[PATCH RFC 00/29] mm/migration: rework movable_ops page migration (part 1)

[PATCH RFC 00/29] mm/migration: rework movable_ops page migration (part 1)

[David Hildenbrand]

Based on mm/mm-new.

In the future, as we decouple "struct page" from "struct folio", pages
that support "non-lru page migration" -- movable_ops page migration
such as memory balloons and zsmalloc -- will no longer be folios. They
will not have ->mapping, ->lru, and likely no refcount and no
page lock. But they will have flags :)

This is the first part (other parts not written yet) of decoupling
movable_ops page migration from folio migration.

In this series, we get rid of the ->mapping usage, and start cleaning up
the code + separating it from folio migration.

Migration core will have to be further reworked to not treat movable_ops
pages like folios. This is the first step into that direction.

RFC because I only did some light testing with virtio-balloon + zsmalloc
page migration, and have to polish some patch descriptions.

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Jerrin Shaji George <jerrin.shaji-george@broadcom.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Cc: "Eugenio Pérez" <eperezma@redhat.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Joshua Hahn <joshua.hahnjy@gmail.com>
Cc: Rakie Kim <rakie.kim@sk.com>
Cc: Byungchul Park <byungchul@sk.com>
Cc: Gregory Price <gourry@gourry.net>
Cc: Ying Huang <ying.huang@linux.alibaba.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Brendan Jackman <jackmanb@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Xu Xin <xu.xin16@zte.com.cn>
Cc: Chengming Zhou <chengming.zhou@linux.dev>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Naoya Horiguchi <nao.horiguchi@gmail.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Rik van Riel <riel@surriel.com>
Cc: Harry Yoo <harry.yoo@oracle.com>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>


David Hildenbrand (29):
  mm/balloon_compaction: we cannot have isolated pages in the balloon
    list
  mm/balloon_compaction: convert balloon_page_delete() to
    balloon_page_finalize()
  mm/zsmalloc: drop PageIsolated() related VM_BUG_ONs
  mm/page_alloc: allow for making page types sticky until freed
  mm/balloon_compaction: make PageOffline sticky
  mm/zsmalloc: make PageZsmalloc() sticky
  mm/migrate: rename isolate_movable_page() to
    isolate_movable_ops_page()
  mm/migrate: rename putback_movable_folio() to
    putback_movable_ops_page()
  mm/migrate: factor out movable_ops page handling into
    migrate_movable_ops_page()
  mm/migrate: remove folio_test_movable() and folio_movable_ops()
  mm/migrate: move movable_ops page handling out of move_to_new_folio()
  mm/zsmalloc: stop using __ClearPageMovable()
  mm/balloon_compaction: stop using __ClearPageMovable()
  mm/migrate: remove __ClearPageMovable()
  mm/migration: remove PageMovable()
  mm: rename __PageMovable() to page_has_movable_ops()
  mm/page_isolation: drop __folio_test_movable() check for large folios
  mm: remove __folio_test_movable()
  mm: stop storing migration_ops in page->mapping
  mm: convert "movable" flag in page->mapping to a page flag
  mm: rename PG_isolated to PG_movable_ops_isolated
  mm/page-flags: rename PAGE_MAPPING_MOVABLE to PAGE_MAPPING_ANON_KSM
  mm/page-alloc: remove PageMappingFlags()
  mm/page-flags: remove folio_mapping_flags()
  mm: simplify folio_expected_ref_count()
  mm: rename PAGE_MAPPING_* to FOLIO_MAPPING_*
  docs/mm: convert from "Non-LRU page migration" to "movable_ops page
    migration"
  mm/balloon_compaction: "movable_ops" doc updates
  mm/balloon_compaction: provide single balloon_page_insert() and
    balloon_mapping_gfp_mask()

 Documentation/mm/page_migration.rst  |  39 ++--
 arch/powerpc/platforms/pseries/cmm.c |   2 +-
 drivers/misc/vmw_balloon.c           |   3 +-
 drivers/virtio/virtio_balloon.c      |   4 +-
 fs/proc/page.c                       |   4 +-
 include/linux/balloon_compaction.h   |  90 ++++-----
 include/linux/fs.h                   |   2 +-
 include/linux/migrate.h              |  42 +----
 include/linux/mm.h                   |   4 +-
 include/linux/mm_types.h             |   1 -
 include/linux/page-flags.h           | 104 +++++++----
 include/linux/pagemap.h              |   2 +-
 include/linux/zsmalloc.h             |   2 +
 mm/balloon_compaction.c              |  21 ++-
 mm/compaction.c                      |  44 +----
 mm/gup.c                             |   4 +-
 mm/internal.h                        |   2 +-
 mm/ksm.c                             |   4 +-
 mm/memory-failure.c                  |   4 +-
 mm/memory_hotplug.c                  |   8 +-
 mm/migrate.c                         | 269 +++++++++++++++------------
 mm/page_alloc.c                      |  12 +-
 mm/page_isolation.c                  |  12 +-
 mm/rmap.c                            |  16 +-
 mm/util.c                            |   6 +-
 mm/vmscan.c                          |   6 +-
 mm/zpdesc.h                          |  15 +-
 mm/zsmalloc.c                        |  29 ++-
 28 files changed, 361 insertions(+), 390 deletions(-)


base-commit: 21f39cc0346ea2081a83f0469995144d75d23075
-- 
2.49.0

[PATCH RFC 01/29] mm/balloon_compaction: we cannot have isolated pages in the balloon list

[David Hildenbrand]

The core will set PG_isolated only after mops->isolate_page() was
called. In case of the balloon, that is where we will remove it from
the balloon list. So we cannot have isolated pages in the balloon list.

Let's drop this unnecessary check.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/balloon_compaction.c | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/mm/balloon_compaction.c b/mm/balloon_compaction.c
index d3e00731e2628..fcb60233aa35d 100644
--- a/mm/balloon_compaction.c
+++ b/mm/balloon_compaction.c
@@ -94,12 +94,6 @@ size_t balloon_page_list_dequeue(struct balloon_dev_info *b_dev_info,
 		if (!trylock_page(page))
 			continue;
 
-		if (IS_ENABLED(CONFIG_BALLOON_COMPACTION) &&
-		    PageIsolated(page)) {
-			/* raced with isolation */
-			unlock_page(page);
-			continue;
-		}
 		balloon_page_delete(page);
 		__count_vm_event(BALLOON_DEFLATE);
 		list_add(&page->lru, pages);
-- 
2.49.0

[PATCH RFC 02/29] mm/balloon_compaction: convert balloon_page_delete() to balloon_page_finalize()

[David Hildenbrand]

Let's move the removal of the page from the balloon list into the single
caller, to remove the dependency on the PG_isolated flag and clarify
locking requirements.

We'll shuffle the operations a bit such that they logically make more sense
(e.g., remove from the list before clearing flags).

In balloon migration functions we can now move the balloon_page_finalize()
out of the balloon lock and perform the finalization just before dropping
the balloon reference.

Document that the page lock is currently required when modifying the
movability aspects of a page; hopefully we can soon decouple this from the
page lock.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 arch/powerpc/platforms/pseries/cmm.c |  2 +-
 drivers/misc/vmw_balloon.c           |  3 +-
 drivers/virtio/virtio_balloon.c      |  4 +--
 include/linux/balloon_compaction.h   | 43 +++++++++++-----------------
 mm/balloon_compaction.c              |  3 +-
 5 files changed, 21 insertions(+), 34 deletions(-)

diff --git a/arch/powerpc/platforms/pseries/cmm.c b/arch/powerpc/platforms/pseries/cmm.c
index 5f4037c1d7fe8..5e0a718d1be7b 100644
--- a/arch/powerpc/platforms/pseries/cmm.c
+++ b/arch/powerpc/platforms/pseries/cmm.c
@@ -532,7 +532,6 @@ static int cmm_migratepage(struct balloon_dev_info *b_dev_info,
 
 	spin_lock_irqsave(&b_dev_info->pages_lock, flags);
 	balloon_page_insert(b_dev_info, newpage);
-	balloon_page_delete(page);
 	b_dev_info->isolated_pages--;
 	spin_unlock_irqrestore(&b_dev_info->pages_lock, flags);
 
@@ -542,6 +541,7 @@ static int cmm_migratepage(struct balloon_dev_info *b_dev_info,
 	 */
 	plpar_page_set_active(page);
 
+	balloon_page_finalize(page);
 	/* balloon page list reference */
 	put_page(page);
 
diff --git a/drivers/misc/vmw_balloon.c b/drivers/misc/vmw_balloon.c
index c817d8c216413..6653fc53c951c 100644
--- a/drivers/misc/vmw_balloon.c
+++ b/drivers/misc/vmw_balloon.c
@@ -1778,8 +1778,7 @@ static int vmballoon_migratepage(struct balloon_dev_info *b_dev_info,
 	 * @pages_lock . We keep holding @comm_lock since we will need it in a
 	 * second.
 	 */
-	balloon_page_delete(page);
-
+	balloon_page_finalize(page);
 	put_page(page);
 
 	/* Inflate */
diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_balloon.c
index 89da052f4f687..e299e18346a30 100644
--- a/drivers/virtio/virtio_balloon.c
+++ b/drivers/virtio/virtio_balloon.c
@@ -866,15 +866,13 @@ static int virtballoon_migratepage(struct balloon_dev_info *vb_dev_info,
 	tell_host(vb, vb->inflate_vq);
 
 	/* balloon's page migration 2nd step -- deflate "page" */
-	spin_lock_irqsave(&vb_dev_info->pages_lock, flags);
-	balloon_page_delete(page);
-	spin_unlock_irqrestore(&vb_dev_info->pages_lock, flags);
 	vb->num_pfns = VIRTIO_BALLOON_PAGES_PER_PAGE;
 	set_page_pfns(vb, vb->pfns, page);
 	tell_host(vb, vb->deflate_vq);
 
 	mutex_unlock(&vb->balloon_lock);
 
+	balloon_page_finalize(page);
 	put_page(page); /* balloon reference */
 
 	return MIGRATEPAGE_SUCCESS;
diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
index 5ca2d56996201..b9f19da37b089 100644
--- a/include/linux/balloon_compaction.h
+++ b/include/linux/balloon_compaction.h
@@ -97,27 +97,6 @@ static inline void balloon_page_insert(struct balloon_dev_info *balloon,
 	list_add(&page->lru, &balloon->pages);
 }
 
-/*
- * balloon_page_delete - delete a page from balloon's page list and clear
- *			 the page->private assignement accordingly.
- * @page    : page to be released from balloon's page list
- *
- * Caller must ensure the page is locked and the spin_lock protecting balloon
- * pages list is held before deleting a page from the balloon device.
- */
-static inline void balloon_page_delete(struct page *page)
-{
-	__ClearPageOffline(page);
-	__ClearPageMovable(page);
-	set_page_private(page, 0);
-	/*
-	 * No touch page.lru field once @page has been isolated
-	 * because VM is using the field.
-	 */
-	if (!PageIsolated(page))
-		list_del(&page->lru);
-}
-
 /*
  * balloon_page_device - get the b_dev_info descriptor for the balloon device
  *			 that enqueues the given page.
@@ -141,12 +120,6 @@ static inline void balloon_page_insert(struct balloon_dev_info *balloon,
 	list_add(&page->lru, &balloon->pages);
 }
 
-static inline void balloon_page_delete(struct page *page)
-{
-	__ClearPageOffline(page);
-	list_del(&page->lru);
-}
-
 static inline gfp_t balloon_mapping_gfp_mask(void)
 {
 	return GFP_HIGHUSER;
@@ -154,6 +127,22 @@ static inline gfp_t balloon_mapping_gfp_mask(void)
 
 #endif /* CONFIG_BALLOON_COMPACTION */
 
+/*
+ * balloon_page_finalize - prepare a balloon page that was removed from the
+ *			   balloon list for release to the page allocator
+ * @page: page to be released to the page allocator
+ *
+ * Caller must ensure that the page is locked.
+ */
+static inline void balloon_page_finalize(struct page *page)
+{
+	if (IS_ENABLED(CONFIG_BALLOON_COMPACTION)) {
+		__ClearPageMovable(page);
+		set_page_private(page, 0);
+	}
+	__ClearPageOffline(page);
+}
+
 /*
  * balloon_page_push - insert a page into a page list.
  * @head : pointer to list
diff --git a/mm/balloon_compaction.c b/mm/balloon_compaction.c
index fcb60233aa35d..ec176bdb8a78b 100644
--- a/mm/balloon_compaction.c
+++ b/mm/balloon_compaction.c
@@ -94,7 +94,8 @@ size_t balloon_page_list_dequeue(struct balloon_dev_info *b_dev_info,
 		if (!trylock_page(page))
 			continue;
 
-		balloon_page_delete(page);
+		list_del(&page->lru);
+		balloon_page_finalize(page);
 		__count_vm_event(BALLOON_DEFLATE);
 		list_add(&page->lru, pages);
 		unlock_page(page);
-- 
2.49.0

[PATCH RFC 03/29] mm/zsmalloc: drop PageIsolated() related VM_BUG_ONs

[David Hildenbrand]

Let's drop these checks; these are conditions the core migration code
must make sure will hold either way, no need to double check.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/zpdesc.h   | 5 -----
 mm/zsmalloc.c | 5 -----
 2 files changed, 10 deletions(-)

diff --git a/mm/zpdesc.h b/mm/zpdesc.h
index d3df316e5bb7b..5cb7e3de43952 100644
--- a/mm/zpdesc.h
+++ b/mm/zpdesc.h
@@ -168,11 +168,6 @@ static inline void __zpdesc_clear_zsmalloc(struct zpdesc *zpdesc)
 	__ClearPageZsmalloc(zpdesc_page(zpdesc));
 }
 
-static inline bool zpdesc_is_isolated(struct zpdesc *zpdesc)
-{
-	return PageIsolated(zpdesc_page(zpdesc));
-}
-
 static inline struct zone *zpdesc_zone(struct zpdesc *zpdesc)
 {
 	return page_zone(zpdesc_page(zpdesc));
diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c
index 999b513c7fdff..7f1431f2be98f 100644
--- a/mm/zsmalloc.c
+++ b/mm/zsmalloc.c
@@ -1719,8 +1719,6 @@ static bool zs_page_isolate(struct page *page, isolate_mode_t mode)
 	 * Page is locked so zspage couldn't be destroyed. For detail, look at
 	 * lock_zspage in free_zspage.
 	 */
-	VM_BUG_ON_PAGE(PageIsolated(page), page);
-
 	return true;
 }
 
@@ -1739,8 +1737,6 @@ static int zs_page_migrate(struct page *newpage, struct page *page,
 	unsigned long old_obj, new_obj;
 	unsigned int obj_idx;
 
-	VM_BUG_ON_PAGE(!zpdesc_is_isolated(zpdesc), zpdesc_page(zpdesc));
-
 	/* The page is locked, so this pointer must remain valid */
 	zspage = get_zspage(zpdesc);
 	pool = zspage->pool;
@@ -1811,7 +1807,6 @@ static int zs_page_migrate(struct page *newpage, struct page *page,
 
 static void zs_page_putback(struct page *page)
 {
-	VM_BUG_ON_PAGE(!PageIsolated(page), page);
 }
 
 static const struct movable_operations zsmalloc_mops = {
-- 
2.49.0

[PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[David Hildenbrand]

Let's allow for not clearing a page type before freeing a page to the
buddy.

We'll focus on having a type set on the first page of a larger
allocation only.

With this change, we can reliably identify typed folios even though
they might be in the process of getting freed, which will come in handy
in migration code (at least in the transition phase).

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/page_alloc.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 858bc17653af9..44e56d31cfeb1 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1380,6 +1380,9 @@ __always_inline bool free_pages_prepare(struct page *page,
 			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
 		page->mapping = NULL;
 	}
+	if (unlikely(page_has_type(page)))
+		page->page_type = UINT_MAX;
+
 	if (is_check_pages_enabled()) {
 		if (free_page_is_bad(page))
 			bad++;
-- 
2.49.0

[PATCH RFC 05/29] mm/balloon_compaction: make PageOffline sticky

[David Hildenbrand]

Let the buddy handle clearing the type.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/balloon_compaction.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
index b9f19da37b089..bfc6e50bd004b 100644
--- a/include/linux/balloon_compaction.h
+++ b/include/linux/balloon_compaction.h
@@ -140,7 +140,7 @@ static inline void balloon_page_finalize(struct page *page)
 		__ClearPageMovable(page);
 		set_page_private(page, 0);
 	}
-	__ClearPageOffline(page);
+	/* PageOffline is sticky until the page is freed to the buddy. */
 }
 
 /*
-- 
2.49.0

[PATCH RFC 06/29] mm/zsmalloc: make PageZsmalloc() sticky

[David Hildenbrand]

Let the buddy handle clearing the type.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/zpdesc.h   | 5 -----
 mm/zsmalloc.c | 3 +--
 2 files changed, 1 insertion(+), 7 deletions(-)

diff --git a/mm/zpdesc.h b/mm/zpdesc.h
index 5cb7e3de43952..5763f36039736 100644
--- a/mm/zpdesc.h
+++ b/mm/zpdesc.h
@@ -163,11 +163,6 @@ static inline void __zpdesc_set_zsmalloc(struct zpdesc *zpdesc)
 	__SetPageZsmalloc(zpdesc_page(zpdesc));
 }
 
-static inline void __zpdesc_clear_zsmalloc(struct zpdesc *zpdesc)
-{
-	__ClearPageZsmalloc(zpdesc_page(zpdesc));
-}
-
 static inline struct zone *zpdesc_zone(struct zpdesc *zpdesc)
 {
 	return page_zone(zpdesc_page(zpdesc));
diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c
index 7f1431f2be98f..f98747aed4330 100644
--- a/mm/zsmalloc.c
+++ b/mm/zsmalloc.c
@@ -880,7 +880,7 @@ static void reset_zpdesc(struct zpdesc *zpdesc)
 	ClearPagePrivate(page);
 	zpdesc->zspage = NULL;
 	zpdesc->next = NULL;
-	__ClearPageZsmalloc(page);
+	/* PageZsmalloc is sticky until the page is freed to the buddy. */
 }
 
 static int trylock_zspage(struct zspage *zspage)
@@ -1055,7 +1055,6 @@ static struct zspage *alloc_zspage(struct zs_pool *pool,
 		if (!zpdesc) {
 			while (--i >= 0) {
 				zpdesc_dec_zone_page_state(zpdescs[i]);
-				__zpdesc_clear_zsmalloc(zpdescs[i]);
 				free_zpdesc(zpdescs[i]);
 			}
 			cache_free_zspage(pool, zspage);
-- 
2.49.0

[PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[David Hildenbrand]

... and start moving back to per-page things that will absolutely not be
folio things in the future. Add documentation and a comment that the
remaining folio stuff (lock, refcount) will have to be reworked as well.

While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
it gracefully (relevant with further changes), and convert a
WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/migrate.h |  4 ++--
 mm/compaction.c         |  2 +-
 mm/migrate.c            | 39 +++++++++++++++++++++++++++++----------
 3 files changed, 32 insertions(+), 13 deletions(-)

diff --git a/include/linux/migrate.h b/include/linux/migrate.h
index aaa2114498d6d..c0ec7422837bd 100644
--- a/include/linux/migrate.h
+++ b/include/linux/migrate.h
@@ -69,7 +69,7 @@ int migrate_pages(struct list_head *l, new_folio_t new, free_folio_t free,
 		  unsigned long private, enum migrate_mode mode, int reason,
 		  unsigned int *ret_succeeded);
 struct folio *alloc_migration_target(struct folio *src, unsigned long private);
-bool isolate_movable_page(struct page *page, isolate_mode_t mode);
+bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode);
 bool isolate_folio_to_list(struct folio *folio, struct list_head *list);
 
 int migrate_huge_page_move_mapping(struct address_space *mapping,
@@ -90,7 +90,7 @@ static inline int migrate_pages(struct list_head *l, new_folio_t new,
 static inline struct folio *alloc_migration_target(struct folio *src,
 		unsigned long private)
 	{ return NULL; }
-static inline bool isolate_movable_page(struct page *page, isolate_mode_t mode)
+static inline bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 	{ return false; }
 static inline bool isolate_folio_to_list(struct folio *folio, struct list_head *list)
 	{ return false; }
diff --git a/mm/compaction.c b/mm/compaction.c
index 3925cb61dbb8f..17455c5a4be05 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -1093,7 +1093,7 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
 					locked = NULL;
 				}
 
-				if (isolate_movable_page(page, mode)) {
+				if (isolate_movable_ops_page(page, mode)) {
 					folio = page_folio(page);
 					goto isolate_success;
 				}
diff --git a/mm/migrate.c b/mm/migrate.c
index ea8c74d996592..6bbb455f8b593 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -51,8 +51,26 @@
 #include "internal.h"
 #include "swap.h"
 
-bool isolate_movable_page(struct page *page, isolate_mode_t mode)
+/**
+ * isolate_movable_ops_page - isolate a movable_ops page for migration
+ * @page: The page.
+ * @mode: The isolation mode.
+ *
+ * Try to isolate a movable_ops page for migration. Will fail if the page is
+ * not a movable_ops page, if the page is already isolated for migration
+ * or if the page was just was released by its owner.
+ *
+ * Once isolated, the page cannot get freed until it is either putback
+ * or migrated.
+ *
+ * Returns true if isolation succeeded, otherwise false.
+ */
+bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 {
+	/*
+	 * TODO: these pages will not be folios in the future. All
+	 * folio dependencies will have to be removed.
+	 */
 	struct folio *folio = folio_get_nontail_page(page);
 	const struct movable_operations *mops;
 
@@ -73,7 +91,7 @@ bool isolate_movable_page(struct page *page, isolate_mode_t mode)
 	 * we use non-atomic bitops on newly allocated page flags so
 	 * unconditionally grabbing the lock ruins page's owner side.
 	 */
-	if (unlikely(!__folio_test_movable(folio)))
+	if (unlikely(!__PageMovable(page)))
 		goto out_putfolio;
 
 	/*
@@ -90,18 +108,19 @@ bool isolate_movable_page(struct page *page, isolate_mode_t mode)
 	if (unlikely(!folio_trylock(folio)))
 		goto out_putfolio;
 
-	if (!folio_test_movable(folio) || folio_test_isolated(folio))
+	if (!PageMovable(page) || PageIsolated(page))
 		goto out_no_isolated;
 
-	mops = folio_movable_ops(folio);
-	VM_BUG_ON_FOLIO(!mops, folio);
+	mops = page_movable_ops(page);
+	if (WARN_ON_ONCE(!mops))
+		goto out_no_isolated;
 
-	if (!mops->isolate_page(&folio->page, mode))
+	if (!mops->isolate_page(page, mode))
 		goto out_no_isolated;
 
 	/* Driver shouldn't use the isolated flag */
-	WARN_ON_ONCE(folio_test_isolated(folio));
-	folio_set_isolated(folio);
+	VM_WARN_ON_ONCE_PAGE(PageIsolated(page), page);
+	SetPageIsolated(page);
 	folio_unlock(folio);
 
 	return true;
@@ -175,8 +194,8 @@ bool isolate_folio_to_list(struct folio *folio, struct list_head *list)
 	if (lru)
 		isolated = folio_isolate_lru(folio);
 	else
-		isolated = isolate_movable_page(&folio->page,
-						ISOLATE_UNEVICTABLE);
+		isolated = isolate_movable_ops_page(&folio->page,
+						    ISOLATE_UNEVICTABLE);
 
 	if (!isolated)
 		return false;
-- 
2.49.0

[PATCH RFC 08/29] mm/migrate: rename putback_movable_folio() to putback_movable_ops_page()

[David Hildenbrand]

... and factor the complete handling of movable_ops pages out.
Convert it similar to isolate_movable_ops_page().

