[PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[Axel Rasmussen]

This change swapped out mod_node_page_state for lruvec_stat_add_folio.
But, these two APIs are not interchangeable: the lruvec version also
increments memcg stats, in addition to "global" pgdat stats.

So after this change, the "pagetables" memcg stat in memory.stat always
yields "0", which is a userspace visible regression.

I tried to look for a refactor where we add a variant of
lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
folio, to try to adhere to the spirit of the original patch. But at the
end of the day this just means we have to call
folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
accomplish much.

This regression is visible in master as well as 6.18 stable, so CC
stable too.

Fixes: f0c92726e89f ("ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()")
Cc: stable@vger.kernel.org
Signed-off-by: Axel Rasmussen <axelrasmussen@google.com>
---
 include/linux/mm.h | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)

diff --git a/include/linux/mm.h b/include/linux/mm.h
index 5be3d8a8f806..abb4963c1f06 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3514,26 +3514,21 @@ static inline bool ptlock_init(struct ptdesc *ptdesc) { return true; }
 static inline void ptlock_free(struct ptdesc *ptdesc) {}
 #endif /* defined(CONFIG_SPLIT_PTE_PTLOCKS) */
 
-static inline unsigned long ptdesc_nr_pages(const struct ptdesc *ptdesc)
-{
-	return compound_nr(ptdesc_page(ptdesc));
-}
-
 static inline void __pagetable_ctor(struct ptdesc *ptdesc)
 {
-	pg_data_t *pgdat = NODE_DATA(memdesc_nid(ptdesc->pt_flags));
+	struct folio *folio = ptdesc_folio(ptdesc);
 
-	__SetPageTable(ptdesc_page(ptdesc));
-	mod_node_page_state(pgdat, NR_PAGETABLE, ptdesc_nr_pages(ptdesc));
+	__folio_set_pgtable(folio);
+	lruvec_stat_add_folio(folio, NR_PAGETABLE);
 }
 
 static inline void pagetable_dtor(struct ptdesc *ptdesc)
 {
-	pg_data_t *pgdat = NODE_DATA(memdesc_nid(ptdesc->pt_flags));
+	struct folio *folio = ptdesc_folio(ptdesc);
 
 	ptlock_free(ptdesc);
-	__ClearPageTable(ptdesc_page(ptdesc));
-	mod_node_page_state(pgdat, NR_PAGETABLE, -ptdesc_nr_pages(ptdesc));
+	__folio_clear_pgtable(folio);
+	lruvec_stat_sub_folio(folio, NR_PAGETABLE);
 }
 
 static inline void pagetable_dtor_free(struct ptdesc *ptdesc)
-- 
2.53.0.414.gf7e9f6c205-goog

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[Matthew Wilcox]

On Tue, Feb 24, 2026 at 04:24:34PM -0800, Axel Rasmussen wrote:
> This change swapped out mod_node_page_state for lruvec_stat_add_folio.
> But, these two APIs are not interchangeable: the lruvec version also
> increments memcg stats, in addition to "global" pgdat stats.
> 
> So after this change, the "pagetables" memcg stat in memory.stat always
> yields "0", which is a userspace visible regression.
> 
> I tried to look for a refactor where we add a variant of
> lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
> folio, to try to adhere to the spirit of the original patch. But at the
> end of the day this just means we have to call
> folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
> accomplish much.

Thank you!  I hadn't been able to get a straight answer on this before.

You're right that there's no good function to call, but that just means
we need to make one.  The principle here is that (eventually) different
memdescs don't need to know about each other.  Obviously we're not there
yet, but we can start disentangling them by not casting ptdescs back to
folios (even though they're created that way).

Here's three patches smooshed together; I have them separately and I'll
post them soon.

