[PATCH v1] mm/ksm: remove page_mapcount() usage in stable_tree_search()

[PATCH v1] mm/ksm: remove page_mapcount() usage in stable_tree_search()

[David Hildenbrand]

We want to limit the use of page_mapcount() to the places where it is
absolutely necessary.

If our folio has a stable node, it is a (small) KSM folio -- see
folio_stable_node(). Let's use folio_mapcount() in stable_tree_search()
instead, which results in no functional change.

The mapcount > 1 check is a bit confusing, because that's usually a check
for page sharing. Looks like the reason is that we are guaranteed to not
exceed ksm_max_page_sharing for the tree KSM folio when merging with
that. Let's update the documentation to make that clearer.

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Alex Shi (tencent) <alexs@kernel.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/ksm.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/mm/ksm.c b/mm/ksm.c
index 486c9974f8e20..159604ad47799 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -1909,12 +1909,15 @@ static struct page *stable_tree_search(struct page *page)
 			if (page_node) {
 				VM_BUG_ON(page_node->head != &migrate_nodes);
 				/*
-				 * Test if the migrated page should be merged
-				 * into a stable node dup. If the mapcount is
-				 * 1 we can migrate it with another KSM page
-				 * without adding it to the chain.
+				 * If the mapcount of our migrated KSM folio is
+				 * at most 1, we can merge it with another
+				 * KSM folio where we know that we have space
+				 * for one more mapping without exceeding the
+				 * ksm_max_page_sharing limit: see
+				 * chain_prune(). This way, we can avoid adding
+				 * this stable node to the chain.
 				 */
-				if (page_mapcount(page) > 1)
+				if (folio_mapcount(folio) > 1)
 					goto chain_append;
 			}
 
-- 
2.44.0

Re: [PATCH v1] mm/ksm: remove page_mapcount() usage in stable_tree_search()

[Alex Shi]

On 4/17/24 1:25 AM, David Hildenbrand wrote:
> We want to limit the use of page_mapcount() to the places where it is
> absolutely necessary.
> 
> If our folio has a stable node, it is a (small) KSM folio -- see
> folio_stable_node(). Let's use folio_mapcount() in stable_tree_search()
> instead, which results in no functional change.
> 
> The mapcount > 1 check is a bit confusing, because that's usually a check
> for page sharing. Looks like the reason is that we are guaranteed to not
> exceed ksm_max_page_sharing for the tree KSM folio when merging with
> that. Let's update the documentation to make that clearer.
> 
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Alex Shi (tencent) <alexs@kernel.org>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/ksm.c | 13 ++++++++-----
>  1 file changed, 8 insertions(+), 5 deletions(-)
> 
> diff --git a/mm/ksm.c b/mm/ksm.c
> index 486c9974f8e20..159604ad47799 100644
> --- a/mm/ksm.c
> +++ b/mm/ksm.c
> @@ -1909,12 +1909,15 @@ static struct page *stable_tree_search(struct page *page)
>  			if (page_node) {
>  				VM_BUG_ON(page_node->head != &migrate_nodes);
>  				/*
> -				 * Test if the migrated page should be merged
> -				 * into a stable node dup. If the mapcount is
> -				 * 1 we can migrate it with another KSM page
> -				 * without adding it to the chain.
> +				 * If the mapcount of our migrated KSM folio is
> +				 * at most 1, we can merge it with another
> +				 * KSM folio where we know that we have space
> +				 * for one more mapping without exceeding the
> +				 * ksm_max_page_sharing limit: see
> +				 * chain_prune(). This way, we can avoid adding
> +				 * this stable node to the chain.
>  				 */
> -				if (page_mapcount(page) > 1)
> +				if (folio_mapcount(folio) > 1)
>  					goto chain_append;
>  			}
>  

LGTM

Reviewed-by: Alex Shi <alexs@kernel.org>