately: so that if it
 	 * turns out to be an mlocked page, remove_migration_ptes() will
@@ -1401,15 +1410,6 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
 	if (old_page_state & PAGE_WAS_MAPPED)
 		remove_migration_ptes(src, dst, 0);
 
-	/*
-	 * Requeue the destination folio on the deferred split queue if
-	 * the source was on the queue.  The source is unqueued in
-	 * __folio_migrate_mapping(), so we recorded the state from
-	 * before move_to_new_folio().
-	 */
-	if (src_deferred_split)
-		deferred_split_folio(dst, src_partially_mapped);
-
 out_unlock_both:
 	folio_unlock(dst);
 	folio_set_owner_migrate_reason(dst, reason);
-- 
2.49.0

Re: [PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[Usama Arif]

On Wed,  1 Apr 2026 21:10:32 +0800 Lance Yang <lance.yang@linux.dev> wrote:

> From: Lance Yang <lance.yang@linux.dev>
> 
> migrate_folio_move() records the deferred split queue state from src and
> replays it on dst. Replaying it after remove_migration_ptes(src, dst, 0)
> makes dst visible before it is requeued, so a concurrent rmap-removal path
> can mark dst partially mapped and trip the WARN in deferred_split_folio().
> 
> Move the requeue before remove_migration_ptes() so dst is back on the
> deferred split queue before it becomes visible again.
> 
> Because migration still holds dst locked at that point, teach
> deferred_split_scan() to requeue a folio when folio_trylock() fails.
> Otherwise a fully mapped underused folio can be dequeued by the shrinker
> and silently lost from split_queue.
> 
> Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
> Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
> Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
> Cc: <stable@vger.kernel.org>
> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Lance Yang <lance.yang@linux.dev>
> ---
> 
> [ Backport note ]
> This patch is a follow-up fix for 8a8ca142a488 ("mm: migrate: requeue
> destination folio on deferred split queue"), which is currently only in
> mm-stable, and should be backported together with it.
> 
> Credit for this fix goes to David, thanks!
> 
>  mm/huge_memory.c | 12 +++++++-----
>  mm/migrate.c     | 18 +++++++++---------
>  2 files changed, 16 insertions(+), 14 deletions(-)
> 


Thanks for the fix! And sorry for introducing the bug in
migrate_folio_move() :)

So I am happy with the migrate_folio_move() change, it makes sense.

The goto next if folio is locked in deferred_split_scan() was actually
on purpose. The reasoning was that if the folio is locked, we consider
it as in use by someone and therefore we shouldnt split it. Eventhough
thp_underused() does a zero-filled check, the whole point of the shrinker
was to split THPs that are "not in use", and in my mind, locked folio
is a folio in use. So not sure about that change..


> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index ff9a42abd1b6..ac6d823e351f 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -4558,7 +4558,7 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
>  				goto next;
>  		}
>  		if (!folio_trylock(folio))
> -			goto next;
> +			goto requeue;
>  		if (!split_folio(folio)) {
>  			did_split = true;
>  			if (underused)
> @@ -4569,11 +4569,13 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
>  next:
>  		if (did_split || !folio_test_partially_mapped(folio))
>  			continue;
> +requeue:
>  		/*
> -		 * Only add back to the queue if folio is partially mapped.
> -		 * If thp_underused returns false, or if split_folio fails
> -		 * in the case it was underused, then consider it used and
> -		 * don't add it back to split_queue.
> +		 * Add back partially mapped folios, or underused folios
> +		 * that we could not lock this round.  If thp_underused()
> +		 * returns false, or if split_folio() succeeds, or if
> +		 * split_folio() fails in the case it was underused, then
> +		 * consider it used and don't add it back to split_queue.
>  		 */
>  		fqueue = folio_split_queue_lock_irqsave(folio, &flags);
>  		if (list_empty(&folio->_deferred_list)) {
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 05cb408846f2..8a64291ab5b4 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -1385,6 +1385,15 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
>  	if (rc)
>  		goto out;
>  
> +	/*
> +	 * Requeue the destination folio on the deferred split queue if
> +	 * the source was on the queue.  The source is unqueued in
> +	 * __folio_migrate_mapping(), so we recorded the state from
> +	 * before move_to_new_folio().
> +	 */
> +	if (src_deferred_split)
> +		deferred_split_folio(dst, src_partially_mapped);
> +
>  	/*
>  	 * When successful, push dst to LRU immediately: so that if it
>  	 * turns out to be an mlocked page, remove_migration_ptes() will
> @@ -1401,15 +1410,6 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
>  	if (old_page_state & PAGE_WAS_MAPPED)
>  		remove_migration_ptes(src, dst, 0);
>  
> -	/*
> -	 * Requeue the destination folio on the deferred split queue if
> -	 * the source was on the queue.  The source is unqueued in
> -	 * __folio_migrate_mapping(), so we recorded the state from
> -	 * before move_to_new_folio().
> -	 */
> -	if (src_deferred_split)
> -		deferred_split_folio(dst, src_partially_mapped);
> -
>  out_unlock_both:
>  	folio_unlock(dst);
>  	folio_set_owner_migrate_reason(dst, reason);
> -- 
> 2.49.0
> 
> 

