[PATCH 0/4] mm, ceph: remove wait_on_page_writeback()

[PATCH 0/4] mm, ceph: remove wait_on_page_writeback()

[Tal Zussman]

wait_on_page_writeback() is a compatibility wrapper for
folio_wait_writeback() and has a single remaining caller in ceph's
writeback path. Convert that caller to use folios and remove the
wrapper.

Along the way, replace ceph's page_snap_context(), which uses
page->private, with a folio-based ceph_folio_snap_context(), dropping
three &folio->page conversions at its call sites.

No functional change intended.

Note: This is currently based on mm-new. It may conflict slightly with
Zi Yan's series "Remove PG_private by using page/folio->private checks
instead" [1], which should allow simplifying ceph_folio_snap_context()
further.

[1] https://lore.kernel.org/linux-mm/20260731-remove-pg_private-v1-0-142c97ba3562@nvidia.com/T/

Signed-off-by: Tal Zussman <tz2294@columbia.edu>
---
Tal Zussman (4):
      ceph: add ceph_folio_snap_context()
      ceph: convert ceph_wait_until_current_writes_complete() to folios
      mm: remove wait_on_page_writeback()
      ceph: remove page_snap_context()

 fs/ceph/addr.c          | 28 ++++++++++++++--------------
 include/linux/pagemap.h |  1 -
 mm/folio-compat.c       |  6 ------
 3 files changed, 14 insertions(+), 21 deletions(-)
---
base-commit: 1dbd7c34bb92dd9c0b0363b75f6ec444299af108
change-id: 20260802-remove-wait-on-page-writeback-8e0cb766d4c7

Best regards,
-- 
Tal Zussman <tz2294@columbia.edu>

[PATCH 1/4] ceph: add ceph_folio_snap_context()

[Tal Zussman]

Add a folio-based counterpart to page_snap_context() that reads the
snap context from folio->private directly.

Convert the three writeback paths that passed &folio->page to
page_snap_context() to ceph_folio_snap_context(), removing three
open-coded folio-to-page conversions. No functional change.

Signed-off-by: Tal Zussman <tz2294@columbia.edu>
---
 fs/ceph/addr.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index ecf33b66610c..a8ed2a14da44 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -75,6 +75,13 @@ static inline struct ceph_snap_context *page_snap_context(struct page *page)
 	return NULL;
 }
 
+static inline struct ceph_snap_context *ceph_folio_snap_context(struct folio *folio)
+{
+	if (folio_test_private(folio))
+		return (void *)folio->private;
+	return NULL;
+}
+
 /*
  * Dirty a page.  Optimistically adjust accounting, on the assumption
  * that we won't race with invalidate.  If we do, readjust.
@@ -745,7 +752,7 @@ static int write_folio_nounlock(struct folio *folio,
 		return -EIO;
 
 	/* verify this is a writeable snap context */
-	snapc = page_snap_context(&folio->page);
+	snapc = ceph_folio_snap_context(folio);
 	if (!snapc) {
 		doutc(cl, "%llx.%llx folio %p not dirty?\n", ceph_vinop(inode),
 		      folio);
@@ -1171,7 +1178,7 @@ int ceph_check_page_before_write(struct address_space *mapping,
 	}
 
 	/* only if matching snap context */
-	pgsnapc = page_snap_context(&folio->page);
+	pgsnapc = ceph_folio_snap_context(folio);
 	if (pgsnapc != ceph_wbc->snapc) {
 		doutc(cl, "folio snapc %p %lld != oldest %p %lld\n",
 		      pgsnapc, pgsnapc->seq,
@@ -1824,7 +1831,7 @@ ceph_find_incompatible(struct folio *folio)
 
 		folio_wait_writeback(folio);
 
-		snapc = page_snap_context(&folio->page);
+		snapc = ceph_folio_snap_context(folio);
 		if (!snapc || snapc == ci->i_head_snapc)
 			break;
 

