[PATCH v2 0/3] mm/migrate: preparatory cleanups for batch copy and offload

[PATCH v2 0/3] mm/migrate: preparatory cleanups for batch copy and offload

[Shivank Garg]

This is a small set of mm/migrate cleanups split out of the batch-copy
and offload RFC [1], so they can be reviewed and merged independently
ahead of that larger series.

No functional change intended.

[1] https://lore.kernel.org/all/20260428155043.39251-2-shivankg@amd.com

Signed-off-by: Shivank Garg <shivankg@amd.com>
---
Changes in v2:
- Rename folio/page_was_mapped to was_mapped and unmap_and_move_huge_page
  to unmap_and_move_hugetlb_folio (David).
- Fix stale "page" wording in comment.
- Rebase to mm-new.
- Link to v1: https://lore.kernel.org/r/20260626-migrate-cleanups-prep-v1-0-a95933af7619@amd.com

---
Shivank Garg (3):
      mm/migrate: rename page to folio leftovers
      mm/migrate: fix stale list name in migrate_folios_move() comment
      mm/migrate: use migrate_info field instead of private

 include/linux/mm_types.h |  1 +
 mm/migrate.c             | 52 +++++++++++++++++++++++-------------------------
 2 files changed, 26 insertions(+), 27 deletions(-)
---
base-commit: 15654b0268cd65b4e673510c0377774e61b86397
change-id: 20260626-migrate-cleanups-prep-77a0ad340cc6

Best regards,
-- 
Shivank Garg <shivankg@amd.com>

[PATCH v2 1/3] mm/migrate: rename page to folio leftovers

[Shivank Garg]

Rename migrate_folio_undo_src()'s page_was_mapped parameter to was_mapped,
unmap_and_move_huge_page() to unmap_and_move_hugetlb_folio(), its
page_was_mapped variable to was_mapped and fix stale "page" wording in its
comments.

Also fix migrate_folio() kerneldoc to say "folio" instead of "page".

Suggested-by: Dev Jain <dev.jain@arm.com>
Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
Signed-off-by: Shivank Garg <shivankg@amd.com>
---
 mm/migrate.c | 36 +++++++++++++++++-------------------
 1 file changed, 17 insertions(+), 19 deletions(-)

diff --git a/mm/migrate.c b/mm/migrate.c
index 49e10feeb094..858350ca58b4 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -884,7 +884,7 @@ static int __migrate_folio(struct address_space *mapping, struct folio *dst,
  * @mapping: The address_space containing the folio.
  * @dst: The folio to migrate the data to.
  * @src: The folio containing the current data.
- * @mode: How to migrate the page.
+ * @mode: How to migrate the folio.
  *
  * Common logic to directly migrate a single LRU folio suitable for
  * folios that do not have private data.
@@ -1157,13 +1157,10 @@ static void __migrate_folio_extract(struct folio *dst,
 }
 
 /* Restore the source folio to the original state upon failure */
-static void migrate_folio_undo_src(struct folio *src,
-				   int page_was_mapped,
-				   struct anon_vma *anon_vma,
-				   bool locked,
-				   struct list_head *ret)
+static void migrate_folio_undo_src(struct folio *src, int was_mapped,
+		struct anon_vma *anon_vma, bool locked, struct list_head *ret)
 {
-	if (page_was_mapped)
+	if (was_mapped)
 		remove_migration_ptes(src, src, 0);
 	/* Drop an anon_vma reference if we took one */
 	if (anon_vma)
@@ -1449,7 +1446,8 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
 }
 
 /*
- * Counterpart of unmap_and_move_page() for hugepage migration.
+ * Counterpart of migrate_folio_unmap() and migrate_folio_move() for hugetlb
+ * folio migration.
  *
  * This function doesn't wait the completion of hugepage I/O
  * because there is no race between I/O and migration for hugepage.
@@ -1466,20 +1464,20 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
  * because then pte is replaced with migration swap entry and direct I/O code
  * will wait in the page fault for migration to complete.
  */
-static int unmap_and_move_huge_page(new_folio_t get_new_folio,
+static int unmap_and_move_hugetlb_folio(new_folio_t get_new_folio,
 		free_folio_t put_new_folio, unsigned long private,
 		struct folio *src, int force, enum migrate_mode mode,
 		enum migrate_reason reason, struct list_head *ret)
 {
 	struct folio *dst;
 	int rc = -EAGAIN;
-	int page_was_mapped = 0;
+	int was_mapped = 0;
 	struct anon_vma *anon_vma = NULL;
 	struct address_space *mapping = NULL;
 	enum ttu_flags ttu = 0;
 
 	if (folio_ref_count(src) == 1) {
-		/* page was freed from under us. So we are done. */
+		/* folio was freed from under us. So we are done. */
 		folio_putback_hugetlb(src);
 		return 0;
 	}
@@ -1501,8 +1499,8 @@ static int unmap_and_move_huge_page(new_folio_t get_new_folio,
 	}
 