Re: [PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[David Hildenbrand (Arm)]

On 4/1/26 18:28, Usama Arif wrote:
> On Wed,  1 Apr 2026 21:10:32 +0800 Lance Yang <lance.yang@linux.dev> wrote:
> 
>> From: Lance Yang <lance.yang@linux.dev>
>>
>> migrate_folio_move() records the deferred split queue state from src and
>> replays it on dst. Replaying it after remove_migration_ptes(src, dst, 0)
>> makes dst visible before it is requeued, so a concurrent rmap-removal path
>> can mark dst partially mapped and trip the WARN in deferred_split_folio().
>>
>> Move the requeue before remove_migration_ptes() so dst is back on the
>> deferred split queue before it becomes visible again.
>>
>> Because migration still holds dst locked at that point, teach
>> deferred_split_scan() to requeue a folio when folio_trylock() fails.
>> Otherwise a fully mapped underused folio can be dequeued by the shrinker
>> and silently lost from split_queue.
>>
>> Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
>> Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
>> Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
>> Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
>> Cc: <stable@vger.kernel.org>
>> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
>> Signed-off-by: Lance Yang <lance.yang@linux.dev>
>> ---
>>
>> [ Backport note ]
>> This patch is a follow-up fix for 8a8ca142a488 ("mm: migrate: requeue
>> destination folio on deferred split queue"), which is currently only in
>> mm-stable, and should be backported together with it.
>>
>> Credit for this fix goes to David, thanks!
>>
>>  mm/huge_memory.c | 12 +++++++-----
>>  mm/migrate.c     | 18 +++++++++---------
>>  2 files changed, 16 insertions(+), 14 deletions(-)
>>
> 
> 
> Thanks for the fix! And sorry for introducing the bug in
> migrate_folio_move() :)
> 
> So I am happy with the migrate_folio_move() change, it makes sense.
> 
> The goto next if folio is locked in deferred_split_scan() was actually
> on purpose. The reasoning was that if the folio is locked, we consider
> it as in use by someone and therefore we shouldnt split it. Eventhough
> thp_underused() does a zero-filled check, the whole point of the shrinker
> was to split THPs that are "not in use", and in my mind, locked folio
> is a folio in use. So not sure about that change..
That is a questionable assessment. It's about checking whether folios
are *underused* not, if they are used, by whoever in the system (e.g.,
migration).

Just take a look when anonymous folios are actually locked :)

So the original locked handling here is just bogus.

-- 
Cheers,

David

Re: [PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[David Hildenbrand (Arm)]

On 4/1/26 15:10, Lance Yang wrote:
> From: Lance Yang <lance.yang@linux.dev>
> 
> migrate_folio_move() records the deferred split queue state from src and
> replays it on dst. Replaying it after remove_migration_ptes(src, dst, 0)
> makes dst visible before it is requeued, so a concurrent rmap-removal path
> can mark dst partially mapped and trip the WARN in deferred_split_folio().
> 
> Move the requeue before remove_migration_ptes() so dst is back on the
> deferred split queue before it becomes visible again.
> 
> Because migration still holds dst locked at that point, teach
> deferred_split_scan() to requeue a folio when folio_trylock() fails.
> Otherwise a fully mapped underused folio can be dequeued by the shrinker
> and silently lost from split_queue.
> 
> Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
> Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
> Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
> Cc: <stable@vger.kernel.org>
> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Lance Yang <lance.yang@linux.dev>
> ---

LGTM, thanks!