-- 
2.39.5

Re: [PATCH 1/4] ceph: add ceph_folio_snap_context()

[Matthew Wilcox]

On Sun, Aug 02, 2026 at 12:50:02PM -0400, Tal Zussman wrote:
> +static inline struct ceph_snap_context *ceph_folio_snap_context(struct folio *folio)
> +{
> +	if (folio_test_private(folio))
> +		return (void *)folio->private;
> +	return NULL;
> +}

Filesystem folios _ought_to have folio->private as NULL when PG_private
is unset and PG_private set when folio->private is non-NULL.  Now, ceph
could be doing something Bad (I believe hugetlbfs does not honour this
for various reasons), but it looks like it's using
folio_attach_private() / folio_detach_private() appropriately.

So I think we should just do:

static inline
struct ceph_snap_context *ceph_folio_snap_context(const struct folio *folio)
{
	return folio->private;
}

(folio->private is already void *, so it doesn't need to be cast).
Splitting the definition where I did puts the unimmportant information
(static inline) on the previous line, so somebody grepping for
ceph_folio_snap_context gets the return type, while not breaking 80
columns.

Re: [PATCH 1/4] ceph: add ceph_folio_snap_context()

[Zi Yan]

On Sun Aug 2, 2026 at 5:52 PM EDT, Matthew Wilcox wrote:
> On Sun, Aug 02, 2026 at 12:50:02PM -0400, Tal Zussman wrote:
>> +static inline struct ceph_snap_context *ceph_folio_snap_context(struct folio *folio)
>> +{
>> +	if (folio_test_private(folio))
>> +		return (void *)folio->private;
>> +	return NULL;
>> +}
>
> Filesystem folios _ought_to have folio->private as NULL when PG_private
> is unset and PG_private set when folio->private is non-NULL.  Now, ceph
> could be doing something Bad (I believe hugetlbfs does not honour this
> for various reasons), but it looks like it's using
> folio_attach_private() / folio_detach_private() appropriately.
>
> So I think we should just do:
>
> static inline
> struct ceph_snap_context *ceph_folio_snap_context(const struct folio *folio)
> {
> 	return folio->private;
> }

Yes, please. So I do not need to handle another exceptional user when I
am trying to remove PG_private[1].

[1] https://lore.kernel.org/all/20260731-remove-pg_private-v1-0-142c97ba3562@nvidia.com/
>
> (folio->private is already void *, so it doesn't need to be cast).
> Splitting the definition where I did puts the unimmportant information
> (static inline) on the previous line, so somebody grepping for
> ceph_folio_snap_context gets the return type, while not breaking 80
> columns.




-- 
Best Regards,
Yan, Zi

Re: [PATCH 1/4] ceph: add ceph_folio_snap_context()

[Tal Zussman]

On 8/2/26 6:44 PM, Zi Yan wrote:
> On Sun Aug 2, 2026 at 5:52 PM EDT, Matthew Wilcox wrote:
>> On Sun, Aug 02, 2026 at 12:50:02PM -0400, Tal Zussman wrote:
>>> +static inline struct ceph_snap_context *ceph_folio_snap_context(struct folio *folio)
>>> +{
>>> +	if (folio_test_private(folio))
>>> +		return (void *)folio->private;
>>> +	return NULL;
>>> +}
>>
>> Filesystem folios _ought_to have folio->private as NULL when PG_private
>> is unset and PG_private set when folio->private is non-NULL.  Now, ceph
>> could be doing something Bad (I believe hugetlbfs does not honour this
>> for various reasons), but it looks like it's using
>> folio_attach_private() / folio_detach_private() appropriately.
>>
>> So I think we should just do:
>>
>> static inline
>> struct ceph_snap_context *ceph_folio_snap_context(const struct folio *folio)
>> {
>> 	return folio->private;
>> }
> 
> Yes, please. So I do not need to handle another exceptional user when I
> am trying to remove PG_private[1].
> 

Sounds good, will do.