While at it, convert the VM_BUG_ON_FOLIO() into a VM_WARN_ON_PAGE().

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/migrate.c | 37 ++++++++++++++++++++++++-------------
 1 file changed, 24 insertions(+), 13 deletions(-)

diff --git a/mm/migrate.c b/mm/migrate.c
index 6bbb455f8b593..32e77898f7d6c 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -133,12 +133,30 @@ bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 	return false;
 }
 
-static void putback_movable_folio(struct folio *folio)
+/**
+ * putback_movable_ops_page - putback an isolated movable_ops page
+ * @page: The isolated page.
+ *
+ * Putback an isolated movable_ops page.
+ *
+ * After the page was putback, it might get freed instantly.
+ */
+static void putback_movable_ops_page(struct page *page)
 {
-	const struct movable_operations *mops = folio_movable_ops(folio);
-
-	mops->putback_page(&folio->page);
-	folio_clear_isolated(folio);
+	/*
+	 * TODO: these pages will not be folios in the future. All
+	 * folio dependencies will have to be removed.
+	 */
+	struct folio *folio = page_folio(page);
+
+	VM_WARN_ON_ONCE_PAGE(!PageIsolated(page), page);
+	folio_lock(folio);
+	/* If the page was released by it's owner, there is nothing to do. */
+	if (PageMovable(page))
+		page_movable_ops(page)->putback_page(page);
+	ClearPageIsolated(page);
+	folio_unlock(folio);
+	folio_put(folio);
 }
 
 /*
@@ -166,14 +184,7 @@ void putback_movable_pages(struct list_head *l)
 		 * have PAGE_MAPPING_MOVABLE.
 		 */
 		if (unlikely(__folio_test_movable(folio))) {
-			VM_BUG_ON_FOLIO(!folio_test_isolated(folio), folio);
-			folio_lock(folio);
-			if (folio_test_movable(folio))
-				putback_movable_folio(folio);
-			else
-				folio_clear_isolated(folio);
-			folio_unlock(folio);
-			folio_put(folio);
+			putback_movable_ops_page(&folio->page);
 		} else {
 			node_stat_mod_folio(folio, NR_ISOLATED_ANON +
 					folio_is_file_lru(folio), -folio_nr_pages(folio));
-- 
2.49.0

[PATCH RFC 09/29] mm/migrate: factor out movable_ops page handling into migrate_movable_ops_page()

[David Hildenbrand]

Let's factor it out, simplifying the calling code.

The assumption is that flush_dcache_page() is not required for
movable_ops pages: as documented for flush_dcache_folio(), it really
only applies when the kernel wrote to pagecache pages / pages in
highmem. movable_ops callbacks should be handling flushing
caches if ever required.

Note that we can now change folio_mapping_flags() to folio_test_anon()
to make it clearer, because movable_ops pages will never take that path.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/migrate.c | 82 ++++++++++++++++++++++++++++------------------------
 1 file changed, 45 insertions(+), 37 deletions(-)

diff --git a/mm/migrate.c b/mm/migrate.c
index 32e77898f7d6c..456e41dad83a2 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -159,6 +159,45 @@ static void putback_movable_ops_page(struct page *page)
 	folio_put(folio);
 }
 
+/**
+ * migrate_movable_ops_page - migrate an isolated movable_ops page
+ * @page: The isolated page.
+ *
+ * Migrate an isolated movable_ops page.
+ *
+ * If the src page was already released by its owner, the src page is
+ * un-isolated (putback) and migration succeeds; the migration core will be the
+ * owner of both pages.
+ *
+ * If the src page was not released by its owner and the migration was
+ * successful, the owner of the src page and the dst page are swapped and
+ * the src page is un-isolated.
+ *
+ * If migration fails, the ownership stays unmodified and the src page
+ * remains isolated: migration may be retried later or the page can be putback.
+ *
+ * TODO: migration core will treat both pages as folios and lock them before
+ * this call to unlock them after this call. Further, the folio refcounts on
+ * src and dst are also released by migration core. These pages will not be
+ * folios in the future, so that must be reworked.
+ *
+ * Returns MIGRATEPAGE_SUCCESS on success, otherwise a negative error
+ * code.
+ */
+static int migrate_movable_ops_page(struct page *dst, struct page *src,
+		enum migrate_mode mode)
+{
+	int rc = MIGRATEPAGE_SUCCESS;
+
+	VM_WARN_ON_ONCE_PAGE(!PageIsolated(src), src);
+	/* If the page was released by it's owner, there is nothing to do. */
+	if (PageMovable(src))
+		rc = page_movable_ops(src)->migrate_page(dst, src, mode);
+	if (rc == MIGRATEPAGE_SUCCESS)
+		ClearPageIsolated(src);
+	return rc;
+}
+
 /*
  * Put previously isolated pages back onto the appropriate lists
  * from where they were once taken off for compaction/migration.
@@ -1023,51 +1062,20 @@ static int move_to_new_folio(struct folio *dst, struct folio *src,
 								mode);
 		else
 			rc = fallback_migrate_folio(mapping, dst, src, mode);
-	} else {
-		const struct movable_operations *mops;
 
-		/*
-		 * In case of non-lru page, it could be released after
-		 * isolation step. In that case, we shouldn't try migration.
-		 */
-		VM_BUG_ON_FOLIO(!folio_test_isolated(src), src);
-		if (!folio_test_movable(src)) {
-			rc = MIGRATEPAGE_SUCCESS;
-			folio_clear_isolated(src);
+		if (rc != MIGRATEPAGE_SUCCESS)
 			goto out;
-		}
-
-		mops = folio_movable_ops(src);
-		rc = mops->migrate_page(&dst->page, &src->page, mode);
-		WARN_ON_ONCE(rc == MIGRATEPAGE_SUCCESS &&
-				!folio_test_isolated(src));
-	}
-
-	/*
-	 * When successful, old pagecache src->mapping must be cleared before
-	 * src is freed; but stats require that PageAnon be left as PageAnon.
-	 */
-	if (rc == MIGRATEPAGE_SUCCESS) {
-		if (__folio_test_movable(src)) {
-			VM_BUG_ON_FOLIO(!folio_test_isolated(src), src);
-
-			/*
-			 * We clear PG_movable under page_lock so any compactor
-			 * cannot try to migrate this page.
-			 */
-			folio_clear_isolated(src);
-		}
-
 		/*
-		 * Anonymous and movable src->mapping will be cleared by
-		 * free_pages_prepare so don't reset it here for keeping
-		 * the type to work PageAnon, for example.
+		 * For pagecache folios, src->mapping must be cleared before src
+		 * is freed. Anonymous folios must stay anonymous until freed.
 		 */
-		if (!folio_mapping_flags(src))
+		if (!folio_test_anon(src))
 			src->mapping = NULL;
 
 		if (likely(!folio_is_zone_device(dst)))
 			flush_dcache_folio(dst);
+	} else {
+		rc = migrate_movable_ops_page(&dst->page, &src->page, mode);
 	}
 out:
 	return rc;
-- 
2.49.0

[PATCH RFC 10/29] mm/migrate: remove folio_test_movable() and folio_movable_ops()

[David Hildenbrand]

Folios will have nothing to do with movable_ops page migration. These
functions are now unused, so let's remove them.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/migrate.h | 14 --------------
 1 file changed, 14 deletions(-)

diff --git a/include/linux/migrate.h b/include/linux/migrate.h
index c0ec7422837bd..c99a00d4ca27d 100644
--- a/include/linux/migrate.h
+++ b/include/linux/migrate.h
@@ -118,20 +118,6 @@ static inline void __ClearPageMovable(struct page *page)
 }
 #endif
 
-static inline bool folio_test_movable(struct folio *folio)
-{
-	return PageMovable(&folio->page);
-}
-
-static inline
-const struct movable_operations *folio_movable_ops(struct folio *folio)
-{
-	VM_BUG_ON(!__folio_test_movable(folio));
-
-	return (const struct movable_operations *)
-		((unsigned long)folio->mapping - PAGE_MAPPING_MOVABLE);
-}
-
 static inline
 const struct movable_operations *page_movable_ops(struct page *page)
 {
-- 
2.49.0

[PATCH RFC 11/29] mm/migrate: move movable_ops page handling out of move_to_new_folio()

[David Hildenbrand]

Let's move that handling directly into migrate_folio_move(), so we can
simplify move_to_new_folio(). While at it, fixup the documentation a
bit.

Note that unmap_and_move_huge_page() does not care, because it only
deals with actual folios. (we only support migration of
individual movable_ops pages)

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/migrate.c | 61 ++++++++++++++++++++++++----------------------------
 1 file changed, 28 insertions(+), 33 deletions(-)

diff --git a/mm/migrate.c b/mm/migrate.c
index 456e41dad83a2..db807f9bbf975 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1024,11 +1024,12 @@ static int fallback_migrate_folio(struct address_space *mapping,
 }
 
 /*
- * Move a page to a newly allocated page
- * The page is locked and all ptes have been successfully removed.
+ * Move a src folio to a newly allocated dst folio.
  *
- * The new page will have replaced the old page if this function
- * is successful.
+ * The src and dst folios are locked and the src folios was unmapped from
+ * the page tables.
+ *
+ * On success, the src folio was replaced by the dst folio.
  *
  * Return value:
  *   < 0 - error code
@@ -1037,34 +1038,30 @@ static int fallback_migrate_folio(struct address_space *mapping,
 static int move_to_new_folio(struct folio *dst, struct folio *src,
 				enum migrate_mode mode)
 {
+	struct address_space *mapping = folio_mapping(src);
 	int rc = -EAGAIN;
-	bool is_lru = !__folio_test_movable(src);
 
 	VM_BUG_ON_FOLIO(!folio_test_locked(src), src);
 	VM_BUG_ON_FOLIO(!folio_test_locked(dst), dst);
 
-	if (likely(is_lru)) {
-		struct address_space *mapping = folio_mapping(src);
-
-		if (!mapping)
-			rc = migrate_folio(mapping, dst, src, mode);
-		else if (mapping_inaccessible(mapping))
-			rc = -EOPNOTSUPP;
-		else if (mapping->a_ops->migrate_folio)
-			/*
-			 * Most folios have a mapping and most filesystems
-			 * provide a migrate_folio callback. Anonymous folios
-			 * are part of swap space which also has its own
-			 * migrate_folio callback. This is the most common path
-			 * for page migration.
-			 */
-			rc = mapping->a_ops->migrate_folio(mapping, dst, src,
-								mode);
-		else
-			rc = fallback_migrate_folio(mapping, dst, src, mode);
+	if (!mapping)
+		rc = migrate_folio(mapping, dst, src, mode);
+	else if (mapping_inaccessible(mapping))
+		rc = -EOPNOTSUPP;
+	else if (mapping->a_ops->migrate_folio)
+		/*
+		 * Most folios have a mapping and most filesystems
+		 * provide a migrate_folio callback. Anonymous folios
+		 * are part of swap space which also has its own
+		 * migrate_folio callback. This is the most common path
+		 * for page migration.
+		 */
+		rc = mapping->a_ops->migrate_folio(mapping, dst, src,
+							mode);
+	else
+		rc = fallback_migrate_folio(mapping, dst, src, mode);
 
-		if (rc != MIGRATEPAGE_SUCCESS)
-			goto out;
+	if (rc == MIGRATEPAGE_SUCCESS) {
 		/*
 		 * For pagecache folios, src->mapping must be cleared before src
 		 * is freed. Anonymous folios must stay anonymous until freed.
@@ -1074,10 +1071,7 @@ static int move_to_new_folio(struct folio *dst, struct folio *src,
 
 		if (likely(!folio_is_zone_device(dst)))
 			flush_dcache_folio(dst);
-	} else {
-		rc = migrate_movable_ops_page(&dst->page, &src->page, mode);
 	}
-out:
 	return rc;
 }
 
@@ -1328,20 +1322,21 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
 	int rc;
 	int old_page_state = 0;
 	struct anon_vma *anon_vma = NULL;
-	bool is_lru = !__folio_test_movable(src);
 	struct list_head *prev;
 
 	__migrate_folio_extract(dst, &old_page_state, &anon_vma);
 	prev = dst->lru.prev;
 	list_del(&dst->lru);
 
+	if (unlikely(__folio_test_movable(src))) {
+		rc = migrate_movable_ops_page(&dst->page, &src->page, mode);
+		goto out_unlock_both;
+	}
+
 	rc = move_to_new_folio(dst, src, mode);
 	if (rc)
 		goto out;
 
-	if (unlikely(!is_lru))
-		goto out_unlock_both;
-
 	/*
 	 * When successful, push dst to LRU immediately: so that if it
 	 * turns out to be an mlocked page, remove_migration_ptes() will
-- 
2.49.0

[PATCH RFC 12/29] mm/zsmalloc: stop using __ClearPageMovable()

[David Hildenbrand]

Instead, let's check in the callbacks if the page was already destroyed,
which can be checked by looking at zpdesc->zspage (see reset_zpdesc()).

If we detect that the page was destroyed:

(1) Fail isolation, just like the migration core would

(2) Fake migration success just like the migration core would

In the putback case there is nothing to do, as we don't do anything just
like the migration core would do.

In the future, we should look into not letting these pages get destroyed
while they are isolated -- and instead delaying that to the
putback/migration call. Add a TODO for that.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/zsmalloc.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c
index f98747aed4330..72c2b7562c511 100644
--- a/mm/zsmalloc.c
+++ b/mm/zsmalloc.c
@@ -876,7 +876,6 @@ static void reset_zpdesc(struct zpdesc *zpdesc)
 {
 	struct page *page = zpdesc_page(zpdesc);
 
-	__ClearPageMovable(page);
 	ClearPagePrivate(page);
 	zpdesc->zspage = NULL;
 	zpdesc->next = NULL;
@@ -1715,10 +1714,11 @@ static void replace_sub_page(struct size_class *class, struct zspage *zspage,
 static bool zs_page_isolate(struct page *page, isolate_mode_t mode)
 {
 	/*
-	 * Page is locked so zspage couldn't be destroyed. For detail, look at
-	 * lock_zspage in free_zspage.
+	 * Page is locked so zspage can't be destroyed concurrently
+	 * (see free_zspage()). But if the page was already destroyed
+	 * (see reset_zpdesc()), refuse isolation here.
 	 */
-	return true;
+	return page_zpdesc(page)->zspage;
 }
 
 static int zs_page_migrate(struct page *newpage, struct page *page,
@@ -1736,6 +1736,13 @@ static int zs_page_migrate(struct page *newpage, struct page *page,
 	unsigned long old_obj, new_obj;
 	unsigned int obj_idx;
 
+	/*
+	 * TODO: nothing prevents a zspage from getting destroyed while
+	 * isolated: we should disallow that and defer it.
+	 */
+	if (!zpdesc->zspage)
+		return MIGRATEPAGE_SUCCESS;
+
 	/* The page is locked, so this pointer must remain valid */
 	zspage = get_zspage(zpdesc);
 	pool = zspage->pool;
-- 
2.49.0

[PATCH RFC 13/29] mm/balloon_compaction: stop using __ClearPageMovable()

[David Hildenbrand]

We can just look at the balloon device (stored in page->private), to see
of the page is still part of the balloon.

As isolated balloon pages cannot get released (they are taken off the
balloon list while isolated), we don't have to worry about this case in
the putback and migration callback. Add a WARN_ON_ONCE for now.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/balloon_compaction.h |  4 +---
 mm/balloon_compaction.c            | 11 +++++++++++
 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
index bfc6e50bd004b..9bce8e9f5018c 100644
--- a/include/linux/balloon_compaction.h
+++ b/include/linux/balloon_compaction.h
@@ -136,10 +136,8 @@ static inline gfp_t balloon_mapping_gfp_mask(void)
  */
 static inline void balloon_page_finalize(struct page *page)
 {
-	if (IS_ENABLED(CONFIG_BALLOON_COMPACTION)) {
-		__ClearPageMovable(page);
+	if (IS_ENABLED(CONFIG_BALLOON_COMPACTION))
 		set_page_private(page, 0);
-	}
 	/* PageOffline is sticky until the page is freed to the buddy. */
 }
 
diff --git a/mm/balloon_compaction.c b/mm/balloon_compaction.c
index ec176bdb8a78b..e4f1a122d786b 100644
--- a/mm/balloon_compaction.c
+++ b/mm/balloon_compaction.c
@@ -206,6 +206,9 @@ static bool balloon_page_isolate(struct page *page, isolate_mode_t mode)
 	struct balloon_dev_info *b_dev_info = balloon_page_device(page);
 	unsigned long flags;
 
+	if (!b_dev_info)
+		return false;
+
 	spin_lock_irqsave(&b_dev_info->pages_lock, flags);
 	list_del(&page->lru);
 	b_dev_info->isolated_pages++;
@@ -219,6 +222,10 @@ static void balloon_page_putback(struct page *page)
 	struct balloon_dev_info *b_dev_info = balloon_page_device(page);
 	unsigned long flags;
 
+	/* Isolated balloon pages cannot get deflated. */
+	if (WARN_ON_ONCE(!b_dev_info))
+		return;
+
 	spin_lock_irqsave(&b_dev_info->pages_lock, flags);
 	list_add(&page->lru, &b_dev_info->pages);
 	b_dev_info->isolated_pages--;
@@ -234,6 +241,10 @@ static int balloon_page_migrate(struct page *newpage, struct page *page,
 	VM_BUG_ON_PAGE(!PageLocked(page), page);
 	VM_BUG_ON_PAGE(!PageLocked(newpage), newpage);
 
+	/* Isolated balloon pages cannot get deflated. */
+	if (WARN_ON_ONCE(!balloon))
+		return -EAGAIN;
+
 	return balloon->migratepage(balloon, newpage, page, mode);
 }
 
-- 
2.49.0

[PATCH RFC 14/29] mm/migrate: remove __ClearPageMovable()

[David Hildenbrand]

Unused, let's remove it.

The Chinese docs in Documentation/translations/zh_CN/mm/page_migration.rst
still mention it, but that whole docs is destined to get outdated and
updated by somebody that actually speaks that language.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/migrate.h |  4 ----
 mm/compaction.c         | 11 -----------
 2 files changed, 15 deletions(-)

diff --git a/include/linux/migrate.h b/include/linux/migrate.h
index c99a00d4ca27d..fb6e9612e9f0b 100644
--- a/include/linux/migrate.h
+++ b/include/linux/migrate.h
@@ -106,16 +106,12 @@ static inline int migrate_huge_page_move_mapping(struct address_space *mapping,
 #ifdef CONFIG_COMPACTION
 bool PageMovable(struct page *page);
 void __SetPageMovable(struct page *page, const struct movable_operations *ops);
-void __ClearPageMovable(struct page *page);
 #else
 static inline bool PageMovable(struct page *page) { return false; }
 static inline void __SetPageMovable(struct page *page,
 		const struct movable_operations *ops)
 {
 }
-static inline void __ClearPageMovable(struct page *page)
-{
-}
 #endif
 
 static inline
diff --git a/mm/compaction.c b/mm/compaction.c
index 17455c5a4be05..889ec696ba96a 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -137,17 +137,6 @@ void __SetPageMovable(struct page *page, const struct movable_operations *mops)
 }
 EXPORT_SYMBOL(__SetPageMovable);
 
-void __ClearPageMovable(struct page *page)
-{
-	VM_BUG_ON_PAGE(!PageMovable(page), page);
-	/*
-	 * This page still has the type of a movable page, but it's
-	 * actually not movable any more.
-	 */
-	page->mapping = (void *)PAGE_MAPPING_MOVABLE;
-}
-EXPORT_SYMBOL(__ClearPageMovable);
-
 /* Do not skip compaction more than 64 times */
 #define COMPACT_MAX_DEFER_SHIFT 6
 
-- 
2.49.0

[PATCH RFC 15/29] mm/migration: remove PageMovable()

[David Hildenbrand]

As __ClearPageMovable() is gone that would have only made
PageMovable()==false but still __PageMovable()==true, now
PageMovable() == __PageMovable().

So we can replace PageMovable() checks by __PageMovable(). In fact,
__PageMovable() cannot change until a page is freed, so we can turn
some PageMovable() into sanity checks for __PageMovable().