Acked-by: David Hildenbrand (Arm) <david@kernel.org>

-- 
Cheers,

David

Re: [PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[Zi Yan]

On 1 Apr 2026, at 9:10, Lance Yang wrote:

> From: Lance Yang <lance.yang@linux.dev>
>
> migrate_folio_move() records the deferred split queue state from src and
> replays it on dst. Replaying it after remove_migration_ptes(src, dst, 0)
> makes dst visible before it is requeued, so a concurrent rmap-removal path
> can mark dst partially mapped and trip the WARN in deferred_split_folio().
>
> Move the requeue before remove_migration_ptes() so dst is back on the
> deferred split queue before it becomes visible again.
>
> Because migration still holds dst locked at that point, teach
> deferred_split_scan() to requeue a folio when folio_trylock() fails.
> Otherwise a fully mapped underused folio can be dequeued by the shrinker
> and silently lost from split_queue.
>
> Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
> Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
> Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
> Cc: <stable@vger.kernel.org>
> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Lance Yang <lance.yang@linux.dev>
> ---
>
> [ Backport note ]
> This patch is a follow-up fix for 8a8ca142a488 ("mm: migrate: requeue
> destination folio on deferred split queue"), which is currently only in
> mm-stable, and should be backported together with it.
>
> Credit for this fix goes to David, thanks!
>
>  mm/huge_memory.c | 12 +++++++-----
>  mm/migrate.c     | 18 +++++++++---------
>  2 files changed, 16 insertions(+), 14 deletions(-)
>
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index ff9a42abd1b6..ac6d823e351f 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -4558,7 +4558,7 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
>  				goto next;
>  		}
>  		if (!folio_trylock(folio))
> -			goto next;
> +			goto requeue;
>  		if (!split_folio(folio)) {
>  			did_split = true;
>  			if (underused)
> @@ -4569,11 +4569,13 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
>  next:
>  		if (did_split || !folio_test_partially_mapped(folio))
>  			continue;
> +requeue:
>  		/*
> -		 * Only add back to the queue if folio is partially mapped.
> -		 * If thp_underused returns false, or if split_folio fails
> -		 * in the case it was underused, then consider it used and
> -		 * don't add it back to split_queue.
> +		 * Add back partially mapped folios, or underused folios
> +		 * that we could not lock this round.  If thp_underused()
> +		 * returns false, or if split_folio() succeeds, or if
> +		 * split_folio() fails in the case it was underused, then
> +		 * consider it used and don't add it back to split_queue.
>  		 */

Should the sentence
“If thp_underused() returns false, or if split_folio() succeeds, or if
split_folio() fails in the case it was underused, then
consider it used and don't add it back to split_queue.”
be moved to below label next?

Since “thp_underused() returns false” is describing “if (!underused) goto next”,
“split_folio() succeeds” is describing “did_split == true in the if”,
“split_folio() fails in the case it was underused” is describing
“did_split == false and !folio_test_partially_mapped(folio) in the if”.

The first sentence matches the goto requeue for folio_trylock().

Otherwise, LGTM.

Acked-by: Zi Yan <ziy@nvidia.com>

>  		fqueue = folio_split_queue_lock_irqsave(folio, &flags);
>  		if (list_empty(&folio->_deferred_list)) {
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 05cb408846f2..8a64291ab5b4 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -1385,6 +1385,15 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
>  	if (rc)
>  		goto out;
>
> +	/*
> +	 * Requeue the destination folio on the deferred split queue if
> +	 * the source was on the queue.  The source is unqueued in
> +	 * __folio_migrate_mapping(), so we recorded the state from
> +	 * before move_to_new_folio().
> +	 */
> +	if (src_deferred_split)
> +		deferred_split_folio(dst, src_partially_mapped);
> +
>  	/*
>  	 * When successful, push dst to LRU immediately: so that if it
>  	 * turns out to be an mlocked page, remove_migration_ptes() will
> @@ -1401,15 +1410,6 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
>  	if (old_page_state & PAGE_WAS_MAPPED)
>  		remove_migration_ptes(src, dst, 0);
>
> -	/*
> -	 * Requeue the destination folio on the deferred split queue if
> -	 * the source was on the queue.  The source is unqueued in
> -	 * __folio_migrate_mapping(), so we recorded the state from
> -	 * before move_to_new_folio().
> -	 */
> -	if (src_deferred_split)
> -		deferred_split_folio(dst, src_partially_mapped);
> -
>  out_unlock_both:
>  	folio_unlock(dst);
>  	folio_set_owner_migrate_reason(dst, reason);
> -- 
> 2.49.0