[PATCH 2/4] ceph: convert ceph_wait_until_current_writes_complete() to folios

[Tal Zussman]

Iterate the writeback batch as folios rather than using
&folios[i]->page. Use ceph_folio_snap_context() for the snap context
check and folio_wait_writeback() instead of wait_on_page_writeback().

This removes one call to compound_head() and the last caller of
wait_on_page_writeback(). No functional change.

Signed-off-by: Tal Zussman <tz2294@columbia.edu>
---
 fs/ceph/addr.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index a8ed2a14da44..f4aaf9a5f196 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -1625,7 +1625,7 @@ void ceph_wait_until_current_writes_complete(struct address_space *mapping,
 					     struct writeback_control *wbc,
 					     struct ceph_writeback_ctl *ceph_wbc)
 {
-	struct page *page;
+	struct folio *folio;
 	unsigned i, nr;
 
 	if (wbc->sync_mode != WB_SYNC_NONE &&
@@ -1640,10 +1640,10 @@ void ceph_wait_until_current_writes_complete(struct address_space *mapping,
 						     PAGECACHE_TAG_WRITEBACK,
 						     &ceph_wbc->fbatch))) {
 			for (i = 0; i < nr; i++) {
-				page = &ceph_wbc->fbatch.folios[i]->page;
-				if (page_snap_context(page) != ceph_wbc->snapc)
+				folio = ceph_wbc->fbatch.folios[i];
+				if (ceph_folio_snap_context(folio) != ceph_wbc->snapc)
 					continue;
-				wait_on_page_writeback(page);
+				folio_wait_writeback(folio);
 			}
 
 			folio_batch_release(&ceph_wbc->fbatch);

-- 
2.39.5

Re: [PATCH 2/4] ceph: convert ceph_wait_until_current_writes_complete() to folios

[Matthew Wilcox]

On Sun, Aug 02, 2026 at 12:50:03PM -0400, Tal Zussman wrote:
> Iterate the writeback batch as folios rather than using
> &folios[i]->page. Use ceph_folio_snap_context() for the snap context
> check and folio_wait_writeback() instead of wait_on_page_writeback().
> 
> This removes one call to compound_head() and the last caller of
> wait_on_page_writeback(). No functional change.
> 
> Signed-off-by: Tal Zussman <tz2294@columbia.edu>

Beautiful.

Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>

[PATCH 3/4] mm: remove wait_on_page_writeback()

[Tal Zussman]

The last caller was converted to folio_wait_writeback(), so drop the
wait_on_page_writeback() compatibility wrapper and its declaration.

Signed-off-by: Tal Zussman <tz2294@columbia.edu>
---
 include/linux/pagemap.h | 1 -
 mm/folio-compat.c       | 6 ------
 2 files changed, 7 deletions(-)

diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
index 4e8b2b29f6d3..c8666fd14d3d 100644
--- a/include/linux/pagemap.h
+++ b/include/linux/pagemap.h
@@ -1335,7 +1335,6 @@ static inline int folio_wait_locked_killable(struct folio *folio)
 }
 
 void folio_end_read(struct folio *folio, bool success);
-void wait_on_page_writeback(struct page *page);
 void folio_wait_writeback(struct folio *folio);
 int folio_wait_writeback_killable(struct folio *folio);
 void end_page_writeback(struct page *page);
diff --git a/mm/folio-compat.c b/mm/folio-compat.c
index a02179a0bded..ae0271591d37 100644
--- a/mm/folio-compat.c
+++ b/mm/folio-compat.c
@@ -23,12 +23,6 @@ void end_page_writeback(struct page *page)
 }
 EXPORT_SYMBOL(end_page_writeback);
 
-void wait_on_page_writeback(struct page *page)
-{
-	return folio_wait_writeback(page_folio(page));
-}
-EXPORT_SYMBOL_GPL(wait_on_page_writeback);
-
 void mark_page_accessed(struct page *page)
 {
 	folio_mark_accessed(page_folio(page));