 	/*
-	 * Check for pages which are in the process of being freed.  Without
-	 * folio_mapping() set, hugetlbfs specific move page routine will not
+	 * Check for folios which are in the process of being freed.  Without
+	 * folio_mapping() set, hugetlbfs specific move folio routine will not
 	 * be called and we could leak usage counts for subpools.
 	 */
 	if (hugetlb_folio_subpool(src) && !folio_mapping(src)) {
@@ -1532,13 +1530,13 @@ static int unmap_and_move_huge_page(new_folio_t get_new_folio,
 		}
 
 		try_to_migrate(src, ttu);
-		page_was_mapped = 1;
+		was_mapped = 1;
 	}
 
 	if (!folio_mapped(src))
 		rc = move_to_new_folio(dst, src, mode);
 
-	if (page_was_mapped)
+	if (was_mapped)
 		remove_migration_ptes(src, !rc ? dst : src, ttu);
 
 	if (ttu & TTU_RMAP_LOCKED)
@@ -1663,10 +1661,10 @@ static int migrate_hugetlbs(struct list_head *from, new_folio_t get_new_folio,
 				continue;
 			}
 
-			rc = unmap_and_move_huge_page(get_new_folio,
-						      put_new_folio, private,
-						      folio, pass > 2, mode,
-						      reason, ret_folios);
+			rc = unmap_and_move_hugetlb_folio(get_new_folio,
+							  put_new_folio, private,
+							  folio, pass > 2, mode,
+							  reason, ret_folios);
 			/*
 			 * The rules are:
 			 *	0: hugetlb folio will be put back

-- 
2.43.0

Re: [PATCH v2 1/3] mm/migrate: rename page to folio leftovers

[Zi Yan]

On 1 Jul 2026, at 1:17, Shivank Garg wrote:

> Rename migrate_folio_undo_src()'s page_was_mapped parameter to was_mapped,
> unmap_and_move_huge_page() to unmap_and_move_hugetlb_folio(), its
> page_was_mapped variable to was_mapped and fix stale "page" wording in its
> comments.
>
> Also fix migrate_folio() kerneldoc to say "folio" instead of "page".
>
> Suggested-by: Dev Jain <dev.jain@arm.com>
> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Shivank Garg <shivankg@amd.com>
> ---
>  mm/migrate.c | 36 +++++++++++++++++-------------------
>  1 file changed, 17 insertions(+), 19 deletions(-)
>
LGTM.

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

Best Regards,
Yan, Zi

Re: [PATCH v2 1/3] mm/migrate: rename page to folio leftovers

[SJ Park]

On Wed, 1 Jul 2026 05:17:20 +0000 Shivank Garg <shivankg@amd.com> wrote:

> Rename migrate_folio_undo_src()'s page_was_mapped parameter to was_mapped,
> unmap_and_move_huge_page() to unmap_and_move_hugetlb_folio(), its
> page_was_mapped variable to was_mapped and fix stale "page" wording in its
> comments.
> 
> Also fix migrate_folio() kerneldoc to say "folio" instead of "page".

Looks cleaner to me!

> 
> Suggested-by: Dev Jain <dev.jain@arm.com>
> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Shivank Garg <shivankg@amd.com>

Reviewed-by: SJ Park <sj@kernel.org>

A couple of trivial loud-thinking comments below.

> ---
>  mm/migrate.c | 36 +++++++++++++++++-------------------
>  1 file changed, 17 insertions(+), 19 deletions(-)
> 
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 49e10feeb094..858350ca58b4 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -884,7 +884,7 @@ static int __migrate_folio(struct address_space *mapping, struct folio *dst,
>   * @mapping: The address_space containing the folio.
>   * @dst: The folio to migrate the data to.
>   * @src: The folio containing the current data.
> - * @mode: How to migrate the page.
> + * @mode: How to migrate the folio.
>   *
>   * Common logic to directly migrate a single LRU folio suitable for
>   * folios that do not have private data.
> @@ -1157,13 +1157,10 @@ static void __migrate_folio_extract(struct folio *dst,
>  }
>  
>  /* Restore the source folio to the original state upon failure */
> -static void migrate_folio_undo_src(struct folio *src,
> -				   int page_was_mapped,
> -				   struct anon_vma *anon_vma,
> -				   bool locked,
> -				   struct list_head *ret)
> +static void migrate_folio_undo_src(struct folio *src, int was_mapped,
> +		struct anon_vma *anon_vma, bool locked, struct list_head *ret)

I like two tabs indentations "slightly" over aligning with tabs + spaces.  So
above looks good to me.

[...]
> @@ -1663,10 +1661,10 @@ static int migrate_hugetlbs(struct list_head *from, new_folio_t get_new_folio,
>  				continue;
>  			}
>  
> -			rc = unmap_and_move_huge_page(get_new_folio,
> -						      put_new_folio, private,
> -						      folio, pass > 2, mode,
> -						      reason, ret_folios);
> +			rc = unmap_and_move_hugetlb_folio(get_new_folio,
> +							  put_new_folio, private,
> +							  folio, pass > 2, mode,
> +							  reason, ret_folios);

And for the same reason I wonder if we could use the two tabs indentation here.
That may reduce lines and make diffstat more cool?  Just loudly wondering.  My
R-b: is valid regardless.