Best Regards,
Yan, Zi

Re: [PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[Andrew Morton]

On Wed,  1 Apr 2026 21:10:32 +0800 Lance Yang <lance.yang@linux.dev> wrote:

> From: Lance Yang <lance.yang@linux.dev>
> 
> migrate_folio_move() records the deferred split queue state from src and
> replays it on dst. Replaying it after remove_migration_ptes(src, dst, 0)
> makes dst visible before it is requeued, so a concurrent rmap-removal path
> can mark dst partially mapped and trip the WARN in deferred_split_folio().
> 
> Move the requeue before remove_migration_ptes() so dst is back on the
> deferred split queue before it becomes visible again.
> 
> Because migration still holds dst locked at that point, teach
> deferred_split_scan() to requeue a folio when folio_trylock() fails.
> Otherwise a fully mapped underused folio can be dequeued by the shrinker
> and silently lost from split_queue.

Thanks.

> Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
> Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
> Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
> Cc: <stable@vger.kernel.org>
> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Lance Yang <lance.yang@linux.dev>

I'll add this to mm-unstable with a plan to move it into the current
mm-stable batch in a few days.  So that 8a8ca142a488 and this
follow-up fix stay in the same bundle.

> [ Backport note ]
> This patch is a follow-up fix for 8a8ca142a488 ("mm: migrate: requeue
> destination folio on deferred split queue"), which is currently only in
> mm-stable, and should be backported together with it.

As far as I understand it, this should happen automatically. 
8a8ca142a488 has cc:stable, this patch has Fixes:8a8ca142a488 and
also cc:stable.

There's enough info here for the -stable people to figure it out!

Re: [PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[Zi Yan]

On 1 Apr 2026, at 15:21, Zi Yan wrote:

> On 1 Apr 2026, at 9:10, Lance Yang wrote:
>
>> From: Lance Yang <lance.yang@linux.dev>
>>
>> migrate_folio_move() records the deferred split queue state from src and
>> replays it on dst. Replaying it after remove_migration_ptes(src, dst, 0)
>> makes dst visible before it is requeued, so a concurrent rmap-removal path
>> can mark dst partially mapped and trip the WARN in deferred_split_folio().
>>
>> Move the requeue before remove_migration_ptes() so dst is back on the
>> deferred split queue before it becomes visible again.
>>
>> Because migration still holds dst locked at that point, teach
>> deferred_split_scan() to requeue a folio when folio_trylock() fails.
>> Otherwise a fully mapped underused folio can be dequeued by the shrinker
>> and silently lost from split_queue.
>>
>> Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
>> Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
>> Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
>> Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
>> Cc: <stable@vger.kernel.org>
>> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
>> Signed-off-by: Lance Yang <lance.yang@linux.dev>
>> ---
>>
>> [ Backport note ]
>> This patch is a follow-up fix for 8a8ca142a488 ("mm: migrate: requeue
>> destination folio on deferred split queue"), which is currently only in
>> mm-stable, and should be backported together with it.
>>
>> Credit for this fix goes to David, thanks!
>>
>>  mm/huge_memory.c | 12 +++++++-----
>>  mm/migrate.c     | 18 +++++++++---------
>>  2 files changed, 16 insertions(+), 14 deletions(-)
>>
>> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
>> index ff9a42abd1b6..ac6d823e351f 100644
>> --- a/mm/huge_memory.c
>> +++ b/mm/huge_memory.c
>> @@ -4558,7 +4558,7 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
>>  				goto next;
>>  		}
>>  		if (!folio_trylock(folio))
>> -			goto next;
>> +			goto requeue;
>>  		if (!split_folio(folio)) {
>>  			did_split = true;
>>  			if (underused)
>> @@ -4569,11 +4569,13 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
>>  next:
>>  		if (did_split || !folio_test_partially_mapped(folio))
>>  			continue;
>> +requeue:
>>  		/*
>> -		 * Only add back to the queue if folio is partially mapped.
>> -		 * If thp_underused returns false, or if split_folio fails
>> -		 * in the case it was underused, then consider it used and
>> -		 * don't add it back to split_queue.
>> +		 * Add back partially mapped folios, or underused folios
>> +		 * that we could not lock this round.  If thp_underused()
>> +		 * returns false, or if split_folio() succeeds, or if
>> +		 * split_folio() fails in the case it was underused, then
>> +		 * consider it used and don't add it back to split_queue.
>>  		 */
>
> Should the sentence
> “If thp_underused() returns false, or if split_folio() succeeds, or if
> split_folio() fails in the case it was underused, then
> consider it used and don't add it back to split_queue.”
> be moved to below label next?
>
> Since “thp_underused() returns false” is describing “if (!underused) goto next”,
> “split_folio() succeeds” is describing “did_split == true in the if”,
> “split_folio() fails in the case it was underused” is describing
> “did_split == false and !folio_test_partially_mapped(folio) in the if”.
>
> The first sentence matches the goto requeue for folio_trylock().