-- 
2.39.5

Re: [PATCH 3/4] mm: remove wait_on_page_writeback()

[Matthew Wilcox]

On Sun, Aug 02, 2026 at 12:50:04PM -0400, Tal Zussman wrote:
> The last caller was converted to folio_wait_writeback(), so drop the
> wait_on_page_writeback() compatibility wrapper and its declaration.
> 
> Signed-off-by: Tal Zussman <tz2294@columbia.edu>

Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>

Re: [PATCH 3/4] mm: remove wait_on_page_writeback()

[David Hildenbrand (Arm)]

On 8/2/26 18:50, Tal Zussman wrote:
> The last caller was converted to folio_wait_writeback(), so drop the
> wait_on_page_writeback() compatibility wrapper and its declaration.
> 
> Signed-off-by: Tal Zussman <tz2294@columbia.edu>
> ---

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

-- 
Cheers,

David

[PATCH 4/4] ceph: remove page_snap_context()

[Tal Zussman]

Convert the final caller in get_writepages_data_length() to use a folio
and ceph_folio_snap_context(), then remove page_snap_context().

This drops the last open-coded use of page->private in ceph's writeback
path.

Signed-off-by: Tal Zussman <tz2294@columbia.edu>
---
 fs/ceph/addr.c | 15 ++++-----------
 1 file changed, 4 insertions(+), 11 deletions(-)

diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index f4aaf9a5f196..702cf5fc1eab 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -29,9 +29,9 @@
  *
  * There are a few funny things going on here.
  *
- * The page->private field is used to reference a struct
- * ceph_snap_context for _every_ dirty page.  This indicates which
- * snapshot the page was logically dirtied in, and thus which snap
+ * The folio->private field is used to reference a struct
+ * ceph_snap_context for _every_ dirty folio.  This indicates which
+ * snapshot the folio was logically dirtied in, and thus which snap
  * context needs to be associated with the osd write during writeback.
  *
  * Similarly, struct ceph_inode_info maintains a set of counters to
@@ -68,13 +68,6 @@
 static int ceph_netfs_check_write_begin(struct file *file, loff_t pos, unsigned int len,
 					struct folio **foliop, void **_fsdata);
 
-static inline struct ceph_snap_context *page_snap_context(struct page *page)
-{
-	if (PagePrivate(page))
-		return (void *)page->private;
-	return NULL;
-}
-
 static inline struct ceph_snap_context *ceph_folio_snap_context(struct folio *folio)
 {
 	if (folio_test_private(folio))
@@ -697,7 +690,7 @@ static u64 get_writepages_data_length(struct inode *inode,
 	u64 end = i_size_read(inode);
 	u64 ret;
 
-	snapc = page_snap_context(ceph_fscrypt_pagecache_page(page));
+	snapc = ceph_folio_snap_context(page_folio(ceph_fscrypt_pagecache_page(page)));
 	if (snapc != ci->i_head_snapc) {
 		bool found = false;
 		spin_lock(&ci->i_ceph_lock);

-- 
2.39.5

Re: [PATCH 4/4] ceph: remove page_snap_context()

[Matthew Wilcox]

On Sun, Aug 02, 2026 at 12:50:05PM -0400, Tal Zussman wrote:
> Convert the final caller in get_writepages_data_length() to use a folio
> and ceph_folio_snap_context(), then remove page_snap_context().
> 
> This drops the last open-coded use of page->private in ceph's writeback
> path.

This one I'm deeply conflicted about.  It's adding an extra call to
compound_head() ... and we're not getting much for it.