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[Matthew Wilcox]

On Wed, Feb 25, 2026 at 04:03:54PM +0000, Matthew Wilcox wrote:
> On Tue, Feb 24, 2026 at 04:24:34PM -0800, Axel Rasmussen wrote:
> > This change swapped out mod_node_page_state for lruvec_stat_add_folio.
> > But, these two APIs are not interchangeable: the lruvec version also
> > increments memcg stats, in addition to "global" pgdat stats.
> > 
> > So after this change, the "pagetables" memcg stat in memory.stat always
> > yields "0", which is a userspace visible regression.
> > 
> > I tried to look for a refactor where we add a variant of
> > lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
> > folio, to try to adhere to the spirit of the original patch. But at the
> > end of the day this just means we have to call
> > folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
> > accomplish much.
> 
> Thank you!  I hadn't been able to get a straight answer on this before.
> 
> You're right that there's no good function to call, but that just means
> we need to make one.  The principle here is that (eventually) different
> memdescs don't need to know about each other.  Obviously we're not there
> yet, but we can start disentangling them by not casting ptdescs back to
> folios (even though they're created that way).
> 
> Here's three patches smooshed together; I have them separately and I'll
> post them soon.

Argh, fatfingered the inclusion and ended up sending ...

diff --git a/include/linux/mm.h b/include/linux/mm.h
index 5be3d8a8f806..34bc6f00ed7b 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3519,21 +3519,32 @@ static inline unsigned long ptdesc_nr_pages(const struct ptdesc *ptdesc)
 	return compound_nr(ptdesc_page(ptdesc));
 }
 
+static inline struct mem_cgroup *pagetable_memcg(const struct ptdesc *ptdesc)
+{
+#ifdef CONFIG_MEMCG
+	return ptdesc->pt_memcg;
+#else
+	return NULL;
+#endif
+}
+
 static inline void __pagetable_ctor(struct ptdesc *ptdesc)
 {
 	pg_data_t *pgdat = NODE_DATA(memdesc_nid(ptdesc->pt_flags));
+	struct mem_cgroup *memcg = pagetable_memcg(ptdesc);
 
 	__SetPageTable(ptdesc_page(ptdesc));
-	mod_node_page_state(pgdat, NR_PAGETABLE, ptdesc_nr_pages(ptdesc));
+	memcg_stat_mod(memcg, pgdat, NR_PAGETABLE, ptdesc_nr_pages(ptdesc));
 }
 
 static inline void pagetable_dtor(struct ptdesc *ptdesc)
 {
 	pg_data_t *pgdat = NODE_DATA(memdesc_nid(ptdesc->pt_flags));
+	struct mem_cgroup *memcg = pagetable_memcg(ptdesc);
 
 	ptlock_free(ptdesc);
 	__ClearPageTable(ptdesc_page(ptdesc));
-	mod_node_page_state(pgdat, NR_PAGETABLE, -ptdesc_nr_pages(ptdesc));
+	memcg_stat_mod(memcg, pgdat, NR_PAGETABLE, -ptdesc_nr_pages(ptdesc));
 }
 
 static inline void pagetable_dtor_free(struct ptdesc *ptdesc)
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index 3cc8ae722886..e9b1da04938a 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -564,7 +564,7 @@ FOLIO_MATCH(compound_head, _head_3);
  * @ptl:              Lock for the page table.
  * @__page_type:      Same as page->page_type. Unused for page tables.
  * @__page_refcount:  Same as page refcount.
- * @pt_memcg_data:    Memcg data. Tracked for page tables here.
+ * @pt_memcg:         Memcg that this page table belongs to.
  *
  * This struct overlays struct page for now. Do not modify without a good
  * understanding of the issues.
@@ -602,7 +602,7 @@ struct ptdesc {
 	unsigned int __page_type;
 	atomic_t __page_refcount;
 #ifdef CONFIG_MEMCG
-	unsigned long pt_memcg_data;
+	struct mem_cgroup *pt_memcg;
 #endif
 };
 