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/migrate.h |  2 --
 mm/compaction.c         | 15 ---------------
 mm/migrate.c            | 18 ++++++++++--------
 3 files changed, 10 insertions(+), 25 deletions(-)

diff --git a/include/linux/migrate.h b/include/linux/migrate.h
index fb6e9612e9f0b..204e89eac998f 100644
--- a/include/linux/migrate.h
+++ b/include/linux/migrate.h
@@ -104,10 +104,8 @@ static inline int migrate_huge_page_move_mapping(struct address_space *mapping,
 #endif /* CONFIG_MIGRATION */
 
 #ifdef CONFIG_COMPACTION
-bool PageMovable(struct page *page);
 void __SetPageMovable(struct page *page, const struct movable_operations *ops);
 #else
-static inline bool PageMovable(struct page *page) { return false; }
 static inline void __SetPageMovable(struct page *page,
 		const struct movable_operations *ops)
 {
diff --git a/mm/compaction.c b/mm/compaction.c
index 889ec696ba96a..5c37373017014 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -114,21 +114,6 @@ static unsigned long release_free_list(struct list_head *freepages)
 }
 
 #ifdef CONFIG_COMPACTION
-bool PageMovable(struct page *page)
-{
-	const struct movable_operations *mops;
-
-	VM_BUG_ON_PAGE(!PageLocked(page), page);
-	if (!__PageMovable(page))
-		return false;
-
-	mops = page_movable_ops(page);
-	if (mops)
-		return true;
-
-	return false;
-}
-
 void __SetPageMovable(struct page *page, const struct movable_operations *mops)
 {
 	VM_BUG_ON_PAGE(!PageLocked(page), page);
diff --git a/mm/migrate.c b/mm/migrate.c
index db807f9bbf975..cf92075108f0c 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -87,9 +87,12 @@ bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 		goto out;
 
 	/*
-	 * Check movable flag before taking the page lock because
+	 * Check for movable_ops pages before taking the page lock because
 	 * we use non-atomic bitops on newly allocated page flags so
 	 * unconditionally grabbing the lock ruins page's owner side.
+	 *
+	 * Note that once a page has movable_ops, it will stay that way
+	 * until the page was freed.
 	 */
 	if (unlikely(!__PageMovable(page)))
 		goto out_putfolio;
@@ -108,7 +111,8 @@ bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 	if (unlikely(!folio_trylock(folio)))
 		goto out_putfolio;
 
-	if (!PageMovable(page) || PageIsolated(page))
+	VM_WARN_ON_ONCE_PAGE(!__PageMovable(page), page);
+	if (PageIsolated(page))
 		goto out_no_isolated;
 
 	mops = page_movable_ops(page);
@@ -149,11 +153,10 @@ static void putback_movable_ops_page(struct page *page)
 	 */
 	struct folio *folio = page_folio(page);
 
+	VM_WARN_ON_ONCE_PAGE(!__PageMovable(page), page);
 	VM_WARN_ON_ONCE_PAGE(!PageIsolated(page), page);
 	folio_lock(folio);
-	/* If the page was released by it's owner, there is nothing to do. */
-	if (PageMovable(page))
-		page_movable_ops(page)->putback_page(page);
+	page_movable_ops(page)->putback_page(page);
 	ClearPageIsolated(page);
 	folio_unlock(folio);
 	folio_put(folio);
@@ -189,10 +192,9 @@ static int migrate_movable_ops_page(struct page *dst, struct page *src,
 {
 	int rc = MIGRATEPAGE_SUCCESS;
 
+	VM_WARN_ON_ONCE_PAGE(!__PageMovable(src), src);
 	VM_WARN_ON_ONCE_PAGE(!PageIsolated(src), src);
-	/* If the page was released by it's owner, there is nothing to do. */
-	if (PageMovable(src))
-		rc = page_movable_ops(src)->migrate_page(dst, src, mode);
+	rc = page_movable_ops(src)->migrate_page(dst, src, mode);
 	if (rc == MIGRATEPAGE_SUCCESS)
 		ClearPageIsolated(src);
 	return rc;
-- 
2.49.0

[PATCH RFC 16/29] mm: rename __PageMovable() to page_has_movable_ops()

[David Hildenbrand]

Let's make it clearer that we are talking about movable_ops pages.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/migrate.h    |  2 +-
 include/linux/page-flags.h |  2 +-
 mm/compaction.c            |  7 ++-----
 mm/memory-failure.c        |  4 ++--
 mm/memory_hotplug.c        |  8 +++-----
 mm/migrate.c               |  8 ++++----
 mm/page_alloc.c            |  2 +-
 mm/page_isolation.c        | 10 +++++-----
 8 files changed, 19 insertions(+), 24 deletions(-)

diff --git a/include/linux/migrate.h b/include/linux/migrate.h
index 204e89eac998f..c575778456f97 100644
--- a/include/linux/migrate.h
+++ b/include/linux/migrate.h
@@ -115,7 +115,7 @@ static inline void __SetPageMovable(struct page *page,
 static inline
 const struct movable_operations *page_movable_ops(struct page *page)
 {
-	VM_BUG_ON(!__PageMovable(page));
+	VM_BUG_ON(!page_has_movable_ops(page));
 
 	return (const struct movable_operations *)
 		((unsigned long)page->mapping - PAGE_MAPPING_MOVABLE);
diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index 4fe5ee67535b2..c67163b73c5ec 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -750,7 +750,7 @@ static __always_inline bool __folio_test_movable(const struct folio *folio)
 			PAGE_MAPPING_MOVABLE;
 }
 
-static __always_inline bool __PageMovable(const struct page *page)
+static __always_inline bool page_has_movable_ops(const struct page *page)
 {
 	return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) ==
 				PAGE_MAPPING_MOVABLE;
diff --git a/mm/compaction.c b/mm/compaction.c
index 5c37373017014..f8b7c09e2e48c 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -1056,11 +1056,8 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
 		 * Skip any other type of page
 		 */
 		if (!PageLRU(page)) {
-			/*
-			 * __PageMovable can return false positive so we need
-			 * to verify it under page_lock.
-			 */
-			if (unlikely(__PageMovable(page)) &&
+			/* Isolation will grab the page lock. */
+			if (unlikely(page_has_movable_ops(page)) &&
 					!PageIsolated(page)) {
 				if (locked) {
 					unlock_page_lruvec_irqrestore(locked, flags);
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index b91a33fb6c694..9e2cff1999347 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1388,8 +1388,8 @@ static inline bool HWPoisonHandlable(struct page *page, unsigned long flags)
 	if (PageSlab(page))
 		return false;
 
-	/* Soft offline could migrate non-LRU movable pages */
-	if ((flags & MF_SOFT_OFFLINE) && __PageMovable(page))
+	/* Soft offline could migrate movable_ops pages */
+	if ((flags & MF_SOFT_OFFLINE) && page_has_movable_ops(page))
 		return true;
 
 	return PageLRU(page) || is_free_buddy_page(page);
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 4d864b4fb8916..1d4d0f093af29 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1739,8 +1739,8 @@ bool mhp_range_allowed(u64 start, u64 size, bool need_mapping)
 
 #ifdef CONFIG_MEMORY_HOTREMOVE
 /*
- * Scan pfn range [start,end) to find movable/migratable pages (LRU pages,
- * non-lru movable pages and hugepages). Will skip over most unmovable
+ * Scan pfn range [start,end) to find movable/migratable pages (LRU and
+ * hugetlb folio, movable_ops pages). Will skip over most unmovable
  * pages (esp., pages that can be skipped when offlining), but bail out on
  * definitely unmovable pages.
  *
@@ -1759,9 +1759,7 @@ static int scan_movable_pages(unsigned long start, unsigned long end,
 		struct folio *folio;
 
 		page = pfn_to_page(pfn);
-		if (PageLRU(page))
-			goto found;
-		if (__PageMovable(page))
+		if (PageLRU(page) || page_has_movable_ops(page))
 			goto found;
 
 		/*
diff --git a/mm/migrate.c b/mm/migrate.c
index cf92075108f0c..5f97369eac2f9 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -94,7 +94,7 @@ bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 	 * Note that once a page has movable_ops, it will stay that way
 	 * until the page was freed.
 	 */
-	if (unlikely(!__PageMovable(page)))
+	if (unlikely(!page_has_movable_ops(page)))
 		goto out_putfolio;
 
 	/*
@@ -111,7 +111,7 @@ bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 	if (unlikely(!folio_trylock(folio)))
 		goto out_putfolio;
 
-	VM_WARN_ON_ONCE_PAGE(!__PageMovable(page), page);
+	VM_WARN_ON_ONCE_PAGE(!page_has_movable_ops(page), page);
 	if (PageIsolated(page))
 		goto out_no_isolated;
 
@@ -153,7 +153,7 @@ static void putback_movable_ops_page(struct page *page)
 	 */
 	struct folio *folio = page_folio(page);
 
-	VM_WARN_ON_ONCE_PAGE(!__PageMovable(page), page);
+	VM_WARN_ON_ONCE_PAGE(!page_has_movable_ops(page), page);
 	VM_WARN_ON_ONCE_PAGE(!PageIsolated(page), page);
 	folio_lock(folio);
 	page_movable_ops(page)->putback_page(page);
@@ -192,7 +192,7 @@ static int migrate_movable_ops_page(struct page *dst, struct page *src,
 {
 	int rc = MIGRATEPAGE_SUCCESS;
 
-	VM_WARN_ON_ONCE_PAGE(!__PageMovable(src), src);
+	VM_WARN_ON_ONCE_PAGE(!page_has_movable_ops(src), src);
 	VM_WARN_ON_ONCE_PAGE(!PageIsolated(src), src);
 	rc = page_movable_ops(src)->migrate_page(dst, src, mode);
 	if (rc == MIGRATEPAGE_SUCCESS)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 44e56d31cfeb1..a134b9fa9520e 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2005,7 +2005,7 @@ static bool prep_move_freepages_block(struct zone *zone, struct page *page,
 			 * migration are movable. But we don't actually try
 			 * isolating, as that would be expensive.
 			 */
-			if (PageLRU(page) || __PageMovable(page))
+			if (PageLRU(page) || page_has_movable_ops(page))
 				(*num_movable)++;
 			pfn++;
 		}
diff --git a/mm/page_isolation.c b/mm/page_isolation.c
index ece3bfc56bcd5..b97b965b3ed01 100644
--- a/mm/page_isolation.c
+++ b/mm/page_isolation.c
@@ -21,9 +21,9 @@
  * consequently belong to a single zone.
  *
  * PageLRU check without isolation or lru_lock could race so that
- * MIGRATE_MOVABLE block might include unmovable pages. And __PageMovable
- * check without lock_page also may miss some movable non-lru pages at
- * race condition. So you can't expect this function should be exact.
+ * MIGRATE_MOVABLE block might include unmovable pages. Similarly, pages
+ * with movable_ops can only be identified some time after they were
+ * allocated. So you can't expect this function should be exact.
  *
  * Returns a page without holding a reference. If the caller wants to
  * dereference that page (e.g., dumping), it has to make sure that it
@@ -133,7 +133,7 @@ static struct page *has_unmovable_pages(unsigned long start_pfn, unsigned long e
 		if ((mode == PB_ISOLATE_MODE_MEM_OFFLINE) && PageOffline(page))
 			continue;
 
-		if (__PageMovable(page) || PageLRU(page))
+		if (PageLRU(page) || page_has_movable_ops(page))
 			continue;
 
 		/*
@@ -421,7 +421,7 @@ static int isolate_single_pageblock(unsigned long boundary_pfn,
 			 * proper free and split handling for them.
 			 */
 			VM_WARN_ON_ONCE_PAGE(PageLRU(page), page);
-			VM_WARN_ON_ONCE_PAGE(__PageMovable(page), page);
+			VM_WARN_ON_ONCE_PAGE(page_has_movable_ops(page), page);
 
 			goto failed;
 		}
-- 
2.49.0

[PATCH RFC 17/29] mm/page_isolation: drop __folio_test_movable() check for large folios

[David Hildenbrand]

Currently, we only support migration of individual non-folio pages, so
we can not run into that.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/page_isolation.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/page_isolation.c b/mm/page_isolation.c
index b97b965b3ed01..f72b6cd38b958 100644
--- a/mm/page_isolation.c
+++ b/mm/page_isolation.c
@@ -92,7 +92,7 @@ static struct page *has_unmovable_pages(unsigned long start_pfn, unsigned long e
 				h = size_to_hstate(folio_size(folio));
 				if (h && !hugepage_migration_supported(h))
 					return page;
-			} else if (!folio_test_lru(folio) && !__folio_test_movable(folio)) {
+			} else if (!folio_test_lru(folio)) {
 				return page;
 			}
 
-- 
2.49.0

[PATCH RFC 18/29] mm: remove __folio_test_movable()

[David Hildenbrand]

Convert to page_has_movable_ops(). While at it, cleanup relevant code
a bit.

The data_race() in migrate_folio_unmap() is questionable: we already
hold a page reference, and concurrent modifications can no longer
happen (iow: __ClearPageMovable() no longer exists). Drop it for now,
we'll rework page_has_movable_ops() soon either way to no longer
rely on page->mapping.

Wherever we cast from folio to page now is a clear sign that this
code has to be decoupled.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/page-flags.h |  6 ------
 mm/migrate.c               | 43 ++++++++++++--------------------------
 mm/vmscan.c                |  6 ++++--
 3 files changed, 17 insertions(+), 38 deletions(-)

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index c67163b73c5ec..4c27ebb689e3c 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -744,12 +744,6 @@ static __always_inline bool PageAnon(const struct page *page)
 	return folio_test_anon(page_folio(page));
 }
 
-static __always_inline bool __folio_test_movable(const struct folio *folio)
-{
-	return ((unsigned long)folio->mapping & PAGE_MAPPING_FLAGS) ==
-			PAGE_MAPPING_MOVABLE;
-}
-
 static __always_inline bool page_has_movable_ops(const struct page *page)
 {
 	return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) ==
diff --git a/mm/migrate.c b/mm/migrate.c
index 5f97369eac2f9..37e722fbfee70 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -219,12 +219,7 @@ void putback_movable_pages(struct list_head *l)
 			continue;
 		}
 		list_del(&folio->lru);
-		/*
-		 * We isolated non-lru movable folio so here we can use
-		 * __folio_test_movable because LRU folio's mapping cannot
-		 * have PAGE_MAPPING_MOVABLE.
-		 */
-		if (unlikely(__folio_test_movable(folio))) {
+		if (unlikely(page_has_movable_ops(&folio->page))) {
 			putback_movable_ops_page(&folio->page);
 		} else {
 			node_stat_mod_folio(folio, NR_ISOLATED_ANON +
@@ -237,26 +232,20 @@ void putback_movable_pages(struct list_head *l)
 /* Must be called with an elevated refcount on the non-hugetlb folio */
 bool isolate_folio_to_list(struct folio *folio, struct list_head *list)
 {
-	bool isolated, lru;
-
 	if (folio_test_hugetlb(folio))
 		return folio_isolate_hugetlb(folio, list);
 
-	lru = !__folio_test_movable(folio);
-	if (lru)
-		isolated = folio_isolate_lru(folio);
-	else
-		isolated = isolate_movable_ops_page(&folio->page,
-						    ISOLATE_UNEVICTABLE);
-
-	if (!isolated)
-		return false;
-
-	list_add(&folio->lru, list);
-	if (lru)
+	if (page_has_movable_ops(&folio->page)) {
+		if (!isolate_movable_ops_page(&folio->page,
+					      ISOLATE_UNEVICTABLE))
+			return false;
+	} else {
+		if (!folio_isolate_lru(folio))
+			return false;
 		node_stat_add_folio(folio, NR_ISOLATED_ANON +
 				    folio_is_file_lru(folio));
-
+	}
+	list_add(&folio->lru, list);
 	return true;
 }
 
@@ -1140,12 +1129,7 @@ static void migrate_folio_undo_dst(struct folio *dst, bool locked,
 static void migrate_folio_done(struct folio *src,
 			       enum migrate_reason reason)
 {
-	/*
-	 * Compaction can migrate also non-LRU pages which are
-	 * not accounted to NR_ISOLATED_*. They can be recognized
-	 * as __folio_test_movable
-	 */
-	if (likely(!__folio_test_movable(src)) && reason != MR_DEMOTION)
+	if (likely(!page_has_movable_ops(&src->page)) && reason != MR_DEMOTION)
 		mod_node_page_state(folio_pgdat(src), NR_ISOLATED_ANON +
 				    folio_is_file_lru(src), -folio_nr_pages(src));
 
@@ -1164,7 +1148,6 @@ static int migrate_folio_unmap(new_folio_t get_new_folio,
 	int rc = -EAGAIN;
 	int old_page_state = 0;
 	struct anon_vma *anon_vma = NULL;
-	bool is_lru = data_race(!__folio_test_movable(src));
 	bool locked = false;
 	bool dst_locked = false;
 
@@ -1265,7 +1248,7 @@ static int migrate_folio_unmap(new_folio_t get_new_folio,
 		goto out;
 	dst_locked = true;
 
-	if (unlikely(!is_lru)) {
+	if (unlikely(page_has_movable_ops(&dst->page))) {
 		__migrate_folio_record(dst, old_page_state, anon_vma);
 		return MIGRATEPAGE_UNMAP;
 	}
@@ -1330,7 +1313,7 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
 	prev = dst->lru.prev;
 	list_del(&dst->lru);
 
-	if (unlikely(__folio_test_movable(src))) {
+	if (unlikely(page_has_movable_ops(&src->page))) {
 		rc = migrate_movable_ops_page(&dst->page, &src->page, mode);
 		goto out_unlock_both;
 	}
diff --git a/mm/vmscan.c b/mm/vmscan.c
index efc818a0bbecb..e5d04c98af976 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1651,9 +1651,11 @@ unsigned int reclaim_clean_pages_from_list(struct zone *zone,
 	unsigned int noreclaim_flag;
 
 	list_for_each_entry_safe(folio, next, folio_list, lru) {
+		/* TODO: these pages should not even appear in this list. */
+		if (page_has_movable_ops(&folio->page))
+			continue;
 		if (!folio_test_hugetlb(folio) && folio_is_file_lru(folio) &&
-		    !folio_test_dirty(folio) && !__folio_test_movable(folio) &&
-		    !folio_test_unevictable(folio)) {
+		    !folio_test_dirty(folio) && !folio_test_unevictable(folio)) {
 			folio_clear_active(folio);
 			list_move(&folio->lru, &clean_folios);
 		}
-- 
2.49.0

[PATCH RFC 19/29] mm: stop storing migration_ops in page->mapping

[David Hildenbrand]

... instead, look them up statically based on the page type. Maybe in the
future we want a registration interface? At least for now, it can be
easily handled using the two page types that actually support page
migration.

The remaining usage of page->mapping is to flag such pages as actually
being movable (having movable_ops), which we will change next.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/balloon_compaction.h |  2 +-
 include/linux/migrate.h            | 14 ++------------
 include/linux/zsmalloc.h           |  2 ++
 mm/balloon_compaction.c            |  1 -
 mm/compaction.c                    |  5 ++---
 mm/migrate.c                       | 23 +++++++++++++++++++++++
 mm/zpdesc.h                        |  5 ++---
 mm/zsmalloc.c                      |  8 +++-----
 8 files changed, 35 insertions(+), 25 deletions(-)

diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
index 9bce8e9f5018c..a8a1706cc56f3 100644
--- a/include/linux/balloon_compaction.h
+++ b/include/linux/balloon_compaction.h
@@ -92,7 +92,7 @@ static inline void balloon_page_insert(struct balloon_dev_info *balloon,
 				       struct page *page)
 {
 	__SetPageOffline(page);
-	__SetPageMovable(page, &balloon_mops);
+	__SetPageMovable(page);
 	set_page_private(page, (unsigned long)balloon);
 	list_add(&page->lru, &balloon->pages);
 }
diff --git a/include/linux/migrate.h b/include/linux/migrate.h
index c575778456f97..64639dab92b91 100644
--- a/include/linux/migrate.h
+++ b/include/linux/migrate.h
@@ -104,23 +104,13 @@ static inline int migrate_huge_page_move_mapping(struct address_space *mapping,
 #endif /* CONFIG_MIGRATION */
 
 #ifdef CONFIG_COMPACTION
-void __SetPageMovable(struct page *page, const struct movable_operations *ops);
+void __SetPageMovable(struct page *page);
 #else
-static inline void __SetPageMovable(struct page *page,
-		const struct movable_operations *ops)
+static inline void __SetPageMovable(struct page *page)
 {
 }
 #endif
 
-static inline
-const struct movable_operations *page_movable_ops(struct page *page)
-{
-	VM_BUG_ON(!page_has_movable_ops(page));
-
-	return (const struct movable_operations *)
-		((unsigned long)page->mapping - PAGE_MAPPING_MOVABLE);
-}
-
 #ifdef CONFIG_NUMA_BALANCING
 int migrate_misplaced_folio_prepare(struct folio *folio,
 		struct vm_area_struct *vma, int node);
diff --git a/include/linux/zsmalloc.h b/include/linux/zsmalloc.h
index 13e9cc5490f71..f3ccff2d966cd 100644
--- a/include/linux/zsmalloc.h
+++ b/include/linux/zsmalloc.h
@@ -46,4 +46,6 @@ void zs_obj_read_end(struct zs_pool *pool, unsigned long handle,
 void zs_obj_write(struct zs_pool *pool, unsigned long handle,
 		  void *handle_mem, size_t mem_len);
 
+extern const struct movable_operations zsmalloc_mops;
+
 #endif
diff --git a/mm/balloon_compaction.c b/mm/balloon_compaction.c
index e4f1a122d786b..2a4a649805c11 100644
--- a/mm/balloon_compaction.c
+++ b/mm/balloon_compaction.c
@@ -253,6 +253,5 @@ const struct movable_operations balloon_mops = {
 	.isolate_page = balloon_page_isolate,
 	.putback_page = balloon_page_putback,
 };
-EXPORT_SYMBOL_GPL(balloon_mops);
 
 #endif /* CONFIG_BALLOON_COMPACTION */
diff --git a/mm/compaction.c b/mm/compaction.c
index f8b7c09e2e48c..a10f23df9396e 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -114,11 +114,10 @@ static unsigned long release_free_list(struct list_head *freepages)
 }
 
 #ifdef CONFIG_COMPACTION
-void __SetPageMovable(struct page *page, const struct movable_operations *mops)
+void __SetPageMovable(struct page *page)
 {
 	VM_BUG_ON_PAGE(!PageLocked(page), page);
-	VM_BUG_ON_PAGE((unsigned long)mops & PAGE_MAPPING_MOVABLE, page);
-	page->mapping = (void *)((unsigned long)mops | PAGE_MAPPING_MOVABLE);
+	page->mapping = (void *)(PAGE_MAPPING_MOVABLE);
 }
 EXPORT_SYMBOL(__SetPageMovable);
 
diff --git a/mm/migrate.c b/mm/migrate.c
index 37e722fbfee70..a36030de94f30 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -43,6 +43,8 @@
 #include <linux/sched/sysctl.h>
 #include <linux/memory-tiers.h>
 #include <linux/pagewalk.h>
+#include <linux/balloon_compaction.h>
+#include <linux/zsmalloc.h>
 
 #include <asm/tlbflush.h>
 
@@ -51,6 +53,27 @@
 #include "internal.h"
 #include "swap.h"
 
+static const struct movable_operations *page_movable_ops(struct page *page)
+{
+	VM_WARN_ON_ONCE_PAGE(!page_has_movable_ops(page), page);
+
+	/*
+	 * If we enable page migration for a page of a certain type by marking
+	 * it as movable, the page type must be sticky until the page gets freed
+	 * back to the buddy.
+	 */
+#ifdef CONFIG_BALLOON_COMPACTION
+	if (PageOffline(page))
+		/* Only balloon compaction sets PageOffline pages movable. */
+		return &balloon_mops;
+#endif /* CONFIG_BALLOON_COMPACTION */
+#ifdef CONFIG_ZSMALLOC
+	if (PageZsmalloc(page))
+		return &zsmalloc_mops;
+#endif /* CONFIG_ZSMALLOC */
+	return NULL;
+}
+
 /**
  * isolate_movable_ops_page - isolate a movable_ops page for migration
  * @page: The page.
diff --git a/mm/zpdesc.h b/mm/zpdesc.h
index 5763f36039736..6855d9e2732d8 100644
--- a/mm/zpdesc.h
+++ b/mm/zpdesc.h
@@ -152,10 +152,9 @@ static inline struct zpdesc *pfn_zpdesc(unsigned long pfn)
 	return page_zpdesc(pfn_to_page(pfn));
 }
 
-static inline void __zpdesc_set_movable(struct zpdesc *zpdesc,
-					const struct movable_operations *mops)
+static inline void __zpdesc_set_movable(struct zpdesc *zpdesc)
 {
-	__SetPageMovable(zpdesc_page(zpdesc), mops);
+	__SetPageMovable(zpdesc_page(zpdesc));
 }
 
 static inline void __zpdesc_set_zsmalloc(struct zpdesc *zpdesc)
diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c
index 72c2b7562c511..7192196b9421d 100644
--- a/mm/zsmalloc.c
+++ b/mm/zsmalloc.c
@@ -1684,8 +1684,6 @@ static void lock_zspage(struct zspage *zspage)
 
 #ifdef CONFIG_COMPACTION
 
-static const struct movable_operations zsmalloc_mops;
-
 static void replace_sub_page(struct size_class *class, struct zspage *zspage,
 				struct zpdesc *newzpdesc, struct zpdesc *oldzpdesc)
 {
@@ -1708,7 +1706,7 @@ static void replace_sub_page(struct size_class *class, struct zspage *zspage,
 	set_first_obj_offset(newzpdesc, first_obj_offset);
 	if (unlikely(ZsHugePage(zspage)))
 		newzpdesc->handle = oldzpdesc->handle;
-	__zpdesc_set_movable(newzpdesc, &zsmalloc_mops);
+	__zpdesc_set_movable(newzpdesc);
 }
 
 static bool zs_page_isolate(struct page *page, isolate_mode_t mode)
@@ -1815,7 +1813,7 @@ static void zs_page_putback(struct page *page)
 {
 }
 
-static const struct movable_operations zsmalloc_mops = {
+const struct movable_operations zsmalloc_mops = {
 	.isolate_page = zs_page_isolate,
 	.migrate_page = zs_page_migrate,
 	.putback_page = zs_page_putback,
@@ -1878,7 +1876,7 @@ static void SetZsPageMovable(struct zs_pool *pool, struct zspage *zspage)
 
 	do {
 		WARN_ON(!zpdesc_trylock(zpdesc));
-		__zpdesc_set_movable(zpdesc, &zsmalloc_mops);
+		__zpdesc_set_movable(zpdesc);
 		zpdesc_unlock(zpdesc);
 	} while ((zpdesc = get_next_zpdesc(zpdesc)) != NULL);
 }
-- 
2.49.0