Hi Andrew,

Can you apply the fixup below to move the comment? Lance told me he
would be away for a while, so he could not send a fixup to move
the comment.

Thanks.


From 6ebeca9f7215cb91905d3f49385dbbafce5a80c2 Mon Sep 17 00:00:00 2001
From: Zi Yan <ziy@nvidia.com>
Date: Wed, 1 Apr 2026 18:52:43 -0400
Subject: [PATCH] move the comment.

Signed-off-by: Zi Yan <ziy@nvidia.com>
---
 mm/huge_memory.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index ac6d823e351ff..970e077019b75 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -4567,15 +4567,18 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
 		}
 		folio_unlock(folio);
 next:
+		/*
+		 * If thp_underused() returns false, or if split_folio()
+		 * succeeds, or if split_folio() fails in the case it was
+		 * underused, then consider it used and don't add it back to
+		 * split_queue.
+		 */
 		if (did_split || !folio_test_partially_mapped(folio))
 			continue;
 requeue:
 		/*
-		 * Add back partially mapped folios, or underused folios
-		 * that we could not lock this round.  If thp_underused()
-		 * returns false, or if split_folio() succeeds, or if
-		 * split_folio() fails in the case it was underused, then
-		 * consider it used and don't add it back to split_queue.
+		 * Add back partially mapped folios, or underused folios that
+		 * we could not lock this round.
 		 */
 		fqueue = folio_split_queue_lock_irqsave(folio, &flags);
 		if (list_empty(&folio->_deferred_list)) {
-- 
2.53.0



>
> Otherwise, LGTM.
>
> Acked-by: Zi Yan <ziy@nvidia.com>
>
>>  		fqueue = folio_split_queue_lock_irqsave(folio, &flags);
>>  		if (list_empty(&folio->_deferred_list)) {
>> diff --git a/mm/migrate.c b/mm/migrate.c
>> index 05cb408846f2..8a64291ab5b4 100644
>> --- a/mm/migrate.c
>> +++ b/mm/migrate.c
>> @@ -1385,6 +1385,15 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
>>  	if (rc)
>>  		goto out;
>>
>> +	/*
>> +	 * Requeue the destination folio on the deferred split queue if
>> +	 * the source was on the queue.  The source is unqueued in
>> +	 * __folio_migrate_mapping(), so we recorded the state from
>> +	 * before move_to_new_folio().
>> +	 */
>> +	if (src_deferred_split)
>> +		deferred_split_folio(dst, src_partially_mapped);
>> +
>>  	/*
>>  	 * When successful, push dst to LRU immediately: so that if it
>>  	 * turns out to be an mlocked page, remove_migration_ptes() will
>> @@ -1401,15 +1410,6 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
>>  	if (old_page_state & PAGE_WAS_MAPPED)
>>  		remove_migration_ptes(src, dst, 0);
>>
>> -	/*
>> -	 * Requeue the destination folio on the deferred split queue if
>> -	 * the source was on the queue.  The source is unqueued in
>> -	 * __folio_migrate_mapping(), so we recorded the state from
>> -	 * before move_to_new_folio().
>> -	 */
>> -	if (src_deferred_split)
>> -		deferred_split_folio(dst, src_partially_mapped);
>> -
>>  out_unlock_both:
>>  	folio_unlock(dst);
>>  	folio_set_owner_migrate_reason(dst, reason);
>> -- 
>> 2.49.0
>
>
> Best Regards,
> Yan, Zi