@@ -617,7 +617,7 @@ TABLE_MATCH(rcu_head, pt_rcu_head);
 TABLE_MATCH(page_type, __page_type);
 TABLE_MATCH(_refcount, __page_refcount);
 #ifdef CONFIG_MEMCG
-TABLE_MATCH(memcg_data, pt_memcg_data);
+TABLE_MATCH(memcg_data, pt_memcg);
 #endif
 #undef TABLE_MATCH
 static_assert(sizeof(struct ptdesc) <= sizeof(struct page));
diff --git a/include/linux/vmstat.h b/include/linux/vmstat.h
index 3c9c266cf782..0da38ea25c97 100644
--- a/include/linux/vmstat.h
+++ b/include/linux/vmstat.h
@@ -518,7 +518,8 @@ static inline const char *vm_event_name(enum vm_event_item item)
 
 void mod_lruvec_state(struct lruvec *lruvec, enum node_stat_item idx,
 			int val);
-
+void memcg_stat_mod(struct mem_cgroup *memcg, pg_data_t *pgdat,
+		enum node_stat_item idx, long val);
 void lruvec_stat_mod_folio(struct folio *folio,
 			     enum node_stat_item idx, int val);
 
@@ -536,6 +537,12 @@ static inline void mod_lruvec_state(struct lruvec *lruvec,
 	mod_node_page_state(lruvec_pgdat(lruvec), idx, val);
 }
 
+static inline void memcg_stat_mod(struct mem_cgroup *memcg, pg_data_t *pgdat,
+		enum node_stat_item idx, long val)
+{
+	mod_node_page_state(pgdat, idx, val);
+}
+
 static inline void lruvec_stat_mod_folio(struct folio *folio,
 					 enum node_stat_item idx, int val)
 {
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index a52da3a5e4fd..8d9e4a42aecf 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -787,24 +787,27 @@ void mod_lruvec_state(struct lruvec *lruvec, enum node_stat_item idx,
 		mod_memcg_lruvec_state(lruvec, idx, val);
 }
 
+void memcg_stat_mod(struct mem_cgroup *memcg, pg_data_t *pgdat,
+		enum node_stat_item idx, long val)
+{
+	/* Untracked pages have no memcg, no lruvec. Update only the node */
+	if (!memcg) {
+		mod_node_page_state(pgdat, idx, val);
+	} else {
+		struct lruvec *lruvec = mem_cgroup_lruvec(memcg, pgdat);
+		mod_lruvec_state(lruvec, idx, val);
+	}
+}
+
 void lruvec_stat_mod_folio(struct folio *folio, enum node_stat_item idx,
 			     int val)
 {
 	struct mem_cgroup *memcg;
 	pg_data_t *pgdat = folio_pgdat(folio);
-	struct lruvec *lruvec;
 
 	rcu_read_lock();
 	memcg = folio_memcg(folio);
-	/* Untracked pages have no memcg, no lruvec. Update only the node */
-	if (!memcg) {
-		rcu_read_unlock();
-		mod_node_page_state(pgdat, idx, val);
-		return;
-	}
-
-	lruvec = mem_cgroup_lruvec(memcg, pgdat);
-	mod_lruvec_state(lruvec, idx, val);
+	memcg_stat_mod(memcg, pgdat, idx, val);
 	rcu_read_unlock();
 }
 EXPORT_SYMBOL(lruvec_stat_mod_folio);
@@ -812,24 +815,9 @@ EXPORT_SYMBOL(lruvec_stat_mod_folio);
 void mod_lruvec_kmem_state(void *p, enum node_stat_item idx, int val)
 {
 	pg_data_t *pgdat = page_pgdat(virt_to_page(p));
-	struct mem_cgroup *memcg;
-	struct lruvec *lruvec;
 
 	rcu_read_lock();
-	memcg = mem_cgroup_from_virt(p);
-
-	/*
-	 * Untracked pages have no memcg, no lruvec. Update only the
-	 * node. If we reparent the slab objects to the root memcg,
-	 * when we free the slab object, we need to update the per-memcg
-	 * vmstats to keep it correct for the root memcg.
-	 */
-	if (!memcg) {
-		mod_node_page_state(pgdat, idx, val);
-	} else {
-		lruvec = mem_cgroup_lruvec(memcg, pgdat);
-		mod_lruvec_state(lruvec, idx, val);
-	}
+	memcg_stat_mod(mem_cgroup_from_virt(p), pgdat, idx, val);
 	rcu_read_unlock();
 }
 

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[David Hildenbrand (Arm)]