[PATCH RFC 20/29] mm: convert "movable" flag in page->mapping to a page flag

[David Hildenbrand]

Instead, let's use a page flag. As the page flag can result in
false-positives, glue it to the page types for which we
support/implement movable_ops page migration.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/balloon_compaction.h |  2 +-
 include/linux/migrate.h            |  8 -----
 include/linux/page-flags.h         | 52 ++++++++++++++++++++++++------
 mm/compaction.c                    |  6 ----
 mm/zpdesc.h                        |  2 +-
 5 files changed, 44 insertions(+), 26 deletions(-)

diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
index a8a1706cc56f3..b222b0737c466 100644
--- a/include/linux/balloon_compaction.h
+++ b/include/linux/balloon_compaction.h
@@ -92,7 +92,7 @@ static inline void balloon_page_insert(struct balloon_dev_info *balloon,
 				       struct page *page)
 {
 	__SetPageOffline(page);
-	__SetPageMovable(page);
+	SetPageMovableOps(page);
 	set_page_private(page, (unsigned long)balloon);
 	list_add(&page->lru, &balloon->pages);
 }
diff --git a/include/linux/migrate.h b/include/linux/migrate.h
index 64639dab92b91..314cbeba1bb91 100644
--- a/include/linux/migrate.h
+++ b/include/linux/migrate.h
@@ -103,14 +103,6 @@ static inline int migrate_huge_page_move_mapping(struct address_space *mapping,
 
 #endif /* CONFIG_MIGRATION */
 
-#ifdef CONFIG_COMPACTION
-void __SetPageMovable(struct page *page);
-#else
-static inline void __SetPageMovable(struct page *page)
-{
-}
-#endif
-
 #ifdef CONFIG_NUMA_BALANCING
 int migrate_misplaced_folio_prepare(struct folio *folio,
 		struct vm_area_struct *vma, int node);
diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index 4c27ebb689e3c..016a6e6fa428a 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -170,6 +170,11 @@ enum pageflags {
 	/* non-lru isolated movable page */
 	PG_isolated = PG_reclaim,
 
+#ifdef CONFIG_MIGRATION
+	/* this is a movable_ops page (for selected typed pages only) */
+	PG_movable_ops = PG_uptodate,
+#endif
+
 	/* Only valid for buddy pages. Used to track pages that are reported */
 	PG_reported = PG_uptodate,
 
@@ -698,9 +703,6 @@ PAGEFLAG_FALSE(VmemmapSelfHosted, vmemmap_self_hosted)
  * bit; and then folio->mapping points, not to an anon_vma, but to a private
  * structure which KSM associates with that merged page.  See ksm.h.
  *
- * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is used for non-lru movable
- * page and then folio->mapping points to a struct movable_operations.
- *
  * Please note that, confusingly, "folio_mapping" refers to the inode
  * address_space which maps the folio from disk; whereas "folio_mapped"
  * refers to user virtual address space into which the folio is mapped.
@@ -743,13 +745,6 @@ static __always_inline bool PageAnon(const struct page *page)
 {
 	return folio_test_anon(page_folio(page));
 }
-
-static __always_inline bool page_has_movable_ops(const struct page *page)
-{
-	return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) ==
-				PAGE_MAPPING_MOVABLE;
-}
-
 #ifdef CONFIG_KSM
 /*
  * A KSM page is one of those write-protected "shared pages" or "merged pages"
@@ -1133,6 +1128,43 @@ bool is_free_buddy_page(const struct page *page);
 
 PAGEFLAG(Isolated, isolated, PF_ANY);
 
+#ifdef CONFIG_MIGRATION
+/*
+ * This page is migratable through movable_ops (for selected typed pages
+ * only).
+ *
+ * Page migration of such pages might fail, for example, if the page is
+ * already isolated by somebody else, or if the page is about to get freed.
+ *
+ * While a subsystem might set selected typed pages that support page migration
+ * as being movable through movable_ops, it must never clear this flag.
+ *
+ * This flag is only cleared when the page is freed back to the buddy.
+ *
+ * Only selected page types support this flag (see page_movable_ops()) and
+ * the flag might be used in other context for other pages. Always use
+ * page_has_movable_ops() instead.
+ */
+PAGEFLAG(MovableOps, movable_ops, PF_NO_TAIL);
+#else
+PAGEFLAG_FALSE(MovableOps, movable_ops);
+#endif
+
+/**
+ * page_has_movable_ops - test for a movable_ops page
+ * @page The page to test.
+ *
+ * Test whether this is a movable_ops page. Such pages will stay that
+ * way until freed.
+ *
+ * Returns true if this is a movable_ops page, otherwise false.
+ */
+static inline bool page_has_movable_ops(const struct page *page)
+{
+	return PageMovableOps(page) &&
+	       (PageOffline(page) || PageZsmalloc(page));
+}
+
 static __always_inline int PageAnonExclusive(const struct page *page)
 {
 	VM_BUG_ON_PGFLAGS(!PageAnon(page), page);
diff --git a/mm/compaction.c b/mm/compaction.c
index a10f23df9396e..86d671a520e91 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -114,12 +114,6 @@ static unsigned long release_free_list(struct list_head *freepages)
 }
 
 #ifdef CONFIG_COMPACTION
-void __SetPageMovable(struct page *page)
-{
-	VM_BUG_ON_PAGE(!PageLocked(page), page);
-	page->mapping = (void *)(PAGE_MAPPING_MOVABLE);
-}
-EXPORT_SYMBOL(__SetPageMovable);
 
 /* Do not skip compaction more than 64 times */
 #define COMPACT_MAX_DEFER_SHIFT 6
diff --git a/mm/zpdesc.h b/mm/zpdesc.h
index 6855d9e2732d8..25bf5ea0beb83 100644
--- a/mm/zpdesc.h
+++ b/mm/zpdesc.h
@@ -154,7 +154,7 @@ static inline struct zpdesc *pfn_zpdesc(unsigned long pfn)
 
 static inline void __zpdesc_set_movable(struct zpdesc *zpdesc)
 {
-	__SetPageMovable(zpdesc_page(zpdesc));
+	SetPageMovableOps(zpdesc_page(zpdesc));
 }
 
 static inline void __zpdesc_set_zsmalloc(struct zpdesc *zpdesc)
-- 
2.49.0

[PATCH RFC 21/29] mm: rename PG_isolated to PG_movable_ops_isolated

[David Hildenbrand]

Let's rename the flag to make it clearer where it applies (not folios
...).

While at it, define the flag only with CONFIG_MIGRATION.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/page-flags.h | 16 +++++++++++-----
 mm/compaction.c            |  2 +-
 mm/migrate.c               | 14 +++++++-------
 3 files changed, 19 insertions(+), 13 deletions(-)

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index 016a6e6fa428a..aa48b05536bca 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -167,10 +167,9 @@ enum pageflags {
 	/* Remapped by swiotlb-xen. */
 	PG_xen_remapped = PG_owner_priv_1,
 
-	/* non-lru isolated movable page */
-	PG_isolated = PG_reclaim,
-
 #ifdef CONFIG_MIGRATION
+	/* movable_ops page that is isolated for migration */
+	PG_movable_ops_isolated = PG_reclaim,
 	/* this is a movable_ops page (for selected typed pages only) */
 	PG_movable_ops = PG_uptodate,
 #endif
@@ -1126,8 +1125,6 @@ static inline bool folio_contain_hwpoisoned_page(struct folio *folio)
 
 bool is_free_buddy_page(const struct page *page);
 
-PAGEFLAG(Isolated, isolated, PF_ANY);
-
 #ifdef CONFIG_MIGRATION
 /*
  * This page is migratable through movable_ops (for selected typed pages
@@ -1146,8 +1143,17 @@ PAGEFLAG(Isolated, isolated, PF_ANY);
  * page_has_movable_ops() instead.
  */
 PAGEFLAG(MovableOps, movable_ops, PF_NO_TAIL);
+/*
+ * A movable_ops page has this flag set while it is isolated for migration.
+ * This flag primarily protects against concurrent migration attempts.
+ *
+ * Once migration ended (success or failure), the flag is cleared. The
+ * flag is managed by the migration core.
+ */
+PAGEFLAG(MovableOpsIsolated, movable_ops_isolated, PF_NO_TAIL);
 #else
 PAGEFLAG_FALSE(MovableOps, movable_ops);
+PAGEFLAG_FALSE(MovableOpsIsolated, movable_ops_isolated);
 #endif
 
 /**
diff --git a/mm/compaction.c b/mm/compaction.c
index 86d671a520e91..809e4395cadfc 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -1051,7 +1051,7 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
 		if (!PageLRU(page)) {
 			/* Isolation will grab the page lock. */
 			if (unlikely(page_has_movable_ops(page)) &&
-					!PageIsolated(page)) {
+			    !PageMovableOpsIsolated(page)) {
 				if (locked) {
 					unlock_page_lruvec_irqrestore(locked, flags);
 					locked = NULL;
diff --git a/mm/migrate.c b/mm/migrate.c
index a36030de94f30..f87a998b696e2 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -135,7 +135,7 @@ bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 		goto out_putfolio;
 
 	VM_WARN_ON_ONCE_PAGE(!page_has_movable_ops(page), page);
-	if (PageIsolated(page))
+	if (PageMovableOpsIsolated(page))
 		goto out_no_isolated;
 
 	mops = page_movable_ops(page);
@@ -146,8 +146,8 @@ bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
 		goto out_no_isolated;
 
 	/* Driver shouldn't use the isolated flag */
-	VM_WARN_ON_ONCE_PAGE(PageIsolated(page), page);
-	SetPageIsolated(page);
+	VM_WARN_ON_ONCE_PAGE(PageMovableOpsIsolated(page), page);
+	SetPageMovableOpsIsolated(page);
 	folio_unlock(folio);
 
 	return true;
@@ -177,10 +177,10 @@ static void putback_movable_ops_page(struct page *page)
 	struct folio *folio = page_folio(page);
 
 	VM_WARN_ON_ONCE_PAGE(!page_has_movable_ops(page), page);
-	VM_WARN_ON_ONCE_PAGE(!PageIsolated(page), page);
+	VM_WARN_ON_ONCE_PAGE(!PageMovableOpsIsolated(page), page);
 	folio_lock(folio);
 	page_movable_ops(page)->putback_page(page);
-	ClearPageIsolated(page);
+	ClearPageMovableOpsIsolated(page);
 	folio_unlock(folio);
 	folio_put(folio);
 }
@@ -216,10 +216,10 @@ static int migrate_movable_ops_page(struct page *dst, struct page *src,
 	int rc = MIGRATEPAGE_SUCCESS;
 
 	VM_WARN_ON_ONCE_PAGE(!page_has_movable_ops(src), src);
-	VM_WARN_ON_ONCE_PAGE(!PageIsolated(src), src);
+	VM_WARN_ON_ONCE_PAGE(!PageMovableOpsIsolated(src), src);
 	rc = page_movable_ops(src)->migrate_page(dst, src, mode);
 	if (rc == MIGRATEPAGE_SUCCESS)
-		ClearPageIsolated(src);
+		ClearPageMovableOpsIsolated(src);
 	return rc;
 }
 
-- 
2.49.0

[PATCH RFC 22/29] mm/page-flags: rename PAGE_MAPPING_MOVABLE to PAGE_MAPPING_ANON_KSM

[David Hildenbrand]

KSM is the only remaining user, let's rename the flag. While at it,
adjust to remaining page -> folio in the doc.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/page-flags.h | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index aa48b05536bca..abed972e902e1 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -697,10 +697,10 @@ PAGEFLAG_FALSE(VmemmapSelfHosted, vmemmap_self_hosted)
  * folio->mapping points to its anon_vma, not to a struct address_space;
  * with the PAGE_MAPPING_ANON bit set to distinguish it.  See rmap.h.
  *
- * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
- * the PAGE_MAPPING_MOVABLE bit may be set along with the PAGE_MAPPING_ANON
+ * On an anonymous folio in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
+ * the PAGE_MAPPING_ANON_KSM bit may be set along with the PAGE_MAPPING_ANON
  * bit; and then folio->mapping points, not to an anon_vma, but to a private
- * structure which KSM associates with that merged page.  See ksm.h.
+ * structure which KSM associates with that merged folio.  See ksm.h.
  *
  * Please note that, confusingly, "folio_mapping" refers to the inode
  * address_space which maps the folio from disk; whereas "folio_mapped"
@@ -714,9 +714,9 @@ PAGEFLAG_FALSE(VmemmapSelfHosted, vmemmap_self_hosted)
  * See mm/slab.h.
  */
 #define PAGE_MAPPING_ANON	0x1
-#define PAGE_MAPPING_MOVABLE	0x2
-#define PAGE_MAPPING_KSM	(PAGE_MAPPING_ANON | PAGE_MAPPING_MOVABLE)
-#define PAGE_MAPPING_FLAGS	(PAGE_MAPPING_ANON | PAGE_MAPPING_MOVABLE)
+#define PAGE_MAPPING_ANON_KSM	0x2
+#define PAGE_MAPPING_KSM	(PAGE_MAPPING_ANON | PAGE_MAPPING_ANON_KSM)
+#define PAGE_MAPPING_FLAGS	(PAGE_MAPPING_ANON | PAGE_MAPPING_ANON_KSM)
 
 static __always_inline bool folio_mapping_flags(const struct folio *folio)
 {
-- 
2.49.0

[PATCH RFC 23/29] mm/page-alloc: remove PageMappingFlags()

[David Hildenbrand]

We can now simply check for PageAnon() and remove PageMappingFlags().

... and while at it, use the folio instead and operate on
folio->mapping.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/page-flags.h | 5 -----
 mm/page_alloc.c            | 7 +++----
 2 files changed, 3 insertions(+), 9 deletions(-)

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index abed972e902e1..f539bd5e14200 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -723,11 +723,6 @@ static __always_inline bool folio_mapping_flags(const struct folio *folio)
 	return ((unsigned long)folio->mapping & PAGE_MAPPING_FLAGS) != 0;
 }
 
-static __always_inline bool PageMappingFlags(const struct page *page)
-{
-	return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) != 0;
-}
-
 static __always_inline bool folio_test_anon(const struct folio *folio)
 {
 	return ((unsigned long)folio->mapping & PAGE_MAPPING_ANON) != 0;
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index a134b9fa9520e..a0ebcc5f54bb2 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1375,10 +1375,9 @@ __always_inline bool free_pages_prepare(struct page *page,
 			(page + i)->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
 		}
 	}
-	if (PageMappingFlags(page)) {
-		if (PageAnon(page))
-			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
-		page->mapping = NULL;
+	if (folio_test_anon(folio)) {
+		mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
+		folio->mapping = NULL;
 	}
 	if (unlikely(page_has_type(page)))
 		page->page_type = UINT_MAX;
-- 
2.49.0

[PATCH RFC 24/29] mm/page-flags: remove folio_mapping_flags()

[David Hildenbrand]

It's unused and the page counterpart is gone, so let's remove it.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/page-flags.h | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index f539bd5e14200..b42986a578b71 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -718,11 +718,6 @@ PAGEFLAG_FALSE(VmemmapSelfHosted, vmemmap_self_hosted)
 #define PAGE_MAPPING_KSM	(PAGE_MAPPING_ANON | PAGE_MAPPING_ANON_KSM)
 #define PAGE_MAPPING_FLAGS	(PAGE_MAPPING_ANON | PAGE_MAPPING_ANON_KSM)
 
-static __always_inline bool folio_mapping_flags(const struct folio *folio)
-{
-	return ((unsigned long)folio->mapping & PAGE_MAPPING_FLAGS) != 0;
-}
-
 static __always_inline bool folio_test_anon(const struct folio *folio)
 {
 	return ((unsigned long)folio->mapping & PAGE_MAPPING_ANON) != 0;
-- 
2.49.0

[PATCH RFC 25/29] mm: simplify folio_expected_ref_count()

[David Hildenbrand]

Now that PAGE_MAPPING_MOVABLE is gone, we can simplify and rely on the
folio_test_anon() test only.

... but staring at the users, this function should never even have been
called on movable_ops pages. E.g.,
* __buffer_migrate_folio() does not make sense for them
* folio_migrate_mapping() does not make sense for them
* migrate_huge_page_move_mapping() does not make sense for them
* __migrate_folio() does not make sense for them
* ... and khugepaged should never stumble over them

Let's simply refuse typed pages (which includes slab) except hugetlb,
and WARN.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/mm.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/include/linux/mm.h b/include/linux/mm.h
index 98a606908307b..61da588dda892 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2167,13 +2167,13 @@ static inline int folio_expected_ref_count(const struct folio *folio)
 	const int order = folio_order(folio);
 	int ref_count = 0;
 
-	if (WARN_ON_ONCE(folio_test_slab(folio)))
+	if (WARN_ON_ONCE(page_has_type(&folio->page) && !folio_test_hugetlb(folio)))
 		return 0;
 
 	if (folio_test_anon(folio)) {
 		/* One reference per page from the swapcache. */
 		ref_count += folio_test_swapcache(folio) << order;
-	} else if (!((unsigned long)folio->mapping & PAGE_MAPPING_FLAGS)) {
+	} else {
 		/* One reference per page from the pagecache. */
 		ref_count += !!folio->mapping << order;
 		/* One reference from PG_private. */
-- 
2.49.0

[PATCH RFC 26/29] mm: rename PAGE_MAPPING_* to FOLIO_MAPPING_*

[David Hildenbrand]

Now that the mapping flags are only used for folios, let's rename the
defines.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 fs/proc/page.c             |  4 ++--
 include/linux/fs.h         |  2 +-
 include/linux/mm_types.h   |  1 -
 include/linux/page-flags.h | 20 ++++++++++----------
 include/linux/pagemap.h    |  2 +-
 mm/gup.c                   |  4 ++--
 mm/internal.h              |  2 +-
 mm/ksm.c                   |  4 ++--
 mm/rmap.c                  | 16 ++++++++--------
 mm/util.c                  |  6 +++---
 10 files changed, 30 insertions(+), 31 deletions(-)

diff --git a/fs/proc/page.c b/fs/proc/page.c
index 999af26c72985..0cdc78c0d23fa 100644
--- a/fs/proc/page.c
+++ b/fs/proc/page.c
@@ -149,7 +149,7 @@ u64 stable_page_flags(const struct page *page)
 
 	k = folio->flags;
 	mapping = (unsigned long)folio->mapping;
-	is_anon = mapping & PAGE_MAPPING_ANON;
+	is_anon = mapping & FOLIO_MAPPING_ANON;
 
 	/*
 	 * pseudo flags for the well known (anonymous) memory mapped pages
@@ -158,7 +158,7 @@ u64 stable_page_flags(const struct page *page)
 		u |= 1 << KPF_MMAP;
 	if (is_anon) {
 		u |= 1 << KPF_ANON;
-		if (mapping & PAGE_MAPPING_KSM)
+		if (mapping & FOLIO_MAPPING_KSM)
 			u |= 1 << KPF_KSM;
 	}
 
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 93ee0d2d6f1ad..f00a073c35c45 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -524,7 +524,7 @@ struct address_space {
 	/*
 	 * On most architectures that alignment is already the case; but
 	 * must be enforced here for CRIS, to let the least significant bit
-	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
+	 * of struct folio's "mapping" pointer be used for FOLIO_MAPPING_ANON.
 	 */
 
 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index d6b91e8a66d6d..f139f47696f63 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -105,7 +105,6 @@ struct page {
 					unsigned int order;
 				};
 			};
-			/* See page-flags.h for PAGE_MAPPING_FLAGS */
 			struct address_space *mapping;
 			union {
 				pgoff_t __folio_index;		/* Our offset within mapping. */
diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index b42986a578b71..23b1e458dfeda 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -695,10 +695,10 @@ PAGEFLAG_FALSE(VmemmapSelfHosted, vmemmap_self_hosted)
 /*
  * On an anonymous folio mapped into a user virtual memory area,
  * folio->mapping points to its anon_vma, not to a struct address_space;
- * with the PAGE_MAPPING_ANON bit set to distinguish it.  See rmap.h.
+ * with the FOLIO_MAPPING_ANON bit set to distinguish it.  See rmap.h.
  *
  * On an anonymous folio in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
- * the PAGE_MAPPING_ANON_KSM bit may be set along with the PAGE_MAPPING_ANON
+ * the FOLIO_MAPPING_ANON_KSM bit may be set along with the FOLIO_MAPPING_ANON
  * bit; and then folio->mapping points, not to an anon_vma, but to a private
  * structure which KSM associates with that merged folio.  See ksm.h.
  *
@@ -713,21 +713,21 @@ PAGEFLAG_FALSE(VmemmapSelfHosted, vmemmap_self_hosted)
  * false before calling the following functions (e.g., folio_test_anon).
  * See mm/slab.h.
  */
-#define PAGE_MAPPING_ANON	0x1
-#define PAGE_MAPPING_ANON_KSM	0x2
-#define PAGE_MAPPING_KSM	(PAGE_MAPPING_ANON | PAGE_MAPPING_ANON_KSM)
-#define PAGE_MAPPING_FLAGS	(PAGE_MAPPING_ANON | PAGE_MAPPING_ANON_KSM)
+#define FOLIO_MAPPING_ANON	0x1
+#define FOLIO_MAPPING_ANON_KSM	0x2
+#define FOLIO_MAPPING_KSM	(FOLIO_MAPPING_ANON | FOLIO_MAPPING_ANON_KSM)
+#define FOLIO_MAPPING_FLAGS	(FOLIO_MAPPING_ANON | FOLIO_MAPPING_ANON_KSM)
 
 static __always_inline bool folio_test_anon(const struct folio *folio)
 {
-	return ((unsigned long)folio->mapping & PAGE_MAPPING_ANON) != 0;
+	return ((unsigned long)folio->mapping & FOLIO_MAPPING_ANON) != 0;
 }
 
 static __always_inline bool PageAnonNotKsm(const struct page *page)
 {
 	unsigned long flags = (unsigned long)page_folio(page)->mapping;
 
-	return (flags & PAGE_MAPPING_FLAGS) == PAGE_MAPPING_ANON;
+	return (flags & FOLIO_MAPPING_FLAGS) == FOLIO_MAPPING_ANON;
 }
 
 static __always_inline bool PageAnon(const struct page *page)
@@ -743,8 +743,8 @@ static __always_inline bool PageAnon(const struct page *page)
  */
 static __always_inline bool folio_test_ksm(const struct folio *folio)
 {
-	return ((unsigned long)folio->mapping & PAGE_MAPPING_FLAGS) ==
-				PAGE_MAPPING_KSM;
+	return ((unsigned long)folio->mapping & FOLIO_MAPPING_FLAGS) ==
+				FOLIO_MAPPING_KSM;
 }
 #else
 FOLIO_TEST_FLAG_FALSE(ksm)
diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
index e63fbfbd5b0f3..10a222e68b851 100644
--- a/include/linux/pagemap.h
+++ b/include/linux/pagemap.h
@@ -502,7 +502,7 @@ static inline pgoff_t mapping_align_index(struct address_space *mapping,
 static inline bool mapping_large_folio_support(struct address_space *mapping)
 {
 	/* AS_FOLIO_ORDER is only reasonable for pagecache folios */
-	VM_WARN_ONCE((unsigned long)mapping & PAGE_MAPPING_ANON,
+	VM_WARN_ONCE((unsigned long)mapping & FOLIO_MAPPING_ANON,
 			"Anonymous mapping always supports large folio");
 
 	return mapping_max_folio_order(mapping) > 0;
diff --git a/mm/gup.c b/mm/gup.c
index 6888e871a74a9..76ef1839a126a 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -2804,9 +2804,9 @@ static bool gup_fast_folio_allowed(struct folio *folio, unsigned int flags)
 		return false;
 