I'd feel better about it if it started with::

 static u64 get_writepages_data_length(struct inode *inode,
                                       struct page *page, u64 start)
 {
+	struct folio *folio = page_folio(page);

and then we had a ceph_fscrypt_pagecache_folio() function and
ceph_fscrypt_folio_offset() (we already have a fscrypt_is_bounce_page())

That way we'd have this function entirely converted except for its
argument, and a future patch can do the conversion with little fuss.
And we'd get rid of one of the four remaining calls to
fscrypt_is_bounce_page()

> Signed-off-by: Tal Zussman <tz2294@columbia.edu>
> ---
>  fs/ceph/addr.c | 15 ++++-----------
>  1 file changed, 4 insertions(+), 11 deletions(-)
> 
> diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
> index f4aaf9a5f196..702cf5fc1eab 100644
> --- a/fs/ceph/addr.c
> +++ b/fs/ceph/addr.c
> @@ -29,9 +29,9 @@
>   *
>   * There are a few funny things going on here.
>   *
> - * The page->private field is used to reference a struct
> - * ceph_snap_context for _every_ dirty page.  This indicates which
> - * snapshot the page was logically dirtied in, and thus which snap
> + * The folio->private field is used to reference a struct
> + * ceph_snap_context for _every_ dirty folio.  This indicates which
> + * snapshot the folio was logically dirtied in, and thus which snap
>   * context needs to be associated with the osd write during writeback.
>   *
>   * Similarly, struct ceph_inode_info maintains a set of counters to
> @@ -68,13 +68,6 @@
>  static int ceph_netfs_check_write_begin(struct file *file, loff_t pos, unsigned int len,
>  					struct folio **foliop, void **_fsdata);
>  
> -static inline struct ceph_snap_context *page_snap_context(struct page *page)
> -{
> -	if (PagePrivate(page))
> -		return (void *)page->private;
> -	return NULL;
> -}
> -
>  static inline struct ceph_snap_context *ceph_folio_snap_context(struct folio *folio)
>  {
>  	if (folio_test_private(folio))
> @@ -697,7 +690,7 @@ static u64 get_writepages_data_length(struct inode *inode,
>  	u64 end = i_size_read(inode);
>  	u64 ret;
>  
> -	snapc = page_snap_context(ceph_fscrypt_pagecache_page(page));
> +	snapc = ceph_folio_snap_context(page_folio(ceph_fscrypt_pagecache_page(page)));
>  	if (snapc != ci->i_head_snapc) {
>  		bool found = false;
>  		spin_lock(&ci->i_ceph_lock);
> 
> -- 
> 2.39.5
> 

Re: [PATCH 4/4] ceph: remove page_snap_context()

[Tal Zussman]

On 8/2/26 10:06 PM, Matthew Wilcox wrote: 
> On Sun, Aug 02, 2026 at 12:50:05PM -0400, Tal Zussman wrote:
>> Convert the final caller in get_writepages_data_length() to use a folio
>> and ceph_folio_snap_context(), then remove page_snap_context().
>> 
>> This drops the last open-coded use of page->private in ceph's writeback
>> path.
> 
> This one I'm deeply conflicted about.  It's adding an extra call to
> compound_head() ... and we're not getting much for it.
> 
> I'd feel better about it if it started with::
> 
>   static u64 get_writepages_data_length(struct inode *inode,
>                                         struct page *page, u64 start)
>   {
> +	struct folio *folio = page_folio(page);
> 
> and then we had a ceph_fscrypt_pagecache_folio() function and
> ceph_fscrypt_folio_offset() (we already have a fscrypt_is_bounce_page())

If we're adding these, I might as well do a few more conversions...

> That way we'd have this function entirely converted except for its
> argument, and a future patch can do the conversion with little fuss.
> And we'd get rid of one of the four remaining calls to
> fscrypt_is_bounce_page()

I'm tempted to convert the argument as well, offload the page_folio
to the single caller (ceph_submit_write), and be done with this function
in one go. Although now that I look at it again, ceph_submit_write can
also be converted with the fscrypt helpers (albeit with a healthy dose
of page_folio() due to ceph's pages arrays...).