Best Regards,
Yan, Zi

Re: [PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[Andrew Morton]

On Wed, 01 Apr 2026 18:55:48 -0400 Zi Yan <ziy@nvidia.com> wrote:

> Can you apply the fixup below to move the comment? Lance told me he
> would be away for a while, so he could not send a fixup to move
> the comment.

Thanks.  I folded that into Lance's base patch so here's the whole
thing:



From: Lance Yang <lance.yang@linux.dev>
Subject: mm: fix deferred split queue races during migration
Date: Wed, 1 Apr 2026 21:10:32 +0800

migrate_folio_move() records the deferred split queue state from src and
replays it on dst.  Replaying it after remove_migration_ptes(src, dst, 0)
makes dst visible before it is requeued, so a concurrent rmap-removal path
can mark dst partially mapped and trip the WARN in deferred_split_folio().

Move the requeue before remove_migration_ptes() so dst is back on the
deferred split queue before it becomes visible again.

Because migration still holds dst locked at that point, teach
deferred_split_scan() to requeue a folio when folio_trylock() fails. 
Otherwise a fully mapped underused folio can be dequeued by the shrinker
and silently lost from split_queue.

[ziy@nvidia.com: move the comment]
  Link: https://lkml.kernel.org/r/FB71A764-0F10-4E5A-B4A0-BA4C7F138408@nvidia.com
Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
Link: https://lkml.kernel.org/r/20260401131032.13011-1-lance.yang@linux.dev
Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
Signed-off-by: Lance Yang <lance.yang@linux.dev>
Signed-off-by: Zi Yan <ziy@nvidia.com>
Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
Acked-by: David Hildenbrand (Arm) <david@kernel.org>
Acked-by: Zi Yan <ziy@nvidia.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Barry Song <baohua@kernel.org>
Cc: Byungchul Park <byungchul@sk.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Deepanshu Kartikey <kartikey406@gmail.com>
Cc: Dev Jain <dev.jain@arm.com>
Cc: Gregory Price <gourry@gourry.net>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Joshua Hahn <joshua.hahnjy@gmail.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Nico Pache <npache@redhat.com>
Cc: Rakie Kim <rakie.kim@sk.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Ying Huang <ying.huang@linux.alibaba.com>
Cc: Usama Arif <usama.arif@linux.dev>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 mm/huge_memory.c |   15 ++++++++++-----
 mm/migrate.c     |   18 +++++++++---------
 2 files changed, 19 insertions(+), 14 deletions(-)

--- a/mm/huge_memory.c~mm-fix-deferred-split-queue-races-during-migration
+++ a/mm/huge_memory.c
@@ -4542,7 +4542,7 @@ retry:
 				goto next;
 		}
 		if (!folio_trylock(folio))
-			goto next;
+			goto requeue;
 		if (!split_folio(folio)) {
 			did_split = true;
 			if (underused)
@@ -4551,13 +4551,18 @@ retry:
 		}
 		folio_unlock(folio);
 next:
+		/*
+		 * If thp_underused() returns false, or if split_folio()
+		 * succeeds, or if split_folio() fails in the case it was
+		 * underused, then consider it used and don't add it back to
+		 * split_queue.
+		 */
 		if (did_split || !folio_test_partially_mapped(folio))
 			continue;
+requeue:
 		/*
-		 * Only add back to the queue if folio is partially mapped.
-		 * If thp_underused returns false, or if split_folio fails
-		 * in the case it was underused, then consider it used and
-		 * don't add it back to split_queue.
+		 * Add back partially mapped folios, or underused folios that
+		 * we could not lock this round.
 		 */
 		fqueue = folio_split_queue_lock_irqsave(folio, &flags);
 		if (list_empty(&folio->_deferred_list)) {
--- a/mm/migrate.c~mm-fix-deferred-split-queue-races-during-migration
+++ a/mm/migrate.c
@@ -1384,6 +1384,15 @@ static int migrate_folio_move(free_folio
 		goto out;
 