On 2/25/26 17:03, Matthew Wilcox wrote:
> On Tue, Feb 24, 2026 at 04:24:34PM -0800, Axel Rasmussen wrote:
>> This change swapped out mod_node_page_state for lruvec_stat_add_folio.
>> But, these two APIs are not interchangeable: the lruvec version also
>> increments memcg stats, in addition to "global" pgdat stats.
>>
>> So after this change, the "pagetables" memcg stat in memory.stat always
>> yields "0", which is a userspace visible regression.
>>
>> I tried to look for a refactor where we add a variant of
>> lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
>> folio, to try to adhere to the spirit of the original patch. But at the
>> end of the day this just means we have to call
>> folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
>> accomplish much.
> 
> Thank you!  I hadn't been able to get a straight answer on this before.
> 
> You're right that there's no good function to call, but that just means
> we need to make one.  The principle here is that (eventually) different
> memdescs don't need to know about each other.  Obviously we're not there
> yet, but we can start disentangling them by not casting ptdescs back to
> folios (even though they're created that way).
> 
> Here's three patches smooshed together; I have them separately and I'll
> post them soon.

Should we just apply + backport the revert for now and re-do it based on
the revert?

-- 
Cheers,

David

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[Shakeel Butt]

+memcg maintainers

On Wed, Feb 25, 2026 at 05:08:28PM +0100, David Hildenbrand (Arm) wrote:
> On 2/25/26 17:03, Matthew Wilcox wrote:
> > On Tue, Feb 24, 2026 at 04:24:34PM -0800, Axel Rasmussen wrote:
> >> This change swapped out mod_node_page_state for lruvec_stat_add_folio.
> >> But, these two APIs are not interchangeable: the lruvec version also
> >> increments memcg stats, in addition to "global" pgdat stats.
> >>
> >> So after this change, the "pagetables" memcg stat in memory.stat always
> >> yields "0", which is a userspace visible regression.
> >>
> >> I tried to look for a refactor where we add a variant of
> >> lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
> >> folio, to try to adhere to the spirit of the original patch. But at the
> >> end of the day this just means we have to call
> >> folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
> >> accomplish much.
> > 
> > Thank you!  I hadn't been able to get a straight answer on this before.
> > 
> > You're right that there's no good function to call, but that just means
> > we need to make one.  The principle here is that (eventually) different
> > memdescs don't need to know about each other.  Obviously we're not there
> > yet, but we can start disentangling them by not casting ptdescs back to
> > folios (even though they're created that way).
> > 
> > Here's three patches smooshed together; I have them separately and I'll
> > post them soon.
> 
> Should we just apply + backport the revert for now and re-do it based on
> the revert?

Yes please.

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[David Hildenbrand (Arm)]

On 2/25/26 01:24, Axel Rasmussen wrote:
> This change swapped out mod_node_page_state for lruvec_stat_add_folio.
> But, these two APIs are not interchangeable: the lruvec version also
> increments memcg stats, in addition to "global" pgdat stats.
> 
> So after this change, the "pagetables" memcg stat in memory.stat always
> yields "0", which is a userspace visible regression.

Argh, missed that these page tables are charged.

> 
> I tried to look for a refactor where we add a variant of
> lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
> folio, to try to adhere to the spirit of the original patch. But at the
> end of the day this just means we have to call
> folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
> accomplish much.

The goal of the original patch is obviously to get rid of any folio
references in ptdesc code.

But if these ptdesc are charged, then we'd also need long-term a
ptdesc->memcg_data.

For now we could still go through page->memcg_data, to avoid going
through the folio.

But that requires some work, so likely we should just directly enlighten
ptdesc to have memcg_data.

Agreed that, as a quick fix, reverting the commit might be best.

But let's wait a bit whether Willy has a comment.