 	/* Anonymous folios pose no problem. */
-	mapping_flags = (unsigned long)mapping & PAGE_MAPPING_FLAGS;
+	mapping_flags = (unsigned long)mapping & FOLIO_MAPPING_FLAGS;
 	if (mapping_flags)
-		return mapping_flags & PAGE_MAPPING_ANON;
+		return mapping_flags & FOLIO_MAPPING_ANON;
 
 	/*
 	 * At this point, we know the mapping is non-null and points to an
diff --git a/mm/internal.h b/mm/internal.h
index feda91c9b3f4e..ef11d341e3915 100644
--- a/mm/internal.h
+++ b/mm/internal.h
@@ -150,7 +150,7 @@ static inline void *folio_raw_mapping(const struct folio *folio)
 {
 	unsigned long mapping = (unsigned long)folio->mapping;
 
-	return (void *)(mapping & ~PAGE_MAPPING_FLAGS);
+	return (void *)(mapping & ~FOLIO_MAPPING_FLAGS);
 }
 
 /*
diff --git a/mm/ksm.c b/mm/ksm.c
index 18b3690bb69ad..660a94964438f 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -893,7 +893,7 @@ static struct folio *ksm_get_folio(struct ksm_stable_node *stable_node,
 	unsigned long kpfn;
 
 	expected_mapping = (void *)((unsigned long)stable_node |
-					PAGE_MAPPING_KSM);
+					FOLIO_MAPPING_KSM);
 again:
 	kpfn = READ_ONCE(stable_node->kpfn); /* Address dependency. */
 	folio = pfn_folio(kpfn);
@@ -1070,7 +1070,7 @@ static inline void folio_set_stable_node(struct folio *folio,
 					 struct ksm_stable_node *stable_node)
 {
 	VM_WARN_ON_FOLIO(folio_test_anon(folio) && PageAnonExclusive(&folio->page), folio);
-	folio->mapping = (void *)((unsigned long)stable_node | PAGE_MAPPING_KSM);
+	folio->mapping = (void *)((unsigned long)stable_node | FOLIO_MAPPING_KSM);
 }
 
 #ifdef CONFIG_SYSFS
diff --git a/mm/rmap.c b/mm/rmap.c
index fd160ddaa9807..c8610ea772683 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -503,12 +503,12 @@ struct anon_vma *folio_get_anon_vma(const struct folio *folio)
 
 	rcu_read_lock();
 	anon_mapping = (unsigned long)READ_ONCE(folio->mapping);
-	if ((anon_mapping & PAGE_MAPPING_FLAGS) != PAGE_MAPPING_ANON)
+	if ((anon_mapping & FOLIO_MAPPING_FLAGS) != FOLIO_MAPPING_ANON)
 		goto out;
 	if (!folio_mapped(folio))
 		goto out;
 
-	anon_vma = (struct anon_vma *) (anon_mapping - PAGE_MAPPING_ANON);
+	anon_vma = (struct anon_vma *) (anon_mapping - FOLIO_MAPPING_ANON);
 	if (!atomic_inc_not_zero(&anon_vma->refcount)) {
 		anon_vma = NULL;
 		goto out;
@@ -550,12 +550,12 @@ struct anon_vma *folio_lock_anon_vma_read(const struct folio *folio,
 retry:
 	rcu_read_lock();
 	anon_mapping = (unsigned long)READ_ONCE(folio->mapping);
-	if ((anon_mapping & PAGE_MAPPING_FLAGS) != PAGE_MAPPING_ANON)
+	if ((anon_mapping & FOLIO_MAPPING_FLAGS) != FOLIO_MAPPING_ANON)
 		goto out;
 	if (!folio_mapped(folio))
 		goto out;
 
-	anon_vma = (struct anon_vma *) (anon_mapping - PAGE_MAPPING_ANON);
+	anon_vma = (struct anon_vma *) (anon_mapping - FOLIO_MAPPING_ANON);
 	root_anon_vma = READ_ONCE(anon_vma->root);
 	if (down_read_trylock(&root_anon_vma->rwsem)) {
 		/*
@@ -1334,9 +1334,9 @@ void folio_move_anon_rmap(struct folio *folio, struct vm_area_struct *vma)
 	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
 	VM_BUG_ON_VMA(!anon_vma, vma);
 
-	anon_vma += PAGE_MAPPING_ANON;
+	anon_vma += FOLIO_MAPPING_ANON;
 	/*
-	 * Ensure that anon_vma and the PAGE_MAPPING_ANON bit are written
+	 * Ensure that anon_vma and the FOLIO_MAPPING_ANON bit are written
 	 * simultaneously, so a concurrent reader (eg folio_referenced()'s
 	 * folio_test_anon()) will not see one without the other.
 	 */
@@ -1367,10 +1367,10 @@ static void __folio_set_anon(struct folio *folio, struct vm_area_struct *vma,
 	/*
 	 * page_idle does a lockless/optimistic rmap scan on folio->mapping.
 	 * Make sure the compiler doesn't split the stores of anon_vma and
-	 * the PAGE_MAPPING_ANON type identifier, otherwise the rmap code
+	 * the FOLIO_MAPPING_ANON type identifier, otherwise the rmap code
 	 * could mistake the mapping for a struct address_space and crash.
 	 */
-	anon_vma = (void *) anon_vma + PAGE_MAPPING_ANON;
+	anon_vma = (void *) anon_vma + FOLIO_MAPPING_ANON;
 	WRITE_ONCE(folio->mapping, (struct address_space *) anon_vma);
 	folio->index = linear_page_index(vma, address);
 }
diff --git a/mm/util.c b/mm/util.c
index 0b270c43d7d12..20bbfe4ce1b8b 100644
--- a/mm/util.c
+++ b/mm/util.c
@@ -670,9 +670,9 @@ struct anon_vma *folio_anon_vma(const struct folio *folio)
 {
 	unsigned long mapping = (unsigned long)folio->mapping;
 
-	if ((mapping & PAGE_MAPPING_FLAGS) != PAGE_MAPPING_ANON)
+	if ((mapping & FOLIO_MAPPING_FLAGS) != FOLIO_MAPPING_ANON)
 		return NULL;
-	return (void *)(mapping - PAGE_MAPPING_ANON);
+	return (void *)(mapping - FOLIO_MAPPING_ANON);
 }
 
 /**
@@ -699,7 +699,7 @@ struct address_space *folio_mapping(struct folio *folio)
 		return swap_address_space(folio->swap);
 
 	mapping = folio->mapping;
-	if ((unsigned long)mapping & PAGE_MAPPING_FLAGS)
+	if ((unsigned long)mapping & FOLIO_MAPPING_FLAGS)
 		return NULL;
 
 	return mapping;
-- 
2.49.0

[PATCH RFC 27/29] docs/mm: convert from "Non-LRU page migration" to "movable_ops page migration"

[David Hildenbrand]

Let's bring the docs up-to-date.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 Documentation/mm/page_migration.rst | 39 ++++++++++++++++++++---------
 1 file changed, 27 insertions(+), 12 deletions(-)

diff --git a/Documentation/mm/page_migration.rst b/Documentation/mm/page_migration.rst
index 519b35a4caf5b..a448e95e0a98e 100644
--- a/Documentation/mm/page_migration.rst
+++ b/Documentation/mm/page_migration.rst
@@ -146,18 +146,33 @@ Steps:
 18. The new page is moved to the LRU and can be scanned by the swapper,
     etc. again.
 
-Non-LRU page migration
-======================
-
-Although migration originally aimed for reducing the latency of memory
-accesses for NUMA, compaction also uses migration to create high-order
-pages.  For compaction purposes, it is also useful to be able to move
-non-LRU pages, such as zsmalloc and virtio-balloon pages.
-
-If a driver wants to make its pages movable, it should define a struct
-movable_operations.  It then needs to call __SetPageMovable() on each
-page that it may be able to move.  This uses the ``page->mapping`` field,
-so this field is not available for the driver to use for other purposes.
+movable_ops page migration
+==========================
+
+Selected typed, non-folio pages (e.g., pages inflated in a memory balloon,
+zsmalloc pages) can be migrated using the movable_ops migration framework.
+
+The "struct movable_operations" provide callbacks specific to a page type
+for isolating, migrating and un-isolating (putback) these pages.
+
+Once a page is indicated as having movable_ops, that condition must not
+change until the page was freed back to the buddy. This includes not
+changing/clearing the page type and not changing/clearing the
+PG_movable_ops page flag.
+
+Arbitrary drivers cannot currently make use of this framework, as it
+requires:
+
+(a) a page type
+(b) indicating them as possibly having movable_ops in page_has_movable_ops()
+    based on the page type
+(c) returning the movable_ops from page_has_movable_ops() based on the page
+    type
+(d) not reusing the PG_movable_ops and PG_movable_ops_isolated page flags
+    for other purposes
+
+For example, balloon drivers can make use of this framework through the
+balloon-compaction framework residing in the core kernel.
 
 Monitoring Migration
 =====================
-- 
2.49.0

[PATCH RFC 28/29] mm/balloon_compaction: "movable_ops" doc updates

[David Hildenbrand]

Let's bring the docs up-to-date. Setting PG_movable_ops + page->private
very likely still requires to be performed under documented locks:
it's complicated.

We will rework this in the future, as we will try avoiding using the
page lock.

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/balloon_compaction.h | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
index b222b0737c466..2fecfead91d26 100644
--- a/include/linux/balloon_compaction.h
+++ b/include/linux/balloon_compaction.h
@@ -4,12 +4,13 @@
  *
  * Common interface definitions for making balloon pages movable by compaction.
  *
- * Balloon page migration makes use of the general non-lru movable page
+ * Balloon page migration makes use of the general "movable_ops page migration"
  * feature.
  *
  * page->private is used to reference the responsible balloon device.
- * page->mapping is used in context of non-lru page migration to reference
- * the address space operations for page isolation/migration/compaction.
+ * That these pages have movable_ops, and which movable_ops apply,
+ * is derived from the page type (PageOffline()) combined with the
+ * PG_movable_ops flag (PageMovableOps()).
  *
  * As the page isolation scanning step a compaction thread does is a lockless
  * procedure (from a page standpoint), it might bring some racy situations while
@@ -17,12 +18,10 @@
  * and safely perform balloon's page compaction and migration we must, always,
  * ensure following these simple rules:
  *
- *   i. when updating a balloon's page ->mapping element, strictly do it under
- *      the following lock order, independently of the far superior
- *      locking scheme (lru_lock, balloon_lock):
+ *   i. Setting the PG_movable_ops flag and page->private with the following
+ *	lock order
  *	    +-page_lock(page);
  *	      +--spin_lock_irq(&b_dev_info->pages_lock);
- *	            ... page->mapping updates here ...
  *
  *  ii. isolation or dequeueing procedure must remove the page from balloon
  *      device page list under b_dev_info->pages_lock.
-- 
2.49.0

[PATCH RFC 29/29] mm/balloon_compaction: provide single balloon_page_insert() and balloon_mapping_gfp_mask()

[David Hildenbrand]

Let's just special-case based on IS_ENABLED(CONFIG_BALLOON_COMPACTION
like we did for balloon_page_finalize().

Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/balloon_compaction.h | 42 +++++++++++-------------------
 1 file changed, 15 insertions(+), 27 deletions(-)

diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
index 2fecfead91d26..7cfe48769239e 100644
--- a/include/linux/balloon_compaction.h
+++ b/include/linux/balloon_compaction.h
@@ -77,6 +77,15 @@ static inline void balloon_devinfo_init(struct balloon_dev_info *balloon)
 
 #ifdef CONFIG_BALLOON_COMPACTION
 extern const struct movable_operations balloon_mops;
+/*
+ * balloon_page_device - get the b_dev_info descriptor for the balloon device
+ *			 that enqueues the given page.
+ */
+static inline struct balloon_dev_info *balloon_page_device(struct page *page)
+{
+	return (struct balloon_dev_info *)page_private(page);
+}
+#endif /* CONFIG_BALLOON_COMPACTION */
 
 /*
  * balloon_page_insert - insert a page into the balloon's page list and make
@@ -91,41 +100,20 @@ static inline void balloon_page_insert(struct balloon_dev_info *balloon,
 				       struct page *page)
 {
 	__SetPageOffline(page);
-	SetPageMovableOps(page);
-	set_page_private(page, (unsigned long)balloon);
-	list_add(&page->lru, &balloon->pages);
-}
-
-/*
- * balloon_page_device - get the b_dev_info descriptor for the balloon device
- *			 that enqueues the given page.
- */
-static inline struct balloon_dev_info *balloon_page_device(struct page *page)
-{
-	return (struct balloon_dev_info *)page_private(page);
-}
-
-static inline gfp_t balloon_mapping_gfp_mask(void)
-{
-	return GFP_HIGHUSER_MOVABLE;
-}
-
-#else /* !CONFIG_BALLOON_COMPACTION */
-
-static inline void balloon_page_insert(struct balloon_dev_info *balloon,
-				       struct page *page)
-{
-	__SetPageOffline(page);
+	if (IS_ENABLED(CONFIG_BALLOON_COMPACTION)) {
+		SetPageMovableOps(page);
+		set_page_private(page, (unsigned long)balloon);
+	}
 	list_add(&page->lru, &balloon->pages);
 }
 
 static inline gfp_t balloon_mapping_gfp_mask(void)
 {
+	if (IS_ENABLED(CONFIG_BALLOON_COMPACTION))
+		return GFP_HIGHUSER_MOVABLE;
 	return GFP_HIGHUSER;
 }
 
-#endif /* CONFIG_BALLOON_COMPACTION */
-
 /*
  * balloon_page_finalize - prepare a balloon page that was removed from the
  *			   balloon list for release to the page allocator
-- 
2.49.0

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Let's allow for not clearing a page type before freeing a page to the
> buddy.
>
> We'll focus on having a type set on the first page of a larger
> allocation only.
>
> With this change, we can reliably identify typed folios even though
> they might be in the process of getting freed, which will come in handy
> in migration code (at least in the transition phase).
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/page_alloc.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 858bc17653af9..44e56d31cfeb1 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -1380,6 +1380,9 @@ __always_inline bool free_pages_prepare(struct page *page,
>  			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
>  		page->mapping = NULL;
>  	}
> +	if (unlikely(page_has_type(page)))
> +		page->page_type = UINT_MAX;
> +
>  	if (is_check_pages_enabled()) {
>  		if (free_page_is_bad(page))
>  			bad++;
> -- 
> 2.49.0

How does this preserve page type? Isn’t page->page_type = UINT_MAX clearing
page_type?

Best Regards,
Yan, Zi

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[Matthew Wilcox]

On Wed, Jun 18, 2025 at 02:04:18PM -0400, Zi Yan wrote:
> > Let's allow for not clearing a page type before freeing a page to the
> > buddy.
> >
> > We'll focus on having a type set on the first page of a larger
> > allocation only.
> >
> > With this change, we can reliably identify typed folios even though
> > they might be in the process of getting freed, which will come in handy
> > in migration code (at least in the transition phase).

> > +	if (unlikely(page_has_type(page)))
> > +		page->page_type = UINT_MAX;
> > +
> >  	if (is_check_pages_enabled()) {
> >  		if (free_page_is_bad(page))
> >  			bad++;
> > -- 
> > 2.49.0
> 
> How does this preserve page type? Isn’t page->page_type = UINT_MAX clearing
> page_type?

The point is that the _caller_ used to have to clear the page type.
This patch allows the caller to free the page without clearing
the page type first.

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[Zi Yan]

On 18 Jun 2025, at 14:04, Zi Yan wrote:

> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>
>> Let's allow for not clearing a page type before freeing a page to the
>> buddy.
>>
>> We'll focus on having a type set on the first page of a larger
>> allocation only.
>>
>> With this change, we can reliably identify typed folios even though
>> they might be in the process of getting freed, which will come in handy
>> in migration code (at least in the transition phase).
>>
>> Signed-off-by: David Hildenbrand <david@redhat.com>
>> ---
>>  mm/page_alloc.c | 3 +++
>>  1 file changed, 3 insertions(+)
>>
>> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
>> index 858bc17653af9..44e56d31cfeb1 100644
>> --- a/mm/page_alloc.c
>> +++ b/mm/page_alloc.c
>> @@ -1380,6 +1380,9 @@ __always_inline bool free_pages_prepare(struct page *page,
>>  			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
>>  		page->mapping = NULL;
>>  	}
>> +	if (unlikely(page_has_type(page)))
>> +		page->page_type = UINT_MAX;
>> +
>>  	if (is_check_pages_enabled()) {
>>  		if (free_page_is_bad(page))
>>  			bad++;
>> -- 
>> 2.49.0
>
> How does this preserve page type? Isn’t page->page_type = UINT_MAX clearing
> page_type?

OK, next patch explains it. free_pages_prepare() clears page_type,
so that caller does not need to.

I think the message is better to be

mm/page_alloc: clear page_type at page free time

page_type is no longer needed to be cleared before a page is freed, as
page free code does that.

With this change, we can reliably identify typed folios even though
they might be in the process of getting freed, which will come in handy
in migration code (at least in the transition phase).

Best Regards,
Yan, Zi

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[Zi Yan]

On 18 Jun 2025, at 14:06, Matthew Wilcox wrote:

> On Wed, Jun 18, 2025 at 02:04:18PM -0400, Zi Yan wrote:
>>> Let's allow for not clearing a page type before freeing a page to the
>>> buddy.
>>>
>>> We'll focus on having a type set on the first page of a larger
>>> allocation only.
>>>
>>> With this change, we can reliably identify typed folios even though
>>> they might be in the process of getting freed, which will come in handy
>>> in migration code (at least in the transition phase).
>
>>> +	if (unlikely(page_has_type(page)))
>>> +		page->page_type = UINT_MAX;
>>> +
>>>  	if (is_check_pages_enabled()) {
>>>  		if (free_page_is_bad(page))
>>>  			bad++;
>>> -- 
>>> 2.49.0
>>
>> How does this preserve page type? Isn’t page->page_type = UINT_MAX clearing
>> page_type?
>
> The point is that the _caller_ used to have to clear the page type.
> This patch allows the caller to free the page without clearing
> the page type first.

Yep, find that out when I read the next patch.

I think the change is fine, but the commit message is unclear.

Best Regards,
Yan, Zi

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> ... and start moving back to per-page things that will absolutely not be
> folio things in the future. Add documentation and a comment that the
> remaining folio stuff (lock, refcount) will have to be reworked as well.
>
> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
> it gracefully (relevant with further changes), and convert a
> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().

The reason is that there is no upstream code, which use movable_ops for
folios? Is there any fundamental reason preventing movable_ops from
being used on folios?

Best Regards,
Yan, Zi

Re: [PATCH RFC 01/29] mm/balloon_compaction: we cannot have isolated pages in the balloon list

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> The core will set PG_isolated only after mops->isolate_page() was
> called. In case of the balloon, that is where we will remove it from
> the balloon list. So we cannot have isolated pages in the balloon list.
>
> Let's drop this unnecessary check.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/balloon_compaction.c | 6 ------
>  1 file changed, 6 deletions(-)
>

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

Best Regards,
Yan, Zi

Re: [PATCH RFC 03/29] mm/zsmalloc: drop PageIsolated() related VM_BUG_ONs

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Let's drop these checks; these are conditions the core migration code
> must make sure will hold either way, no need to double check.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/zpdesc.h   | 5 -----
>  mm/zsmalloc.c | 5 -----
>  2 files changed, 10 deletions(-)
>

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

Best Regards,
Yan, Zi

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Matthew Wilcox]

On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> 
> > ... and start moving back to per-page things that will absolutely not be
> > folio things in the future. Add documentation and a comment that the
> > remaining folio stuff (lock, refcount) will have to be reworked as well.
> >
> > While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
> > it gracefully (relevant with further changes), and convert a
> > WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
> 
> The reason is that there is no upstream code, which use movable_ops for
> folios? Is there any fundamental reason preventing movable_ops from
> being used on folios?

folios either belong to a filesystem or they are anonymous memory, and
so either the filesystem knows how to migrate them (through its a_ops)
or the migration code knows how to handle anon folios directly.

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Let's allow for not clearing a page type before freeing a page to the
> buddy.
>
> We'll focus on having a type set on the first page of a larger
> allocation only.
>
> With this change, we can reliably identify typed folios even though
> they might be in the process of getting freed, which will come in handy
> in migration code (at least in the transition phase).
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/page_alloc.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 858bc17653af9..44e56d31cfeb1 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -1380,6 +1380,9 @@ __always_inline bool free_pages_prepare(struct page *page,
>  			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
>  		page->mapping = NULL;
>  	}
> +	if (unlikely(page_has_type(page)))
> +		page->page_type = UINT_MAX;
> +
>  	if (is_check_pages_enabled()) {
>  		if (free_page_is_bad(page))
>  			bad++;

Should we be pedantic to only do this for PageOffline and PageZsmalloc
and warn for the rest page types?

Something like:

if (unlikely(page_has_type(page))) {
	if (PageOffline(page) || PageZsmalloc(page))
		page->page_type = UINT_MAX;
	else
		VM_WARN_ONCE_PAGE(1, page);
}

Best Regards,
Yan, Zi

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Zi Yan]

On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:

> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>
>>> ... and start moving back to per-page things that will absolutely not be
>>> folio things in the future. Add documentation and a comment that the
>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
>>>
>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
>>> it gracefully (relevant with further changes), and convert a
>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
>>
>> The reason is that there is no upstream code, which use movable_ops for
>> folios? Is there any fundamental reason preventing movable_ops from
>> being used on folios?
>
> folios either belong to a filesystem or they are anonymous memory, and
> so either the filesystem knows how to migrate them (through its a_ops)
> or the migration code knows how to handle anon folios directly.

for device private pages, to support migrating >0 order anon or fs folios
to device, how should we represent them for devices? if you think folio is
only for anon and fs.

Best Regards,
Yan, Zi

Re: [PATCH RFC 05/29] mm/balloon_compaction: make PageOffline sticky

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Let the buddy handle clearing the type.

The below might be more precise? Since page type is cleared
before the page reaches buddy code.

Let the free page routine handle clearing the type.

>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/balloon_compaction.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
> index b9f19da37b089..bfc6e50bd004b 100644
> --- a/include/linux/balloon_compaction.h
> +++ b/include/linux/balloon_compaction.h
> @@ -140,7 +140,7 @@ static inline void balloon_page_finalize(struct page *page)
>  		__ClearPageMovable(page);
>  		set_page_private(page, 0);
>  	}
> -	__ClearPageOffline(page);
> +	/* PageOffline is sticky until the page is freed to the buddy. */
>  }
>
>  /*
> -- 
> 2.49.0

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

Best Regards,
Yan, Zi

Re: [PATCH RFC 06/29] mm/zsmalloc: make PageZsmalloc() sticky

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Let the buddy handle clearing the type.

Same comment as Patch 5.