 	/*
+	 * Requeue the destination folio on the deferred split queue if
+	 * the source was on the queue.  The source is unqueued in
+	 * __folio_migrate_mapping(), so we recorded the state from
+	 * before move_to_new_folio().
+	 */
+	if (src_deferred_split)
+		deferred_split_folio(dst, src_partially_mapped);
+
+	/*
 	 * When successful, push dst to LRU immediately: so that if it
 	 * turns out to be an mlocked page, remove_migration_ptes() will
 	 * automatically build up the correct dst->mlock_count for it.
@@ -1399,15 +1408,6 @@ static int migrate_folio_move(free_folio
 	if (old_page_state & PAGE_WAS_MAPPED)
 		remove_migration_ptes(src, dst, 0);
 
-	/*
-	 * Requeue the destination folio on the deferred split queue if
-	 * the source was on the queue.  The source is unqueued in
-	 * __folio_migrate_mapping(), so we recorded the state from
-	 * before move_to_new_folio().
-	 */
-	if (src_deferred_split)
-		deferred_split_folio(dst, src_partially_mapped);
-
 out_unlock_both:
 	folio_unlock(dst);
 	folio_set_owner_migrate_reason(dst, reason);
_

Re: [PATCH mm-unstable 1/1] mm: fix deferred split queue races during migration

[Lance Yang]

On 2026/4/2 07:19, Andrew Morton wrote:
> On Wed, 01 Apr 2026 18:55:48 -0400 Zi Yan <ziy@nvidia.com> wrote:
> 
>> Can you apply the fixup below to move the comment? Lance told me he
>> would be away for a while, so he could not send a fixup to move
>> the comment.
> 
> Thanks.  I folded that into Lance's base patch so here's the whole
> thing:
> 