> 
> This regression is visible in master as well as 6.18 stable, so CC
> stable too.
> 
> Fixes: f0c92726e89f ("ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()")
> Cc: stable@vger.kernel.org
> Signed-off-by: Axel Rasmussen <axelrasmussen@google.com>
> ---
>  include/linux/mm.h | 17 ++++++-----------
>  1 file changed, 6 insertions(+), 11 deletions(-)
> 
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 5be3d8a8f806..abb4963c1f06 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -3514,26 +3514,21 @@ static inline bool ptlock_init(struct ptdesc *ptdesc) { return true; }
>  static inline void ptlock_free(struct ptdesc *ptdesc) {}
>  #endif /* defined(CONFIG_SPLIT_PTE_PTLOCKS) */
>  
> -static inline unsigned long ptdesc_nr_pages(const struct ptdesc *ptdesc)
> -{
> -	return compound_nr(ptdesc_page(ptdesc));
> -}
> -
>  static inline void __pagetable_ctor(struct ptdesc *ptdesc)
>  {
> -	pg_data_t *pgdat = NODE_DATA(memdesc_nid(ptdesc->pt_flags));
> +	struct folio *folio = ptdesc_folio(ptdesc);
>  
> -	__SetPageTable(ptdesc_page(ptdesc));
> -	mod_node_page_state(pgdat, NR_PAGETABLE, ptdesc_nr_pages(ptdesc));
> +	__folio_set_pgtable(folio);
> +	lruvec_stat_add_folio(folio, NR_PAGETABLE);
>  }
>  
>  static inline void pagetable_dtor(struct ptdesc *ptdesc)
>  {
> -	pg_data_t *pgdat = NODE_DATA(memdesc_nid(ptdesc->pt_flags));
> +	struct folio *folio = ptdesc_folio(ptdesc);
>  
>  	ptlock_free(ptdesc);
> -	__ClearPageTable(ptdesc_page(ptdesc));
> -	mod_node_page_state(pgdat, NR_PAGETABLE, -ptdesc_nr_pages(ptdesc));
> +	__folio_clear_pgtable(folio);
> +	lruvec_stat_sub_folio(folio, NR_PAGETABLE);
>  }
>  
>  static inline void pagetable_dtor_free(struct ptdesc *ptdesc)


-- 
Cheers,

David

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[David Hildenbrand (Arm)]

On 2/25/26 17:04, David Hildenbrand (Arm) wrote:
> On 2/25/26 01:24, Axel Rasmussen wrote:
>> This change swapped out mod_node_page_state for lruvec_stat_add_folio.
>> But, these two APIs are not interchangeable: the lruvec version also
>> increments memcg stats, in addition to "global" pgdat stats.
>>
>> So after this change, the "pagetables" memcg stat in memory.stat always
>> yields "0", which is a userspace visible regression.
> 
> Argh, missed that these page tables are charged.
> 
>>
>> I tried to look for a refactor where we add a variant of
>> lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
>> folio, to try to adhere to the spirit of the original patch. But at the
>> end of the day this just means we have to call
>> folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
>> accomplish much.
> 
> The goal of the original patch is obviously to get rid of any folio
> references in ptdesc code.
> 
> But if these ptdesc are charged, then we'd also need long-term a
> ptdesc->memcg_data.
> 
> For now we could still go through page->memcg_data, to avoid going
> through the folio.
> 
> But that requires some work, so likely we should just directly enlighten
> ptdesc to have memcg_data.
> 
> Agreed that, as a quick fix, reverting the commit might be best.
> 
> But let's wait a bit whether Willy has a comment.

Heh, Willy just replied while composing the mail. :)

-- 
Cheers,

David

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[Shakeel Butt]

On Tue, Feb 24, 2026 at 04:24:34PM -0800, Axel Rasmussen wrote:
> This change swapped out mod_node_page_state for lruvec_stat_add_folio.
> But, these two APIs are not interchangeable: the lruvec version also
> increments memcg stats, in addition to "global" pgdat stats.
> 
> So after this change, the "pagetables" memcg stat in memory.stat always
> yields "0", which is a userspace visible regression.
> 
> I tried to look for a refactor where we add a variant of
> lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
> folio, to try to adhere to the spirit of the original patch. But at the
> end of the day this just means we have to call
> folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
> accomplish much.
> 
> This regression is visible in master as well as 6.18 stable, so CC
> stable too.
> 
> Fixes: f0c92726e89f ("ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()")
> Cc: stable@vger.kernel.org
> Signed-off-by: Axel Rasmussen <axelrasmussen@google.com>

