diff mbox series

[v2,1/4] mm: migrate: Move the page count validation to the proper place

Message ID 644a666e1e177c57a39a4c37c6e2ca64052b9d7e.1629008158.git.baolin.wang@linux.alibaba.com (mailing list archive)
State New
Headers show
Series Some cleanup for page migration | expand

Commit Message

Baolin Wang Aug. 15, 2021, 6:23 a.m. UTC
We've got the expected count for anonymous page or file page by
expected_page_refs() at the beginning of migrate_page_move_mapping(),
thus we should move the page count validation a little forward to
reduce duplicated code.

Moreover the i_pages lock is not used to guarantee the page refcount
safety in migrate_page_move_mapping(), so we can move the getting page
count out of the i_pages lock. Since now the migration page has
established a migration pte under the page lock now, with the page
refcount freezing validation, to ensure that the page references
meet the migration requirement.

Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
---
 mm/migrate.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

Comments

Matthew Wilcox Aug. 15, 2021, 10:34 a.m. UTC | #1
On Sun, Aug 15, 2021 at 02:23:03PM +0800, Baolin Wang wrote:
> We've got the expected count for anonymous page or file page by
> expected_page_refs() at the beginning of migrate_page_move_mapping(),
> thus we should move the page count validation a little forward to
> reduce duplicated code.
> 
> Moreover the i_pages lock is not used to guarantee the page refcount
> safety in migrate_page_move_mapping(), so we can move the getting page
> count out of the i_pages lock. Since now the migration page has
> established a migration pte under the page lock now, with the page
> refcount freezing validation, to ensure that the page references
> meet the migration requirement.

I remain unconvinced by this.  

Looking at folio_migrate_mapping() a little more deeply, I don't
understand why we first check folio_ref_count() and then attempt
to free the refcount.  Why not just try to freeze it directly?

ie instead of your patch, this:

+++ b/mm/migrate.c
@@ -403,13 +403,8 @@ int folio_migrate_mapping(struct address_space *mapping,
        newzone = folio_zone(newfolio);

        xas_lock_irq(&xas);
-       if (folio_ref_count(folio) != expected_count ||
-           xas_load(&xas) != folio) {
-               xas_unlock_irq(&xas);
-               return -EAGAIN;
-       }
-
-       if (!folio_ref_freeze(folio, expected_count)) {
+       if (xas_load(&xas) != folio ||
+           !folio_ref_freeze(folio, expected_count)) {
                xas_unlock_irq(&xas);
                return -EAGAIN;
        }

And since we've got the lock on the page, how can somebody else be
removing it from the page cache?  I think that xas_load() can be
removed too.  So even more simply,

+++ b/mm/migrate.c
@@ -403,12 +403,6 @@ int folio_migrate_mapping(struct address_space *mapping,
        newzone = folio_zone(newfolio);

        xas_lock_irq(&xas);
-       if (folio_ref_count(folio) != expected_count ||
-           xas_load(&xas) != folio) {
-               xas_unlock_irq(&xas);
-               return -EAGAIN;
-       }
-
        if (!folio_ref_freeze(folio, expected_count)) {
                xas_unlock_irq(&xas);
                return -EAGAIN;

but I'm not really set up to test page migration.  Does your test suite
test migrating file-backed pages?
Baolin Wang Aug. 16, 2021, 10:58 a.m. UTC | #2
On 2021/8/15 18:34, Matthew Wilcox wrote:
> On Sun, Aug 15, 2021 at 02:23:03PM +0800, Baolin Wang wrote:
>> We've got the expected count for anonymous page or file page by
>> expected_page_refs() at the beginning of migrate_page_move_mapping(),
>> thus we should move the page count validation a little forward to
>> reduce duplicated code.
>>
>> Moreover the i_pages lock is not used to guarantee the page refcount
>> safety in migrate_page_move_mapping(), so we can move the getting page
>> count out of the i_pages lock. Since now the migration page has
>> established a migration pte under the page lock now, with the page
>> refcount freezing validation, to ensure that the page references
>> meet the migration requirement.
> 
> I remain unconvinced by this.
> 
> Looking at folio_migrate_mapping() a little more deeply, I don't
> understand why we first check folio_ref_count() and then attempt
> to free the refcount.  Why not just try to freeze it directly?
> 
> ie instead of your patch, this:
> 
> +++ b/mm/migrate.c
> @@ -403,13 +403,8 @@ int folio_migrate_mapping(struct address_space *mapping,
>          newzone = folio_zone(newfolio);
> 
>          xas_lock_irq(&xas);
> -       if (folio_ref_count(folio) != expected_count ||
> -           xas_load(&xas) != folio) {
> -               xas_unlock_irq(&xas);
> -               return -EAGAIN;
> -       }
> -
> -       if (!folio_ref_freeze(folio, expected_count)) {
> +       if (xas_load(&xas) != folio ||
> +           !folio_ref_freeze(folio, expected_count)) {
>                  xas_unlock_irq(&xas);
>                  return -EAGAIN;
>          }

I think this is reasonable, like what we've done in __remove_mapping().

> And since we've got the lock on the page, how can somebody else be
> removing it from the page cache?  I think that xas_load() can be
> removed too.  So even more simply,

Good point, this is more simply.

> 
> +++ b/mm/migrate.c
> @@ -403,12 +403,6 @@ int folio_migrate_mapping(struct address_space *mapping,
>          newzone = folio_zone(newfolio);
> 
>          xas_lock_irq(&xas);
> -       if (folio_ref_count(folio) != expected_count ||
> -           xas_load(&xas) != folio) {
> -               xas_unlock_irq(&xas);
> -               return -EAGAIN;
> -       }
> -
>          if (!folio_ref_freeze(folio, expected_count)) {
>                  xas_unlock_irq(&xas);
>                  return -EAGAIN;
> 
> but I'm not really set up to test page migration.  Does your test suite
> test migrating file-backed pages?

Yes, I've tested above changes, and the mapped file pages migration 
works well. So can I resend this patch set with your Suggested-by tag? 
Thanks.
diff mbox series

Patch

diff --git a/mm/migrate.c b/mm/migrate.c
index c6eb2a8..8e9282f 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -386,11 +386,10 @@  int folio_migrate_mapping(struct address_space *mapping,
 	int expected_count = expected_page_refs(mapping, &folio->page) + extra_count;
 	long nr = folio_nr_pages(folio);
 
-	if (!mapping) {
-		/* Anonymous page without mapping */
-		if (folio_ref_count(folio) != expected_count)
-			return -EAGAIN;
+	if (folio_ref_count(folio) != expected_count)
+		return -EAGAIN;
 
+	if (!mapping) {
 		/* No turning back from here */
 		newfolio->index = folio->index;
 		newfolio->mapping = folio->mapping;
@@ -404,8 +403,7 @@  int folio_migrate_mapping(struct address_space *mapping,
 	newzone = folio_zone(newfolio);
 
 	xas_lock_irq(&xas);
-	if (folio_ref_count(folio) != expected_count ||
-	    xas_load(&xas) != folio) {
+	if (xas_load(&xas) != folio) {
 		xas_unlock_irq(&xas);
 		return -EAGAIN;
 	}