>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/zpdesc.h   | 5 -----
>  mm/zsmalloc.c | 3 +--
>  2 files changed, 1 insertion(+), 7 deletions(-)
>
Acked-by: Zi Yan <ziy@nvidia.com>

Best Regards,
Yan, Zi

Re: [PATCH RFC 08/29] mm/migrate: rename putback_movable_folio() to putback_movable_ops_page()

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> ... and factor the complete handling of movable_ops pages out.
> Convert it similar to isolate_movable_ops_page().
>
> While at it, convert the VM_BUG_ON_FOLIO() into a VM_WARN_ON_PAGE().
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/migrate.c | 37 ++++++++++++++++++++++++-------------
>  1 file changed, 24 insertions(+), 13 deletions(-)
>
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 6bbb455f8b593..32e77898f7d6c 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -133,12 +133,30 @@ bool isolate_movable_ops_page(struct page *page, isolate_mode_t mode)
>  	return false;
>  }
>
> -static void putback_movable_folio(struct folio *folio)
> +/**
> + * putback_movable_ops_page - putback an isolated movable_ops page
> + * @page: The isolated page.
> + *
> + * Putback an isolated movable_ops page.
> + *
> + * After the page was putback, it might get freed instantly.
> + */
> +static void putback_movable_ops_page(struct page *page)
>  {
> -	const struct movable_operations *mops = folio_movable_ops(folio);
> -
> -	mops->putback_page(&folio->page);
> -	folio_clear_isolated(folio);
> +	/*
> +	 * TODO: these pages will not be folios in the future. All
> +	 * folio dependencies will have to be removed.
> +	 */
> +	struct folio *folio = page_folio(page);
> +
> +	VM_WARN_ON_ONCE_PAGE(!PageIsolated(page), page);
> +	folio_lock(folio);
> +	/* If the page was released by it's owner, there is nothing to do. */
> +	if (PageMovable(page))
> +		page_movable_ops(page)->putback_page(page);
> +	ClearPageIsolated(page);
> +	folio_unlock(folio);
> +	folio_put(folio);

Why not use page version of lock, unlock, and put? Especially you are
thinking about not using folio for these pages. Just a question,
I am OK with current patch.

>  }
>
>  /*
> @@ -166,14 +184,7 @@ void putback_movable_pages(struct list_head *l)
>  		 * have PAGE_MAPPING_MOVABLE.
>  		 */
>  		if (unlikely(__folio_test_movable(folio))) {
> -			VM_BUG_ON_FOLIO(!folio_test_isolated(folio), folio);
> -			folio_lock(folio);
> -			if (folio_test_movable(folio))
> -				putback_movable_folio(folio);
> -			else
> -				folio_clear_isolated(folio);
> -			folio_unlock(folio);
> -			folio_put(folio);
> +			putback_movable_ops_page(&folio->page);
>  		} else {
>  			node_stat_mod_folio(folio, NR_ISOLATED_ANON +
>  					folio_is_file_lru(folio), -folio_nr_pages(folio));
> -- 
> 2.49.0

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

Best Regards,
Yan, Zi

Re: [PATCH RFC 08/29] mm/migrate: rename putback_movable_folio() to putback_movable_ops_page()

[Matthew Wilcox]

On Wed, Jun 18, 2025 at 03:10:10PM -0400, Zi Yan wrote:
> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> > +	/*
> > +	 * TODO: these pages will not be folios in the future. All
> > +	 * folio dependencies will have to be removed.
> > +	 */
> > +	struct folio *folio = page_folio(page);
> > +
> > +	VM_WARN_ON_ONCE_PAGE(!PageIsolated(page), page);
> > +	folio_lock(folio);
> > +	/* If the page was released by it's owner, there is nothing to do. */
> > +	if (PageMovable(page))
> > +		page_movable_ops(page)->putback_page(page);
> > +	ClearPageIsolated(page);
> > +	folio_unlock(folio);
> > +	folio_put(folio);
> 
> Why not use page version of lock, unlock, and put? Especially you are
> thinking about not using folio for these pages. Just a question,
> I am OK with current patch.

That would reintroduce unnecessary calls to compound_head().

Re: [PATCH RFC 08/29] mm/migrate: rename putback_movable_folio() to putback_movable_ops_page()

[Zi Yan]

On 18 Jun 2025, at 15:18, Matthew Wilcox wrote:

> On Wed, Jun 18, 2025 at 03:10:10PM -0400, Zi Yan wrote:
>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>> +	/*
>>> +	 * TODO: these pages will not be folios in the future. All
>>> +	 * folio dependencies will have to be removed.
>>> +	 */
>>> +	struct folio *folio = page_folio(page);
>>> +
>>> +	VM_WARN_ON_ONCE_PAGE(!PageIsolated(page), page);
>>> +	folio_lock(folio);
>>> +	/* If the page was released by it's owner, there is nothing to do. */
>>> +	if (PageMovable(page))
>>> +		page_movable_ops(page)->putback_page(page);
>>> +	ClearPageIsolated(page);
>>> +	folio_unlock(folio);
>>> +	folio_put(folio);
>>
>> Why not use page version of lock, unlock, and put? Especially you are
>> thinking about not using folio for these pages. Just a question,
>> I am OK with current patch.
>
> That would reintroduce unnecessary calls to compound_head().

Got it. But here page is not folio, so it cannot be a compound page.
Then, we will need page versions without compound_head() for
non compound pages. Could that happen in the future when only folio
can be compound and page is only order-0?

Best Regards,
Yan, Zi

Re: [PATCH RFC 08/29] mm/migrate: rename putback_movable_folio() to putback_movable_ops_page()

[Matthew Wilcox]

On Wed, Jun 18, 2025 at 03:25:46PM -0400, Zi Yan wrote:
> On 18 Jun 2025, at 15:18, Matthew Wilcox wrote:
> >> Why not use page version of lock, unlock, and put? Especially you are
> >> thinking about not using folio for these pages. Just a question,
> >> I am OK with current patch.
> >
> > That would reintroduce unnecessary calls to compound_head().
> 
> Got it. But here page is not folio, so it cannot be a compound page.
> Then, we will need page versions without compound_head() for
> non compound pages. Could that happen in the future when only folio
> can be compound and page is only order-0?

I think the assumption that we'll only see compound pages as part of
folios is untrue.  For example, slabs will still allocate multiple
pages (though slabs aren't migratable at this point).  The sketch at
https://kernelnewbies.org/MatthewWilcox/Memdescs supports "misc pages"
with an order stored in bits 12-17 of the memdesc.  I don't know
how useful that will turn out to be; maybe we'll never implement that.

Re: [PATCH RFC 03/29] mm/zsmalloc: drop PageIsolated() related VM_BUG_ONs

[Sergey Senozhatsky]

On (25/06/18 19:39), David Hildenbrand wrote:
> Let's drop these checks; these are conditions the core migration code
> must make sure will hold either way, no need to double check.
> 
> Signed-off-by: David Hildenbrand <david@redhat.com>

Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>

Re: [PATCH RFC 06/29] mm/zsmalloc: make PageZsmalloc() sticky

[Sergey Senozhatsky]

On (25/06/18 19:39), David Hildenbrand wrote:
> Let the buddy handle clearing the type.
> 
> Signed-off-by: David Hildenbrand <david@redhat.com>

FWIW,
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>

Re: [PATCH RFC 09/29] mm/migrate: factor out movable_ops page handling into migrate_movable_ops_page()

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Let's factor it out, simplifying the calling code.
>
> The assumption is that flush_dcache_page() is not required for
> movable_ops pages: as documented for flush_dcache_folio(), it really
> only applies when the kernel wrote to pagecache pages / pages in
> highmem. movable_ops callbacks should be handling flushing
> caches if ever required.
>
> Note that we can now change folio_mapping_flags() to folio_test_anon()
> to make it clearer, because movable_ops pages will never take that path.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/migrate.c | 82 ++++++++++++++++++++++++++++------------------------
>  1 file changed, 45 insertions(+), 37 deletions(-)
>
LGTM. Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 10/29] mm/migrate: remove folio_test_movable() and folio_movable_ops()

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Folios will have nothing to do with movable_ops page migration. These
> functions are now unused, so let's remove them.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/migrate.h | 14 --------------
>  1 file changed, 14 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 11/29] mm/migrate: move movable_ops page handling out of move_to_new_folio()

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Let's move that handling directly into migrate_folio_move(), so we can
> simplify move_to_new_folio(). While at it, fixup the documentation a
> bit.
>
> Note that unmap_and_move_huge_page() does not care, because it only
> deals with actual folios. (we only support migration of
> individual movable_ops pages)
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/migrate.c | 61 ++++++++++++++++++++++++----------------------------
>  1 file changed, 28 insertions(+), 33 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 14/29] mm/migrate: remove __ClearPageMovable()

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Unused, let's remove it.
>
> The Chinese docs in Documentation/translations/zh_CN/mm/page_migration.rst
> still mention it, but that whole docs is destined to get outdated and
> updated by somebody that actually speaks that language.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/migrate.h |  4 ----
>  mm/compaction.c         | 11 -----------
>  2 files changed, 15 deletions(-)
>

The comment for struct movable_operations needs an update too.
"
If page migration is successful, the driver should call
__ClearPageMovable(@src) and return MIGRATEPAGE_SUCCESS.
"

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

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 15/29] mm/migration: remove PageMovable()

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> As __ClearPageMovable() is gone that would have only made
> PageMovable()==false but still __PageMovable()==true, now
> PageMovable() == __PageMovable().
>
> So we can replace PageMovable() checks by __PageMovable(). In fact,
> __PageMovable() cannot change until a page is freed, so we can turn
> some PageMovable() into sanity checks for __PageMovable().
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/migrate.h |  2 --
>  mm/compaction.c         | 15 ---------------
>  mm/migrate.c            | 18 ++++++++++--------
>  3 files changed, 10 insertions(+), 25 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 16/29] mm: rename __PageMovable() to page_has_movable_ops()

[Zi Yan]

On 18 Jun 2025, at 13:39, David Hildenbrand wrote:

> Let's make it clearer that we are talking about movable_ops pages.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/migrate.h    |  2 +-
>  include/linux/page-flags.h |  2 +-
>  mm/compaction.c            |  7 ++-----
>  mm/memory-failure.c        |  4 ++--
>  mm/memory_hotplug.c        |  8 +++-----
>  mm/migrate.c               |  8 ++++----
>  mm/page_alloc.c            |  2 +-
>  mm/page_isolation.c        | 10 +++++-----
>  8 files changed, 19 insertions(+), 24 deletions(-)
>
> diff --git a/include/linux/migrate.h b/include/linux/migrate.h
> index 204e89eac998f..c575778456f97 100644
> --- a/include/linux/migrate.h
> +++ b/include/linux/migrate.h
> @@ -115,7 +115,7 @@ static inline void __SetPageMovable(struct page *page,
>  static inline
>  const struct movable_operations *page_movable_ops(struct page *page)
>  {
> -	VM_BUG_ON(!__PageMovable(page));
> +	VM_BUG_ON(!page_has_movable_ops(page));
>
>  	return (const struct movable_operations *)
>  		((unsigned long)page->mapping - PAGE_MAPPING_MOVABLE);
> diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
> index 4fe5ee67535b2..c67163b73c5ec 100644
> --- a/include/linux/page-flags.h
> +++ b/include/linux/page-flags.h
> @@ -750,7 +750,7 @@ static __always_inline bool __folio_test_movable(const struct folio *folio)
>  			PAGE_MAPPING_MOVABLE;
>  }
>
> -static __always_inline bool __PageMovable(const struct page *page)
> +static __always_inline bool page_has_movable_ops(const struct page *page)
>  {
>  	return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) ==
>  				PAGE_MAPPING_MOVABLE;
> diff --git a/mm/compaction.c b/mm/compaction.c
> index 5c37373017014..f8b7c09e2e48c 100644
> --- a/mm/compaction.c
> +++ b/mm/compaction.c
> @@ -1056,11 +1056,8 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
>  		 * Skip any other type of page
>  		 */
>  		if (!PageLRU(page)) {
> -			/*
> -			 * __PageMovable can return false positive so we need
> -			 * to verify it under page_lock.
> -			 */
> -			if (unlikely(__PageMovable(page)) &&
> +			/* Isolation will grab the page lock. */

I feel that the removed comment should stay, since the current comment
makes no sense when I read it alone.

In addition, why is __PageMovable() is renamed to page_has_movable_ops() but
__SetPageMovable() stays the same? page_has_movable_ops() and __SetPageMovable()
are functions for checking and setting PAGE_MAPPING_MOVABLE. The naming just
does not look symmetric.

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 17/29] mm/page_isolation: drop __folio_test_movable() check for large folios

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> Currently, we only support migration of individual non-folio pages, so
> we can not run into that.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/page_isolation.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 18/29] mm: remove __folio_test_movable()

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> Convert to page_has_movable_ops(). While at it, cleanup relevant code
> a bit.
>
> The data_race() in migrate_folio_unmap() is questionable: we already
> hold a page reference, and concurrent modifications can no longer
> happen (iow: __ClearPageMovable() no longer exists). Drop it for now,
> we'll rework page_has_movable_ops() soon either way to no longer
> rely on page->mapping.
>
> Wherever we cast from folio to page now is a clear sign that this
> code has to be decoupled.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/page-flags.h |  6 ------
>  mm/migrate.c               | 43 ++++++++++++--------------------------
>  mm/vmscan.c                |  6 ++++--
>  3 files changed, 17 insertions(+), 38 deletions(-)
>

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

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 19/29] mm: stop storing migration_ops in page->mapping

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> ... instead, look them up statically based on the page type. Maybe in the
> future we want a registration interface? At least for now, it can be
> easily handled using the two page types that actually support page
> migration.
>
> The remaining usage of page->mapping is to flag such pages as actually
> being movable (having movable_ops), which we will change next.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/balloon_compaction.h |  2 +-
>  include/linux/migrate.h            | 14 ++------------
>  include/linux/zsmalloc.h           |  2 ++
>  mm/balloon_compaction.c            |  1 -
>  mm/compaction.c                    |  5 ++---
>  mm/migrate.c                       | 23 +++++++++++++++++++++++
>  mm/zpdesc.h                        |  5 ++---
>  mm/zsmalloc.c                      |  8 +++-----
>  8 files changed, 35 insertions(+), 25 deletions(-)
>

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

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 20/29] mm: convert "movable" flag in page->mapping to a page flag

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> Instead, let's use a page flag. As the page flag can result in
> false-positives, glue it to the page types for which we
> support/implement movable_ops page migration.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/balloon_compaction.h |  2 +-
>  include/linux/migrate.h            |  8 -----
>  include/linux/page-flags.h         | 52 ++++++++++++++++++++++++------
>  mm/compaction.c                    |  6 ----
>  mm/zpdesc.h                        |  2 +-
>  5 files changed, 44 insertions(+), 26 deletions(-)
>

<snip>

> +
> +/**
> + * page_has_movable_ops - test for a movable_ops page
> + * @page The page to test.
> + *
> + * Test whether this is a movable_ops page. Such pages will stay that
> + * way until freed.
> + *
> + * Returns true if this is a movable_ops page, otherwise false.
> + */
> +static inline bool page_has_movable_ops(const struct page *page)
> +{
> +	return PageMovableOps(page) &&
> +	       (PageOffline(page) || PageZsmalloc(page));
> +}
> +

Should we do the code below in case PageMovableOps is set on pages
other than PageOffline and PageZsmalloc?

return PageMovableOps(page) &&
	   !VM_WARN_ON_ONCE_PAGE(!(PageOffline(page) || PageZsmalloc(page)),
							 page);

>  static __always_inline int PageAnonExclusive(const struct page *page)
>  {
>  	VM_BUG_ON_PGFLAGS(!PageAnon(page), page);
> diff --git a/mm/compaction.c b/mm/compaction.c
> index a10f23df9396e..86d671a520e91 100644
> --- a/mm/compaction.c
> +++ b/mm/compaction.c
> @@ -114,12 +114,6 @@ static unsigned long release_free_list(struct list_head *freepages)
>  }
>
>  #ifdef CONFIG_COMPACTION
> -void __SetPageMovable(struct page *page)
> -{
> -	VM_BUG_ON_PAGE(!PageLocked(page), page);
> -	page->mapping = (void *)(PAGE_MAPPING_MOVABLE);
> -}
> -EXPORT_SYMBOL(__SetPageMovable);
>
>  /* Do not skip compaction more than 64 times */
>  #define COMPACT_MAX_DEFER_SHIFT 6
> diff --git a/mm/zpdesc.h b/mm/zpdesc.h
> index 6855d9e2732d8..25bf5ea0beb83 100644
> --- a/mm/zpdesc.h
> +++ b/mm/zpdesc.h
> @@ -154,7 +154,7 @@ static inline struct zpdesc *pfn_zpdesc(unsigned long pfn)
>
>  static inline void __zpdesc_set_movable(struct zpdesc *zpdesc)
>  {
> -	__SetPageMovable(zpdesc_page(zpdesc));
> +	SetPageMovableOps(zpdesc_page(zpdesc));
>  }
>
>  static inline void __zpdesc_set_zsmalloc(struct zpdesc *zpdesc)
> -- 
> 2.49.0

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


--
Best Regards,
Yan, Zi

Re: [PATCH RFC 21/29] mm: rename PG_isolated to PG_movable_ops_isolated

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> Let's rename the flag to make it clearer where it applies (not folios
> ...).
>
> While at it, define the flag only with CONFIG_MIGRATION.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/page-flags.h | 16 +++++++++++-----
>  mm/compaction.c            |  2 +-
>  mm/migrate.c               | 14 +++++++-------
>  3 files changed, 19 insertions(+), 13 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>
--
Best Regards,
Yan, Zi

Re: [PATCH RFC 22/29] mm/page-flags: rename PAGE_MAPPING_MOVABLE to PAGE_MAPPING_ANON_KSM

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> KSM is the only remaining user, let's rename the flag. While at it,
> adjust to remaining page -> folio in the doc.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/page-flags.h | 12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 23/29] mm/page-alloc: remove PageMappingFlags()

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> We can now simply check for PageAnon() and remove PageMappingFlags().
>
> ... and while at it, use the folio instead and operate on
> folio->mapping.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/page-flags.h | 5 -----
>  mm/page_alloc.c            | 7 +++----
>  2 files changed, 3 insertions(+), 9 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 24/29] mm/page-flags: remove folio_mapping_flags()

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> It's unused and the page counterpart is gone, so let's remove it.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/page-flags.h | 5 -----
>  1 file changed, 5 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 25/29] mm: simplify folio_expected_ref_count()

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> Now that PAGE_MAPPING_MOVABLE is gone, we can simplify and rely on the
> folio_test_anon() test only.
>
> ... but staring at the users, this function should never even have been
> called on movable_ops pages. E.g.,
> * __buffer_migrate_folio() does not make sense for them
> * folio_migrate_mapping() does not make sense for them
> * migrate_huge_page_move_mapping() does not make sense for them
> * __migrate_folio() does not make sense for them
> * ... and khugepaged should never stumble over them
>
> Let's simply refuse typed pages (which includes slab) except hugetlb,
> and WARN.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  include/linux/mm.h | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>
--
Best Regards,
Yan, Zi

Re: [PATCH RFC 26/29] mm: rename PAGE_MAPPING_* to FOLIO_MAPPING_*

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> Now that the mapping flags are only used for folios, let's rename the
> defines.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  fs/proc/page.c             |  4 ++--
>  include/linux/fs.h         |  2 +-
>  include/linux/mm_types.h   |  1 -
>  include/linux/page-flags.h | 20 ++++++++++----------
>  include/linux/pagemap.h    |  2 +-
>  mm/gup.c                   |  4 ++--
>  mm/internal.h              |  2 +-
>  mm/ksm.c                   |  4 ++--
>  mm/rmap.c                  | 16 ++++++++--------
>  mm/util.c                  |  6 +++---
>  10 files changed, 30 insertions(+), 31 deletions(-)
>
Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 27/29] docs/mm: convert from "Non-LRU page migration" to "movable_ops page migration"

[Zi Yan]

On 18 Jun 2025, at 13:40, David Hildenbrand wrote:

> Let's bring the docs up-to-date.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  Documentation/mm/page_migration.rst | 39 ++++++++++++++++++++---------
>  1 file changed, 27 insertions(+), 12 deletions(-)
>
> diff --git a/Documentation/mm/page_migration.rst b/Documentation/mm/page_migration.rst
> index 519b35a4caf5b..a448e95e0a98e 100644
> --- a/Documentation/mm/page_migration.rst
> +++ b/Documentation/mm/page_migration.rst
> @@ -146,18 +146,33 @@ Steps:
>  18. The new page is moved to the LRU and can be scanned by the swapper,
>      etc. again.
>
> -Non-LRU page migration
> -======================
> -
> -Although migration originally aimed for reducing the latency of memory
> -accesses for NUMA, compaction also uses migration to create high-order
> -pages.  For compaction purposes, it is also useful to be able to move
> -non-LRU pages, such as zsmalloc and virtio-balloon pages.
> -
> -If a driver wants to make its pages movable, it should define a struct
> -movable_operations.  It then needs to call __SetPageMovable() on each
> -page that it may be able to move.  This uses the ``page->mapping`` field,
> -so this field is not available for the driver to use for other purposes.
> +movable_ops page migration
> +==========================
> +
> +Selected typed, non-folio pages (e.g., pages inflated in a memory balloon,

This is a great clarification. Thanks.

> +zsmalloc pages) can be migrated using the movable_ops migration framework.
> +
> +The "struct movable_operations" provide callbacks specific to a page type
> +for isolating, migrating and un-isolating (putback) these pages.
> +
> +Once a page is indicated as having movable_ops, that condition must not
> +change until the page was freed back to the buddy. This includes not
> +changing/clearing the page type and not changing/clearing the
> +PG_movable_ops page flag.
> +
> +Arbitrary drivers cannot currently make use of this framework, as it
> +requires:
> +
> +(a) a page type
> +(b) indicating them as possibly having movable_ops in page_has_movable_ops()
> +    based on the page type
> +(c) returning the movable_ops from page_has_movable_ops() based on the page
> +    type
> +(d) not reusing the PG_movable_ops and PG_movable_ops_isolated page flags
> +    for other purposes
> +
> +For example, balloon drivers can make use of this framework through the
> +balloon-compaction framework residing in the core kernel.
>
>  Monitoring Migration
>  =====================

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

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[David Hildenbrand]

On 18.06.25 20:08, Zi Yan wrote:
> On 18 Jun 2025, at 14:04, Zi Yan wrote:
> 
>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>
>>> Let's allow for not clearing a page type before freeing a page to the
>>> buddy.
>>>
>>> We'll focus on having a type set on the first page of a larger
>>> allocation only.
>>>
>>> With this change, we can reliably identify typed folios even though
>>> they might be in the process of getting freed, which will come in handy
>>> in migration code (at least in the transition phase).
>>>
>>> Signed-off-by: David Hildenbrand <david@redhat.com>
>>> ---
>>>   mm/page_alloc.c | 3 +++
>>>   1 file changed, 3 insertions(+)
>>>
>>> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
>>> index 858bc17653af9..44e56d31cfeb1 100644
>>> --- a/mm/page_alloc.c
>>> +++ b/mm/page_alloc.c
>>> @@ -1380,6 +1380,9 @@ __always_inline bool free_pages_prepare(struct page *page,
>>>   			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
>>>   		page->mapping = NULL;
>>>   	}
>>> +	if (unlikely(page_has_type(page)))
>>> +		page->page_type = UINT_MAX;
>>> +
>>>   	if (is_check_pages_enabled()) {
>>>   		if (free_page_is_bad(page))
>>>   			bad++;
>>> -- 
>>> 2.49.0
>>
>> How does this preserve page type? Isn’t page->page_type = UINT_MAX clearing
>> page_type?
> 
> OK, next patch explains it. free_pages_prepare() clears page_type,
> so that caller does not need to.
> 
> I think the message is better to be
> 
> mm/page_alloc: clear page_type at page free time
> 
> page_type is no longer needed to be cleared before a page is freed, as
> page free code does that.
> 
> With this change, we can reliably identify typed folios even though
> they might be in the process of getting freed, which will come in handy
> in migration code (at least in the transition phase).