Acked-by: Shakeel Butt <shakeel.butt@linux.dev>

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[Johannes Weiner]

On Tue, Feb 24, 2026 at 04:24:34PM -0800, Axel Rasmussen wrote:
> This change swapped out mod_node_page_state for lruvec_stat_add_folio.
> But, these two APIs are not interchangeable: the lruvec version also
> increments memcg stats, in addition to "global" pgdat stats.
> 
> So after this change, the "pagetables" memcg stat in memory.stat always
> yields "0", which is a userspace visible regression.
> 
> I tried to look for a refactor where we add a variant of
> lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
> folio, to try to adhere to the spirit of the original patch. But at the
> end of the day this just means we have to call
> folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
> accomplish much.
> 
> This regression is visible in master as well as 6.18 stable, so CC
> stable too.
> 
> Fixes: f0c92726e89f ("ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()")
> Cc: stable@vger.kernel.org
> Signed-off-by: Axel Rasmussen <axelrasmussen@google.com>

Acked-by: Johannes Weiner <hannes@cmpxchg.org>

Willy's cleanup proposal looks good to me too, but this is more
straight forward to backport to stable.

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[Axel Rasmussen]

On Wed, Feb 25, 2026 at 11:31 AM Johannes Weiner <hannes@cmpxchg.org> wrote:
>
> On Tue, Feb 24, 2026 at 04:24:34PM -0800, Axel Rasmussen wrote:
> > This change swapped out mod_node_page_state for lruvec_stat_add_folio.
> > But, these two APIs are not interchangeable: the lruvec version also
> > increments memcg stats, in addition to "global" pgdat stats.
> >
> > So after this change, the "pagetables" memcg stat in memory.stat always
> > yields "0", which is a userspace visible regression.
> >
> > I tried to look for a refactor where we add a variant of
> > lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
> > folio, to try to adhere to the spirit of the original patch. But at the
> > end of the day this just means we have to call
> > folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
> > accomplish much.
> >
> > This regression is visible in master as well as 6.18 stable, so CC
> > stable too.
> >
> > Fixes: f0c92726e89f ("ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()")
> > Cc: stable@vger.kernel.org
> > Signed-off-by: Axel Rasmussen <axelrasmussen@google.com>
>
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>
>
> Willy's cleanup proposal looks good to me too, but this is more
> straight forward to backport to stable.

Thanks all for taking a look! I feel similarly, the other series looks
reasonable (modulo a couple of fixups pointed out in the other
thread), but at least for 6.18 / stable I would mildly prefer to just
do the simpler thing and put the status quo back. (Especially since
the other approach needs a bit more tweaking.)

Re: [PATCH] Revert "ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()"

[Vishal Moola (Oracle)]

On Tue, Feb 24, 2026 at 04:24:34PM -0800, Axel Rasmussen wrote:
> This change swapped out mod_node_page_state for lruvec_stat_add_folio.
> But, these two APIs are not interchangeable: the lruvec version also
> increments memcg stats, in addition to "global" pgdat stats.
> 
> So after this change, the "pagetables" memcg stat in memory.stat always
> yields "0", which is a userspace visible regression.
> 
> I tried to look for a refactor where we add a variant of
> lruvec_stat_mod_folio which takes a pgdat and a memcg instead of a
> folio, to try to adhere to the spirit of the original patch. But at the
> end of the day this just means we have to call
> folio_memcg(ptdesc_folio(ptdesc)) anyway, which doesn't really
> accomplish much.

I recall making sure we had a memcg_data member in ptdesc so that ptdesc
could directly reference it. Until we have a generic version, reverting
this patch should be the right way to go.

> This regression is visible in master as well as 6.18 stable, so CC
> stable too.
> 
> Fixes: f0c92726e89f ("ptdesc: remove references to folios from __pagetable_ctor() and pagetable_dtor()")
> Cc: stable@vger.kernel.org
> Signed-off-by: Axel Rasmussen <axelrasmussen@google.com>

Reviewed-by: Vishal Moola (Oracle) <vishal.moola@gmail.com>