Thanks,
SJ

[...]

[PATCH v2 2/3] mm/migrate: fix stale list name in migrate_folios_move() comment

[Shivank Garg]

The return-value description in migrate_folios_move() still refers to
unmap_folios, but that list no longer exists. Update this name to
src_folios.

Reviewed-by: Zi Yan <ziy@nvidia.com>
Acked-by: David Hildenbrand (Arm) <david@kernel.org>
Signed-off-by: Shivank Garg <shivankg@amd.com>
---
 mm/migrate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/migrate.c b/mm/migrate.c
index 858350ca58b4..5bc81d61116f 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1739,7 +1739,7 @@ static void migrate_folios_move(struct list_head *src_folios,
 		/*
 		 * The rules are:
 		 *	0: folio will be freed
-		 *	-EAGAIN: stay on the unmap_folios list
+		 *	-EAGAIN: stay on the src_folios list
 		 *	Other errno: put on ret_folios list
 		 */
 		switch (rc) {

-- 
2.43.0

[PATCH v2 3/3] mm/migrate: use migrate_info field instead of private

[Shivank Garg]

Add an unsigned long migrate_info member to the struct folio union and
use it to store migration state (anon_vma pointer and FOLIO_WAS_*
markers) instead of using folio->private. While at it, switch to bitwise
OR.

No functional change.

Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
Reviewed-by: Jonathan Cameron <jic23@kernel.org>
Acked-by: David Hildenbrand (Arm) <david@kernel.org>
Reviewed-by: Huang Ying <ying.huang@linux.alibaba.com>
Acked-by: Zi Yan <ziy@nvidia.com>
Signed-off-by: Shivank Garg <shivankg@amd.com>
---
 include/linux/mm_types.h |  1 +
 mm/migrate.c             | 14 +++++++-------
 2 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index b18c2b2e7d2c..ae9bca4eda5c 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -427,6 +427,7 @@ struct folio {
 			union {
 				void *private;
 				swp_entry_t swap;
+				unsigned long migrate_info;
 			};
 			atomic_t _mapcount;
 			atomic_t _refcount;
diff --git a/mm/migrate.c b/mm/migrate.c
index 5bc81d61116f..6fb6f46eea5f 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1130,7 +1130,7 @@ static int move_to_new_folio(struct folio *dst, struct folio *src,
 }
 
 /*
- * To record some information during migration, we use unused private
+ * To record some information during migration, we use the migrate_info
  * field of struct folio of the newly allocated destination folio.
  * This is safe because nobody is using it except us.
  */
@@ -1143,17 +1143,17 @@ enum {
 static void __migrate_folio_record(struct folio *dst,
 		int old_folio_state, struct anon_vma *anon_vma)
 {
-	dst->private = (void *)anon_vma + old_folio_state;
+	dst->migrate_info = (unsigned long)anon_vma | old_folio_state;
 }
 
 static void __migrate_folio_extract(struct folio *dst,
 		int *old_folio_state, struct anon_vma **anon_vmap)
 {
-	unsigned long private = (unsigned long)dst->private;
+	unsigned long info = dst->migrate_info;
 
-	*anon_vmap = (struct anon_vma *)(private & ~FOLIO_OLD_STATES);
-	*old_folio_state = private & FOLIO_OLD_STATES;
-	dst->private = NULL;
+	*anon_vmap = (struct anon_vma *)(info & ~FOLIO_OLD_STATES);
+	*old_folio_state = info & FOLIO_OLD_STATES;
+	dst->migrate_info = 0;
 }
 
 /* Restore the source folio to the original state upon failure */
@@ -1214,7 +1214,7 @@ static int migrate_folio_unmap(new_folio_t get_new_folio,
 		return -ENOMEM;
 	*dstp = dst;
 
-	dst->private = NULL;
+	dst->migrate_info = 0;
 
 	if (!folio_trylock(src)) {
 		if (mode == MIGRATE_ASYNC)

-- 
2.43.0

Re: [PATCH v2 3/3] mm/migrate: use migrate_info field instead of private

[SJ Park]

On Wed, 1 Jul 2026 05:17:22 +0000 Shivank Garg <shivankg@amd.com> wrote:

> Add an unsigned long migrate_info member to the struct folio union and
> use it to store migration state (anon_vma pointer and FOLIO_WAS_*
> markers) instead of using folio->private. While at it, switch to bitwise
> OR.
> 
> No functional change.
> 
> Suggested-by: David Hildenbrand (Arm) <david@kernel.org>
> Reviewed-by: Jonathan Cameron <jic23@kernel.org>
> Acked-by: David Hildenbrand (Arm) <david@kernel.org>
> Reviewed-by: Huang Ying <ying.huang@linux.alibaba.com>
> Acked-by: Zi Yan <ziy@nvidia.com>
> Signed-off-by: Shivank Garg <shivankg@amd.com>

Reviewed-by: SJ Park <sj@kernel.org>


Thanks,
SJ

[...]