Thank you all!
Lance

> 
> From: Lance Yang <lance.yang@linux.dev>
> Subject: mm: fix deferred split queue races during migration
> Date: Wed, 1 Apr 2026 21:10:32 +0800
> 
> migrate_folio_move() records the deferred split queue state from src and
> replays it on dst.  Replaying it after remove_migration_ptes(src, dst, 0)
> makes dst visible before it is requeued, so a concurrent rmap-removal path
> can mark dst partially mapped and trip the WARN in deferred_split_folio().
> 
> Move the requeue before remove_migration_ptes() so dst is back on the
> deferred split queue before it becomes visible again.
> 
> Because migration still holds dst locked at that point, teach
> deferred_split_scan() to requeue a folio when folio_trylock() fails.
> Otherwise a fully mapped underused folio can be dequeued by the shrinker
> and silently lost from split_queue.
> 
> [ziy@nvidia.com: move the comment]
>    Link: https://lkml.kernel.org/r/FB71A764-0F10-4E5A-B4A0-BA4C7F138408@nvidia.com
> Link: https://syzkaller.appspot.com/bug?extid=a7067a757858ac8eb085
> Link: https://lkml.kernel.org/r/20260401131032.13011-1-lance.yang@linux.dev
> Fixes: 8a8ca142a488 ("mm: migrate: requeue destination folio on deferred split queue")
> Signed-off-by: Lance Yang <lance.yang@linux.dev>
> Signed-off-by: Zi Yan <ziy@nvidia.com>
> Reported-by: syzbot+a7067a757858ac8eb085@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/linux-mm/69ccb65b.050a0220.183828.003a.GAE@google.com/
> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
> Acked-by: David Hildenbrand (Arm) <david@kernel.org>
> Acked-by: Zi Yan <ziy@nvidia.com>
> Cc: Alistair Popple <apopple@nvidia.com>
> Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
> Cc: Barry Song <baohua@kernel.org>
> Cc: Byungchul Park <byungchul@sk.com>
> Cc: David Hildenbrand <david@kernel.org>
> Cc: Deepanshu Kartikey <kartikey406@gmail.com>
> Cc: Dev Jain <dev.jain@arm.com>
> Cc: Gregory Price <gourry@gourry.net>
> Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
> Cc: Joshua Hahn <joshua.hahnjy@gmail.com>
> Cc: Lance Yang <lance.yang@linux.dev>
> Cc: Liam Howlett <liam.howlett@oracle.com>
> Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
> Cc: Matthew Brost <matthew.brost@intel.com>
> Cc: Nico Pache <npache@redhat.com>
> Cc: Rakie Kim <rakie.kim@sk.com>
> Cc: Ryan Roberts <ryan.roberts@arm.com>
> Cc: Wei Yang <richard.weiyang@gmail.com>
> Cc: Ying Huang <ying.huang@linux.alibaba.com>
> Cc: Usama Arif <usama.arif@linux.dev>
> Cc: <stable@vger.kernel.org>
> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
> ---
> 
>   mm/huge_memory.c |   15 ++++++++++-----
>   mm/migrate.c     |   18 +++++++++---------
>   2 files changed, 19 insertions(+), 14 deletions(-)
> 
> --- a/mm/huge_memory.c~mm-fix-deferred-split-queue-races-during-migration
> +++ a/mm/huge_memory.c
> @@ -4542,7 +4542,7 @@ retry:
>   				goto next;
>   		}
>   		if (!folio_trylock(folio))
> -			goto next;
> +			goto requeue;
>   		if (!split_folio(folio)) {
>   			did_split = true;
>   			if (underused)
> @@ -4551,13 +4551,18 @@ retry:
>   		}
>   		folio_unlock(folio);
>   next:
> +		/*
> +		 * If thp_underused() returns false, or if split_folio()
> +		 * succeeds, or if split_folio() fails in the case it was
> +		 * underused, then consider it used and don't add it back to
> +		 * split_queue.
> +		 */
>   		if (did_split || !folio_test_partially_mapped(folio))
>   			continue;
> +requeue:
>   		/*
> -		 * Only add back to the queue if folio is partially mapped.
> -		 * If thp_underused returns false, or if split_folio fails
> -		 * in the case it was underused, then consider it used and
> -		 * don't add it back to split_queue.
> +		 * Add back partially mapped folios, or underused folios that
> +		 * we could not lock this round.
>   		 */
>   		fqueue = folio_split_queue_lock_irqsave(folio, &flags);
>   		if (list_empty(&folio->_deferred_list)) {
> --- a/mm/migrate.c~mm-fix-deferred-split-queue-races-during-migration
> +++ a/mm/migrate.c
> @@ -1384,6 +1384,15 @@ static int migrate_folio_move(free_folio
>   		goto out;
>   
>   	/*
> +	 * Requeue the destination folio on the deferred split queue if
> +	 * the source was on the queue.  The source is unqueued in
> +	 * __folio_migrate_mapping(), so we recorded the state from
> +	 * before move_to_new_folio().
> +	 */
> +	if (src_deferred_split)
> +		deferred_split_folio(dst, src_partially_mapped);
> +
> +	/*
>   	 * When successful, push dst to LRU immediately: so that if it
>   	 * turns out to be an mlocked page, remove_migration_ptes() will
>   	 * automatically build up the correct dst->mlock_count for it.
> @@ -1399,15 +1408,6 @@ static int migrate_folio_move(free_folio
>   	if (old_page_state & PAGE_WAS_MAPPED)
>   		remove_migration_ptes(src, dst, 0);
>   
> -	/*
> -	 * Requeue the destination folio on the deferred split queue if
> -	 * the source was on the queue.  The source is unqueued in
> -	 * __folio_migrate_mapping(), so we recorded the state from
> -	 * before move_to_new_folio().
> -	 */
> -	if (src_deferred_split)
> -		deferred_split_folio(dst, src_partially_mapped);
> -
>   out_unlock_both:
>   	folio_unlock(dst);
>   	folio_set_owner_migrate_reason(dst, reason);
> _
>