I'll change it to

     mm/page_alloc: let page freeing clear any set page type
     
     Currently, any user of page types must clear that type before freeing
     a page back to the buddy, otherwise we'll run into mapcount related
     sanity checks (because the page type currently overlays the page
     mapcount).
     
     Let's allow for not clearing the page type by page type users by letting
     the buddy handle it instead.
     
     We'll focus on having a page type set on the first page of a larger
     allocation only.
     
     With this change, we can reliably identify typed folios even though
     they might be in the process of getting freed, which will come in handy
     in migration code (at least in the transition phase).
     



-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[David Hildenbrand]

On 18.06.25 20:43, Zi Yan wrote:
> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> 
>> Let's allow for not clearing a page type before freeing a page to the
>> buddy.
>>
>> We'll focus on having a type set on the first page of a larger
>> allocation only.
>>
>> With this change, we can reliably identify typed folios even though
>> they might be in the process of getting freed, which will come in handy
>> in migration code (at least in the transition phase).
>>
>> Signed-off-by: David Hildenbrand <david@redhat.com>
>> ---
>>   mm/page_alloc.c | 3 +++
>>   1 file changed, 3 insertions(+)
>>
>> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
>> index 858bc17653af9..44e56d31cfeb1 100644
>> --- a/mm/page_alloc.c
>> +++ b/mm/page_alloc.c
>> @@ -1380,6 +1380,9 @@ __always_inline bool free_pages_prepare(struct page *page,
>>   			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
>>   		page->mapping = NULL;
>>   	}
>> +	if (unlikely(page_has_type(page)))
>> +		page->page_type = UINT_MAX;
>> +
>>   	if (is_check_pages_enabled()) {
>>   		if (free_page_is_bad(page))
>>   			bad++;
> 
> Should we be pedantic to only do this for PageOffline and PageZsmalloc
> and warn for the rest page types?

I think we should just allow any page types. Limiting it to specific 
types sounds like some use-after-free check that probably shouldn't be 
handled that way.

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[Zi Yan]

On 23 Jun 2025, at 11:26, David Hildenbrand wrote:

> On 18.06.25 20:08, Zi Yan wrote:
>> On 18 Jun 2025, at 14:04, Zi Yan wrote:
>>
>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>>
>>>> Let's allow for not clearing a page type before freeing a page to the
>>>> buddy.
>>>>
>>>> We'll focus on having a type set on the first page of a larger
>>>> allocation only.
>>>>
>>>> With this change, we can reliably identify typed folios even though
>>>> they might be in the process of getting freed, which will come in handy
>>>> in migration code (at least in the transition phase).
>>>>
>>>> Signed-off-by: David Hildenbrand <david@redhat.com>
>>>> ---
>>>>   mm/page_alloc.c | 3 +++
>>>>   1 file changed, 3 insertions(+)
>>>>
>>>> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
>>>> index 858bc17653af9..44e56d31cfeb1 100644
>>>> --- a/mm/page_alloc.c
>>>> +++ b/mm/page_alloc.c
>>>> @@ -1380,6 +1380,9 @@ __always_inline bool free_pages_prepare(struct page *page,
>>>>   			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
>>>>   		page->mapping = NULL;
>>>>   	}
>>>> +	if (unlikely(page_has_type(page)))
>>>> +		page->page_type = UINT_MAX;
>>>> +
>>>>   	if (is_check_pages_enabled()) {
>>>>   		if (free_page_is_bad(page))
>>>>   			bad++;
>>>> -- 
>>>> 2.49.0
>>>
>>> How does this preserve page type? Isn’t page->page_type = UINT_MAX clearing
>>> page_type?
>>
>> OK, next patch explains it. free_pages_prepare() clears page_type,
>> so that caller does not need to.
>>
>> I think the message is better to be
>>
>> mm/page_alloc: clear page_type at page free time
>>
>> page_type is no longer needed to be cleared before a page is freed, as
>> page free code does that.
>>
>> With this change, we can reliably identify typed folios even though
>> they might be in the process of getting freed, which will come in handy
>> in migration code (at least in the transition phase).
>
>
> I'll change it to
>
>     mm/page_alloc: let page freeing clear any set page type
>         Currently, any user of page types must clear that type before freeing
>     a page back to the buddy, otherwise we'll run into mapcount related
>     sanity checks (because the page type currently overlays the page
>     mapcount).
>         Let's allow for not clearing the page type by page type users by letting
>     the buddy handle it instead.
>         We'll focus on having a page type set on the first page of a larger
>     allocation only.
>         With this change, we can reliably identify typed folios even though
>     they might be in the process of getting freed, which will come in handy
>     in migration code (at least in the transition phase).

Thanks.

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

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 05/29] mm/balloon_compaction: make PageOffline sticky

[David Hildenbrand]

On 18.06.25 20:50, Zi Yan wrote:
> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> 
>> Let the buddy handle clearing the type.
> 
> The below might be more precise? Since page type is cleared
> before the page reaches buddy code.
> 
> Let the free page routine handle clearing the type.

"Let the page freeing code handle clearing the page type."

?

Thanks!

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 05/29] mm/balloon_compaction: make PageOffline sticky

[Zi Yan]

On 23 Jun 2025, at 11:28, David Hildenbrand wrote:

> On 18.06.25 20:50, Zi Yan wrote:
>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>
>>> Let the buddy handle clearing the type.
>>
>> The below might be more precise? Since page type is cleared
>> before the page reaches buddy code.
>>
>> Let the free page routine handle clearing the type.
>
> "Let the page freeing code handle clearing the page type."
>
> ?
>

Sounds good to me. :)

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[David Hildenbrand]

On 18.06.25 20:48, Zi Yan wrote:
> On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:
> 
>> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>>
>>>> ... and start moving back to per-page things that will absolutely not be
>>>> folio things in the future. Add documentation and a comment that the
>>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
>>>>
>>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
>>>> it gracefully (relevant with further changes), and convert a
>>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
>>>
>>> The reason is that there is no upstream code, which use movable_ops for
>>> folios? Is there any fundamental reason preventing movable_ops from
>>> being used on folios?
>>
>> folios either belong to a filesystem or they are anonymous memory, and
>> so either the filesystem knows how to migrate them (through its a_ops)
>> or the migration code knows how to handle anon folios directly.

Right, migration of folios will be handled by migration core.

> 
> for device private pages, to support migrating >0 order anon or fs folios
> to device, how should we represent them for devices? if you think folio is
> only for anon and fs.

I assume they are proper folios, so yes. Just like they are handled 
today (-> folios)

I was asking a related question at LSF/MM in Alistair's session: are we 
sure these things will be folios even before they are assigned to a 
filesystem? I recall the answer was "yes".

So we don't (and will not) support movable_ops for folios.

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 08/29] mm/migrate: rename putback_movable_folio() to putback_movable_ops_page()

[David Hildenbrand]

On 18.06.25 22:04, Matthew Wilcox wrote:
> On Wed, Jun 18, 2025 at 03:25:46PM -0400, Zi Yan wrote:
>> On 18 Jun 2025, at 15:18, Matthew Wilcox wrote:
>>>> Why not use page version of lock, unlock, and put? Especially you are
>>>> thinking about not using folio for these pages. Just a question,
>>>> I am OK with current patch.
>>>
>>> That would reintroduce unnecessary calls to compound_head().

Right. And I want to make it clear that everything that uses "folio" 
here must be reworked: not necessarily switching to the "page" variant, 
but actually by implementing it entirely differently.

(e.g., store them in an array instead of a list, get rid of the lock 
bit, try getting rid of the refcount as well)

>>
>> Got it. But here page is not folio, so it cannot be a compound page.
>> Then, we will need page versions without compound_head() for
>> non compound pages. Could that happen in the future when only folio
>> can be compound and page is only order-0?
> 
> I think the assumption that we'll only see compound pages as part of
> folios is untrue.  For example, slabs will still allocate multiple
> pages (though slabs aren't migratable at this point).  The sketch at
> https://kernelnewbies.org/MatthewWilcox/Memdescs supports "misc pages"
> with an order stored in bits 12-17 of the memdesc.  I don't know
> how useful that will turn out to be; maybe we'll never implement that.

Once we have to support that for movable_ops, it will be an interesting 
question how to handle that. Most certainly, these things will not be 
folios. :)

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Zi Yan]

On 23 Jun 2025, at 11:33, David Hildenbrand wrote:

> On 18.06.25 20:48, Zi Yan wrote:
>> On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:
>>
>>> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
>>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>>>
>>>>> ... and start moving back to per-page things that will absolutely not be
>>>>> folio things in the future. Add documentation and a comment that the
>>>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
>>>>>
>>>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
>>>>> it gracefully (relevant with further changes), and convert a
>>>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
>>>>
>>>> The reason is that there is no upstream code, which use movable_ops for
>>>> folios? Is there any fundamental reason preventing movable_ops from
>>>> being used on folios?
>>>
>>> folios either belong to a filesystem or they are anonymous memory, and
>>> so either the filesystem knows how to migrate them (through its a_ops)
>>> or the migration code knows how to handle anon folios directly.
>
> Right, migration of folios will be handled by migration core.
>
>>
>> for device private pages, to support migrating >0 order anon or fs folios
>> to device, how should we represent them for devices? if you think folio is
>> only for anon and fs.
>
> I assume they are proper folios, so yes. Just like they are handled today (-> folios)
>
> I was asking a related question at LSF/MM in Alistair's session: are we sure these things will be folios even before they are assigned to a filesystem? I recall the answer was "yes".
>
> So we don't (and will not) support movable_ops for folios.

Got it. (I was abusing it to help develop alloc_contig_range() at pageblock
granularity, since it was easy to write a driver to allocate a compound page
at a specific PFN and claim the page is movable, then do page online/offline
the range containing the PFNs. :) )

For the patch, Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 14/29] mm/migrate: remove __ClearPageMovable()

[David Hildenbrand]

On 20.06.25 22:15, Zi Yan wrote:
> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> 
>> Unused, let's remove it.
>>
>> The Chinese docs in Documentation/translations/zh_CN/mm/page_migration.rst
>> still mention it, but that whole docs is destined to get outdated and
>> updated by somebody that actually speaks that language.
>>
>> Signed-off-by: David Hildenbrand <david@redhat.com>
>> ---
>>   include/linux/migrate.h |  4 ----
>>   mm/compaction.c         | 11 -----------
>>   2 files changed, 15 deletions(-)
>>
> 
> The comment for struct movable_operations needs an update too.
> "
> If page migration is successful, the driver should call
> __ClearPageMovable(@src) and return MIGRATEPAGE_SUCCESS.
 > "

Ah, thanks, it will simply be "should return MIGRATEPAGE_SUCCESS".

There is more magic to it (the driver must setup the dst page as
movable), but probably that might all be reworked soon.

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 16/29] mm: rename __PageMovable() to page_has_movable_ops()

[David Hildenbrand]

On 20.06.25 22:37, Zi Yan wrote:
> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> 
>> Let's make it clearer that we are talking about movable_ops pages.
>>
>> Signed-off-by: David Hildenbrand <david@redhat.com>
>> ---
>>   include/linux/migrate.h    |  2 +-
>>   include/linux/page-flags.h |  2 +-
>>   mm/compaction.c            |  7 ++-----
>>   mm/memory-failure.c        |  4 ++--
>>   mm/memory_hotplug.c        |  8 +++-----
>>   mm/migrate.c               |  8 ++++----
>>   mm/page_alloc.c            |  2 +-
>>   mm/page_isolation.c        | 10 +++++-----
>>   8 files changed, 19 insertions(+), 24 deletions(-)
>>
>> diff --git a/include/linux/migrate.h b/include/linux/migrate.h
>> index 204e89eac998f..c575778456f97 100644
>> --- a/include/linux/migrate.h
>> +++ b/include/linux/migrate.h
>> @@ -115,7 +115,7 @@ static inline void __SetPageMovable(struct page *page,
>>   static inline
>>   const struct movable_operations *page_movable_ops(struct page *page)
>>   {
>> -	VM_BUG_ON(!__PageMovable(page));
>> +	VM_BUG_ON(!page_has_movable_ops(page));
>>
>>   	return (const struct movable_operations *)
>>   		((unsigned long)page->mapping - PAGE_MAPPING_MOVABLE);
>> diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
>> index 4fe5ee67535b2..c67163b73c5ec 100644
>> --- a/include/linux/page-flags.h
>> +++ b/include/linux/page-flags.h
>> @@ -750,7 +750,7 @@ static __always_inline bool __folio_test_movable(const struct folio *folio)
>>   			PAGE_MAPPING_MOVABLE;
>>   }
>>
>> -static __always_inline bool __PageMovable(const struct page *page)
>> +static __always_inline bool page_has_movable_ops(const struct page *page)
>>   {
>>   	return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) ==
>>   				PAGE_MAPPING_MOVABLE;
>> diff --git a/mm/compaction.c b/mm/compaction.c
>> index 5c37373017014..f8b7c09e2e48c 100644
>> --- a/mm/compaction.c
>> +++ b/mm/compaction.c
>> @@ -1056,11 +1056,8 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
>>   		 * Skip any other type of page
>>   		 */
>>   		if (!PageLRU(page)) {
>> -			/*
>> -			 * __PageMovable can return false positive so we need
>> -			 * to verify it under page_lock.
>> -			 */
>> -			if (unlikely(__PageMovable(page)) &&
>> +			/* Isolation will grab the page lock. */
> 
> I feel that the removed comment should stay, since the current comment
> makes no sense when I read it alone.

Well, talking about the page lock is moot either way. The thing is, 
anything can change while we don't hold a page reference. So should we 
change the comment to

/* isolation code will deal with any races. */

... or drop it completely?

> 
> In addition, why is __PageMovable() is renamed to page_has_movable_ops() but
> __SetPageMovable() stays the same? page_has_movable_ops() and __SetPageMovable()
> are functions for checking and setting PAGE_MAPPING_MOVABLE. The naming just
> does not look symmetric.

See follow-up commits where __SetPageMovable() is cleaned up.

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 20/29] mm: convert "movable" flag in page->mapping to a page flag

[David Hildenbrand]

On 23.06.25 16:14, Zi Yan wrote:
> On 18 Jun 2025, at 13:40, David Hildenbrand wrote:
> 
>> Instead, let's use a page flag. As the page flag can result in
>> false-positives, glue it to the page types for which we
>> support/implement movable_ops page migration.
>>
>> Signed-off-by: David Hildenbrand <david@redhat.com>
>> ---
>>   include/linux/balloon_compaction.h |  2 +-
>>   include/linux/migrate.h            |  8 -----
>>   include/linux/page-flags.h         | 52 ++++++++++++++++++++++++------
>>   mm/compaction.c                    |  6 ----
>>   mm/zpdesc.h                        |  2 +-
>>   5 files changed, 44 insertions(+), 26 deletions(-)
>>
> 
> <snip>
> 
>> +
>> +/**
>> + * page_has_movable_ops - test for a movable_ops page
>> + * @page The page to test.
>> + *
>> + * Test whether this is a movable_ops page. Such pages will stay that
>> + * way until freed.
>> + *
>> + * Returns true if this is a movable_ops page, otherwise false.
>> + */
>> +static inline bool page_has_movable_ops(const struct page *page)
>> +{
>> +	return PageMovableOps(page) &&
>> +	       (PageOffline(page) || PageZsmalloc(page));
>> +}
>> +
> 
> Should we do the code below in case PageMovableOps is set on pages
> other than PageOffline and PageZsmalloc?
> 
> return PageMovableOps(page) &&
> 	   !VM_WARN_ON_ONCE_PAGE(!(PageOffline(page) || PageZsmalloc(page)),
> 							 page);

I think this could trigger in some corner when other pages (esp. 
folios?) reuse the flag and we run into some weird races while we don't 
hold any reference on the page.

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[David Hildenbrand]

On 23.06.25 17:42, Zi Yan wrote:
> On 23 Jun 2025, at 11:33, David Hildenbrand wrote:
> 
>> On 18.06.25 20:48, Zi Yan wrote:
>>> On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:
>>>
>>>> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
>>>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>>>>
>>>>>> ... and start moving back to per-page things that will absolutely not be
>>>>>> folio things in the future. Add documentation and a comment that the
>>>>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
>>>>>>
>>>>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
>>>>>> it gracefully (relevant with further changes), and convert a
>>>>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
>>>>>
>>>>> The reason is that there is no upstream code, which use movable_ops for
>>>>> folios? Is there any fundamental reason preventing movable_ops from
>>>>> being used on folios?
>>>>
>>>> folios either belong to a filesystem or they are anonymous memory, and
>>>> so either the filesystem knows how to migrate them (through its a_ops)
>>>> or the migration code knows how to handle anon folios directly.
>>
>> Right, migration of folios will be handled by migration core.
>>
>>>
>>> for device private pages, to support migrating >0 order anon or fs folios
>>> to device, how should we represent them for devices? if you think folio is
>>> only for anon and fs.
>>
>> I assume they are proper folios, so yes. Just like they are handled today (-> folios)
>>
>> I was asking a related question at LSF/MM in Alistair's session: are we sure these things will be folios even before they are assigned to a filesystem? I recall the answer was "yes".
>>
>> So we don't (and will not) support movable_ops for folios.
> 
> Got it. (I was abusing it to help develop alloc_contig_range() at pageblock
> granularity, since it was easy to write a driver to allocate a compound page
> at a specific PFN and claim the page is movable, then do page online/offline
> the range containing the PFNs. :) )

Oh, interesting :)

Yeah, with a folio that will be a bit harder ...

> 
> For the patch, Reviewed-by: Zi Yan <ziy@nvidia.com>

Thanks!

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 16/29] mm: rename __PageMovable() to page_has_movable_ops()

[Zi Yan]

On 23 Jun 2025, at 11:47, David Hildenbrand wrote:

> On 20.06.25 22:37, Zi Yan wrote:
>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>
>>> Let's make it clearer that we are talking about movable_ops pages.
>>>
>>> Signed-off-by: David Hildenbrand <david@redhat.com>
>>> ---
>>>   include/linux/migrate.h    |  2 +-
>>>   include/linux/page-flags.h |  2 +-
>>>   mm/compaction.c            |  7 ++-----
>>>   mm/memory-failure.c        |  4 ++--
>>>   mm/memory_hotplug.c        |  8 +++-----
>>>   mm/migrate.c               |  8 ++++----
>>>   mm/page_alloc.c            |  2 +-
>>>   mm/page_isolation.c        | 10 +++++-----
>>>   8 files changed, 19 insertions(+), 24 deletions(-)
>>>
>>> diff --git a/include/linux/migrate.h b/include/linux/migrate.h
>>> index 204e89eac998f..c575778456f97 100644
>>> --- a/include/linux/migrate.h
>>> +++ b/include/linux/migrate.h
>>> @@ -115,7 +115,7 @@ static inline void __SetPageMovable(struct page *page,
>>>   static inline
>>>   const struct movable_operations *page_movable_ops(struct page *page)
>>>   {
>>> -	VM_BUG_ON(!__PageMovable(page));
>>> +	VM_BUG_ON(!page_has_movable_ops(page));
>>>
>>>   	return (const struct movable_operations *)
>>>   		((unsigned long)page->mapping - PAGE_MAPPING_MOVABLE);
>>> diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
>>> index 4fe5ee67535b2..c67163b73c5ec 100644
>>> --- a/include/linux/page-flags.h
>>> +++ b/include/linux/page-flags.h
>>> @@ -750,7 +750,7 @@ static __always_inline bool __folio_test_movable(const struct folio *folio)
>>>   			PAGE_MAPPING_MOVABLE;
>>>   }
>>>
>>> -static __always_inline bool __PageMovable(const struct page *page)
>>> +static __always_inline bool page_has_movable_ops(const struct page *page)
>>>   {
>>>   	return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) ==
>>>   				PAGE_MAPPING_MOVABLE;
>>> diff --git a/mm/compaction.c b/mm/compaction.c
>>> index 5c37373017014..f8b7c09e2e48c 100644
>>> --- a/mm/compaction.c
>>> +++ b/mm/compaction.c
>>> @@ -1056,11 +1056,8 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
>>>   		 * Skip any other type of page
>>>   		 */
>>>   		if (!PageLRU(page)) {
>>> -			/*
>>> -			 * __PageMovable can return false positive so we need
>>> -			 * to verify it under page_lock.
>>> -			 */
>>> -			if (unlikely(__PageMovable(page)) &&
>>> +			/* Isolation will grab the page lock. */
>>
>> I feel that the removed comment should stay, since the current comment
>> makes no sense when I read it alone.
>
> Well, talking about the page lock is moot either way. The thing is, anything can change while we don't hold a page reference. So should we change the comment to
>
> /* isolation code will deal with any races. */

Sounds good to me.

>
> ... or drop it completely?
>
>>
>> In addition, why is __PageMovable() is renamed to page_has_movable_ops() but
>> __SetPageMovable() stays the same? page_has_movable_ops() and __SetPageMovable()
>> are functions for checking and setting PAGE_MAPPING_MOVABLE. The naming just
>> does not look symmetric.
>
> See follow-up commits where __SetPageMovable() is cleaned up.

Right. It becomes clear at Patch 20.

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

--
Best Regards,
Yan, Zi

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[David Hildenbrand]

On 23.06.25 17:42, Zi Yan wrote:
> On 23 Jun 2025, at 11:33, David Hildenbrand wrote:
> 
>> On 18.06.25 20:48, Zi Yan wrote:
>>> On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:
>>>
>>>> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
>>>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>>>>
>>>>>> ... and start moving back to per-page things that will absolutely not be
>>>>>> folio things in the future. Add documentation and a comment that the
>>>>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
>>>>>>
>>>>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
>>>>>> it gracefully (relevant with further changes), and convert a
>>>>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
>>>>>
>>>>> The reason is that there is no upstream code, which use movable_ops for
>>>>> folios? Is there any fundamental reason preventing movable_ops from
>>>>> being used on folios?
>>>>
>>>> folios either belong to a filesystem or they are anonymous memory, and
>>>> so either the filesystem knows how to migrate them (through its a_ops)
>>>> or the migration code knows how to handle anon folios directly.
>>
>> Right, migration of folios will be handled by migration core.
>>
>>>
>>> for device private pages, to support migrating >0 order anon or fs folios
>>> to device, how should we represent them for devices? if you think folio is
>>> only for anon and fs.
>>
>> I assume they are proper folios, so yes. Just like they are handled today (-> folios)
>>
>> I was asking a related question at LSF/MM in Alistair's session: are we sure these things will be folios even before they are assigned to a filesystem? I recall the answer was "yes".
>>
>> So we don't (and will not) support movable_ops for folios.
> 
> Got it. (I was abusing it to help develop alloc_contig_range() at pageblock
> granularity, since it was easy to write a driver to allocate a compound page
> at a specific PFN and claim the page is movable, then do page online/offline
> the range containing the PFNs. :) )
> 
> For the patch, Reviewed-by: Zi Yan <ziy@nvidia.com>

BTW, thinking about it, I think we could handle compound pages quite 
easily, we'd just have to lookup the head at some point -- but we 
wouldn't be using folios for that.

BUT, I am note sure how compound pages without a memdesc would look like 
in a memdesc world, and if it would actually be "compound pages" in the 
traditional sense.

So, I think there would be ways to handle that, once we get to it.

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Huang, Ying]

David Hildenbrand <david@redhat.com> writes:

> On 18.06.25 20:48, Zi Yan wrote:
>> On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:
>> 
>>> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
>>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>>>>
>>>>> ... and start moving back to per-page things that will absolutely not be
>>>>> folio things in the future. Add documentation and a comment that the
>>>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
>>>>>
>>>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
>>>>> it gracefully (relevant with further changes), and convert a
>>>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
>>>>
>>>> The reason is that there is no upstream code, which use movable_ops for
>>>> folios? Is there any fundamental reason preventing movable_ops from
>>>> being used on folios?
>>>
>>> folios either belong to a filesystem or they are anonymous memory, and
>>> so either the filesystem knows how to migrate them (through its a_ops)
>>> or the migration code knows how to handle anon folios directly.
>
> Right, migration of folios will be handled by migration core.
>
>> for device private pages, to support migrating >0 order anon or fs
>> folios
>> to device, how should we represent them for devices? if you think folio is
>> only for anon and fs.
>
> I assume they are proper folios, so yes. Just like they are handled
> today (-> folios)
>
> I was asking a related question at LSF/MM in Alistair's session: are
> we sure these things will be folios even before they are assigned to a
> filesystem? I recall the answer was "yes".
>
> So we don't (and will not) support movable_ops for folios.

Is it possible to use some device specific callbacks (DMA?) to copy
from/to the device private folios (or pages) to/from the normal
file/anon folios in the future?

---
Best Regards,
Huang, Ying

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Alistair Popple]

On Sun, Jun 29, 2025 at 07:28:50PM +0800, Huang, Ying wrote:
> David Hildenbrand <david@redhat.com> writes:
> 
> > On 18.06.25 20:48, Zi Yan wrote:
> >> On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:
> >> 
> >>> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
> >>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> >>>>
> >>>>> ... and start moving back to per-page things that will absolutely not be
> >>>>> folio things in the future. Add documentation and a comment that the
> >>>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
> >>>>>
> >>>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
> >>>>> it gracefully (relevant with further changes), and convert a
> >>>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
> >>>>
> >>>> The reason is that there is no upstream code, which use movable_ops for
> >>>> folios? Is there any fundamental reason preventing movable_ops from
> >>>> being used on folios?
> >>>
> >>> folios either belong to a filesystem or they are anonymous memory, and
> >>> so either the filesystem knows how to migrate them (through its a_ops)
> >>> or the migration code knows how to handle anon folios directly.
> >
> > Right, migration of folios will be handled by migration core.
> >
> >> for device private pages, to support migrating >0 order anon or fs
> >> folios
> >> to device, how should we represent them for devices? if you think folio is
> >> only for anon and fs.
> >
> > I assume they are proper folios, so yes. Just like they are handled
> > today (-> folios)

Yes, they should be proper folios.

> > I was asking a related question at LSF/MM in Alistair's session: are
> > we sure these things will be folios even before they are assigned to a
> > filesystem? I recall the answer was "yes".
> >
> > So we don't (and will not) support movable_ops for folios.
> 
> Is it possible to use some device specific callbacks (DMA?) to copy
> from/to the device private folios (or pages) to/from the normal
> file/anon folios in the future?

I guess we could put such callbacks on the folio->pgmap, but I'm not sure why
we would want to. Currently all migration to/from device private (or coherent)
folios is managed by the device, which is one of the features of ZONE_DEVICE.
Did you have some particular reason/idea for why we might want to do this?

> ---
> Best Regards,
> Huang, Ying

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Huang, Ying]

Alistair Popple <apopple@nvidia.com> writes:

> On Sun, Jun 29, 2025 at 07:28:50PM +0800, Huang, Ying wrote:
>> David Hildenbrand <david@redhat.com> writes:
>> 
>> > On 18.06.25 20:48, Zi Yan wrote:
>> >> On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:
>> >> 
>> >>> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
>> >>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
>> >>>>
>> >>>>> ... and start moving back to per-page things that will absolutely not be
>> >>>>> folio things in the future. Add documentation and a comment that the
>> >>>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
>> >>>>>
>> >>>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
>> >>>>> it gracefully (relevant with further changes), and convert a
>> >>>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
>> >>>>
>> >>>> The reason is that there is no upstream code, which use movable_ops for
>> >>>> folios? Is there any fundamental reason preventing movable_ops from
>> >>>> being used on folios?
>> >>>
>> >>> folios either belong to a filesystem or they are anonymous memory, and
>> >>> so either the filesystem knows how to migrate them (through its a_ops)
>> >>> or the migration code knows how to handle anon folios directly.
>> >
>> > Right, migration of folios will be handled by migration core.
>> >
>> >> for device private pages, to support migrating >0 order anon or fs
>> >> folios
>> >> to device, how should we represent them for devices? if you think folio is
>> >> only for anon and fs.
>> >
>> > I assume they are proper folios, so yes. Just like they are handled
>> > today (-> folios)
>
> Yes, they should be proper folios.

So, folios include file cache, anonymous, and some device private.

>> > I was asking a related question at LSF/MM in Alistair's session: are
>> > we sure these things will be folios even before they are assigned to a
>> > filesystem? I recall the answer was "yes".
>> >
>> > So we don't (and will not) support movable_ops for folios.
>> 
>> Is it possible to use some device specific callbacks (DMA?) to copy
>> from/to the device private folios (or pages) to/from the normal
>> file/anon folios in the future?
>
> I guess we could put such callbacks on the folio->pgmap, but I'm not sure why
> we would want to. Currently all migration to/from device private (or coherent)
> folios is managed by the device, which is one of the features of ZONE_DEVICE.

Yes.  The is the current behavior per my understanding too.

> Did you have some particular reason/idea for why we might want to do this?

No.  Just want to check whether there are some requirements for that.  I
think that it's just another way to organize code.

---
Best Regards,
Huang, Ying

Re: [PATCH RFC 13/29] mm/balloon_compaction: stop using __ClearPageMovable()

[Huang, Ying]

David Hildenbrand <david@redhat.com> writes:

> We can just look at the balloon device (stored in page->private), to see
> of the page is still part of the balloon.

s/of/if/

?

just a trivial issue if I'm not wrong.

> As isolated balloon pages cannot get released (they are taken off the
> balloon list while isolated), we don't have to worry about this case in
> the putback and migration callback. Add a WARN_ON_ONCE for now.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>

[snip]

---
Best Regards,
Huang, Ying

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Alistair Popple]

On Mon, Jun 30, 2025 at 08:58:03AM +0800, Huang, Ying wrote:
> Alistair Popple <apopple@nvidia.com> writes:
> 
> > On Sun, Jun 29, 2025 at 07:28:50PM +0800, Huang, Ying wrote:
> >> David Hildenbrand <david@redhat.com> writes:
> >> 
> >> > On 18.06.25 20:48, Zi Yan wrote:
> >> >> On 18 Jun 2025, at 14:39, Matthew Wilcox wrote:
> >> >> 
> >> >>> On Wed, Jun 18, 2025 at 02:14:15PM -0400, Zi Yan wrote:
> >> >>>> On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> >> >>>>
> >> >>>>> ... and start moving back to per-page things that will absolutely not be
> >> >>>>> folio things in the future. Add documentation and a comment that the
> >> >>>>> remaining folio stuff (lock, refcount) will have to be reworked as well.
> >> >>>>>
> >> >>>>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
> >> >>>>> it gracefully (relevant with further changes), and convert a
> >> >>>>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
> >> >>>>
> >> >>>> The reason is that there is no upstream code, which use movable_ops for
> >> >>>> folios? Is there any fundamental reason preventing movable_ops from
> >> >>>> being used on folios?
> >> >>>
> >> >>> folios either belong to a filesystem or they are anonymous memory, and
> >> >>> so either the filesystem knows how to migrate them (through its a_ops)
> >> >>> or the migration code knows how to handle anon folios directly.
> >> >
> >> > Right, migration of folios will be handled by migration core.
> >> >
> >> >> for device private pages, to support migrating >0 order anon or fs
> >> >> folios
> >> >> to device, how should we represent them for devices? if you think folio is
> >> >> only for anon and fs.
> >> >
> >> > I assume they are proper folios, so yes. Just like they are handled
> >> > today (-> folios)
> >
> > Yes, they should be proper folios.
> 
> So, folios include file cache, anonymous, and some device private.

Oh maybe I misunderstood what you were asking. We have anon and fs folios, and
we currently have device private versions of the former. However ideally I think
in a memdesc world we would have anon/fs folios, and the device private bit
would be in the memdesc or some such and so at the level of a folio we'd only
be dealing with "proper" anon or fs folios (of course in practice we may never
permit device private versions of the latter).

In my earlier answer I just wanted to highlight the fact that order >0 device
folios now look the same as normal higher order anon or fs folios. Ie. we don't
do any of the special pgmap refcounting, etc. that we used to do for higher
order device folios.

> >> > I was asking a related question at LSF/MM in Alistair's session: are
> >> > we sure these things will be folios even before they are assigned to a
> >> > filesystem? I recall the answer was "yes".
> >> >
> >> > So we don't (and will not) support movable_ops for folios.
> >> 
> >> Is it possible to use some device specific callbacks (DMA?) to copy
> >> from/to the device private folios (or pages) to/from the normal
> >> file/anon folios in the future?
> >
> > I guess we could put such callbacks on the folio->pgmap, but I'm not sure why
> > we would want to. Currently all migration to/from device private (or coherent)
> > folios is managed by the device, which is one of the features of ZONE_DEVICE.
> 
> Yes.  The is the current behavior per my understanding too.
> 
> > Did you have some particular reason/idea for why we might want to do this?
> 
> No.  Just want to check whether there are some requirements for that.  I
> think that it's just another way to organize code.
> 
> ---
> Best Regards,
> Huang, Ying

Re: [PATCH RFC 03/29] mm/zsmalloc: drop PageIsolated() related VM_BUG_ONs

[Harry Yoo]

On Wed, Jun 18, 2025 at 07:39:46PM +0200, David Hildenbrand wrote:
> Let's drop these checks; these are conditions the core migration code
> must make sure will hold either way, no need to double check.
> 
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---

Acked-by: Harry Yoo <harry.yoo@oracle.com>

-- 
Cheers,
Harry / Hyeonggon

Re: [PATCH RFC 04/29] mm/page_alloc: allow for making page types sticky until freed

[Harry Yoo]

On Mon, Jun 23, 2025 at 05:26:05PM +0200, David Hildenbrand wrote:
> On 18.06.25 20:08, Zi Yan wrote:
> > On 18 Jun 2025, at 14:04, Zi Yan wrote:
> > 
> > > On 18 Jun 2025, at 13:39, David Hildenbrand wrote:
> > > 
> > > > Let's allow for not clearing a page type before freeing a page to the
> > > > buddy.
> > > > 
> > > > We'll focus on having a type set on the first page of a larger
> > > > allocation only.
> > > > 
> > > > With this change, we can reliably identify typed folios even though
> > > > they might be in the process of getting freed, which will come in handy
> > > > in migration code (at least in the transition phase).
> > > > 
> > > > Signed-off-by: David Hildenbrand <david@redhat.com>
> > > > ---
> > > >   mm/page_alloc.c | 3 +++
> > > >   1 file changed, 3 insertions(+)
> > > > 
> > > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > > index 858bc17653af9..44e56d31cfeb1 100644
> > > > --- a/mm/page_alloc.c
> > > > +++ b/mm/page_alloc.c
> > > > @@ -1380,6 +1380,9 @@ __always_inline bool free_pages_prepare(struct page *page,
> > > >   			mod_mthp_stat(order, MTHP_STAT_NR_ANON, -1);
> > > >   		page->mapping = NULL;
> > > >   	}
> > > > +	if (unlikely(page_has_type(page)))
> > > > +		page->page_type = UINT_MAX;
> > > > +
> > > >   	if (is_check_pages_enabled()) {
> > > >   		if (free_page_is_bad(page))
> > > >   			bad++;
> > > > -- 
> > > > 2.49.0
> > > 
> > > How does this preserve page type? Isn’t page->page_type = UINT_MAX clearing
> > > page_type?
> > 
> > OK, next patch explains it. free_pages_prepare() clears page_type,
> > so that caller does not need to.
> > 
> > I think the message is better to be
> > 
> > mm/page_alloc: clear page_type at page free time
> > 
> > page_type is no longer needed to be cleared before a page is freed, as
> > page free code does that.
> > 
> > With this change, we can reliably identify typed folios even though
> > they might be in the process of getting freed, which will come in handy
> > in migration code (at least in the transition phase).
> 
> 
> I'll change it to
> 
>     mm/page_alloc: let page freeing clear any set page type
>     Currently, any user of page types must clear that type before freeing
>     a page back to the buddy, otherwise we'll run into mapcount related
>     sanity checks (because the page type currently overlays the page
>     mapcount).
>     Let's allow for not clearing the page type by page type users by letting
>     the buddy handle it instead.
>     We'll focus on having a page type set on the first page of a larger
>     allocation only.
>     With this change, we can reliably identify typed folios even though
>     they might be in the process of getting freed, which will come in handy
>     in migration code (at least in the transition phase).

Acked-by: Harry Yoo <harry.yoo@oracle.com>

-- 
Cheers,
Harry / Hyeonggon

Re: [PATCH RFC 05/29] mm/balloon_compaction: make PageOffline sticky

[Harry Yoo]

On Wed, Jun 18, 2025 at 07:39:48PM +0200, David Hildenbrand wrote:
> Let the buddy handle clearing the type.
> 
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---

with the updated commit message,
Acked-by: Harry Yoo <harry.yoo@oracle.com>

>  include/linux/balloon_compaction.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/linux/balloon_compaction.h b/include/linux/balloon_compaction.h
> index b9f19da37b089..bfc6e50bd004b 100644
> --- a/include/linux/balloon_compaction.h
> +++ b/include/linux/balloon_compaction.h
> @@ -140,7 +140,7 @@ static inline void balloon_page_finalize(struct page *page)
>  		__ClearPageMovable(page);
>  		set_page_private(page, 0);
>  	}
> -	__ClearPageOffline(page);
> +	/* PageOffline is sticky until the page is freed to the buddy. */
>  }
>  
>  /*
> -- 
> 2.49.0
> 

-- 
Cheers,
Harry / Hyeonggon

Re: [PATCH RFC 06/29] mm/zsmalloc: make PageZsmalloc() sticky

[Harry Yoo]

On Wed, Jun 18, 2025 at 07:39:49PM +0200, David Hildenbrand wrote:
> Let the buddy handle clearing the type.
> 
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---

with the updated commit message,
Acked-by: Harry Yoo <harry.yoo@oracle.com>

-- 
Cheers,
Harry / Hyeonggon

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[Harry Yoo]

On Wed, Jun 18, 2025 at 07:39:50PM +0200, David Hildenbrand wrote:
> ... and start moving back to per-page things that will absolutely not be
> folio things in the future. Add documentation and a comment that the
> remaining folio stuff (lock, refcount) will have to be reworked as well.
> 
> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
> it gracefully (relevant with further changes), and convert a
> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
> 
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---

Haha yeah, back to pages after folio conversion :P
But makes sense.

Reviewed-by: Harry Yoo <harry.yoo@oracle.com>

Side question: In the future, maybe we will be unable to tell whether
a page is compound or not, without first inspecting page->memdesc?
(e.g., struct slab could have an order and a pointer to the head page...
just imagining).

-- 
Cheers,
Harry / Hyeonggon

Re: [PATCH RFC 07/29] mm/migrate: rename isolate_movable_page() to isolate_movable_ops_page()

[David Hildenbrand]

On 30.06.25 10:04, Harry Yoo wrote:
> On Wed, Jun 18, 2025 at 07:39:50PM +0200, David Hildenbrand wrote:
>> ... and start moving back to per-page things that will absolutely not be
>> folio things in the future. Add documentation and a comment that the
>> remaining folio stuff (lock, refcount) will have to be reworked as well.
>>
>> While at it, convert the VM_BUG_ON() into a WARN_ON_ONCE() and handle
>> it gracefully (relevant with further changes), and convert a
>> WARN_ON_ONCE() into a VM_WARN_ON_ONCE_PAGE().
>>
>> Signed-off-by: David Hildenbrand <david@redhat.com>
>> ---
> 
> Haha yeah, back to pages after folio conversion :P
> But makes sense.
> 
> Reviewed-by: Harry Yoo <harry.yoo@oracle.com>
> 
> Side question: In the future, maybe we will be unable to tell whether
> a page is compound or not, without first inspecting page->memdesc?
> (e.g., struct slab could have an order and a pointer to the head page...
> just imagining).

Right, it's not really clear what we would do in the future, but for 
anything that allocates a memdesc, that is possible.

We wouldn't even need the "head" + "tail" bit indication.

For things without an allocated memdesc (e.g., PageOffline), maybe 
compound pages will simply not apply.

Likely, the concept of compound pages as we knew it will go away.

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 13/29] mm/balloon_compaction: stop using __ClearPageMovable()

[David Hildenbrand]

On 30.06.25 03:18, Huang, Ying wrote:
> David Hildenbrand <david@redhat.com> writes:
> 
>> We can just look at the balloon device (stored in page->private), to see
>> of the page is still part of the balloon.
> 
> s/of/if/
> 
> ?

Yes, thanks!

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 18/29] mm: remove __folio_test_movable()

[David Hildenbrand]

> @@ -1164,7 +1148,6 @@ static int migrate_folio_unmap(new_folio_t get_new_folio,
>   	int rc = -EAGAIN;
>   	int old_page_state = 0;
>   	struct anon_vma *anon_vma = NULL;
> -	bool is_lru = data_race(!__folio_test_movable(src));
>   	bool locked = false;
>   	bool dst_locked = false;
>   
> @@ -1265,7 +1248,7 @@ static int migrate_folio_unmap(new_folio_t get_new_folio,
>   		goto out;
>   	dst_locked = true;
>   
> -	if (unlikely(!is_lru)) {
> +	if (unlikely(page_has_movable_ops(&dst->page))) {

While doing more testing, I found a weird bug where we end up having 
balloon pages isolated forever, and never putback/migrated. So neither 
the balloon can deflate them nor the migration migrate them.

This line here must be

if (unlikely(page_has_movable_ops(&src->page))) {

But I am not sure yet if that's the issue I am chasing :)

-- 
Cheers,

David / dhildenb

Re: [PATCH RFC 11/29] mm/migrate: move movable_ops page handling out of move_to_new_folio()

[David Hildenbrand]

On 18.06.25 19:39, David Hildenbrand wrote:
> Let's move that handling directly into migrate_folio_move(), so we can
> simplify move_to_new_folio(). While at it, fixup the documentation a
> bit.
> 
> Note that unmap_and_move_huge_page() does not care, because it only
> deals with actual folios. (we only support migration of
> individual movable_ops pages)
> 
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>   mm/migrate.c | 61 ++++++++++++++++++++++++----------------------------
>   1 file changed, 28 insertions(+), 33 deletions(-)
> 
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 456e41dad83a2..db807f9bbf975 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -1024,11 +1024,12 @@ static int fallback_migrate_folio(struct address_space *mapping,
>   }
>   
>   /*
> - * Move a page to a newly allocated page
> - * The page is locked and all ptes have been successfully removed.
> + * Move a src folio to a newly allocated dst folio.
>    *
> - * The new page will have replaced the old page if this function
> - * is successful.
> + * The src and dst folios are locked and the src folios was unmapped from
> + * the page tables.
> + *
> + * On success, the src folio was replaced by the dst folio.
>    *
>    * Return value:
>    *   < 0 - error code
> @@ -1037,34 +1038,30 @@ static int fallback_migrate_folio(struct address_space *mapping,
>   static int move_to_new_folio(struct folio *dst, struct folio *src,
>   				enum migrate_mode mode)
>   {
> +	struct address_space *mapping = folio_mapping(src);
>   	int rc = -EAGAIN;
> -	bool is_lru = !__folio_test_movable(src);
>   
>   	VM_BUG_ON_FOLIO(!folio_test_locked(src), src);
>   	VM_BUG_ON_FOLIO(!folio_test_locked(dst), dst);
>   
> -	if (likely(is_lru)) {
> -		struct address_space *mapping = folio_mapping(src);
> -
> -		if (!mapping)
> -			rc = migrate_folio(mapping, dst, src, mode);
> -		else if (mapping_inaccessible(mapping))
> -			rc = -EOPNOTSUPP;
> -		else if (mapping->a_ops->migrate_folio)
> -			/*
> -			 * Most folios have a mapping and most filesystems
> -			 * provide a migrate_folio callback. Anonymous folios
> -			 * are part of swap space which also has its own
> -			 * migrate_folio callback. This is the most common path
> -			 * for page migration.
> -			 */
> -			rc = mapping->a_ops->migrate_folio(mapping, dst, src,
> -								mode);
> -		else
> -			rc = fallback_migrate_folio(mapping, dst, src, mode);
> +	if (!mapping)
> +		rc = migrate_folio(mapping, dst, src, mode);
> +	else if (mapping_inaccessible(mapping))
> +		rc = -EOPNOTSUPP;
> +	else if (mapping->a_ops->migrate_folio)
> +		/*
> +		 * Most folios have a mapping and most filesystems
> +		 * provide a migrate_folio callback. Anonymous folios
> +		 * are part of swap space which also has its own
> +		 * migrate_folio callback. This is the most common path
> +		 * for page migration.
> +		 */
> +		rc = mapping->a_ops->migrate_folio(mapping, dst, src,
> +							mode);
> +	else
> +		rc = fallback_migrate_folio(mapping, dst, src, mode);
>   
> -		if (rc != MIGRATEPAGE_SUCCESS)
> -			goto out;
> +	if (rc == MIGRATEPAGE_SUCCESS) {
>   		/*
>   		 * For pagecache folios, src->mapping must be cleared before src
>   		 * is freed. Anonymous folios must stay anonymous until freed.
> @@ -1074,10 +1071,7 @@ static int move_to_new_folio(struct folio *dst, struct folio *src,
>   
>   		if (likely(!folio_is_zone_device(dst)))
>   			flush_dcache_folio(dst);
> -	} else {
> -		rc = migrate_movable_ops_page(&dst->page, &src->page, mode);
>   	}
> -out:
>   	return rc;
>   }
>   
> @@ -1328,20 +1322,21 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
>   	int rc;
>   	int old_page_state = 0;
>   	struct anon_vma *anon_vma = NULL;
> -	bool is_lru = !__folio_test_movable(src);
>   	struct list_head *prev;
>   
>   	__migrate_folio_extract(dst, &old_page_state, &anon_vma);
>   	prev = dst->lru.prev;
>   	list_del(&dst->lru);
>   
> +	if (unlikely(__folio_test_movable(src))) {
> +		rc = migrate_movable_ops_page(&dst->page, &src->page, mode);
> +		goto out_unlock_both;
> +	}
> +


I suspect this was the issue I was chasing:

diff --git a/mm/migrate.c b/mm/migrate.c
index 9dd327a63abc5..22c115710d0e2 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1330,6 +1330,8 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
  
         if (unlikely(__folio_test_movable(src))) {
                 rc = migrate_movable_ops_page(&dst->page, &src->page, mode);
+               if (rc)
+                       goto out;
                 goto out_unlock_both;
         }
  

Testing ...

-- 
Cheers,

David / dhildenb